CN119047833B - Project risk level assessment method based on approximate personalized transfer - Google Patents
Project risk level assessment method based on approximate personalized transfer Download PDFInfo
- Publication number
- CN119047833B CN119047833B CN202411175384.0A CN202411175384A CN119047833B CN 119047833 B CN119047833 B CN 119047833B CN 202411175384 A CN202411175384 A CN 202411175384A CN 119047833 B CN119047833 B CN 119047833B
- Authority
- CN
- China
- Prior art keywords
- data
- sample
- approximate
- project
- risk level
- Prior art date
- Legal status (The legal status is an assumption and is not a legal conclusion. Google has not performed a legal analysis and makes no representation as to the accuracy of the status listed.)
- Active
Links
Classifications
-
- G—PHYSICS
- G06—COMPUTING OR CALCULATING; COUNTING
- G06Q—INFORMATION AND COMMUNICATION TECHNOLOGY [ICT] SPECIALLY ADAPTED FOR ADMINISTRATIVE, COMMERCIAL, FINANCIAL, MANAGERIAL OR SUPERVISORY PURPOSES; SYSTEMS OR METHODS SPECIALLY ADAPTED FOR ADMINISTRATIVE, COMMERCIAL, FINANCIAL, MANAGERIAL OR SUPERVISORY PURPOSES, NOT OTHERWISE PROVIDED FOR
- G06Q10/00—Administration; Management
- G06Q10/06—Resources, workflows, human or project management; Enterprise or organisation planning; Enterprise or organisation modelling
- G06Q10/063—Operations research, analysis or management
- G06Q10/0635—Risk analysis of enterprise or organisation activities
-
- G—PHYSICS
- G06—COMPUTING OR CALCULATING; COUNTING
- G06F—ELECTRIC DIGITAL DATA PROCESSING
- G06F18/00—Pattern recognition
- G06F18/20—Analysing
- G06F18/21—Design or setup of recognition systems or techniques; Extraction of features in feature space; Blind source separation
- G06F18/214—Generating training patterns; Bootstrap methods, e.g. bagging or boosting
-
- G—PHYSICS
- G06—COMPUTING OR CALCULATING; COUNTING
- G06N—COMPUTING ARRANGEMENTS BASED ON SPECIFIC COMPUTATIONAL MODELS
- G06N3/00—Computing arrangements based on biological models
- G06N3/02—Neural networks
- G06N3/04—Architecture, e.g. interconnection topology
- G06N3/042—Knowledge-based neural networks; Logical representations of neural networks
-
- G—PHYSICS
- G06—COMPUTING OR CALCULATING; COUNTING
- G06N—COMPUTING ARRANGEMENTS BASED ON SPECIFIC COMPUTATIONAL MODELS
- G06N3/00—Computing arrangements based on biological models
- G06N3/02—Neural networks
- G06N3/08—Learning methods
-
- G—PHYSICS
- G06—COMPUTING OR CALCULATING; COUNTING
- G06Q—INFORMATION AND COMMUNICATION TECHNOLOGY [ICT] SPECIALLY ADAPTED FOR ADMINISTRATIVE, COMMERCIAL, FINANCIAL, MANAGERIAL OR SUPERVISORY PURPOSES; SYSTEMS OR METHODS SPECIALLY ADAPTED FOR ADMINISTRATIVE, COMMERCIAL, FINANCIAL, MANAGERIAL OR SUPERVISORY PURPOSES, NOT OTHERWISE PROVIDED FOR
- G06Q50/00—Information and communication technology [ICT] specially adapted for implementation of business processes of specific business sectors, e.g. utilities or tourism
- G06Q50/06—Energy or water supply
-
- Y—GENERAL TAGGING OF NEW TECHNOLOGICAL DEVELOPMENTS; GENERAL TAGGING OF CROSS-SECTIONAL TECHNOLOGIES SPANNING OVER SEVERAL SECTIONS OF THE IPC; TECHNICAL SUBJECTS COVERED BY FORMER USPC CROSS-REFERENCE ART COLLECTIONS [XRACs] AND DIGESTS
- Y04—INFORMATION OR COMMUNICATION TECHNOLOGIES HAVING AN IMPACT ON OTHER TECHNOLOGY AREAS
- Y04S—SYSTEMS INTEGRATING TECHNOLOGIES RELATED TO POWER NETWORK OPERATION, COMMUNICATION OR INFORMATION TECHNOLOGIES FOR IMPROVING THE ELECTRICAL POWER GENERATION, TRANSMISSION, DISTRIBUTION, MANAGEMENT OR USAGE, i.e. SMART GRIDS
- Y04S10/00—Systems supporting electrical power generation, transmission or distribution
- Y04S10/50—Systems or methods supporting the power network operation or management, involving a certain degree of interaction with the load-side end user applications
Landscapes
- Engineering & Computer Science (AREA)
- Business, Economics & Management (AREA)
- Theoretical Computer Science (AREA)
- Physics & Mathematics (AREA)
- Human Resources & Organizations (AREA)
- General Physics & Mathematics (AREA)
- Data Mining & Analysis (AREA)
- Economics (AREA)
- Health & Medical Sciences (AREA)
- Artificial Intelligence (AREA)
- Strategic Management (AREA)
- Life Sciences & Earth Sciences (AREA)
- Evolutionary Computation (AREA)
- General Health & Medical Sciences (AREA)
- General Engineering & Computer Science (AREA)
- Marketing (AREA)
- General Business, Economics & Management (AREA)
- Mathematical Physics (AREA)
- Entrepreneurship & Innovation (AREA)
- Software Systems (AREA)
- Tourism & Hospitality (AREA)
- Biomedical Technology (AREA)
- Computing Systems (AREA)
- Molecular Biology (AREA)
- Biophysics (AREA)
- Computational Linguistics (AREA)
- Development Economics (AREA)
- Public Health (AREA)
- Primary Health Care (AREA)
- Evolutionary Biology (AREA)
- Computer Vision & Pattern Recognition (AREA)
- Bioinformatics & Computational Biology (AREA)
- Water Supply & Treatment (AREA)
- Educational Administration (AREA)
- Bioinformatics & Cheminformatics (AREA)
- Game Theory and Decision Science (AREA)
- Operations Research (AREA)
- Quality & Reliability (AREA)
- Management, Administration, Business Operations System, And Electronic Commerce (AREA)
Abstract
The application discloses a project risk level assessment method based on approximate personalized transfer, which relates to the technical field of project risk assessment and comprises the steps of obtaining a project data set of a power project construction project; the method comprises the steps of determining data division indexes of a risk level assessment structure corresponding to each item data in a project data set, determining the data division indexes as data features of the project data, inputting each item data and the data features into a pre-trained risk level assessment model to obtain risk scores corresponding to power engineering construction projects output by the risk level assessment model, and determining project risk levels corresponding to the risk scores based on corresponding relations between the pre-created risk scores and the project risk levels. The method and the device are used for solving the problems of poor accuracy and low evaluation efficiency in the project risk level evaluation in the prior art, and improving the accuracy and the evaluation efficiency of the project risk level evaluation.
