CN112035286A - Method and device for determining cause of failure, storage medium, and electronic device - Google Patents

Abstract

The present invention provides a method and device, a storage medium and an electronic device for determining the cause of a failure, wherein the method includes: inputting the characteristic attributes collected when the program fails into a classification network model, so as to obtain the characteristic attributes corresponding to different Multiple probability values of categories, wherein the classification network model is trained by using multiple sets of data through machine learning, and each set of data in the multiple sets of data includes: feature attributes, and different Multiple probability values of the category; the target category corresponding to the maximum probability value among the determined multiple probability values is used as the failure cause of the program failure, and the above technical solution is used to solve the problem of the program failure in the related art. The response is slow, and the failure cause of the program failure cannot be found out in time.

Description

Method and device for determining cause of failure, storage medium, and electronic device

technical field

The present invention relates to the field of communications, and in particular, to a method and device for determining a cause of a failure, a storage medium, and an electronic device.

Background technique

For a program, the most important thing is the high availability of the program. The service brought by the program is directly facing the user. If the program is unavailable, it will have a bad impact on the user experience. In the current technology, the abnormality of the program can often be discovered in time when the program is abnormal, for example, the server resource is unavailable, or the table is locked. At this time, the fault has occurred. And the current solutions for program problems, most of which rely on manual inspection of error logs to troubleshoot the problem. Sometimes the same error log has different causes, and it needs to be analyzed according to the scene to find out. The correct reason for the error, it may take a lot of time in the middle, and the service of the program is unavailable during this time, which will greatly affect the customer experience.

In the related art, an effective technical solution has not been proposed for problems such as slow response when a program failure occurs, and failure to find out the failure cause of the program failure in time.

SUMMARY OF THE INVENTION

Embodiments of the present invention provide a method and device, a storage medium, and an electronic device for determining the cause of a failure, so as to at least solve the problems in the related art that the response to a program failure is slow and the failure cause of the program failure cannot be found in time.

According to an embodiment of the present invention, a method for determining the cause of a fault is provided, which includes: inputting a characteristic attribute collected when a program fails into a classification network model, so as to obtain multiple probability values corresponding to different categories of the characteristic attribute , wherein the classification network model is trained by using multiple sets of data through machine learning, and each set of data in the multiple sets of data includes: feature attributes, and multiple probability values of different categories corresponding to the feature attributes ; Use the target category corresponding to the largest probability value among the determined plurality of probability values as the failure cause of the program failure.

Optionally, before inputting the feature attributes collected when the program fails into the classification network model to obtain multiple probability values corresponding to different categories of the feature attributes, the above method further includes: configuring a class library for the classification network model. , wherein the category library includes multiple categories corresponding to feature attributes; the classification network is classified according to the feature attributes corresponding to the program failure and a plurality of categories selected from the category library corresponding to the program failure The model is trained.

Optionally, inputting the characteristic attributes collected when the program fails into the classification network model to obtain multiple probability values corresponding to different categories of the characteristic attributes, including: a determining step, including: determining each classification network model according to the classification network model. The first probability, the second probability, and the third probability of a category, where the first probability is the probability that the feature attribute exists under the condition that the target category occurs, and the second probability is the target category The probability of occurrence, the third probability is the probability of the existence of the characteristic attribute; determine the product of the first probability and the second probability, and use the ratio of the product and the third probability as the probability of the target category value; the above determination steps are performed cyclically to determine a plurality of probability values of the different categories.

Optionally, after using the target category corresponding to the largest probability value among the determined plurality of probability values as the failure cause of the program failure, the above method further includes: selecting the failure cause according to the failure processing record of the program. Corresponding fault handling records; process the program faults according to the fault handling records.

Optionally, processing the program fault according to the fault processing record includes: in the case that the program fault is not successfully processed according to the fault processing record, processing the first probability value among the plurality of probability values. The target category corresponding to the two major probability values is used as the failure cause of the program failure.

Optionally, in the case that the program failure is successfully processed according to the failure handling record, the above method further includes: correspondingly saving the failure handling record and the program cause in the target device running the program.

According to an embodiment of the present invention, a device for determining a cause of a fault is provided, including: a first processing module, configured to input the characteristic attribute collected when the program fails into a classification network model, so as to obtain the corresponding characteristic attribute Multiple probability values of different categories, wherein the classification network model is trained by using multiple sets of data through machine learning, and each set of data in the multiple sets of data includes: feature attributes, and the corresponding feature attributes. multiple probability values of different categories; the determining module uses the target category corresponding to the maximum probability value among the determined multiple probability values as the failure cause of the program failure.

