CN114997491A - Method, device, equipment and storage medium for optimizing gasoline octane number loss - Google Patents

Abstract

The invention relates to a method, a device, equipment and a storage medium for optimizing the loss of gasoline octane number, wherein the method comprises the following steps: acquiring a gasoline refining data sample set, wherein the gasoline refining data sample set comprises an operation variable sample set and a production factor sample set; screening out a target operation variable sample set which meets a preset requirement from the operation variable sample set based on a random forest algorithm; establishing a gasoline octane number loss prediction model, and training the gasoline octane number loss prediction model according to the target operation variable sample set and the production factor sample set to obtain a target gasoline octane number loss prediction model with complete training; and setting constraint conditions, and determining the optimization conditions of the target operating variables meeting the constraint conditions based on the well-trained target gasoline octane number loss prediction model. The invention relates to a method, a device, equipment and a storage medium for optimizing the octane number loss of gasoline, which are used for screening operating variables from a sample set and improving the optimization efficiency.

Description

A kind of gasoline octane loss optimization method, device, equipment and storage medium

technical field

The invention relates to the technical field of gasoline refining, in particular to a method, device, equipment and storage medium for optimizing the loss of gasoline octane number.

Background technique

Gasoline is the main fuel for small vehicles, and the exhaust emissions from gasoline combustion have an important impact on the atmospheric environment. To this end, countries around the world have developed increasingly stringent gasoline quality standards. The focus of gasoline cleaning is to reduce the sulfur and olefin content in gasoline while maintaining its octane number as much as possible. At present, domestic refining process equipment and processes are not unified, the composition of raw materials is complex, and there are many uncontrollable factors, making it difficult to achieve the goal of continuous capacity expansion and optimized production. Therefore, under the current strict national standards, how to reduce sulfur, olefins and other substances in gasoline so that chemical plants can obtain gasoline with the highest octane number as possible has become the focus and difficulty in the field of gasoline production.

The focus of gasoline cleaning is to reduce the sulfur and olefin content in gasoline while maintaining its octane number as much as possible. However, the modeling of chemical processes in the prior art is generally realized by means of data association or mechanism modeling, and there is a linear relationship between their operating variables.

The traditional data association model has relatively few variables, and the mechanism modeling requires high analysis of raw materials, and the response to process optimization is not timely, so the effect is not ideal. Therefore, how to establish an octane loss model in the gasoline refining process and optimize the operation is an urgent problem to be solved.

SUMMARY OF THE INVENTION

In view of this, it is necessary to provide a method, device, equipment and storage medium for optimizing gasoline octane number loss, so as to solve the problems of less related variables and poor optimization effect in the prior art when optimizing gasoline octane number loss.

In order to achieve the above-mentioned technical purpose, the present invention has adopted the following technical solutions:

In a first aspect, the present invention provides a method for optimizing gasoline octane number loss, comprising:

Obtain a gasoline refining data sample set, which includes an operational variable sample set and a production factor sample set;

Based on the random forest algorithm, select the target operating variable sample set that meets the preset requirements from the operating variable sample set;

Establish a gasoline octane loss prediction model, train the gasoline octane loss prediction model according to the target operating variable sample set and the production factor sample set, and obtain a fully trained target gasoline octane loss prediction model;

Constraints are set, and based on the well-trained target gasoline octane loss prediction model, the optimization conditions of the target operating variables that satisfy the constraints are determined.

Preferably, based on the random forest algorithm, the target operating variable sample set that meets the preset requirements is screened from the operating variable sample set, including:

Preprocess the gasoline refining data sample set to obtain a reliable gasoline refining data sample set;

Build multiple decision trees, calculate the reliable gasoline refined data sample set error of each decision tree by OOB method, and obtain the OOB error;

Reorder the manipulated variables, calculate the error of the manipulated variables again, and get the reordering error;

Calculate the importance of each manipulated variable based on the OOB error and the reordering error;

According to the importance of the manipulated variables, the manipulated variables that meet the preset requirements are screened out, and the target manipulated variable sample set is obtained.

