A software fault pre-diagnosis method and system suitable for domestic servers
By building a software fault prediagnosis model based on differential operation and convolutional neural network, real-time prediction and prevention of domestic server software failures is achieved, problems of complex operation and high professional requirements in the existing technology are solved, and fault diagnosis efficiency is improved and server stability.
Patent Information
- Application Number
- CN202411207565.7
- Authority / Receiving Office
- CN · China
- Patent Type
- Patents(China)
- Current Assignee / Owner
- Filing Date
- 2024-08-30
- Publication Date
- 2024-11-22
- Estimated Expiration
- 2044-08-30
AI Technical Summary
The existing domestic server software fault prediagnosis methods require programmers to manually write functions, which are complex in operation and high professional requirements, making it difficult to achieve efficient software fault prediction and prevention.
By obtaining the server's history and normal operation parameters, performing differential operations and preprocessing, building a software fault prediagnosis model, and using the convolutional neural network model to predict real-time software faults, realizing the advance prediction and maintenance of software faults.
It improves the server software diagnosis efficiency, reduces the probability of software failure, ensures the normal operation of the server, and simplifies the fault handling process.
Smart Images

Figure CN118708499B_ABST
Abstract
Description
Technical Field
[0001] The present invention belongs to the technical field of computer fault diagnosis, and in particular relates to a software fault pre-diagnosis method and system applicable to domestic servers. Background Art
[0002] Software failures in servers account for the highest proportion of server failures, up to 70%. There are many reasons for server software failures. The most common ones are that the server's BIOS version is too low, the server's management software or the server's driver has a bug, or the application has a conflict, which may cause a software failure. Different software failures will cause different effects. For example, a driver bug can cause the server to freeze, conflict with certain software, and unstable disk operation. A management program bug can cause the server system to run slower, the CPU usage rate to increase, and some functions to be unable to be used normally. Therefore, pre-diagnosis of server software failures plays a vital role in the normal operation of the server.
[0003] In a large server environment, software failures may be caused by reasons such as software errors, configuration problems, and resource exhaustion. In order to effectively perform fault diagnosis, a series of technologies and methods need to be adopted. Existing server software fault detection methods include: Logging and analysis: Server software usually generates a large number of logs to record various operations and events. By analyzing these logs, abnormal behaviors and potential causes of failures can be identified. Real-time monitoring and real-time log analysis tools can help detect abnormal situations in a timely manner.
[0004] At present, the software fault pre-diagnosis method of the domestic server mentioned above still requires programmers to write corresponding functions and add them to the corresponding log system. Or when adding new functions, it is necessary to enter the underlying operating system to analyze the system, which requires high professionalism.
[0005] Therefore, the present invention mounts a software fault pre-diagnosis system of a domestic server to predict software faults before they occur, thereby improving the efficiency of server software diagnosis, so as to implement software and server maintenance processing before software faults occur, thereby reducing the probability of software faults occurring. Summary of the invention
[0006] The purpose of the present invention is to provide a software fault pre-diagnosis method and system by mounting a domestic server, so as to predict software faults before they occur, improve the efficiency of server software diagnosis, and implement software and server maintenance processing before software faults occur, thereby reducing the probability of software faults occurring.
[0007] In a first aspect, a software fault pre-diagnosis method applicable to a domestic server is provided, comprising the following steps:
[0008] S1: obtaining first software operating parameters and first server operating parameters of each software of the domestic server within a first specified period before a failure occurs in a historical operating process, and preprocessing the obtained first software operating parameters and first server operating parameters;
[0009] S2: obtaining second software operating parameters and second server operating parameters during a second specified period of time during normal operation of each software of the domestic server, and preprocessing the obtained second software operating parameters and second server operating parameters;
[0010] S3: performing a differential operation on the first software operation parameter and the second software operation parameter to obtain software operation parameter differential data, performing a differential operation on the first server operation parameter and the second server operation parameter to obtain server operation parameter differential data, and mapping and marking the software operation parameter differential data and the server operation parameter differential data according to their corresponding fault types and fault severity, and forming a differential data set with the marked differential data;
[0011] S4: constructing a software fault pre-diagnosis model, and dividing the differential data set into a training set, a test set, and a validation set;
[0012] S5: input the training set, the test set and the validation set into the software fault prediction model in sequence, train, test and validate the software fault prediction model in sequence, and judge whether the software fault prediction model meets the expected requirements through the validation results, if so, execute step S6, if not, re-execute step S1;
[0013] S6: Collecting real-time software running parameters and real-time server running parameters during the running of each software;
[0014] S7: Input the acquired real-time software operating parameters and real-time server operating parameters into the software fault pre-diagnosis model, and the software fault pre-diagnosis model outputs the software fault pre-diagnosis results in real time according to the input real-time software operating parameters and real-time server operating parameters.