Optionally, the above-mentioned apparatus further includes: a configuration module configured to configure a class library for the classification network model, wherein the class library includes a plurality of categories corresponding to characteristic attributes; according to the characteristic attributes corresponding to the program failure, and The classification network model is trained from a plurality of classes selected from the class library corresponding to the program failure.

Optionally, the above-mentioned first processing module is further configured to perform a determining step, including: determining a first probability, a second probability, and a third probability of each category according to the classification network model, wherein the first probability is the probability that the feature attribute exists under the condition that the target category occurs, the second probability is the probability that the target category occurs, and the third probability is the probability that the feature attribute exists; determine the first probability The product of the probability and the second probability is used, and the ratio of the product and the third probability is used as the probability value of the target category; the above determination steps are performed cyclically to determine multiple probability values of the different categories.

Optionally, the above-mentioned device further includes: a selection module for selecting a fault processing record corresponding to the fault cause according to the fault processing record of the program; a second processing module for selecting the program fault according to the fault processing record. to be processed.

Optionally, the above-mentioned second processing module is further configured to, in the case that the program fault is not successfully processed according to the fault processing record, the corresponding value corresponding to the second largest probability value in the plurality of probability values. The target class serves as the failure cause for the program failure.

Optionally, the above-mentioned second processing module is further configured to store the fault processing record and the program cause correspondingly in the target of the running program under the condition that the program fault is successfully processed according to the fault processing record. in the device.

According to another embodiment of the present invention, a storage medium is also provided, and a computer program is stored in the storage medium, wherein the computer program is configured to execute the steps in any one of the above method embodiments when running.

According to yet another embodiment of the present invention, there is also provided an electronic device comprising a memory and a processor, wherein the memory stores a computer program, the processor is configured to run the computer program to execute any of the above Steps in Method Examples.

Through the present invention, the feature attributes collected when the program fails are input into the classification network model, so as to obtain multiple probability values corresponding to different categories of the feature attributes, wherein the classification network model uses multiple sets of data through machine learning After training, each set of data in the multiple sets of data includes: feature attributes, and multiple probability values of different categories corresponding to the feature attributes; The corresponding target category is used as the failure cause of the program failure, that is, the probability value result output by the classified network model reduces the processing time of the program failure, and the above technical solution solves the problem that the response to the program failure is slow in the related art. It is impossible to find out the fault cause of the program failure in time. Through the classification network model, the failure problem can be predicted for the possible program failure, and the failure problem can be quickly responded and confirmed, avoiding the long-term analysis of the program failure cause. , which improves the processing efficiency for program failures.

Description of drawings

The accompanying drawings described herein are used to provide a further understanding of the present invention and constitute a part of the present application. The exemplary embodiments of the present invention and their descriptions are used to explain the present invention and do not constitute an improper limitation of the present invention. In the attached image:

Fig. 1 is a hardware structure block diagram of a computer terminal of a method for determining a fault cause according to an embodiment of the present invention;

2 is a flowchart of a method for determining a cause of a failure according to an embodiment of the present invention;

3 is a process flow diagram of a classification network model for program fault handling and prediction based on Naive Bayes classification according to an optional embodiment of the present invention;

4 is a structural block diagram (1) of a device for determining a cause of a fault according to an embodiment of the present invention;

FIG. 5 is a structural block diagram (2) of an apparatus for determining a cause of a failure according to an embodiment of the present invention.

Detailed ways

Hereinafter, the present invention will be described in detail with reference to the accompanying drawings and in conjunction with embodiments. It should be noted that the embodiments in the present application and the features of the embodiments may be combined with each other in the case of no conflict.

It should be noted that the terms "first", "second" and the like in the description and claims of the present invention and the above drawings are used to distinguish similar objects, and are not necessarily used to describe a specific sequence or sequence.

The methods provided by the embodiments of the present application may be executed in a mobile terminal, a computer terminal, or a similar computing device. Taking running on a computer terminal as an example, FIG. 1 is a hardware structural block diagram of a computer terminal according to a method for determining a cause of a fault according to an embodiment of the present invention. As shown in FIG. 1 , the computer terminal may include one or more (only one is shown in FIG. 1 ) processor 102 (the processor 102 may include, but is not limited to, a processing device such as a microprocessor MCU or a programmable logic device FPGA, etc.) and a memory 104 for storing data, optionally, the above-mentioned computer terminal may further include a transmission device 106 and an input and output device 108 for communication functions. Those of ordinary skill in the art can understand that the structure shown in FIG. 1 is only a schematic diagram, which does not limit the structure of the above-mentioned computer terminal.