Preferably, a gasoline octane loss prediction model is established, and the gasoline octane loss prediction model is trained according to the target operating variable sample set and the production factor sample set to obtain a fully trained target gasoline octane loss prediction model, including:

Perform Bayesian normalization on the target operating variable sample set and the production factor sample set;

Establish a gasoline octane loss prediction model, and set the parameters of the gasoline octane loss prediction model;

Input the normalized target operating variable sample set and production factor sample set into the gasoline octane loss prediction model, and use the octane loss as the output to train the gasoline octane loss prediction model to obtain a fully trained target. Gasoline octane loss prediction model.

Preferably, a gasoline octane loss prediction model is established, and parameters of the gasoline octane loss prediction model are set, including:

Set the excitation function, network training function and network performance function type of the gasoline octane loss prediction model;

Set the number of neurons in the hidden layer of the gasoline octane loss prediction model;

Set the network parameters of the gasoline octane loss prediction model.

Preferably, constraints are set, and based on the well-trained target gasoline octane loss prediction model, the optimization conditions of the target operating variables that satisfy the constraints are determined, including:

Establish an associative neural network model and set constraints;

According to the correlation neural network model, establish the functional relationship f(net1) between octane loss and operating variables, and the functional relationship f(net2) between gasoline sulfur content and operating variables;

According to the functional relationship f(net1) and the functional relationship f(net2), the optimization conditions of the optimal operating variables satisfying the constraints are determined.

Preferably, according to the functional relationship f(net1) and the functional relationship f(net2), the optimization conditions of the target operating variables that satisfy the constraints are determined, including:

According to the functional relationship f(net1) and the functional relationship f(net2), determine the octane number loss and sulfur content corresponding to each operating variable;

According to the octane number loss and sulfur content corresponding to the operating variables, and record the optimization conditions corresponding to the target operating variables;

The above operations are repeated until the optimization conditions corresponding to all target operating variables are determined.

Preferably, the gasoline refining data sample set is preprocessed to obtain a reliable gasoline refining data sample set, including:

Determine whether there is incomplete data in the gasoline refining data sample set;

Delete the incomplete data that reaches the incomplete threshold, and replace the incomplete data that does not reach the incomplete threshold;

Determine whether there is abnormal data in the gasoline refining data sample set;

According to the preset criteria, the abnormal data is eliminated to obtain a reliable gasoline refining data sample set.

In a second aspect, the present invention also provides a gasoline octane number loss optimization device, comprising:

The data acquisition module is used to acquire the gasoline refining data sample set, and the gasoline refining data sample set includes the operation variable sample set and the production factor sample set;

The screening module is used to screen out the target operating variable sample set that meets the preset requirements from the operating variable sample set based on the random forest algorithm;

The modeling module is used to establish a gasoline octane loss prediction model, and train the gasoline octane loss prediction model according to the target operating variable sample set and the production factor sample set, and obtain a fully trained target gasoline octane loss prediction model ;

The optimization module is used to set constraints, and based on the well-trained target gasoline octane loss prediction model, determine the optimization conditions of the target operating variables that satisfy the constraints.

In a third aspect, the present invention also provides an electronic device including a memory and a processor, wherein,

memory for storing programs;

A processor, coupled with the memory, is configured to execute the program stored in the memory, so as to implement the steps in the method for optimizing gasoline octane number loss in any one of the above implementation manners.

In a fourth aspect, the present invention also provides a computer-readable storage medium for storing computer-readable programs or instructions. When the programs or instructions are executed by a processor, the gasoline-integrated storage medium in any of the above-mentioned implementations can be implemented. Steps in the Alkane Loss Optimization Method.

The beneficial effects of adopting the above embodiments are: a gasoline octane loss optimization method, device, equipment and storage medium provided by the present invention can obtain a gasoline refining data sample set, and the gasoline refining data sample set has multiple sets of data, which increases gasoline The relevant variables in the octane loss optimization, and screen out the key operating variables to reduce the gasoline octane loss, and determine the optimal target under the set constraints by training a complete target gasoline octane loss prediction model The optimization conditions of operating variables improve the effect of gasoline octane loss optimization under certain constraints.