[0015] Preferably, the first server operating parameters, the second server operating parameters and the real-time server operating parameters all include server throughput, CPU utilization, memory utilization, disk space, and network traffic, and the server throughput includes the average throughput, peak throughput and minimum throughput during a specified period; the first software operating parameters, the second software operating parameters and the real-time software operating parameters all include response time, software throughput, resource utilization, number of clicks and number of concurrent users.
[0016] Preferably, the first designated period and the second designated period have the same length of time.
[0017] Preferably, the pre-processing process of step S1 and step S2 includes the following steps:
[0018] S11: Perform data cleaning on the data;
[0019] S12: performing data transformation processing on the data, through function transformation or mean normalization transformation or discretization transformation;
[0020] S13: performing data normalization processing on the data.
[0021] Preferably, the data cleaning process of step S11 is as follows:
[0022] S111: Identify missing data for all data;
[0023] S112: Deleting, replacing or interpolating the identified missing data;
[0024] S113: Identify abnormal data for all data;
[0025] S114: Delete the identified abnormal data or correct the average value.
[0026] Preferably, step S3 includes the following specific process:
[0027] S31: dividing the first software running parameter and the second software running parameter into a first software running data block and a second software running data block according to time;
[0028] S32: performing differential operations respectively according to the corresponding relationship between the first software running data block and the second software running data block, and merging them to obtain software running differential data;
[0029] S33: dividing the first server operation parameter and the second server operation parameter into different first server operation data blocks and second server operation data blocks according to time;
[0030] S34: performing differential operations respectively according to the correspondence between the first server operation data block and the second server operation data block, and merging them to obtain server operation differential data;
[0031] S35: Map and mark the software operation parameter differential data and the server operation parameter differential data according to their corresponding fault types and fault severity levels.
[0032] Preferably, step S5 includes the following specific process:
[0033] S51: inputting the training set into a software fault pre-diagnosis model to train the software fault pre-diagnosis model;
[0034] S52: input the test set into the trained software fault pre-diagnosis model, test the trained software fault pre-diagnosis model, and determine whether the test result meets the test requirements. If so, execute step S53; if not, obtain a different training set and execute step S51 again;
[0035] S53: verifying the software fault pre-diagnosis model after the test of the verification set data, and judging whether the verification result meets the preset requirements.
[0036] Preferably, the software fault pre-diagnosis model is a convolutional neural network model, and the input layer of the convolutional neural network model is provided with dual input channels, wherein one input channel is used to input software operating parameters, and the other channel is used to input server operating parameters.
[0037] In a second aspect, a software fault pre-diagnosis system applicable to a domestic server is provided, which is used to implement any one of the software fault pre-diagnosis methods applicable to a domestic server, including a data acquisition module, a data block division module, a data difference module, a data pre-processing module, a model creation module, a fault tolerance module and a software fault pre-diagnosis model;
[0038] The data acquisition module is used to obtain server operating parameters and software operating parameters during the operation of the domestic server;
[0039] The data block division module is used to divide the first software operation parameter and the second software operation parameter into the first software operation data block and the second software operation data block according to time, and divide the first server operation parameter and the second server operation parameter into different first server operation data blocks and second server operation data blocks according to time;
[0040] The data difference module is used to perform difference processing on the corresponding data blocks in the first software running data block and the second software running data block, and to perform difference processing on the corresponding data blocks in the first server running data block and the second server running data block;
[0041] The data preprocessing module is used to perform data preprocessing on each acquired data;
[0042] The model creation module is used to build a software fault pre-diagnosis model;
[0043] The fault-tolerant module includes a redundancy module, a backup module and an automatic failover module, which ensures the normal operation of the server when a software failure occurs;
[0044] The software fault pre-diagnosis model is used to receive real-time software operating parameters and real-time server operating parameters and output software fault pre-diagnosis results in real time.
[0045] Preferably, it also includes a communication module, which is used to establish a communication protocol between the CPU of the server and the single-chip microcomputer, and communicate based on the established communication protocol, wherein the communication protocol includes the format of commands and the transmission method of data.