Description
Technical Field
The application relates to the technical field of project risk assessment, in particular to a project risk level assessment method based on approximate personalized transfer.
Background
Risk management of electrical engineering projects is important in project management. The risk level evaluation of the current power engineering projects is mainly carried out by capital construction workers to subjectively judge the risk level according to past experience. In the project construction process, the problems of difficult identification of key information, few consideration factors, insufficient detail reflection and the like exist, so that the accuracy of manually evaluating the risk level is low, and the efficiency of a manual evaluation mode is low.
The risk level assessment of the current power engineering project also partially adopts a machine learning algorithm, and a convolutional neural network is utilized to carry out iterative training model, so that the risk level assessment is carried out through the obtained model. However, the method has strong dependence on the model hyper-parameters, which may cause poor accuracy of the predicted risk level, complex calculation process, high requirement on calculation resources and low evaluation efficiency.
Disclosure of Invention
Aiming at the problems and the technical requirements, the inventor provides a project risk level assessment method based on approximate personalized transfer, which is used for solving the problems of poor accuracy and low assessment efficiency in project risk level assessment in the prior art and improving the accuracy and the assessment efficiency of project risk level assessment.
The embodiment of the application provides a project risk level assessment method based on approximate personalized delivery,
Acquiring a project data set of an electric power engineering construction project;
determining data division indexes of risk level assessment structures corresponding to each item of data in the item data set respectively, and determining the data division indexes as data features of the item data, wherein the risk level assessment structures are created based on item risk requirements of electric power engineering construction items;
Inputting the project data and the data characteristics into a pre-trained risk level assessment model to obtain risk scores corresponding to the power engineering construction projects output by the risk level assessment model, wherein the risk level assessment model is a graph neural network model, approximate personalized transfer values are transmitted in all nodes of the graph neural network model through an approximate personalized transfer mechanism, and the risk level assessment model is obtained by training based on project data samples, data characteristic samples, risk score samples and the approximate personalized transfer values;
and determining the item risk level corresponding to the risk score based on the corresponding relation between the pre-created risk score and the item risk level.
According to the project risk level assessment method based on approximate personalized transfer, which is provided by the embodiment of the application, the risk level assessment model comprises at least two nodes;
the training process for training the risk level assessment model based on the approximate personalized delivery mechanism comprises the following steps:
Acquiring at least two training data sample sets, wherein each training data sample set comprises a project data sample corresponding to an electric power engineering construction project, a data characteristic sample corresponding to the project data sample and a risk score sample corresponding to the electric power engineering construction project, and a node corresponds to one training data sample set;
Inputting the training data sample into the risk level assessment model, and performing the following data processing operation on each node through the risk level assessment model:
obtaining a current prediction risk score based on a project data sample corresponding to a current node, the data characteristic sample and a last approximate personalized transfer value sample transferred by a previous node; obtaining a current approximate personalized transfer value sample based on the project data sample, the data characteristic sample and the current prediction risk score, and inputting the current approximate personalized transfer value sample into a next node;
And optimizing model parameters of the risk level assessment model based on the predicted risk score corresponding to each node and the risk score sample corresponding to each node until the training of the risk level assessment model is completed under the condition that the node jump times reach the preset times.
According to the project risk level assessment method based on approximate personalized delivery provided by the embodiment of the application, the data processing operation performed on each node further comprises the following steps:
And obtaining a current approximate personalized transfer value sample based on the item data sample, the data feature sample and the current predicted risk score, and inputting the current approximate personalized transfer value sample into a next node.
According to the project risk level assessment method based on approximate personalized delivery provided by the embodiment of the application, the data processing operation performed on each node further comprises the following steps:
and under the condition that the current node is the last node, obtaining the current prediction risk score based on the item data sample corresponding to the last node, the data characteristic sample and the last approximate personalized transfer value sample transferred by the last node.
According to the project risk level assessment method based on approximate personalized delivery provided by the embodiment of the application, the determination process of the node jump times pi (i x) comprises the following steps:
Vectorizing the project data samples and the data feature samples to obtain a node feature matrix ix;
inputting the node characteristic matrix into a calculation formula of the number of hops to obtain the number of hops output by the calculation formula of the number of hops;
the calculation formula of the jump times comprises the following steps:
where pi (i x) represents the number of hops, alpha represents the probability of a hop to travel from the current node to the next node, Representing a preset matrix, i x representing a node characteristic matrix.
According to the method for evaluating the risk level of the project based on the approximate personalized delivery, which is provided by the embodiment of the application, the method for obtaining the current approximate personalized delivery value sample based on the project data sample, the data characteristic sample and the current predicted risk score comprises the following steps:
inputting the project data sample, the data characteristic sample and the current prediction risk score into a pre-established personalized transfer value calculation formula to obtain a personalized transfer value sample output by the personalized transfer value calculation formula;
Performing approximation processing on the personalized transfer value sample to obtain an approximate personalized transfer value sample;
wherein, individualized transmission value formula includes:
where Z represents a personalized transfer value sample, softmax represents a normalized exponential function, alpha represents a probability of a jump to walk from a current node to a next node, Representing a preset matrix, wherein H represents a current prediction risk score, and i x represents a node characteristic matrix.