Description of drawings

Fig. 1 is the schematic flow sheet of an embodiment of the gasoline octane number loss optimization method provided by the invention;

2 is a schematic flowchart of an embodiment of screening operational variables provided by the present invention;

Fig. 3 is the importance degree diagram of an embodiment of the operation variable sorting provided by the present invention;

4 is a schematic structural diagram of an embodiment of a gasoline octane number loss optimization device provided by the present invention;

FIG. 5 is a schematic structural diagram of an electronic device provided by an embodiment of the present invention.

Detailed ways

The preferred embodiments of the present invention are specifically described below with reference to the accompanying drawings, wherein the accompanying drawings constitute a part of the present application, and together with the embodiments of the present invention, are used to explain the principles of the present invention, but are not used to limit the scope of the present invention.

In the description of the present application, "plurality" means two or more, unless otherwise expressly and specifically defined.

Reference herein to an "embodiment" means that a particular feature, structure, or characteristic described in connection with the embodiment can be included in at least one embodiment of the present invention. The appearances of the phrase in various places in the specification are not necessarily all referring to the same embodiment, nor a separate or alternative embodiment that is mutually exclusive of other embodiments. It is explicitly and implicitly understood by those skilled in the art that the embodiments described herein may be combined with other embodiments.

The present invention provides a method, device, equipment and storage medium for optimizing the loss of gasoline octane number, which will be described separately below.

Please refer to FIG. 1. FIG. 1 is a schematic flowchart of an embodiment of a gasoline octane number loss optimization method provided by the present invention. A specific embodiment of the present invention discloses a gasoline octane number loss optimization method, including:

S101. Obtain a gasoline refining data sample set, where the gasoline refining data sample set includes an operational variable sample set and a production factor sample set;

S102. Based on the random forest algorithm, select a target operating variable sample set that meets preset requirements from the operating variable sample set;

S103, establishing a gasoline octane number loss prediction model, and training the gasoline octane number loss prediction model according to the target operation variable sample set and the production factor sample set, to obtain a target gasoline octane number loss prediction model with complete training;

S104 , setting constraints, and determining optimization conditions of target operating variables that satisfy the constraints based on the well-trained target gasoline octane loss prediction model.

In a specific embodiment of the present invention, step S101 refers to the preprocessing results of industrial data in the past 4 years, and selects variables such as 7 raw material properties, 2 raw adsorbent properties, 2 regenerated adsorbent properties, and 2 product properties. and another 354 operating variables (total 367 variables) to analyze their effects on gasoline octane loss value, and construct gasoline refining data sample set based on the preprocessing results of industrial data in the past 4 years.

In a specific embodiment of the present invention, step S102 calculates the importance of each operational variable in the industrial data in the past 4 years through the random forest algorithm, and the preset requirement is to arrange the importance from high to low, and the highest 30 operational variables . The selected sample set of target operating variables has a more obvious effect on the optimization of gasoline octane loss.

In a specific embodiment of the present invention, the gasoline octane loss prediction model established in step S103 is a BP neural network model, the standard operation variable sample set and the production factor sample set are used as the input of the BP neural network model, and the gasoline octane number is used as the input of the BP neural network model. As the output, the BP neural network model is trained to obtain a fully trained target gasoline octane loss prediction model, that is, a model that meets the prediction accuracy of gasoline octane loss.

In a specific embodiment of the present invention, step S104 adjusts 30 target operating variables by training a complete target gasoline octane loss prediction model under constraints, and records their influence on gasoline octane loss, Find the optimal optimization conditions for each operating variable, record them, and determine the optimal optimization conditions for all operating variables. According to the constraints, the optimization conditions of the optimal target operating variables under the constraints are determined, which improves the optimization effect of the gasoline octane loss under the specific constraints.