For example, the computer terminal may also include more or fewer components than those shown in FIG. 1 , or have a different configuration with equivalent or more functions than those shown in FIG. 1 . The memory 104 can be used to store computer programs, for example, software programs and modules of application software, such as the computer programs corresponding to the method for determining the cause of the failure in the embodiment of the present invention. Executing various functional applications and data processing implements the above-mentioned methods. Memory 104 may include high-speed random access memory, and may also include non-volatile memory, such as one or more magnetic storage devices, flash memory, or other non-volatile solid-state memory.

In some instances, memory 104 may further include memory located remotely from processor 102, which may be connected to a computer terminal through a network. Examples of such networks include, but are not limited to, the Internet, an intranet, a local area network, a mobile communication network, and combinations thereof. Transmission means 106 are used to receive or transmit data via a network. The specific example of the above-mentioned network may include a wireless network provided by the communication provider of the computer terminal. In one example, the transmission device 106 includes a network adapter (Network Interface Controller, NIC for short), which can be connected to other network devices through a base station so as to communicate with the Internet. In one example, the transmission device 106 may be a radio frequency (Radio Frequency, RF for short) module, which is used to communicate with the Internet in a wireless manner.

An embodiment of the present invention provides a method for determining a cause of a failure, which is applied to the above-mentioned computer terminal. FIG. 2 is a flowchart of a method for determining a cause of a failure according to an embodiment of the present invention. As shown in FIG. 2 , the process includes the following steps :

Step S202, input the feature attributes collected when the program fails into the classification network model, so as to obtain multiple probability values corresponding to different categories of the feature attributes, wherein the classification network model is trained by using multiple sets of data through machine learning , each set of data in the multiple sets of data includes: feature attributes, and multiple probability values of different categories corresponding to the feature attributes;

Step S204, taking the target category corresponding to the largest probability value among the determined multiple probability values as the failure cause of the program failure.

Through the above steps, the feature attributes collected when the program fails are input into the classification network model, so as to obtain multiple probability values corresponding to different categories of the feature attributes, wherein the classification network model uses multiple sets of data through machine learning. After training, each set of data in the multiple sets of data includes: feature attributes, and multiple probability values of different categories corresponding to the feature attributes; The corresponding target category is used as the failure cause of the program failure, that is, the probability value result output by the classified network model reduces the processing time of the program failure, and the above technical solution solves the problem that the response to the program failure is slow in the related art. It is impossible to find out the fault cause of the program failure in time. Through the classification network model, the failure problem can be predicted for the possible program failure, and the failure problem can be quickly responded and confirmed, avoiding the long-term analysis of the program failure cause. , which improves the processing efficiency for program failures.

It should be noted that, the above-mentioned feature attributes input into the classification network model may be one or multiple at the same time, which is determined according to the training process of the classification network model, and can be flexibly transformed. Don't be too restrictive.

In step S202, there are multiple implementations for the training method of the classification network model. Optionally, a category library is configured for the classification network model, wherein the category library includes multiple categories corresponding to the feature attributes; according to the program failure The classification network model is trained with the corresponding feature attributes at the time, and a plurality of categories corresponding to the program failure selected from the category library.

That is to say, in order to facilitate the identification of multiple categories through the classification network model, and to achieve category distinction for different program faults, when training the classification network model, configure a class library for the classification network model, and then from the category library Obtain multiple categories corresponding to program failures to train the classification network model.

Optionally, inputting the characteristic attributes collected when the program fails into the classification network model to obtain multiple probability values corresponding to different categories of the characteristic attributes, including: a determining step, including: determining each classification network model according to the classification network model. The first probability, the second probability, and the third probability of a category, where the first probability is the probability that the feature attribute exists under the condition that the target category occurs, and the second probability is the target category The probability of occurrence, the third probability is the probability of the existence of the characteristic attribute; determine the product of the first probability and the second probability, and use the ratio of the product and the third probability as the probability of the target category value; the above determination steps are performed cyclically to determine a plurality of probability values of the different categories.

In short, in order to make the determined multiple probability values more accurate, when obtaining multiple probability values of feature attributes corresponding to different categories through the classification network model, it is necessary to determine through the classification network model that under the condition that the target category occurs, The first probability of the existence of characteristic attributes; the second probability of the occurrence of the target category when the program fails; the third independent probability of the existence of multiple different characteristic attributes, and further combining the conditional probability and the Bayesian formula, the first probability is calculated as The second probability is multiplied by the third probability to perform a proportional operation, and then the ratio result of the proportional operation is used as the probability value of the target category, so that the above proportional operation is performed cyclically to determine the probability values corresponding to multiple different categories one-to-one.