According to the item risk level assessment method based on approximate personalized transfer provided by the embodiment of the application, the method for performing the approximation processing on the personalized transfer value sample to obtain the approximate personalized transfer value sample comprises the following steps:
acquiring an initial value of an approximate personalized transfer value sample;
Optimizing model parameters of the graph neural network model based on node skipping until the node skipping times reach preset times;
obtaining an approximate personalized transfer value sample calculation formula of the current node based on each iteration, and obtaining an approximate personalized transfer value sample through the approximate personalized transfer value sample calculation formula;
Wherein the approximate personalized transfer value sample calculation formula comprises:
Wherein Z ′ (k) represents an approximate personalized delivery value sample corresponding to the current node, and Z ′ (k-1) represents an approximate personalized delivery value sample corresponding to the previous node.
According to the project risk level assessment method based on approximate personalized transfer provided by the embodiment of the application, the model parameters of the risk level assessment model are optimized, and the method comprises the following steps:
optimizing the parameter weight corresponding to the model parameter, including:
calculating the parameter weight corresponding to the model parameter in the current iteration process by using a preset weight calculation formula;
wherein, weight calculation formula includes:
Wherein Q t represents the current parameter weight corresponding to the current iteration process, Q t-1 represents the last parameter weight corresponding to the last iteration process, gamma represents the parameter multiplier, beta represents the learning rate, Representing the first order momentum of the t-gradient,Represents the second order momentum of the t-gradient, e represents a constant, and λ represents a regularization coefficient.
According to the project risk level assessment method based on approximate personalized delivery provided by the embodiment of the application, the determining process of the risk score sample comprises the following steps:
Calculating risk scores of the project data samples based on preset risk score grade standards to obtain calculation results;
And converting the calculation result into a corresponding risk coding sample, and taking the risk coding sample as the risk score sample.
According to the project risk level assessment method based on approximate personalized transfer provided by the embodiment of the application, the risk level assessment model training is determined to be completed, and the method comprises the following steps:
acquiring at least two test data sample sets, wherein each test data sample set comprises test item data and test data characteristics corresponding to the test data sample;
the following test procedure is performed for each test data sample set:
Inputting the test item data and the test data sample into the risk level assessment model, wherein the risk level assessment model outputs a test risk score;
Determining the accuracy of the test risk scores corresponding to all the test data sample sets, and determining that the risk level assessment model training is completed under the condition that the accuracy is determined to be larger than the preset accuracy.
The project risk level assessment method based on approximate individuation transfer provided by the embodiment of the application is characterized by comprising the steps of acquiring a project data set of an electric power project construction project, determining a data division index of a risk level assessment structure corresponding to each project data in the project data set, determining the data division index as a data characteristic of the project data, wherein the risk level assessment structure is created based on project requirements of the electric power project construction project, and provides an effective data basis for subsequent project risk level assessment through the project data and the data characteristic of the project risk requirement of the electric power project, further inputting each project data and the data characteristic into a pre-trained risk level assessment model to obtain a risk score corresponding to the electric power project construction project output by the risk level assessment model, wherein the risk level assessment model is a graph neural network model, transmitting approximate individuation transfer values in each node of the graph neural network model through an approximate individuation transfer mechanism, determining the corresponding risk score of the project data and the corresponding to the project through the approximate individuation transfer mechanism, and accurately processing the graph neural network by utilizing the relation between the pre-created risk score and the project risk level data according to the corresponding relation between the project risk score and the project risk level assessment model, and the method can be accurately processed by utilizing the training information of the neural network model, and the key network can be processed relative to the data of the data can be processed by the training mechanism, and the key network can be accurately processed by only by the node of the training information, the method reduces the workload of data processing, improves the training speed of the graphic neural network model, predicts the risk score by utilizing the graphic neural network after training, finally obtains the project risk grade of the power engineering construction project, and achieves the purposes of improving the accuracy and the evaluation efficiency of project risk grade evaluation.
Drawings
In order to more clearly illustrate the embodiments of the present application or the technical solutions in the prior art, the drawings that are required in the embodiments or the description of the prior art will be briefly described, and it is obvious that the drawings in the following description are some embodiments of the present application, and other drawings may be obtained according to these drawings without inventive effort for a person skilled in the art.
FIG. 1 is a schematic flow chart of a method for evaluating risk level of an item based on approximate personalized delivery according to an embodiment of the present application;
Fig. 2 is a schematic structural diagram of an item risk level assessment device based on approximate personalized delivery according to an embodiment of the present application;
fig. 3 is a schematic structural diagram of an electronic device according to an embodiment of the present application.
Detailed Description
For the purpose of making the objects, technical solutions and advantages of the embodiments of the present application more apparent, the technical solutions of the embodiments of the present application will be clearly and completely described below with reference to the accompanying drawings in the embodiments of the present application, and it is apparent that the described embodiments are some embodiments of the present application, but not all embodiments of the present application. All other embodiments, which can be made by those skilled in the art based on the embodiments of the application without making any inventive effort, are intended to be within the scope of the application.
The embodiment of the application provides a project risk level assessment method based on approximate personalized transfer. The method can be applied to the intelligent terminal and the server. Some other descriptions of the embodiments of the present application are for illustration, not for limiting the scope of the application, and will not be described in detail later. The specific implementation of the method is shown in fig. 1:
step 101, acquiring a project data set of a power engineering construction project.
Step 102, determining data division indexes of risk level evaluation structures corresponding to the item data in the item data set respectively, and determining the data division indexes as data features of the item data.
The risk level assessment structure is created based on project risk requirements of the power engineering construction project.
And step 103, inputting the data and the data characteristics of each item into a pre-trained risk level assessment model to obtain risk scores corresponding to the power engineering construction items output by the risk level assessment model.
The risk level assessment model is a graph neural network model, an approximate personalized transfer value is transmitted in each node of the graph neural network model through an approximate personalized transfer mechanism, and the risk level assessment model is obtained through training based on a project data sample, a data characteristic sample, a risk score sample and the approximate personalized transfer value.