Compared with the prior art, the present embodiment provides a gasoline octane loss optimization method, which obtains a gasoline refining data sample set, and the gasoline refining data sample set has multiple sets of data, which increases the correlation when optimizing gasoline octane loss. variable, and screen out the key operating variables to reduce gasoline octane loss, and by training a complete target gasoline octane loss prediction model to determine the optimal optimization conditions of the target operating variables under the set constraints, improve the The effect of gasoline octane loss optimization under specific constraints.

Please refer to FIG. 2. FIG. 2 is a schematic flowchart of an embodiment of screening operating variables provided by the present invention. In some embodiments of the present invention, based on a random forest algorithm, targets that meet preset requirements are screened from a sample set of operating variables. A sample set of manipulated variables, including:

S201. Preprocess the gasoline refining data sample set to obtain a reliable gasoline refining data sample set;

S202, constructing multiple decision trees, and calculating the error of the reliable gasoline refined data sample set of each decision tree by the OOB method, and obtaining the OOB error;

S203, reordering the operating variables, and calculating the error of the operating variables again to obtain the reordering error;

S204, calculate the importance of each operating variable according to the OOB error and the reordering error;

S205 , according to the importance of the operating variables, screen out the operating variables that meet the preset requirements, and obtain a target operating variable sample set.

In a specific embodiment of the present invention, the gasoline refining data sample set in step S201 includes the preprocessing results of industrial data in the past four years. The data is relatively large, and some data may have defects or even errors. If the gasoline refining data sample set is directly used , which will cause serious errors in the final optimization results. Therefore, a reliable gasoline refining data sample set is obtained through preprocessing.

In a specific embodiment of the present invention, the random forest algorithm in step S202 actually obtains the prediction result by relying on the constructed multiple decision trees. Random forest can calculate the importance of features and rank them. In this paper, the OOB (Out-Of-Bag) method will be used to calculate the error value, and the OOB error will be obtained, which are respectively recorded as Err _oob 1, Err _oob 2,..., Err _oob k.

In a specific embodiment of the present invention, step S203 reorders 367 data of 30 kinds of operating variables, and calculates the errors of the operating variables after the reordering, and obtains the reordering errors, which are respectively denoted as Err _i 1 and Err _i 2 ,…,Err _i k. It should be noted that the reordering error and OOB error of the same manipulated variable are corresponding.

计算每个操作变量的重要性，其中，I为操作变量的重要性，k为误差的总数。In a specific embodiment of the present invention, step S204 is based on the formula

Calculate the importance of each manipulated variable, where I is the importance of the manipulated variable and k is the total number of errors.

In a specific embodiment of the present invention, step S205 sorts the calculated importance of each operating variable, and selects the top 30 optimal features as main features, that is, target operating variables. It should be noted that a sample set of 30 corresponding target manipulated variables that are screened out from the manipulated variable sample set is the target manipulated variable sample set. Please refer to FIG. 3 . FIG. 3 is an importance level diagram of an embodiment of the operation variable sorting provided by the present invention.

In some embodiments of the present invention, a gasoline octane loss prediction model is established, and the gasoline octane loss prediction model is trained according to the target operating variable sample set and the production factor sample set to obtain a fully trained target gasoline octane number Loss prediction models, including:

Perform Bayesian normalization on the target operating variable sample set and the production factor sample set;

Establish a gasoline octane loss prediction model, and set the parameters of the gasoline octane loss prediction model;

Input the normalized target operating variable sample set and production factor sample set into the gasoline octane loss prediction model, and use the octane loss as the output to train the gasoline octane loss prediction model to obtain a fully trained target. Gasoline octane loss prediction model.