[0046] The beneficial effects of the present invention include:
[0047] The software fault pre-diagnosis method and system applicable to domestic servers provided by the present invention first obtain the software operating parameters and server operating parameters before the software fault occurs on the server; obtain the software operating parameters and the second server operating parameters in normal operation; use differential operation to obtain software operating parameter differential data, and use differential operation of the first server operating parameter and the second server operating parameter to obtain server operating parameter differential data; build a software fault pre-diagnosis model, and train, test and verify the software fault pre-diagnosis model; collect real-time software operating parameters and real-time server operating parameters in the running process of each software; input the obtained real-time software operating parameters and real-time server operating parameters into the model to output the software fault pre-diagnosis results in real time, so as to realize real-time prediction of software faults.
[0048] By collecting various parameters of the server and software in real time and based on the software fault prediction model, we can predict server software failures in advance, improve the efficiency of server software diagnosis, and perform software and server maintenance before software failures occur, ultimately reducing the probability of software failures.
[0049] By setting dual input channels in the input layer of the convolutional neural network model, each input channel has its own convolution kernel for extracting features from its own input, one input channel is used to input software operating parameters, and the other channel is used to input server operating parameters, so that the software fault pre-diagnosis model can learn the features of data from multiple angles or sources, thereby increasing the representation ability of the model. BRIEF DESCRIPTION OF THE DRAWINGS
[0050] Figure 1 The figure is a flow chart of the software fault pre-diagnosis method applicable to domestic servers of the present invention.
[0051] Figure 2 The figure is a schematic diagram of the software fault prediction process in the prior art.
[0052] Figure 3 The schematic diagram of the software fault pre-diagnosis system architecture applicable to domestic servers of the present invention. DETAILED DESCRIPTION
[0053] The following is combined with Figure 1-Figure 3 The present invention is further described in detail:
[0054] Example 1
[0055] See also Figure 1 , a software fault pre-diagnosis method applicable to domestic servers, comprising the following steps:
[0056] S1: obtaining first software operating parameters and first server operating parameters of each software of the domestic server within a first specified period before a failure occurs in a historical operating process, and preprocessing the obtained first software operating parameters and first server operating parameters;
[0057] S2: obtaining second software operating parameters and second server operating parameters within a second specified period during normal operation of each software of the domestic server, and preprocessing the obtained second software operating parameters and second server operating parameters, wherein the first specified period and the second specified period have the same time length;
[0058] S3: performing a differential operation on the first software operation parameter and the second software operation parameter to obtain software operation parameter differential data, performing a differential operation on the first server operation parameter and the second server operation parameter to obtain server operation parameter differential data, and mapping and marking the software operation parameter differential data and the server operation parameter differential data according to their corresponding fault types and fault severity, and forming a differential data set with the marked differential data;
[0059] S4: constructing a software fault pre-diagnosis model, and dividing the differential data set into a training set, a test set, and a validation set, respectively used for training, testing, and validating the software fault pre-diagnosis model, so that the software fault pre-diagnosis model has a certain prediction ability and ensures the accuracy of the prediction result;
[0060] S5: input the training set, the test set and the validation set into the software fault prediction model in sequence, train, test and validate the software fault prediction model in sequence, and judge whether the software fault prediction model meets the expected requirements through the validation results, if so, execute step S6, if not, re-execute step S1;
[0061] S6: Collecting real-time software running parameters and real-time server running parameters during the running of each software;
[0062] S7: Input the acquired real-time software operating parameters and real-time server operating parameters into the software fault pre-diagnosis model, and the software fault pre-diagnosis model outputs the software fault pre-diagnosis results in real time according to the input real-time software operating parameters and real-time server operating parameters.
[0063] See also Figure 2 The software fault process in the prior art first creates a software analysis entity, initializes the alarm state, that is, presets the alarm condition, and then determines whether the preset alarm condition is currently met. If so, executes the recovery logic and determines whether the recovery condition is met. If the recovery condition is met, the alarm is restored and the alarm state is updated. If the recovery condition is not met, it returns to determine whether the preset alarm condition is currently met; if not, the alarm logic is executed and the alarm condition is determined. If the alarm condition is met, an alarm command is generated and the alarm state is updated. If the alarm condition is not met, it returns to determine whether the preset alarm condition is currently met. The entire process only determines whether to alarm by setting the alarm condition, so the accuracy of diagnosing software faults is obviously not high.