Step 104, determining the item risk level corresponding to the risk score based on the corresponding relation between the pre-created risk score and the item risk level.
The risk score is converted to obtain a corresponding risk code, and the item risk grade is obtained based on a corresponding relation between the pre-established risk code and the item risk grade.
The project risk level assessment method based on approximate individuation transfer provided by the embodiment of the application is characterized by comprising the steps of acquiring a project data set of an electric power project construction project, determining a data division index of a risk level assessment structure corresponding to each project data in the project data set, determining the data division index as a data characteristic of the project data, wherein the risk level assessment structure is created based on project requirements of the electric power project construction project, and provides an effective data basis for subsequent project risk level assessment through the project data and the data characteristic of the project risk requirement of the electric power project, further inputting each project data and the data characteristic into a pre-trained risk level assessment model to obtain a risk score corresponding to the electric power project construction project output by the risk level assessment model, wherein the risk level assessment model is a graph neural network model, transmitting approximate individuation transfer values in each node of the graph neural network model through an approximate individuation transfer mechanism, determining the corresponding risk score of the project data and the corresponding to the project through the approximate individuation transfer mechanism, and accurately processing the graph neural network by utilizing the relation between the pre-created risk score and the project risk level data according to the corresponding relation between the project risk score and the project risk level assessment model, and the method can be accurately processed by utilizing the training information of the neural network model, and the key network can be processed relative to the data of the data can be processed by the training mechanism, and the key network can be accurately processed by only by the node of the training information, the method reduces the workload of data processing, improves the training speed of the graphic neural network model, predicts the risk score by utilizing the graphic neural network after training, finally obtains the project risk grade of the power engineering construction project, and achieves the purposes of improving the accuracy and the evaluation efficiency of project risk grade evaluation.
Specifically, deconstructing the project data by the risk level evaluation structure. And obtaining a risk grade evaluation structure through grade evaluation structure discretization processing and weight addition processing.
The risk level evaluation structure is set to be a three-level evaluation structure, and corresponds to three-level division indexes, wherein the three-level division indexes respectively comprise a first-level index, a second-level index and a third-level index. Wherein the three-level evaluation structure is a subordinate structure.
The primary indexes comprise 5 indexes of personnel factors, mechanical equipment factors, material factors, operation method factors and environmental factors corresponding to the electric power engineering construction project.
The secondary indexes comprise 12 indexes of supervision units, construction mechanical equipment, safety systems and tools, construction materials, file materials, operation/technical methods, management methods, operation environments, natural environments and social environments corresponding to the electric power engineering construction projects.
The three-level index comprises: safety inspection level, quality inspection level, recheck execution force, drawing, fund delivery timeliness, construction requirement change, worker basic quality status, worker number, worker age status, worker training time, team leader basic quality status, team leader qualification, equipment maintenance condition, equipment load operation condition (equipment performance), equipment safety protection device inspection, equipment service life, equipment transportation, equipment involved site and time, system stability, system accuracy, tool and protection product spot inspection condition, material spot inspection quality, material replenishment and transportation time, material preservation status and applicability, material hazard level, material safety protection level, and equipment safety protection device inspection project related files, procedure information, qualification approval handling time, file arrangement and backup conditions, construction process level, number of operators (single person/multiple person, etc.), operation properties (general/temporary/rush repair, etc.), overtime construction conditions, accident potential correction and investigation frequency, project quality, emergency plan and exercise preparation, perfection of regulation and safety inspection, payroll reward and punishment system, hydropower supply conditions, construction site layout (building position, material storage position, etc.), operation site (high altitude, crossing railway, etc.), operation period (day, night), geological and hydrologic conditions, climate, epidemic situation and natural disaster, temperature, law, policy, resident opinion, and related department opinion 44 indexes.
Specifically, the weight of each data division index is obtained through a judgment matrix, and the method specifically comprises the following steps:
For example, the judgment matrix is
Wherein ABC represents the data division indicator, the range of values of x, y, z is (1, 3,5,7, 9), the important values corresponding to the respective data division indicators, for example, 1 represents less important, 3 represents slightly important, 5 represents significantly important, 7 represents strongly important and 9 represents extremely important.
The subscripts of x, y and z represent the data division indexes, and the judgment matrix is a schematic illustration, namely, the three data division indexes of ABC are used for illustration, and the specific data division indexes are specific indexes of the first-level index, the second-level index and the third-level index.
Of course, the user may also set the weight of each data division index based on the actual situation person.
In one particular embodiment, the risk level assessment model includes at least two nodes. The training process for training the risk level assessment model based on the approximate personalized delivery mechanism comprises the following steps:
The method comprises the steps of obtaining at least two training data sample sets, inputting the training data samples into a risk level assessment model, carrying out data processing operation on each node through the risk level assessment model, obtaining a current predicted risk score based on a project data sample corresponding to a current node, a data characteristic sample and a last approximate personalized transfer value sample transferred by a previous node, obtaining a current approximate personalized transfer value sample based on the project data sample, the data characteristic sample and the current predicted risk score, inputting the current approximate personalized transfer value sample into a next node, optimizing model parameters of the risk level assessment model based on the predicted risk score corresponding to each node and the risk score sample corresponding to each node, and determining that training of the risk level assessment model is completed under the condition that the node jump times reach preset times.
Each training data sample set comprises a project data sample corresponding to a power engineering construction project, a data characteristic sample corresponding to the project data sample and a risk score sample corresponding to the power engineering construction project. One node corresponds to each set of training data samples.
Specifically, the association relationship between nodes is established based on the engineering category and the units of each power engineering project.
The engineering category comprises a thermal power engineering project, a hydropower engineering project, a wind power engineering project, a power supply line engineering project, a photovoltaic project, a nuclear power project and a power supply bureau facility project.
Inputting samples in a plurality of training data sample sets configured with association relations between nodes into a risk level assessment model, performing data processing operation on the samples in each node through the risk level assessment model, comparing the predicted risk score and the risk score samples of each node, optimizing model parameters of the risk level assessment model based on comparison results, and determining that the training of the risk level assessment model is completed under the condition that the node jump times reach preset times.