Optionally, after using the target category corresponding to the largest probability value among the determined plurality of probability values as the failure cause of the program failure, the above method further includes: selecting the failure cause according to the failure processing record of the program. Corresponding fault handling records; process the program faults according to the fault handling records.

That is to say, in order to improve the processing efficiency of the program failure that determines the cause of the failure, after completing the target category confirmation of the program failure through the classification model, select from the target devices that run the program before the program failure, the failure cause and the current program failure The fault processing records of the same or similar cause of the target category of the target category, the program fault of the currently confirmed target category is processed, thereby speeding up the processing response to the program fault.

Optionally, processing the program fault according to the fault processing record includes: in the case that the program fault is not successfully processed according to the fault processing record, processing the first probability value among the plurality of probability values. The target category corresponding to the two major probability values is used as the failure cause of the program failure.

Due to the diversity of program failure types, the selected failure handling records may not be able to successfully handle program failures. The target category corresponding to the high probability value is used as the fault cause of the program fault, and then the fault processing record corresponding to the second largest probability value is reselected in the fault processing record to process the program fault.

Optionally, in the case that the program failure is successfully processed according to the failure processing record, the above method further includes: correspondingly saving the failure processing record and the program failure cause in the target device running the program.

That is to say, after the program fault processing is successfully completed by using the fault processing record, in order to quickly handle the same program fault next time, it is necessary to store the fault processing record and the program fault cause in the target device running the program correspondingly.

In order to better understand the process of determining the cause of the above-mentioned fault, the following description is given in conjunction with optional embodiments, but is not used to limit the technical solutions of the embodiments of the present invention.

The technical background of the optional embodiment of the present invention is based on the Naive Bayes algorithm. Naive Bayes classification is the simplest and common classification method in Bayesian classification. From a mathematical point of view, the classification problem can be done by Defined as follows: Given the sets C=y ₁ , y ₂ ,....,y _n and I=x ₁ ,x ₂ ,....,x _n determine the mapping rule y=f(x) such that any x ∈I has one and only one y∈C, such that y∈f(x)I holds, where C is called a category set, where each element is a category, and I is called an item set (feature set), where each element is An item to be classified, f is called a classifier. The task of a classification algorithm is to construct a classifier f.

The errors that occur during the running of the program and the on-site situation (equivalent to the program failure in the embodiment of the present invention) can be used as a characteristic attribute of the program, and the different errors that occur each time are regarded as a characteristic set, for example, each time the program reports The error log, the time when the error occurs, the resource information of the server when the error occurs, the QoS of the program (Quality of Service, quality of service, QoS for short), etc., can be used as a feature attribute to form a feature set.

The specific reason for each occurrence of an error in the program can be used as a category (equivalent to the target category in the embodiment of the present invention), for example, insufficient server resources, excessive concurrency, etc., can be used as a category set.

In order to facilitate the understanding of the classification network model in the embodiment of the present invention, the working process is now explained as follows:

(1) Let D be a set of training tuples (equivalent to feature attributes in the embodiment of the present invention) and their associated class labels (equivalent to categories in the embodiment of the present invention). Each tuple is represented by an n-dimensional attribute vector X={x _1, x ₂ ,...,x _n }.

(2) Assuming there are m class labels C _{1 ,} C ₂ ,...C _m , given a tuple X, the classification method will predict that X belongs to the class with the highest posterior probability. That is, naive Bayesian classification predicts that X belongs to class C _i if and only if P(C _i |X)>P(C _j |X), 1≤j≤m, j≠i; thus, P The class C ₁ with the largest (C _i |X) is called the maximum posterior probability. According to Bayes' theorem:

其中,式中P(X|C_i)为在模式属于C_i类的条件下出现X的概率密度,称为X的类条件概率密度；P(C_i)表示在所研究的识别问题中出现C_i类的概率,又称为先验概率；P(X)是特征向量X的概率密度。

where P(X|C _i ) is the probability density of X appearing under the condition that the pattern belongs to class C _{i ,} which is called the class conditional probability density of X; The probability of class C _i , also known as prior probability; P(X) is the probability density of feature vector X.