[0064] The data preprocessing process in steps S1 and S2 of the present application is to make all the acquired data valid, avoid duplicate data and abnormal data in the acquired data, and through normalization processing, all data are limited to a certain range, so as to have the same characterization. Step S3 performs differential operation on the first software operation parameter and the second software operation parameter, the first server operation parameter and the second server operation parameter according to the corresponding data block, and obtains the differential data between the first software operation parameter and the first server operation parameter in the specified period before the software failure occurs in the historical operation process of each server, and the second software operation parameter and the second server operation parameter in the specified period of normal software operation, and then finds the difference between normal operation and each parameter and each parameter before the software failure, and inputs the server operation parameters and software operation parameters collected in real time in the later stage into the software failure prediction model to predict the software failure.
[0065] The first server operating parameters, the second server operating parameters and the real-time server operating parameters all include server throughput, CPU utilization, memory utilization, disk space, and network traffic. The server throughput includes the average throughput, peak throughput, and minimum throughput during a specified period; the first software operating parameters, the second software operating parameters and the real-time software operating parameters all include response time, software throughput, resource utilization, number of clicks, and number of concurrent users.
[0066] Example 2
[0067] Based on Example 1, the pretreatment process of step S1 and step S2 includes the following steps:
[0068] S11: Perform data cleaning on the data to remove abnormal data, duplicate data, etc. from all the data. The specific data cleaning process is as follows:
[0069] S111: Identify missing data for all data;
[0070] S112: Deleting, replacing or interpolating the identified missing data;
[0071] S113: Identify abnormal data for all data;
[0072] S114: Delete the identified abnormal data or correct the average value.
[0073] S12: performing data transformation processing on the data, through function transformation or mean normalization transformation or discretization transformation. Data transformation processing is the process of converting data from one format or structure to another format or structure, so as to facilitate the subsequent data analysis;
[0074] S13: performing data normalization processing on the data.
[0075] Step S3 includes the following specific processes:
[0076] S31: dividing the first software running parameter and the second software running parameter into a first software running data block and a second software running data block according to time;
[0077] S32: performing differential operations respectively according to the corresponding relationship between the first software running data block and the second software running data block, and merging them to obtain software running differential data;
[0078] S33: dividing the first server operation parameter and the second server operation parameter into different first server operation data blocks and second server operation data blocks according to time;
[0079] S34: performing differential operations respectively according to the correspondence between the first server operation data block and the second server operation data block, and merging them to obtain server operation differential data;
[0080] S35: Map and mark the software operation parameter differential data and the server operation parameter differential data according to their corresponding fault types and fault severity levels.
[0081] The first software operation parameter and the second software operation parameter are divided into different data blocks for differential processing, which improves the efficiency of differential processing on the one hand and reduces the error of differential processing on the other hand. The data blocks are divided according to time to provide certain reference indicators for the subsequent software fault prediction model to predict the time of fault occurrence.
[0082] Example 3
[0083] Based on Example 1 or Example 2, step S5 includes the following specific process:
[0084] S51: inputting the training set into a software fault pre-diagnosis model to train the software fault pre-diagnosis model;
[0085] S52: input the test set into the trained software fault pre-diagnosis model, test the trained software fault pre-diagnosis model, and determine whether the test result meets the test requirements. If so, execute step S53; if not, obtain a different training set and execute step S51 again;
[0086] S53: verifying the software fault pre-diagnosis model after the test of the verification set data, and judging whether the verification result meets the preset requirements.
[0087] Only when the software fault prediction model meets certain requirements after training and testing, and the subsequent verification results also meet certain requirements, can the software fault prediction model be used for subsequent real-time software fault prediction, that is, the subsequent real-time software fault prediction results have a certain degree of accuracy. If the subsequent test results or verification results do not meet the specified requirements, it is necessary to reselect the training set, test set and verification set for training, testing and verification, so that the final software fault prediction model must have the corresponding learning ability and can achieve accurate prediction of software faults.
[0088] The software fault pre-diagnosis model is a convolutional neural network model, and the input layer of the convolutional neural network model is provided with dual input channels, one of which is used to input software operating parameters, and the other is used to input server operating parameters. By providing dual input channels in the input layer of the convolutional neural network model, each input channel has its own convolution kernel, which is used to extract features from respective inputs, one of which is used to input software operating parameters, and the other is used to input server operating parameters, so that the software fault pre-diagnosis model can learn features of data from multiple angles or sources, thereby increasing the representation capability of the model. Each output channel identifies a specific pattern, and the input channel kernel identifies a pattern in a combined input. Assuming there are multiple output channels, each output channel identifies a pattern in a combined input through its corresponding input channel, thereby achieving more complex feature extraction and pattern recognition. By combining different input information, the model can learn richer feature representations, thereby improving the accuracy of prediction and the generalization capability of the model.