The association relation between the nodes is also used as a data characteristic sample of the samples in the association nodes.
Wherein, samples in a plurality of training data sample sets configured with association relations between nodes are input into a risk level assessment model in the form of a graph structure.
In a specific embodiment, the data processing operation performed on each node further includes:
And obtaining a current approximate personalized transfer value sample based on the item data sample, the data feature sample and the current prediction risk score, and inputting the current approximate personalized transfer value sample into the next node.
Specifically, after inputting samples in a plurality of training data sample sets configured with association relations between nodes into a risk level evaluation model, one node is randomly determined and taken as the first node. When the risk score is predicted for the sample corresponding to the first node, no last approximate personalized transfer value sample transferred by the last node is used, and the current predicted risk score is obtained based on the project data sample and the data characteristic sample.
The data characteristic sample comprises each data division index, a node mark of each node with an association relation with a first node and each association relation. All the data characteristic samples corresponding to the nodes comprise the content.
In a specific embodiment, the data processing operation performed on each node further includes:
And under the condition that the current node is the last node, obtaining the current prediction risk score based on the item data sample, the data characteristic sample corresponding to the last node and the last approximate personalized transfer value sample transferred by the last node.
Specifically, as the last node does not transfer the nodes, the last node does not need to calculate the approximate personalized transfer value sample any more, and the prediction risk score corresponding to the last node is obtained.
In a specific embodiment, the determining process of the node jump number includes:
And inputting the node characteristic matrix into a calculation formula of the number of hops to obtain the number of hops output by the calculation formula of the number of hops.
Wherein, the calculation formula of the jump times is shown in formula (1):
Where pi (i x) represents the number of node hops, alpha represents the probability of a hop to travel from the current node to the next node, Representing a preset matrix, i x representing a node characteristic matrix.
The preset matrix is normalized through a preset self-annular adjacent matrix, and then is subjected to inversion with a preset degree matrix to obtain a matrix value.
In one embodiment, a risk score sample is predetermined, and the risk score sample determining process includes:
the risk score samples are divided into five risk score classes, for example as illustrated by table 1:
Table 1 risk score ranking table
Specifically, calculating risk scores of project data of the power engineering construction project based on the risk score grade standard, and converting the risk scores into risk coding samples corresponding to the risk scores to obtain risk score samples.
Specifically, the risk score of the power engineering construction project is obtained by multiplying the weight corresponding to the data division index corresponding to each project data by the score corresponding to each data division index and carrying out summation processing.
In one embodiment, the specific implementation of obtaining the current approximate personalized delivery value sample includes:
Inputting the project data sample, the data characteristic sample and the current prediction risk score into a personalized transfer value calculation formula to obtain a personalized transfer value sample output by the personalized transfer value calculation formula, and performing approximation processing on the personalized transfer value sample to obtain an approximate personalized transfer value sample.
Wherein, the calculation formula of the personalized transfer value is shown in formula (2):
where Z represents a personalized transfer value sample, softmax represents a normalized exponential function, alpha represents a probability of a jump to walk from a current node to a next node, Representing a preset matrix, wherein H represents a current prediction risk score, and i x represents a node characteristic matrix.
The node characteristic matrix is obtained by vectorizing the project data samples and the data characteristic samples.
Specifically, the specific implementation of the approximating process for the personalized delivery value sample includes:
initializing the personalized transfer value, and specifically referring to a formula (3):
Z′ (0)＝H＝fθ(X).................................(3)
wherein Z ′ (0) represents an initial value of the approximate personalized delivery value sample, and f θ (X) represents a graph neural network model with a model parameter θ.
Optimizing the graph neural network model through node jump, and obtaining a k+1st approximate personalized transfer value sample through a formula (4) after k iterations:
where Z' (k+1) represents the k+1 times approximately personalized transfer value sample.
Normalizing the last two iterations to obtain an approximate personalized transfer value sample calculation formula of the current node, see formula (5):
Wherein Z ′ (k) represents an approximate personalized delivery value sample corresponding to the current node, and Z ′ (k-1) represents an approximate personalized delivery value sample corresponding to the previous node.
Specifically, the approximate personalized transfer value sample, the jump probability and the prediction score sample corresponding to the previous node can be directly input into the formula (5) to obtain the approximate personalized transfer value sample corresponding to the current node.
Specifically, the approximate personalized transfer value sample corresponding to the previous node is set to zero for the first node, and the approximate personalized transfer value sample is not required to be calculated for the last node.
In a specific embodiment, optimizing the model parameters of the risk level assessment model includes optimizing the parameter weights corresponding to the model parameters.
The specific optimization mode of the parameter weight is that a preset weight calculation formula is utilized to calculate the parameter weight corresponding to the model parameter in the current iteration process.
Wherein, the weight calculation formula is shown in formula (6):
Wherein Q t represents the current parameter weight corresponding to the current iteration process, Q t-1 represents the last parameter weight corresponding to the last iteration process, gamma represents the parameter multiplier, beta represents the learning rate, Representing the first order momentum of the t-gradient,Represents the second order momentum of the t-gradient, e represents a constant, and λ represents a regularization coefficient.
In one embodiment, determining the specific implementation of the risk level assessment model training completion includes:
The method comprises the steps of obtaining at least two test data sample sets, inputting test item data and test data samples into a risk level assessment model, outputting test risk scores by the risk level assessment model, determining the accuracy of the test risk scores corresponding to all the test data sample sets, and determining that training of the risk level assessment model is completed under the condition that the accuracy is larger than a preset accuracy.
Each test data sample set comprises test item data and test data characteristics corresponding to the test data samples.
The method decouples the characteristic transformation and data transmission processes, so that the depth of the graph neural network model is completely independent of a propagation algorithm, and the nodes can obtain farther distance information without excessive parameterization. In addition, the relation among the nodes can be considered in the process of acquiring the graph structure data, so that the problem of low prediction efficiency caused by excessive model parameters due to a large amount of lost node information in the calculation process of the convolutional neural network is solved. And because less parameter information is used in the calculation by approximate personalized transfer, the result can more truly reflect the original data information, the probability of overfitting is reduced, and the result has more practical application value.
