Missile computer AI test and maintenance system and method based on big data learning

Abstract

The invention provides a learnable missile-borne computer AI test and maintenance system and method based on big data. Comprising a hardware data acquisition module and a software data test analysis fault maintenance module, the hardware data acquisition module comprises a power supply control unit, a time sequence control unit and a data acquisition unit, and the power supply control unit is responsible for providing a power supply for the missile-borne computer intelligent AI test maintenance system; the time sequence control unit is mainly used for processing the clock domain crossing problem of eight channels; the data acquisition unit performs data sampling, the software data test analysis and fault maintenance module comprises a main control program assembly, a data test analysis assembly and a data fault maintenance AI assembly, and the main control program assembly is used for overall operation of the system and scheduling and collaboration of functions of all parts; the data test analysis component is mainly used for performing batch analysis on the acquired data; the data fault maintenance AI assembly is mainly used for deep learning of test big data, and compared with the prior art, the system has the remarkable advantage of being high in integration level.

Description

Technical Field

[0001] The present invention belongs to the field of onboard computer data test batch processing, and specifically relates to an onboard computer AI test and maintenance system and method based on big data learning, especially an onboard computer intelligent AI test and maintenance system based on big data learning. Background Art

[0002] The onboard computer is a key component in the missile system and the core part of the seeker information processing. With the continuous development and updating of electronic technology and integration processes, the size of onboard computers is getting smaller and smaller, and the integration of components is getting higher and higher. The efficiency of the original test system can no longer meet the current mass production scale.

[0003] For example, patent document CN117315408A discloses a method and system for constructing a fault criterion model based on an AI target detection model. The method includes: obtaining historical operating data of the AI target detection model, including historical fault data, to provide basic information for the establishment of a fault criterion model. Secondly, the historical fault data of the AI target detection model is cleaned and integrated into a data fault set to ensure the quality and availability of the data. Then, operating parameters are extracted from the data fault set, and key functional parameters are further screened out. Subsequently, a functional model is established and fitting experiments are performed to improve model performance. In this process, data processing is performed on the functional data set to determine the optimal functional data, thereby improving the accuracy of the model. Finally, the fault criterion model of the AI target detection model is determined through the functional model to provide support for the reliability of the establishment of the fault criterion model.

[0004] For example, patent document CN119762052A discloses a digital AI maintenance and support system for armored equipment, which includes a fault information receiving module, an AI solution generation module, an intelligent auxiliary execution module, and a management feedback module. The fault information receiving module is used to receive fault information of armored equipment, the AI solution generation module generates a maintenance solution based on the fault information, the intelligent auxiliary execution module is used to output auxiliary information of the maintenance solution, and the management feedback module is used to manage data information during the maintenance process; the system can intelligently analyze and generate maintenance solutions based on fault information, and control maintenance equipment to perform maintenance, thereby improving the automated maintenance capabilities of armored equipment.

[0005] However, the test system in the prior art has the following disadvantages: 1. The size of the missile-mounted computer has been reduced to about one-third of its original size. In the originally designed missile-mounted computer data processing system, the fixtures used in the hardware testing part were fastened with screws. Each disassembly and assembly required manpower, which was not only inefficient but also unable to meet the needs of mass production.

[0006] 2. The data processing of the on-board computer test system adopts a linear processing mode for a single set of products. The system needs to complete the testing of all data such as frequency, spectrum, AD / DA, etc. of the first set of products before starting the test process for the second set of products. During this process, when testing a certain data of the first set of products, other acquisition and testing modules are in standby state, resulting in a serious waste of computing power resources. Especially when faced with massive test samples, this processing mode greatly extends the overall test time.

[0007] 3. The data interpretation, analysis, organization and summary work in the onboard computer test system mainly relies on manpower. During the specific operation, the staff needs to open the data test form of each set of products one by one, manually judge whether the data meets the standards, and carry out subsequent organization and summary. In the past, when each batch had only a dozen sets of products, this manual processing method could still cope with it. However, in the face of the current situation where the number of batch products has increased significantly, it is obviously unable to cope with it.

[0008] 4. The onboard computer test system has accumulated a massive amount of test data during its operation. However, the maintenance work for faulty products still mainly relies on manual operation. Summary of the Invention

[0009] In view of the defects in the prior art, the purpose of the present invention is to provide an on-board computer AI testing and maintenance system based on big data learning.

[0010] According to the present invention, a missile-borne computer AI testing and maintenance system based on big data learning is provided, which includes: a hardware data acquisition module and a software data testing, analysis and fault repair module; the software data testing, analysis and fault repair module can input the missile-borne data collected by the hardware data acquisition module in parallel, and then analyze, interpret and summarize the missile-borne data, and perform preliminary repairs on unqualified products.

[0011] Preferably, the hardware data acquisition module is used to test the onboard computer product and collect onboard data; the hardware data acquisition module includes: a power supply control unit, a timing control unit, and a data acquisition unit.