(3) Since P(X) is a constant for all classes, it is only necessary that P(C _i |X)P(C _i ) is the largest. If the prior probabilities of the classes are unknown, the classes are usually assumed to be equally probable, ie P(C ₁ )=P(C ₂ )=...=P(C _m ), and P(C _i | X) maximizes, otherwise maximizes P(C _i |X)P(C _i );

考察该属性是分类的还是连续值的，例如，为了计算P(X|C_i)，考虑如下两种情况：(4) Given a dataset with many attributes, the cost of computing P(C _i |X) is very large. In order to reduce the computational cost, a naive assumption of class condition independence can be made. Given a tuple of class labels, attribute values are assumed to be conditionally independent of each other. Therefore, get

Consider whether the attribute is categorical or continuous-valued, for example, to compute P(X|C _i ), consider the following two cases:

(a) If A _k is a categorical attribute, then P(x _k |C _{i )} is the number of tuples of class C _i with the value of attribute A _k in D divided by the number of tuples of class C _i in D |C _i,D |

(b) If A _k is a continuous-valued attribute, it is assumed that the continuous-valued attribute obeys a Gaussian distribution with mean η and standard deviation σ, which is defined by the following formula:

即P(x_k|C_i)＝g(x_k,η_ci,σ_ci)；

That is, P(x _k |C _i )=g(x _k ,η _ci ,σ _ci );

(5) In order to predict the class label of X, for each class C _i , calculate P(C _i |X)P(C _i ). This taxonomy predicts that the class of the input tuple X is C _i if and only if, P(X|C _i )P(C _i )>P(X|C _j )P(C _j ), 1≤j≤m , j≠i. That is, the predicted class label is the class C _i that maximizes P(X|C _i )P(C _i ).

An optional embodiment of the present invention proposes a classification network model for program fault processing and prediction based on Naive Bayes classification. As shown in Figure 3, the processing flow of the network model is as follows:

Step S302: In the preparation stage, determine the characteristic attributes, for example, the characteristic attributes such as "error log, access time, QoS of the program" that are common in the program, and at the same time clarify what the predicted value is, and divide each characteristic attribute appropriately, and then use Manually classify a part of the data to form training samples. This stage is the only stage in the entire Naive Bayes classification that needs to be done manually. Its quality will have an important impact on the entire process. The quality of the classifier is largely determined by the feature attributes. , feature attribute division and training sample quality determination.

Step S304: In the training stage, input feature attributes and training samples to obtain classifiers with different outputs; this stage is to generate classifiers, and the main job is to calculate the frequency of occurrence of each category in the training samples and the division of each feature attribute for each. The conditional probability P(y _i ) of the class.

Step S306: In the application stage, the maximum item of P(x|y _i )P(y _i ) is used as the category to which x belongs, and the value of P(x|y _i )P(y _i ) is calculated for each category. The classifier classifies the new data, the input is the classifier and the new data, and the output is the classification result of the new data.

It should be noted that this phenomenon occurs when P(x _k |C _i )=0, that is, when a certain feature attribute item under a certain category does not appear, then the situation will be: although there is no such zero probability, it is still possible to get a high probability that X _belongs to class Ci. At this time, the occurrence of this phenomenon can be avoided by Laplace calibration or Laplace estimation. Specifically, it is assumed that the training database D is large, so that the change in the estimated probability caused by adding 1 to each count can be Negligible, thus conveniently avoiding a probability value of 0.

In order to better understand the above-mentioned classification network model for program failure processing and prediction based on Naive Bayes classification, an actual case is used to illustrate the following.

After the classification network model has been trained, a new set of data is processed, assuming that the new set of data has two categories: C ₁ : insufficient server resources; C ₂ : program deadlock; the characteristic attributes of this set of new data are: A1 _: Error log, A2 _: Access time, A3: QoS _of the program.

由于该组新数据存在两种类别，因此，需要求得P(C₁|A₁A₂A₃)和P(C₂|A₁A₂A₃)的概率，进而通过比较C₁、C₂哪个分类的可能性大，从而可以得出该组新数据的分类结果。也就是说，该组新数据的分类结果等价于P(A₁A₂A₃|C_j)P(C_j)最大值的求取。When it is necessary to determine the probability of category C _j under the premise of the existence of three characteristic attributes such as A ₁ , A ₂ , and A ₃ , the conditional probability is expressed as P(C _j |A ₁ A ₂ A ₃ ), combined with the above Bayesian formula can be obtained:

Since there are two categories of new data in this group, it is necessary to obtain the probabilities of P(C ₁ |A ₁ A ₂ A ₃ ) and P(C ₂ |A ₁ A ₂ A ₃ ), and then by comparing C ₁ and C ₂ Which classification is more likely, so that the classification results of this new set of data can be obtained. That is to say, the classification result of this group of new data is equivalent to finding the maximum value of P(A ₁ A ₂ A ₃ |C _j )P(C _j ).