The embodiment of the application also provides a device for evaluating the project risk level based on approximate personalized transfer, the specific implementation of the device can refer to the description of the method for evaluating the project risk level based on approximate personalized transfer, and the repetition is not repeated, as shown in fig. 2, the device comprises:
An acquisition module 201, configured to acquire a project data set of a power engineering construction project;
The first determining module 202 is configured to determine data division indicators of risk level assessment structures corresponding to each item data in the item data set, and determine the data division indicators as data features of the item data, where the risk level assessment structures are created based on item risk requirements of the power engineering construction item;
The prediction module 203 is configured to input each item data and data features into a pre-trained risk level assessment model, and obtain a risk score corresponding to the power engineering construction item output by the risk level assessment model, where the risk level assessment model is a graph neural network model, and transmit an approximate personalized transfer value in each node of the graph neural network model through an approximate personalized transfer mechanism, and the risk level assessment model is obtained by training based on the item data sample, the data feature sample, the risk score sample and the approximate personalized transfer value;
The second determining module 204 is configured to determine, based on a pre-created correspondence between risk scores and item risk levels, an item risk level corresponding to the risk score.
Fig. 3 illustrates a physical schematic diagram of an electronic device, which may include a processor 301, a communication interface (Communications Interface) 302, a memory 303, and a communication bus 304, where the processor 301, the communication interface 302, and the memory 303 perform communication with each other through the communication bus 304, as shown in fig. 3. The processor 301 may call a logic instruction in the memory 303 to execute a project risk level assessment method based on approximate personalized transfer, where the method includes obtaining a project data set of an electric power project construction project, determining data division indexes of a risk level assessment structure corresponding to each project data in the project data set, respectively, and determining the data division indexes as data features of the project data, where the risk level assessment structure is created based on project risk requirements of the electric power project construction project, inputting each project data and the data features into a pre-trained risk level assessment model to obtain a risk score corresponding to the electric power project construction project output by the risk level assessment model, where the risk level assessment model is a graph neural network model, transmitting approximate personalized transfer values in each node of the graph neural network model through an approximate personalized transfer mechanism, training the risk level assessment model based on a project data sample, a data feature sample, a risk score sample and the approximate personalized transfer values, and determining a project risk level corresponding to the risk score based on a corresponding relation between the pre-created risk score and the project risk level.
Further, the logic instructions in the memory 303 may be implemented in the form of software functional units and stored in a computer readable storage medium when sold or used as a stand alone product. Based on this understanding, the technical solution of the present invention may be embodied essentially or in a part contributing to the prior art or in a part of the technical solution, in the form of a software product stored in a storage medium, comprising several instructions for causing a computer device (which may be a personal computer, a server, a network device, etc.) to perform all or part of the steps of the method according to the embodiments of the present invention. The storage medium includes a U disk, a removable hard disk, a Read-Only Memory (ROM), a random access Memory (RAM, random Access Memory), a magnetic disk, an optical disk, or other various media capable of storing program codes.
On the other hand, the invention also provides a computer program product, which comprises a computer program stored on a non-transitory computer readable storage medium, wherein the computer program comprises program instructions, when the program instructions are executed by a computer, the computer can execute the project risk level assessment method based on approximate personalized transfer provided by the methods, the method comprises the steps of acquiring a project data set of a power project construction project, determining a data division index of a risk level assessment structure corresponding to each project data in the project data set, determining the data division index as a data feature of the project data, wherein the risk level assessment structure is created based on project risk requirements of the power project construction project, inputting each project data and the data feature into a pre-trained risk level assessment model, obtaining a risk score corresponding to the power project construction project output by the risk level assessment model, wherein the risk level assessment model is a graph neural network model, transmitting approximate transfer values in each node of the graph neural network model through an approximate personalized transfer mechanism, the risk level assessment model is based on project data samples, data features, risk scores and approximate personalized transfer values, and risk level training relations are created based on the corresponding risk scores of the project data, and the risk level training relations are obtained.
In still another aspect, the present invention further provides a non-transitory computer readable storage medium, on which a computer program is stored, the computer program being implemented when executed by a processor to perform the method for evaluating a risk level of an item based on approximate personalized delivery provided in the foregoing embodiments, where the method includes obtaining a set of item data of an electrical engineering construction item, determining data division indicators of a risk level evaluation structure corresponding to each item data in the set of item data, and determining the data division indicators as data features of the item data, where the risk level evaluation structure is created based on item risk requirements of the electrical engineering construction item, inputting each item data and the data features into a pre-trained risk level evaluation model, obtaining a risk score corresponding to the electrical engineering construction item output by the risk level evaluation model, where the risk level evaluation model is a graph neural network model, transmitting an approximate personalized delivery value in each node of the graph neural network model by an approximate personalized delivery mechanism, where the risk level evaluation model is obtained by training based on the item data samples, the data feature samples, the score samples, and the approximate personalized delivery value, and determining a risk level corresponding to the item score based on a pre-created risk score and a corresponding relationship of the item risk level.
The apparatus embodiments described above are merely illustrative, wherein the elements illustrated as separate elements may or may not be physically separate, and the elements shown as elements may or may not be physical elements, may be located in one place, or may be distributed over a plurality of network elements. Some or all of the modules may be selected according to actual needs to achieve the purpose of the solution of this embodiment. Those of ordinary skill in the art will understand and implement the present invention without undue burden.
From the above description of the embodiments, it will be apparent to those skilled in the art that the embodiments may be implemented by means of software plus necessary general hardware platforms, or of course may be implemented by means of hardware. Based on this understanding, the foregoing technical solution may be embodied essentially or in a part contributing to the prior art in the form of a software product, which may be stored in a computer readable storage medium, such as ROM/RAM, a magnetic disk, an optical disk, etc., including several instructions for causing a computer device (which may be a personal computer, a server, or a network device, etc.) to execute the method described in the respective embodiments or some parts of the embodiments.