[0089] See also Figure 3A software fault pre-diagnosis system applicable to domestic servers is used to implement any one of the software fault pre-diagnosis methods applicable to domestic servers, including a data acquisition module, a data block division module, a data difference module, a data pre-processing module, a model creation module and a fault tolerance module. The data acquisition module is used to obtain server operating parameters and software operating parameters during the operation of the domestic server.
[0090] The data block division module is used to divide the first software operation parameter and the second software operation parameter into the first software operation data block and the second software operation data block according to time, and divide the first server operation parameter and the second server operation parameter into different first server operation data blocks and second server operation data blocks according to time. The data difference module is used to perform difference processing on the corresponding data blocks in the first software operation data block and the second software operation data block, and to perform difference processing on the corresponding data blocks in the first server operation data block and the second server operation data block. The software fault pre-diagnosis model is used to receive real-time software operation parameters and real-time server operation parameters and output software fault pre-diagnosis results in real time.
[0091] The data preprocessing module is used to preprocess the acquired data. The fault-tolerant module includes a redundancy module, a backup module and an automatic failover module to ensure the normal operation of the server when a software failure occurs. It also includes a communication module, which is used to establish a communication protocol between the CPU and the single-chip microcomputer of the server and communicate based on the established communication protocol, and the communication protocol includes the format of the command and the transmission method of the data.
[0092] The software fault prediction system applicable to domestic servers is also provided with a remote management module, which is used for the administrator to remotely connect to the management server to achieve management of the server to be predicted.
[0093] In summary, the software fault pre-diagnosis method and system applicable to domestic servers provided by the present invention first obtain the software operating parameters and server operating parameters before the software fault occurs on the server; obtain the software operating parameters and the second server operating parameters in normal operation; use differential operation to obtain software operating parameter differential data, and obtain server operating parameter differential data by differential operation of the first server operating parameter and the second server operating parameter; construct a software fault pre-diagnosis model, and train, test and verify the software fault pre-diagnosis model; collect real-time software operating parameters and real-time server operating parameters during the operation of each software; input the obtained real-time software operating parameters and real-time server operating parameters into the model to output the software fault pre-diagnosis results in real time, so as to realize real-time prediction of software faults.
[0094] By collecting various parameters of the server and software in real time, based on the software fault prediction model, the server software fault is predicted in advance, the server software diagnosis efficiency is improved, and software and server maintenance processing is performed before the software fault occurs, which ultimately reduces the probability of software fault occurrence. By setting dual input channels in the input layer of the convolutional neural network model, each input channel has its own convolution kernel for extracting features from the respective inputs, one of which is used to input software operating parameters and the other is used to input server operating parameters, so that the software fault pre-diagnosis model can learn the features of data from multiple angles or sources, increasing the representation ability of the model.
Claims
1. A software fault pre-diagnosis method applicable to domestic servers, characterized in that: The following steps are involved: S1: obtaining first software operating parameters and first server operating parameters of each software of the domestic server within a first specified period before a failure occurs in a historical operating process, and preprocessing the obtained first software operating parameters and first server operating parameters; S2: obtaining second software operating parameters and second server operating parameters during a second specified period of time during normal operation of each software of the domestic server, and preprocessing the obtained second software operating parameters and second server operating parameters; S3: performing a differential operation on the first software operation parameter and the second software operation parameter to obtain software operation parameter differential data, performing a differential operation on the first server operation parameter and the second server operation parameter to obtain server operation parameter differential data, and mapping and marking the software operation parameter differential data and the server operation parameter differential data according to their corresponding fault types and fault severity, and forming a differential data set with the marked differential data; S4: constructing a software fault pre-diagnosis model, and dividing the differential data set into a training set, a test set, and a validation set; S5: input the training set, the test set and the validation set into the software fault prediction model in sequence, train, test and validate the software fault prediction model in sequence, and judge whether the software fault prediction model meets the expected requirements through the validation results, if so, execute step S6, if not, re-execute step S1; S6: Collecting real-time software running parameters and real-time server running parameters during the running of each software; S7: inputting the acquired real-time software operating parameters and real-time server operating parameters into the software fault pre-diagnosis model, and the software fault pre-diagnosis model outputs the software fault pre-diagnosis result in real time according to the input real-time software operating parameters and real-time server operating parameters; Among them, the software fault pre-diagnosis model is a convolutional neural network model, and the input layer of the convolutional neural network model is set with dual input channels, one of which is used to input software operating parameters, and the other is used to input server operating parameters.