其中，n为输入层神经元个数，m为输出层神经元个数，a为[1,10]之间的常数。将目标操作变量样本集以及生产因素样本集进行贝叶斯归一化到[-1,1]范围内，设输入层的节点数为30，输出层的节点数为1，建立汽油辛烷值损失预测模型。In the above embodiment, according to the structure of the neural network, if there is a neural network with a hidden layer, as long as it has enough hidden nodes, a nonlinear function can be infinitely approximated with arbitrary precision. Therefore, the prediction of the mathematical model of the present invention adopts a prediction model established by a three-layer multi-input single-output BP neural network including one hidden layer. The neural network prediction of this mathematical model refers to the following empirical formula for selecting the number of neurons in the hidden layer:

Among them, n is the number of neurons in the input layer, m is the number of neurons in the output layer, and a is a constant between [1, 10]. The target operating variable sample set and the production factor sample set are Bayesian normalized to the range of [-1, 1], the number of nodes in the input layer is 30, and the number of nodes in the output layer is 1, and the gasoline octane number is established. loss prediction model.

The normalized target operating variable sample set and production factor samples are used to predict the gasoline octane loss prediction model, and the gasoline octane loss prediction model that meets the prediction requirements is the well-trained target gasoline octane loss prediction model.

In some embodiments of the present invention, a gasoline octane loss prediction model is established, and parameters of the gasoline octane loss prediction model are set, including:

Set the excitation function, network training function and network performance function type of the gasoline octane loss prediction model;

Set the number of neurons in the hidden layer of the gasoline octane loss prediction model;

Set the network parameters of the gasoline octane loss prediction model.

In the above-mentioned embodiment, the BP neural network usually adopts the sigmoid differentiable function and the linear function as the excitation function of the network. The model used in this problem is to select the sigmoid tangent function tansig as the excitation function of the hidden layer neurons, and set the output of the network to be normalized to the range of [-1, 1]. Therefore, the prediction model selects the sigmoid logarithmic function tansig as the output. The activation function of the layer neurons.

The sigmoid activation function formula is as follows:

The tansig activation function formula is as follows:

The formula of the logsig activation function is as follows:

Set the excitation functions of the hidden layer and output layer of the neural network as tansig and logsig functions respectively, the network training function as trainingdx, and the network performance function as mse, and then set the number of neurons in the hidden layer to 10 initially. Adjust the situation appropriately, and then set the network parameters. The number of network iterations epochs is 244, the expected error goal is 0.00000001, and the learning rate lr is 0.01. After setting the parameters, start training the neural network, which may take a long time to wait.

In some embodiments of the present invention, constraints are set, and based on a well-trained target gasoline octane loss prediction model, the optimization conditions of target operating variables that satisfy the constraints are determined, including:

Establish an associative neural network model and set constraints;

According to the correlation neural network model, establish the functional relationship f(net1) between octane loss and operating variables, and the functional relationship f(net2) between gasoline sulfur content and operating variables;

According to the functional relationship f(net1) and the functional relationship f(net2), the optimization conditions of the optimal operating variables satisfying the constraints are determined.

In the above embodiments, the sulfur content has many adverse effects on gasoline, and in the embodiment of the present invention, the constraint is that the sulfur content is not greater than 5ug/g. Establish a correlation neural network model, and use it to establish a functional relationship between the loss of octane number (RON) and the 30 main operating variables after screening, denoted as f(net1). Then, a neural network is used to establish the functional relationship between the product sulfur content (not the raw material sulfur content) and 30 variables, denoted as f(net2), and the octane number (RON) is the loss of octane number before adjustment. And according to the functional relationship f(net1) and the functional relationship f(net2), the optimization conditions of the optimal operating variables that satisfy the constraints are further determined.

In some embodiments of the present invention, according to the functional relationship f(net1) and the functional relationship f(net2), determining the optimization condition of the target operating variable that satisfies the constraint condition, including:

According to the functional relationship f(net1) and the functional relationship f(net2), determine the octane number loss and sulfur content corresponding to each operating variable;

According to the octane number loss and sulfur content corresponding to the operating variables, and record the optimization conditions corresponding to the target operating variables;

The above operations are repeated until the optimization conditions corresponding to all target operating variables are determined.

In the above embodiment, the target plan 1 is defined as f(net1)<0.7*RON, and the target plan 2 is defined as f(net2)<=5. According to the well-trained target gasoline octane loss prediction model, the calculation of each The octane loss and sulfur content corresponding to an operating variable, as well as the acceptance probability, and record the optimization conditions corresponding to the target operating variable. It can be understood that, the functional relationship f(net1) and the functional relationship f(net2) in the present invention can also be adjusted according to the actual situation.