Finally, it should be noted that the above description is only of a preferred embodiment of the application, and the application is not limited to the above examples. It is to be understood that other modifications and variations which may be directly derived or contemplated by those skilled in the art without departing from the spirit and concepts of the present application are deemed to be included within the scope of the present application.
Claims (9)
1. A method for evaluating a risk level of an item based on approximate personalized delivery, the method comprising:
acquiring a project data set of an electric power engineering construction project;
determining data division indexes of risk level assessment structures corresponding to each item of data in the item data set respectively, and determining the data division indexes as data features of the item data, wherein the risk level assessment structures are created based on item risk requirements of electric power engineering construction items;
Inputting the project data and the data characteristics into a pre-trained risk level assessment model to obtain risk scores corresponding to the power engineering construction projects output by the risk level assessment model, wherein the risk level assessment model is a graph neural network model, approximate personalized transfer values are transmitted in all nodes of the graph neural network model through an approximate personalized transfer mechanism, and the risk level assessment model is obtained by training based on project data samples, data characteristic samples, risk score samples and the approximate personalized transfer values;
Determining a project risk level corresponding to the risk score based on a corresponding relation between the pre-created risk score and the project risk level;
The risk level assessment model comprises at least two nodes;
the training process for training the risk level assessment model based on the approximate personalized delivery mechanism comprises the following steps:
Acquiring at least two training data sample sets, wherein each training data sample set comprises a project data sample corresponding to an electric power engineering construction project, a data characteristic sample corresponding to the project data sample and a risk score sample corresponding to the electric power engineering construction project, and a node corresponds to one training data sample set;
Inputting the training data sample into the risk level assessment model, and performing the following data processing operation on each node through the risk level assessment model:
obtaining a current prediction risk score based on a project data sample corresponding to a current node, the data characteristic sample and a last approximate personalized transfer value sample transferred by a previous node; obtaining a current approximate personalized transfer value sample based on the project data sample, the data characteristic sample and the current prediction risk score, and inputting the current approximate personalized transfer value sample into a next node;
And optimizing model parameters of the risk level assessment model based on the predicted risk score corresponding to each node and the risk score sample corresponding to each node until the training of the risk level assessment model is completed under the condition that the node jump times reach the preset times.
2. The method for item risk level assessment based on approximate personalized delivery of claim 1, wherein the data processing operation performed on each node further comprises:
And obtaining a current approximate personalized transfer value sample based on the item data sample, the data feature sample and the current predicted risk score, and inputting the current approximate personalized transfer value sample into a next node.
3. The method for item risk level assessment based on approximate personalized delivery of claim 2, wherein the data processing operation performed on each node further comprises:
and under the condition that the current node is the last node, obtaining the current prediction risk score based on the item data sample corresponding to the last node, the data characteristic sample and the last approximate personalized transfer value sample transferred by the last node.
4. The method for evaluating the risk level of an item based on approximate personalized delivery according to claim 1, wherein the determining process of the node hop count pi (i x) comprises:
Vectorizing the project data samples and the data feature samples to obtain a node feature matrix i x;
inputting the node characteristic matrix into a calculation formula of the number of hops to obtain the number of hops output by the calculation formula of the number of hops;
the calculation formula of the jump times comprises the following steps:
where pi (i x) represents the number of hops, alpha represents the probability of a hop to travel from the current node to the next node, Representing a preset matrix, i x representing a node characteristic matrix.
5. The method for evaluating a risk level of an item based on approximate personalized delivery according to claim 1, wherein said obtaining a current approximate personalized delivery value sample based on said item data sample, said data feature sample, and said current predicted risk score comprises:
inputting the project data sample, the data characteristic sample and the current prediction risk score into a pre-established personalized transfer value calculation formula to obtain a personalized transfer value sample output by the personalized transfer value calculation formula;
Performing approximation processing on the personalized transfer value sample to obtain an approximate personalized transfer value sample;
wherein, individualized transmission value formula includes:
where Z represents a personalized transfer value sample, softmax represents a normalized exponential function, alpha represents a probability of a jump to walk from a current node to a next node, Representing a preset matrix, wherein H represents a current prediction risk score, and i x represents a node characteristic matrix.
6. The method for evaluating the risk level of an item based on approximate personalized delivery according to claim 5, wherein the approximating the personalized delivery value sample to obtain the approximate personalized delivery value sample comprises:
acquiring an initial value of an approximate personalized transfer value sample;
Optimizing model parameters of the graph neural network model based on node skipping until the node skipping times reach preset times;
obtaining an approximate personalized transfer value sample calculation formula of the current node based on each iteration, and obtaining an approximate personalized transfer value sample through the approximate personalized transfer value sample calculation formula;
Wherein the approximate personalized transfer value sample calculation formula comprises:
Wherein Z ′ (k) represents an approximate personalized delivery value sample corresponding to the current node, and Z ′ (k-1) represents an approximate personalized delivery value sample corresponding to the previous node.
7. The method for item risk level assessment based on approximate personalized delivery of claim 1, wherein optimizing model parameters of the risk level assessment model comprises:
optimizing the parameter weight corresponding to the model parameter, including:
calculating the parameter weight corresponding to the model parameter in the current iteration process by using a preset weight calculation formula;
wherein, weight calculation formula includes:
Wherein Q t represents the current parameter weight corresponding to the current iteration process, Q t-1 represents the last parameter weight corresponding to the last iteration process, gamma represents the parameter multiplier, beta represents the learning rate, Representing the first order momentum of the t-gradient,Represents the second order momentum of the t-gradient, e represents a constant, and λ represents a regularization coefficient.
8. The method for item risk level assessment based on approximate personalized delivery of claim 1, wherein the process of determining the risk score sample comprises:
Calculating risk scores of the project data samples based on preset risk score grade standards to obtain calculation results;
And converting the calculation result into a corresponding risk coding sample, and taking the risk coding sample as the risk score sample.
9. The method for risk level assessment of an item based on approximate personalized delivery of claim 1, wherein determining that the risk level assessment model training is complete comprises:
obtaining at least two test data sample sets, wherein each test data sample set comprises:
Test item data and test data characteristics corresponding to test data samples;
the following test procedure is performed for each test data sample set:
Inputting the test item data and the test data sample into the risk level assessment model, wherein the risk level assessment model outputs a test risk score;
Determining the accuracy of the test risk scores corresponding to all the test data sample sets, and determining that the risk level assessment model training is completed under the condition that the accuracy is determined to be larger than the preset accuracy.