2. A software fault pre-diagnosis method applicable to domestic servers according to claim 1, characterized in that: The first server operating parameters, the second server operating parameters and the real-time server operating parameters all include server throughput, CPU utilization, memory utilization, disk space, and network traffic. The server throughput includes the average throughput, peak throughput, and minimum throughput during a specified period; the first software operating parameters, the second software operating parameters and the real-time software operating parameters all include response time, software throughput, resource utilization, number of clicks, and number of concurrent users.
3. A software fault pre-diagnosis method applicable to domestic servers according to claim 2, characterized in that: The first designated period and the second designated period have the same length of time.
4. A software fault pre-diagnosis method applicable to domestic servers according to claim 1, characterized in that: The preprocessing process of step S1 and step S2 includes the following steps: S11: Perform data cleaning on the data; S12: performing data transformation processing on the data, through function transformation or mean normalization transformation or discretization transformation; S13: performing data normalization processing on the data.
5. A software fault pre-diagnosis method applicable to domestic servers according to claim 4, characterized in that: The data cleaning process of step S11 is as follows: S111: Identify missing data for all data; S112: Deleting, replacing or interpolating the identified missing data; S113: Identify abnormal data for all data; S114: Delete the identified abnormal data or correct the average value.
6. A software fault pre-diagnosis method applicable to domestic servers according to claim 1, characterized in that: Step S3 includes the following specific processes: S31: dividing the first software running parameter and the second software running parameter into a first software running data block and a second software running data block according to time; S32: performing differential operations respectively according to the corresponding relationship between the first software running data block and the second software running data block, and merging them to obtain software running differential data; S33: dividing the first server operation parameter and the second server operation parameter into different first server operation data blocks and second server operation data blocks according to time; S34: performing differential operations respectively according to the correspondence between the first server operation data block and the second server operation data block, and merging them to obtain server operation differential data; S35: Map and mark the software operation parameter differential data and the server operation parameter differential data according to their corresponding fault types and fault severity levels.
7. A software fault pre-diagnosis method applicable to domestic servers according to claim 1, characterized in that: Step S5 includes the following specific processes: S51: inputting the training set into a software fault pre-diagnosis model to train the software fault pre-diagnosis model; S52: input the test set into the trained software fault pre-diagnosis model, test the trained software fault pre-diagnosis model, and determine whether the test result meets the test requirements. If so, execute step S53; if not, obtain a different training set and execute step S51 again; S53: verifying the software fault pre-diagnosis model after the test of the verification set data, and judging whether the verification result meets the preset requirements.
8. A software fault pre-diagnosis system applicable to domestic servers, used to implement a software fault pre-diagnosis method applicable to domestic servers as described in any one of claims 1 to 7, characterized in that: It includes data acquisition module, data block division module, data difference module, data preprocessing module, model creation module, fault tolerance module and software fault pre-diagnosis model; The data acquisition module is used to obtain server operating parameters and software operating parameters during the operation of the domestic server; The data block division module is used to divide the first software operation parameter and the second software operation parameter into the first software operation data block and the second software operation data block according to time, and divide the first server operation parameter and the second server operation parameter into different first server operation data blocks and second server operation data blocks according to time; The data difference module is used to perform difference processing on the corresponding data blocks in the first software running data block and the second software running data block, and to perform difference processing on the corresponding data blocks in the first server running data block and the second server running data block; The data preprocessing module is used to perform data preprocessing on each acquired data; The model creation module is used to build a software fault pre-diagnosis model; The fault-tolerant module includes a redundancy module, a backup module and an automatic failover module, which ensures the normal operation of the server when a software failure occurs; The software fault pre-diagnosis model is used to receive real-time software operating parameters and real-time server operating parameters and output software fault pre-diagnosis results in real time; The software fault pre-diagnosis model is a convolutional neural network model, and the input layer of the convolutional neural network model is provided with dual input channels, one of which is used to input software operation parameters, and the other is used to input server operation parameters.
9. A software fault pre-diagnosis system suitable for domestic servers according to claim 8, characterized in that: It also includes a communication module, which is used to establish a communication protocol between the CPU of the server and the single-chip microcomputer, and communicate based on the established communication protocol. The communication protocol includes the format of commands and the transmission method of data.