[0012] Preferably, the power supply control unit is used to supply power to the onboard computer intelligent AI testing and maintenance system; the power supply control unit includes: a ripple suppression component, which can effectively reduce the impact of ripple and noise on subsequent precision data and rate data acquisition, while improving the long-term stability of the system; the power supply control unit adopts a plug-in slide design, which balances the plug-in and pull-out forces to avoid distortion, shrinkage and breakage of the connector core pins due to uneven plug-in and pull-out forces or plug-in and pull-out shaking and twisting; the power supply control unit adopts a two-level buffer overcurrent protection design. Specifically, when the load current exceeds the first warning line, the voltage stabilizer in the device continues to work, and the onboard computer data processing system enters the half-load working mode. When the load current exceeds the second current warning line, the voltage stabilizer in the device is interrupted, and the onboard computer data processing system stops working.

[0013] Preferably, the timing control unit can handle the cross-clock domain of 8 channels; the timing control unit adopts multiple measures to solve the problems of metastability, clock offset and jitter, and the multiple measures include: asynchronous FIFO buffering method, bilateral handshake protocol confirmation method, and clock domain crossing method.

[0014] Preferably, the data acquisition unit is capable of performing analog digital data sampling, digital analog data sampling, and temperature and humidity data sampling; the data acquisition unit includes: a noise suppression component, a digital processing component, and a temperature drift compensation component, and the data acquisition unit can effectively improve the reliability of the collected data; under the control of the data acquisition unit, the analog digital data, digital analog data, and temperature and humidity data are respectively transmitted to the noise suppression component, and after the digital processing component receives the data from the noise suppression component, it summarizes and sends it to the temperature drift compensation component.

[0015] Preferably, the software data testing and analysis fault repair module is used to analyze, interpret, and summarize the sampled data, and perform preliminary repairs on unqualified products; the software data testing and analysis fault repair module includes: a main control program component, a data testing and analysis component, and a data fault repair AI component; after the main control program component receives the parallel input data, it sends the processed product data to the data testing and analysis component, and sends the analysis of the fault cause and preliminary repair suggestions to the data fault repair AI component; after the data testing and analysis component receives the processed product data from the main control program component, it returns the correct data to the main control program component, and at the same time sends the correct product data to the data fault repair AI component to feed the AI, and obtains incorrect product data from the data fault repair AI component; the data fault repair AI component receives the analysis of the fault cause and preliminary repair suggestions from the main control program component, receives the correct product data from the data testing and analysis component to feed the AI, and obtains incorrect product data from the data testing and analysis component; Preferably, the main control program component is used for the overall operation of the system and the scheduling and coordination of various functions; the main control program component has the function of communicating and transmitting data with the onboard computer via the CAN bus, and can test the working status of received instructions to determine whether the instruction reception is normal; the main control program component is responsible for global resource mobilization, can monitor system resources in real time, and perform data analysis with high computational complexity when resources are surplus, which can effectively improve test efficiency; Preferably, the data test and analysis component can analyze, interpret and integrate the collected data in batches; the data test and analysis component can realize cascade calls to multiple software, and the multiple software include: Matlab, Excel; Preferably, the data test analysis component calls Excel through the C# main program, operates Excel, automatically fills in the qualification judgment basis of various indicators, and performs preliminary interpretation of the test data, filters the data of unqualified products, and records the numbers; Preferably, the data fault repair AI component is capable of deep learning of test big data, preliminary troubleshooting of data faults, and early situation-based maintenance recommendations for faults that have not yet occurred; the data fault repair AI component adopts a supervised learning model; the data fault repair AI component adopts an algorithm that combines logistic regression and decision trees, and is capable of self-iteration and continuous learning; the data fault repair AI component first optimizes the time for AI to search the database through an algorithm, and understands the user's query intention through natural language processing, and the data fault repair AI component includes a set of query decision instructions; the data fault repair AI component first performs data integration, conversion and anomaly detection through intelligent AI, and then analyzes abnormal data through a regression algorithm to establish a fault tree including probability distribution to help users make maintenance decisions; the fault tree includes: excessive AD error, failure of the AD device itself, abnormal AD device power supply, failure of the power supply ripple suppression module, abnormal digital signal processing algorithm, DC / DC abnormality, power supply chip abnormality, and voltage regulator abnormality; the data fault repair AI component can issue early warnings for abnormal and excessive data and provide situation-based maintenance recommendations.

[0016] Preferably, under the control of the data troubleshooting AI component, the input data is processed through natural language processing and keyword extraction is performed to determine whether it is a legal natural language; if it is determined to be a legal natural language, keyword retrieval is performed and mapped to the database primary key while establishing a logic tree. If the judgment is passed, the recursive logic processing mode is entered, in which iterative learning is performed to adjust the branch weights and update the decision instruction set, and an action is taken, and then it is determined whether the action is correct; if the action is correct, iterative learning is performed again to adjust the branch weights and update the decision instruction set. If the action is incorrect, it is determined again whether the query is understood correctly; if the query is determined to be incorrect before the recursive logic processing mode, it is determined again whether the query is understood correctly; if it is determined not to be a legal natural language after the initial keyword extraction, it is terminated directly.