Further, let A _i be independent of each other, then: P(A ₁ A ₂ A ₃ |C _j )=P(A ₁ |C _j )P(A ₂ |C _j )P(A ₃ |C _j );

P(A ₁ |C ₁ )=Y ₁ ; P(A ₂ |C ₁ )=Y ₂ ; P(A ₃ |C ₁ )=Y ₃ ;

P(A ₁ |C ₂ )=Y ₃ ; P(A ₂ |C ₂ )=Y ₄ ; P(A ₃ |C ₅ )=Y ₆ ;

P(A ₁ A ₂ A ₃ |C ₁ )=Y ₁ Y ₂ Y ₃ ;

P(A ₁ A ₂ A ₃ |C ₂ )=Y ₄ Y ₅ Y ₆ ;

If Y ₁ Y ₂ Y ₃ > Y ₄ Y ₅ Y ₆ , it means that the category of the new data after processing by the classification network model is C _{1 with insufficient server resources,} if Y ₁ Y ₂ Y ₃ <Y ₄ Y ₅ Y ₆ , it means that this group of new data is classified as C ₂ of program deadlock after being processed by the classification network model.

The optional embodiment of the present invention solves the problems in the related art, such as the slow response to the program failure and the inability to find out the failure cause of the program failure in time. Predicting failure problems for program failures, and quickly responding to and confirming failure problems, avoiding long-term analysis of the causes of program failures, and improving the processing efficiency of program failures, and the classification provided by optional embodiments of the present invention The network model structure makes it possible to quickly respond to sudden program problems, find the point of the problem in time, respond to the treatment, and make predictions and feedback corresponding processing methods for possible problems in time, thereby improving the developer's program. The efficiency of abnormal problem processing and the usability of program code are improved, and there is no need to manually extract feature attributes.

From the description of the above embodiments, those skilled in the art can clearly understand that the method according to the above embodiment can be implemented by means of software plus a necessary general hardware platform, and of course can also be implemented by hardware, but in many cases the former is better implementation. Based on this understanding, the technical solutions of the present invention can be embodied in the form of software products in essence or the parts that make contributions to the prior art, and the computer software products are stored in a storage medium (such as ROM/RAM, magnetic disk, CD-ROM), including several instructions to make a terminal device (which may be a mobile phone, a computer, a server, or a network device, etc.) to execute the methods of the various embodiments of the present invention.

In this embodiment, an apparatus for determining a fault cause is also provided, and the apparatus is used to implement the above-mentioned embodiments and preferred implementations, and what has been described will not be repeated. As used below, the term "module" may be a combination of software and/or hardware that implements a predetermined function. Although the apparatus described in the following embodiments is preferably implemented in software, implementations in hardware, or a combination of software and hardware, are also possible and contemplated.

FIG. 4 is a structural block diagram of an apparatus for determining a cause of a fault according to an embodiment of the present invention. As shown in FIG. 4 , the apparatus includes:

(1) The first processing module 42 is used to input the characteristic attributes collected when the program fails into the classification network model, so as to obtain multiple probability values corresponding to different categories of the characteristic attributes, wherein the classification network model is Trained by using multiple sets of data through machine learning, each set of data in the multiple sets of data includes: feature attributes, and multiple probability values of different categories corresponding to the feature attributes;

(2) The determination module 44 takes the target category corresponding to the largest probability value among the determined plurality of probability values as the failure cause of the program failure.

Through the above device, the feature attributes collected when the program fails are input into the classification network model, so as to obtain multiple probability values corresponding to different categories of the feature attributes, wherein the classification network model uses multiple sets of data through machine learning. After training, each set of data in the multiple sets of data includes: feature attributes, and multiple probability values of different categories corresponding to the feature attributes; The corresponding target category is used as the failure cause of the program failure, that is, the probability value result output by the classified network model reduces the processing time of the program failure, and the above technical solution solves the problem that the response to the program failure is slow in the related art. It is impossible to find out the fault cause of the program failure in time. Through the classification network model, the failure problem can be predicted for the possible program failure, and the failure problem can be quickly responded and confirmed, avoiding the long-term analysis of the program failure cause. , which improves the processing efficiency for program failures.

It should be noted that, the above-mentioned feature attributes input into the classification network model may be one or multiple at the same time, which is determined according to the training process of the classification network model, and can be flexibly transformed. Don't be too restrictive.

FIG. 5 is another device for determining the cause of a fault according to an embodiment of the present invention. As shown in FIG. 5 , the device includes all the modules shown in FIG. 4 , and further includes:

Optionally, the above-mentioned apparatus further includes: a configuration module 40, configured to configure a class library for the classification network model, wherein the class library includes a plurality of categories corresponding to characteristic attributes; according to the characteristic attributes corresponding to the program failure, and training the classification network model with a plurality of categories selected from the category library and corresponding to the program failure.

That is to say, in order to facilitate the identification of multiple categories through the classification network model, and to achieve category distinction for different program faults, when training the classification network model, configure a class library for the classification network model, and then from the category library Obtain multiple categories corresponding to program failures to train the classification network model.