Priority Applications (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN202411175384.0A CN119047833B (en) | 2024-08-26 | 2024-08-26 | Project risk level assessment method based on approximate personalized transfer |
Applications Claiming Priority (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN202411175384.0A CN119047833B (en) | 2024-08-26 | 2024-08-26 | Project risk level assessment method based on approximate personalized transfer |
Publications (2)
| Publication Number | Publication Date |
|---|---|
| CN119047833A CN119047833A (en) | 2024-11-29 |
| CN119047833B true CN119047833B (en) | 2025-07-08 |
Family
ID=93571624
Family Applications (1)
| Application Number | Title | Priority Date | Filing Date |
|---|---|---|---|
| CN202411175384.0A Active CN119047833B (en) | 2024-08-26 | 2024-08-26 | Project risk level assessment method based on approximate personalized transfer |
Country Status (1)
| Country | Link |
|---|---|
| CN (1) | CN119047833B (en) |
Citations (2)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN112241920A (en) * | 2020-09-25 | 2021-01-19 | 徐佳慧 | Investment and financing organization evaluation method, system and equipment based on graph neural network |
| CN118153952A (en) * | 2024-03-20 | 2024-06-07 | 南京航空航天大学 | Power resource infrastructure project risk level assessment method under attention mechanism optimization |
Family Cites Families (6)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| US10740396B2 (en) * | 2013-05-24 | 2020-08-11 | Sap Se | Representing enterprise data in a knowledge graph |
| US10380896B2 (en) * | 2017-12-27 | 2019-08-13 | Intel Corporation | Characterizing proximity risks within a radio mesh |
| US11704600B2 (en) * | 2019-06-27 | 2023-07-18 | Microsoft Technology Licensing, Llc | Multistage feed ranking system with methodology providing scalable multi-objective model approximation |
| CN114510405B (en) * | 2022-01-28 | 2024-07-09 | 腾讯科技（深圳）有限公司 | Index data evaluation method, apparatus, device, storage medium, and program product |
| CN115145626A (en) * | 2022-06-24 | 2022-10-04 | 中国航空结算有限责任公司 | Project configuration management method and device and electronic equipment |
| AU2023342945A1 (en) * | 2022-09-13 | 2025-04-03 | Richard ISAACSON | Apparatus for generating a personalized risk assessment for neurodegenerative disease |
-
2024
- 2024-08-26 CN CN202411175384.0A patent/CN119047833B/en active Active
Patent Citations (2)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN112241920A (en) * | 2020-09-25 | 2021-01-19 | 徐佳慧 | Investment and financing organization evaluation method, system and equipment based on graph neural network |
| CN118153952A (en) * | 2024-03-20 | 2024-06-07 | 南京航空航天大学 | Power resource infrastructure project risk level assessment method under attention mechanism optimization |
Also Published As
| Publication number | Publication date |
|---|---|
| CN119047833A (en) | 2024-11-29 |
Similar Documents
| Publication | Publication Date | Title |
|---|---|---|
| Abrishami et al. | BN-SLIM: A Bayesian Network methodology for human reliability assessment based on Success Likelihood Index Method (SLIM) | |
| CN116628220A (en) | Electric power work ticket generation method and system based on relation graph convolution neural network | |
| CN113891342B (en) | Base station inspection method, device, electronic equipment and storage medium | |
| CN111737486B (en) | Person post matching method and storage device based on knowledge graph and deep learning | |
| CN115392627A (en) | An online risk assessment method and system for distribution network | |
| CN118153952B (en) | Power resource infrastructure project risk level assessment method under attention mechanism optimization | |
| CN111025041A (en) | Electric vehicle charging pile monitoring method and system, computer equipment and medium | |
| CN117312791A (en) | Current transformer error state trend prediction method, device, equipment and medium | |
| CN114091908A (en) | Comprehensive evaluation method, device and equipment for distribution network considering multi-mode energy storage station | |
| CN114548634A (en) | A method, device, storage medium and electronic device for comprehensive assessment of offshore wind resources | |
| CN119066668A (en) | Safety operation management system based on normalization | |
| CN116401372A (en) | Knowledge graph representation learning method and device, electronic equipment and readable storage medium | |
| CN119047833B (en) | Project risk level assessment method based on approximate personalized transfer | |
| Fan et al. | Adaptive RAS/SPS System Setting for Improving Grid Reliability and Asset Utilization through Predictive Simulation and Controls. A Use Case for Transformative Remedial Action Scheme Tool (TRAST): Jim Bridger RAS Evaluation and Analysis | |
| CN115526393A (en) | Construction cost prediction method based on transformer project key influence factor screening | |
| CN109193779B (en) | Distributed wind power generation maximum capacity evaluation method | |
| CN116703075B (en) | Power scheduling decision method, device, electronic equipment and storage medium | |
| CN117788016A (en) | Carbon emission comprehensive efficiency evaluation method, device, equipment and storage medium | |
| CN117196257A (en) | Auxiliary settlement system for power distribution network engineering | |
| CN117252305A (en) | A house risk assessment method, device, equipment and medium | |
| CN116957324A (en) | Risk control method, risk control device, electronic equipment and storage medium | |
| Borisov et al. | The System of Fuzzy Cognitive Analysis and Modeling of System Dynamics | |
| CN115545533A (en) | Data processing method, device, equipment, medium and product | |
| Ma et al. | Highway construction investment risk evaluation using BP neural network model | |
| Syaputra et al. | Wind Speed Modelling using Some Modified Gamma Distribution |
Legal Events
| Date | Code | Title | Description |
|---|---|---|---|
| PB01 | Publication | ||
| PB01 | Publication | ||
| SE01 | Entry into force of request for substantive examination | ||
| SE01 | Entry into force of request for substantive examination | ||
| GR01 | Patent grant | ||
| GR01 | Patent grant |