In some embodiments of the present invention, the gasoline refining data sample set is preprocessed to obtain a reliable gasoline refining data sample set, including:

Determine whether there is incomplete data in the gasoline refining data sample set;

Delete the incomplete data that reaches the incomplete threshold, and replace the incomplete data that does not reach the incomplete threshold;

Determine whether there is abnormal data in the gasoline refining data sample set;

According to the preset criteria, the abnormal data is eliminated to obtain a reliable gasoline refining data sample set.

In the above embodiment, for sites that only contain part of the time points, if there are too many incomplete data and cannot be supplemented, such sites can be deleted; for sites with all data empty, such sites can be deleted Point deletion; for some sites with empty data, you can replace them with the average data of the two hours before and after; by summarizing the FCC gasoline, you can summarize the operating range of the original data variables, and then remove a part that is not here. The location of the sample in the operating range; the default criterion is the Laida criterion (3σ criterion), and outliers are removed by the Laida criterion.

The Laida Criterion is specifically:

First, it is assumed that the estimated manipulated variables are measured with equal precision to obtain x ₁ , x ₂ , ..., x _n , and the arithmetic mean x and residual error vi = x _i -x ( _i =1,2 , 3,...,n), and calculate the standard error σ according to the Bessel formula, if the residual error v _b (1<=b<=n) of an estimated variable value x _b satisfies |v _b | =|x _b -x|>3σ, then it is considered that x _b is a bad value with a gross error value and should be eliminated. The Bessel formula is as follows:

where n is the total number of manipulated variables, v is the error, and x is the manipulated variable.

In order to better implement the method for optimizing the loss of gasoline octane number in the embodiment of the present invention, on the basis of the method for optimizing the loss of gasoline octane number, correspondingly, please refer to FIG. 4 , which is the loss of gasoline octane number provided by the present invention. A schematic structural diagram of an embodiment of an optimization device, an embodiment of the present invention provides a gasoline octane loss optimization device 400, including:

The data acquisition module 401 is used to acquire a gasoline refining data sample set, and the gasoline refining data sample set includes an operation variable sample set and a production factor sample set;

The screening module 402 is configured to, based on the random forest algorithm, screen out the target operating variable sample set that meets the preset requirements from the operating variable sample set;

The modeling module 403 is used for establishing a gasoline octane loss prediction model, and according to the target operating variable sample set and the production factor sample set, the gasoline octane loss prediction model is trained to obtain a fully trained target gasoline octane loss prediction Model;

The optimization module 404 is configured to set constraints, and based on the well-trained target gasoline octane loss prediction model, determine the optimization conditions of the target operating variables that satisfy the constraints.

It should be noted here that the apparatus 400 provided in the above embodiments can implement the technical solutions described in the above method embodiments, and the specific implementation principles of the above modules or units can refer to the corresponding content in the above method embodiments, which are not repeated here. Repeat.

Please refer to FIG. 5 , which is a schematic structural diagram of an electronic device according to an embodiment of the present invention. Based on the above-mentioned method for optimizing the loss of gasoline octane number, the present invention also provides a device for optimizing the loss of gasoline octane number, which can be a mobile terminal, a desktop computer, a notebook, a palmtop computer, a server, etc. computing equipment. The gasoline octane loss optimization apparatus includes a processor 510 , a memory 520 and a display 530 . FIG. 5 shows only some components of the electronic device, but it should be understood that implementation of all of the illustrated components is not required, and more or fewer components may be implemented instead.