Optionally, the above-mentioned first processing module 42 is further configured to perform a determining step, including: determining a first probability, a second probability, and a third probability of each category according to the classification network model, wherein the first probability The probability is the probability that the characteristic attribute exists under the condition that the target category occurs, the second probability is the probability that the target category occurs, and the third probability is the probability that the characteristic attribute exists; A product of a probability and a second probability, and the ratio of the product and the third probability is used as the probability value of the target category; the above determination steps are performed cyclically to determine multiple probability values of the different categories.

In short, in order to make the determined multiple probability values more accurate, when obtaining multiple probability values of feature attributes corresponding to different categories through the classification network model, it is necessary to determine through the classification network model that under the condition that the target category occurs, The first probability of the existence of characteristic attributes; the second probability of the occurrence of the target category when the program fails; the third independent probability of the existence of multiple different characteristic attributes, and further combining the conditional probability and the Bayesian formula, the first probability is calculated as The second probability is multiplied by the third probability to perform a proportional operation, and then the ratio result of the proportional operation is used as the probability value of the target category, so that the above proportional operation is performed cyclically to determine the probability values corresponding to multiple different categories one-to-one.

Optionally, the above-mentioned device further includes: a selection module 46 for selecting a fault processing record corresponding to the fault cause according to the fault processing record of the program; a second processing module for selecting the program according to the fault processing record. Troubleshoot.

That is to say, in order to improve the processing efficiency of the program failure that determines the cause of the failure, after completing the target category confirmation of the program failure through the classification model, select from the target devices that run the program before the program failure, the failure cause and the current program failure The fault processing records of the same or similar cause of the target category of the target category, the program fault of the currently confirmed target category is processed, thereby speeding up the processing response to the program fault.

Optionally, the above-mentioned second processing module 48 is further configured to, in the case that the program fault is not successfully processed according to the fault processing record, the corresponding second largest probability value in the plurality of probability values. target class as the failure cause of the program failure.

Due to the diversity of program failure types, the selected failure handling records may not be able to successfully handle program failures. The target category corresponding to the high probability value is used as the fault cause of the program fault, and then the fault processing record corresponding to the second largest probability value is reselected in the fault processing record to process the program fault.

Optionally, the above-mentioned second processing module 48 is further configured to store the fault processing record and the program cause correspondingly in the running program under the condition that the program fault is successfully processed according to the fault processing record. in the target device.

That is to say, after the program fault processing is successfully completed by using the fault processing record, in order to quickly handle the same program fault next time, it is necessary to store the fault processing record and the program fault cause in the target device running the program correspondingly.

It should be noted that the above modules can be implemented by software or hardware, and the latter can be implemented in the following ways, but not limited to this: the above modules are all located in the same processor; or, the above modules can be combined in any combination The forms are located in different processors.

An embodiment of the present invention further provides a storage medium, where a computer program is stored in the storage medium, wherein the computer program is configured to execute the steps in any one of the above method embodiments when running.

In an exemplary embodiment, in this embodiment, the above-mentioned storage medium may be configured to store a computer program for performing the following steps:

S1, input the feature attributes collected when the program fails into a classification network model, so as to obtain multiple probability values corresponding to different categories of the feature attributes, wherein the classification network model is obtained by using multiple sets of data through machine learning training Each group of data in the multiple sets of data includes: feature attributes, and multiple probability values of different categories corresponding to the feature attributes;

S2. The target category corresponding to the largest probability value among the determined multiple probability values is used as the failure cause of the program failure.

An embodiment of the present invention further provides a storage medium, where the storage medium includes a stored program, wherein the above-mentioned program executes any one of the above-mentioned methods when running.

An embodiment of the present invention also provides an electronic device, comprising a memory and a processor, where a computer program is stored in the memory, and the processor is configured to run the computer program to execute the steps in any of the above method embodiments.

Optionally, the above-mentioned electronic device may further include a transmission device and an input-output device, wherein the transmission device is connected to the above-mentioned processor, and the input-output device is connected to the above-mentioned processor.

Optionally, in this embodiment, the above-mentioned processor may be configured to execute the following steps through a computer program:

S1, input the feature attributes collected when the program fails into a classification network model, so as to obtain multiple probability values corresponding to different categories of the feature attributes, wherein the classification network model is obtained by using multiple sets of data through machine learning training Each group of data in the multiple sets of data includes: feature attributes, and multiple probability values of different categories corresponding to the feature attributes;

S2. The target category corresponding to the largest probability value among the determined multiple probability values is used as the failure cause of the program failure.

Optionally, in this embodiment, the above-mentioned storage medium may include but is not limited to: a USB flash drive, a read-only memory (Read-Only Memory, referred to as ROM), a random access memory (Random Access Memory, referred to as RAM), Various media that can store program codes, such as removable hard disks, magnetic disks, or optical disks.

Optionally, for specific examples in this embodiment, reference may be made to the examples described in the foregoing embodiments and optional implementation manners, and details are not described herein again in this embodiment.

Obviously, those skilled in the art should understand that the above-mentioned modules or steps of the present invention can be implemented by a general-purpose computing device, which can be centralized on a single computing device, or distributed in a network composed of multiple computing devices Alternatively, they may be implemented in program code executable by a computing device, such that they may be stored in a storage device and executed by the computing device, and in some cases, in a different order than here The steps shown or described are performed either by fabricating them separately into individual integrated circuit modules, or by fabricating multiple modules or steps of them into a single integrated circuit module. As such, the present invention is not limited to any particular combination of hardware and software.

The above descriptions are only preferred embodiments of the present invention, and are not intended to limit the present invention. For those skilled in the art, the present invention may have various modifications and changes. Any modification, equivalent replacement, improvement, etc. made within the principle of the present invention shall be included within the protection scope of the present invention.

Claims (10)

1. a method for determining a cause of failure, characterized in that, comprising: The feature attributes collected when the program fails are input into the classification network model to obtain multiple probability values corresponding to different categories of the feature attributes, wherein the classification network model is trained by using multiple sets of data through machine learning, Each of the multiple sets of data includes: feature attributes, and multiple probability values of different categories corresponding to the feature attributes; The target category corresponding to the largest probability value among the determined plurality of probability values is used as the failure cause of the program failure. 2. The method according to claim 1, characterized in that, before inputting the characteristic attributes collected when the program fails into a classification network model to obtain multiple probability values corresponding to different categories of the characteristic attributes, the method Also includes: Configuring a category library for the classification network model, wherein the category library includes multiple categories corresponding to feature attributes; The classification network model is trained according to the characteristic attributes corresponding to the program failure and a plurality of categories corresponding to the program failure selected from the category library. 3. The method according to claim 1, wherein the characteristic attributes collected when the program fails are input into the classification network model, so as to obtain a plurality of probability values corresponding to different categories of the characteristic attributes, including: The determining step includes: determining a first probability, a second probability, and a third probability of each category according to the classification network model, wherein the first probability is that the feature attribute exists under the condition that the target category occurs probability, the second probability is the probability of occurrence of the target category, and the third probability is the probability of the existence of the characteristic attribute; determine the product of the first probability and the second probability, and combine the product with the The ratio of the third probability is used as the probability value of the target category; The above determination steps are performed cyclically to determine a plurality of probability values for the different categories. 4 . The method according to claim 1 , wherein after taking the target category corresponding to the determined maximum probability value among the plurality of probability values as the failure cause of the program failure, the method further comprises: 5 . Select the fault processing record corresponding to the fault cause according to the fault processing record of the program; The program fault is processed according to the fault processing record. 5. The method according to claim 4, wherein processing the program fault according to the fault processing record comprises: In the case that the program failure is not successfully processed according to the failure processing record, the target category corresponding to the second largest probability value among the plurality of probability values is used as the failure cause of the program failure. 6. The method according to claim 5, characterized in that, in the case that the program fault is successfully processed according to the fault processing record, the method further comprises: The fault handling record and the program failure cause are correspondingly stored in the target device running the program. 7. A device for determining a cause of failure, characterized in that, comprising: The first processing module is used to input the characteristic attributes collected when the program fails into the classification network model, so as to obtain multiple probability values corresponding to different categories of the characteristic attributes, wherein the classification network model uses multiple sets of data Trained through machine learning, each of the multiple sets of data includes: feature attributes, and multiple probability values of different categories corresponding to the feature attributes; The determining module uses the determined target category corresponding to the largest probability value among the plurality of probability values as the failure cause of the program failure. 8. The device according to claim 7, wherein the device further comprises: a configuration module configured to configure a class library for the classification network model, wherein the class library includes a plurality of corresponding feature attributes category; the classification network model is trained according to the characteristic attributes corresponding to the program failure, and a plurality of categories corresponding to the program failure selected from the category library. 9. A computer-readable storage medium, wherein a computer program is stored in the storage medium, wherein the computer program is configured to execute the program described in any one of claims 1 to 6 when running. Methods. 10. An electronic device comprising a memory and a processor, wherein a computer program is stored in the memory, and the processor is configured to run the computer program to execute any one of claims 1 to 6 method described in.

Images