The memory 520 may in some embodiments be an internal storage unit of the gasoline octane loss optimization device, such as a hard disk or memory of the gasoline octane loss optimization device. In other embodiments, the memory 520 may also be an external storage device of the gasoline octane loss optimization device, such as a plug-in hard disk equipped on the gasoline octane loss optimization device, a smart memory card (Smart Media Card, SMC), Secure digital (Secure Digital, SD) card, flash memory card (FlashCard) and so on. Further, the memory 520 may also include both an internal storage unit of the gasoline octane loss optimization device and an external storage device. The memory 520 is used to store application software and various data installed in the gasoline octane loss optimization device, such as program codes installed in the gasoline octane loss optimization device. The memory 520 may also be used to temporarily store data that has been output or is to be output. In one embodiment, the memory 520 stores a gasoline octane loss optimization program 540, and the gasoline octane loss optimization program 540 can be executed by the processor 510, so as to realize the gasoline octane loss in various embodiments of the present application. Optimization.

The processor 510 may be a central processing unit (CPU), a microprocessor or other data processing chips in some embodiments, and is used to execute program codes or process data stored in the memory 520, such as executing gasoline octane Value loss optimization methods, etc.

In some embodiments, the display 530 may be an LED display, a liquid crystal display, a touch-sensitive liquid crystal display, an OLED (Organic Light-Emitting Diode, organic light-emitting diode) touch device, and the like. Display 530 is used to display information at the gasoline octane loss optimization device and to display a user interface for visualization. The components 510-530 of the gasoline octane loss optimization apparatus communicate with each other via the system bus.

In one embodiment, the steps in the above gasoline octane loss optimization method are implemented when the processor 510 executes the gasoline octane loss optimization program 540 in the memory 520 .

The present embodiment also provides a computer-readable storage medium on which a gasoline octane loss optimization program is stored, and when the gasoline octane loss optimization program is executed by a processor, the following steps are implemented:

Obtain a gasoline refining data sample set, which includes an operational variable sample set and a production factor sample set;

Based on the random forest algorithm, select the target operating variable sample set that meets the preset requirements from the operating variable sample set;

Establish a gasoline octane loss prediction model, train the gasoline octane loss prediction model according to the target operating variable sample set and the production factor sample set, and obtain a fully trained target gasoline octane loss prediction model;

Constraints are set, and based on the well-trained target gasoline octane loss prediction model, the optimization conditions of the target operating variables that satisfy the constraints are determined.

To sum up, the present embodiment provides a method, device, equipment and storage medium for optimizing gasoline octane number loss, and obtains a gasoline refining data sample set. The gasoline refining data sample set has multiple sets of data, which increases the optimization of gasoline octane number loss. The key operating variables that reduce the gasoline octane loss are screened out, and the optimal target operating variable optimization conditions are determined under the set constraints by training a complete target gasoline octane loss prediction model. , which improves the effect of gasoline octane loss optimization under certain constraints.

The above description is only a preferred embodiment of the present invention, but the protection scope of the present invention is not limited to this. Substitutions should be covered within the protection scope of the present invention.

Claims (10)

1. A method for optimizing octane number loss of gasoline, comprising:

acquiring a gasoline refining data sample set, wherein the gasoline refining data sample set comprises an operation variable sample set and a production factor sample set;

screening out a target operation variable sample set which meets a preset requirement from the operation variable sample set based on a random forest algorithm;

establishing a gasoline octane number loss prediction model, and training the gasoline octane number loss prediction model according to the target operation variable sample set and the production factor sample set to obtain a target gasoline octane number loss prediction model with complete training;

and setting constraint conditions, and determining the optimization conditions of the target operating variables meeting the constraint conditions based on the well-trained target gasoline octane number loss prediction model.

2. The gasoline octane number loss optimization method according to claim 1, wherein the screening out a target operational variable sample set from the operational variable sample set based on a random forest algorithm, which meets preset requirements, comprises:

preprocessing the gasoline refining data sample set to obtain a reliable gasoline refining data sample set;

constructing a plurality of decision trees, and calculating the error of the reliable gasoline refining data sample set of each decision tree in an OOB (object-oriented object) mode to obtain an OOB error;

reordering the operation variables, and calculating the error of the operation variables again to obtain a reordering error;

calculating the importance of each operation variable according to the OOB error and the reordering error;

and screening out the operation variables meeting the preset requirements according to the importance of the operation variables to obtain a target operation variable sample set.

3. The method of claim 1, wherein the establishing a gasoline octane number loss prediction model, and training the gasoline octane number loss prediction model according to the target operating variable sample set and the production factor sample set to obtain a well-trained target gasoline octane number loss prediction model comprises:

carrying out Bayesian normalization on the target operation variable sample set and the production factor sample set;

establishing a gasoline octane number loss prediction model, and setting parameters of the gasoline octane number loss prediction model;

and inputting the normalized target operation variable sample set and the normalized production factor sample set into the gasoline octane number loss prediction model, and training the gasoline octane number loss prediction model by taking octane number loss as output to obtain a target gasoline octane number loss prediction model with complete training.

4. The method of claim 3, wherein the establishing a gasoline octane number loss prediction model, setting parameters of the gasoline octane number loss prediction model, comprises:

setting an excitation function, a network training function and a network performance function type of the gasoline octane number loss prediction model;

setting the neuron number of the hidden layer of the gasoline octane number loss prediction model;

and setting network parameters of the gasoline octane number loss prediction model.

5. The gasoline octane number loss optimization method of claim 3, wherein the setting of the constraint condition and the determination of the optimization condition of the target operating variable satisfying the constraint condition based on the well-trained target gasoline octane number loss prediction model comprise:

establishing a relevant neural network model and setting constraint conditions;

establishing a functional relation f (net1) between octane number loss and the operation variable and a functional relation f (net2) between gasoline sulfur content and the operation variable according to the associated neural network model;

and determining the optimization condition of the optimal operating variable meeting the constraint condition according to the functional relation f (net1) and the functional relation f (net 2).

6. The method for optimizing gasoline octane number loss according to claim 5, wherein said determining, according to said functional relationship f (net1) and said functional relationship f (net2), the optimization conditions of the target operating variables that satisfy said constraints comprises:

determining the octane number loss amount and the sulfur content corresponding to each operation variable according to the functional relation f (net1) and the functional relation f (net 2);

according to the octane value loss and the sulfur content corresponding to the operation variables, recording optimization conditions corresponding to target operation variables;

and repeating the operation until the optimization conditions corresponding to all the target operation variables are determined.

7. The method of claim 2, wherein the preprocessing the sample set of gasoline refinery data to obtain a sample set of reliable gasoline refinery data comprises:

judging whether incomplete data exist in the gasoline refining data sample set or not;

deleting the incomplete data reaching the incomplete threshold value, and replacing the incomplete data not reaching the incomplete threshold value;

judging whether abnormal data exist in the gasoline refining data sample set or not;

and according to a preset criterion, removing the abnormal data to obtain a reliable gasoline refining data sample set.

8. A gasoline octane number loss optimization device, comprising:

the system comprises a data acquisition module, a data processing module and a data processing module, wherein the data acquisition module is used for acquiring a gasoline refining data sample set, and the gasoline refining data sample set comprises an operation variable sample set and a production factor sample set;

the screening module is used for screening out a target operation variable sample set meeting preset requirements from the operation variable sample set based on a random forest algorithm;

the modeling module is used for establishing a gasoline octane number loss prediction model, and training the gasoline octane number loss prediction model according to the target operating variable sample set and the production factor sample set to obtain a target gasoline octane number loss prediction model with complete training;

and the optimization module is used for setting constraint conditions and determining the optimization conditions of the target operating variables meeting the constraint conditions based on the target gasoline octane number loss prediction model which is trained completely.

9. An electronic device comprising a memory and a processor, wherein,

the memory is used for storing programs;

the processor, coupled to the memory, is configured to execute the program stored in the memory to implement the steps in the method for optimizing gasoline octane number loss according to any one of the preceding claims 1 to 7.

10. A computer-readable storage medium storing a computer-readable program or instructions, which when executed by a processor, is capable of performing the steps of the method for optimizing the octane number loss of gasoline according to any one of claims 1 to 7.

Images