[0017] The present invention provides a method for testing and repairing missile-mounted computers using AI based on big data learning, comprising: Step S1: First, turn on the power switch, then make the device current self-check, and finally make the power supply control unit and the timing control unit start working; Step S2: First, the main control software is started, then the one-key test function is run, and finally the hardware data acquisition module starts working and transmits the collected data to the main control program component through the controller local area network communication; Step S3: instructing the main control program component in step S2 to send the collected data to the software data test analysis and troubleshooting module, and call the data interpretation parameter setting; Step S4: The software data testing, analysis, and troubleshooting module analyzes and interprets the data based on preset parameter indicators, summarizes and outputs qualified product data, and draws a data envelopment diagram for the batch products. Finally, the qualified product data is fed to the AI for learning, and the unqualified products are sent to the data troubleshooting AI component. Step S5: The data fault repair AI component will analyze the fault cause corresponding to the fault phenomenon based on previous cases, automatically generate a fault tree, conduct troubleshooting according to the fault tree, and finally provide maintenance recommendations; Step S6: The data troubleshooting AI component analyzes the qualified product data and analyzes the data of products that are marked as qualified because they do not exceed the calibrated correct data range, but are significantly different from other products in the same batch. Based on previous cases, it provides repair recommendations based on the situation. Step S7: issuing a data test report based on the maintenance recommendations made in step S6.

[0018] Preferably, step S4 includes: Step S401: Execute data testing and analysis steps; Step S402: executing data integration step; Step S403: executing data output step; Furthermore, step S401 specifically refers to, assuming that the number of the first batch of weapons and equipment is 500 sets, first, opening the Excel file through the C# main program and completing the relevant initialization settings; then, using the built-in If function of Excel to set the embodiment, thereby interpreting the legitimacy of the data; finally, executing the operation 500 times through the For loop program, thereby interpreting all the test data of the 500 sets of weapons and equipment, and carrying out analysis and screening work; Furthermore, step S402 specifically refers to, first, opening the 500 Excel files of the 500 sets of products in step S401 one by one through the corresponding code, then copying and summarizing the key data therein into a data summary table, and finally analyzing and obtaining the envelope and data distribution of the test data of the entire batch; Furthermore, the step S403 specifically refers to drawing the data in the data summary table summarized in step S402 into an envelope diagram by setting C# code.

[0019] The present invention provides an on-board computer AI testing and maintenance equipment based on big data learning, which achieves significant improvement in testing efficiency and performance.

[0020] Compared with the prior art, the present invention has the following beneficial effects: 1. The present invention uses tooling to carry out batch installation of products, which can greatly reduce the test time and significantly improve the test efficiency. It effectively solves the disadvantages of traditional screw fastening of single-set products, and designs an insertable slide tooling, which has the characteristics of fast and convenient product plugging and unplugging, uniform force, supports hot plugging function, and successfully avoids the risk of misoperation of live plugging and unplugging.

[0021] 2. The present invention adopts a high-precision power supply and anti-interference circuit design, and adopts multiple measures at the hardware and software levels at the input and output ends. It can simultaneously collect high-precision and high-speed data from multiple sets of products, achieving a significant improvement in test efficiency and performance.

[0022] 3. The present invention automatically analyzes test data through intelligent AI and can conveniently query various indicators of batch products in natural language, including data distribution and comparison with previous batch product data. This not only improves data accuracy and shortens data processing time, but also assists in the subsequent repair work of unqualified products.

[0023] 4. The present invention adopts a multi-channel parallel working mode and uses a number of measures such as asynchronous FIFO buffering, bilateral handshake protocol confirmation, and clock domain crossing (CDC), which can efficiently utilize idle resources and realize the processing of multiple products at the same time.

[0024] 5. The present invention has the advantages of high integration, fast testing, and intelligent data analysis, and is particularly suitable for testing large quantities of products. BRIEF DESCRIPTION OF THE DRAWINGS

[0025] Other features, objects and advantages of the present invention will become more apparent upon reading the detailed description of non-limiting embodiments with reference to the following drawings: Figure 1 This is a schematic diagram of the architecture of the missile-mounted computer AI testing and maintenance system based on big data learning proposed by the present invention; Figure 2 This is a schematic diagram of the power supply control unit fixing fixture in the missile-mounted computer AI testing and maintenance system based on big data learning proposed by the present invention; Figure 3 This is a schematic diagram of the software data testing, analysis and troubleshooting module in the AI testing and maintenance system for missile-mounted computers based on big data learning proposed by the present invention; Figure 4 This is a flow chart of the AI testing and maintenance system for missile-mounted computers based on big data learning proposed by the present invention; Figure 5 This is a learning logic diagram of the missile-mounted computer AI testing and maintenance system based on big data learning proposed by the present invention; Figure 6This is a schematic diagram of the maintenance fault tree of the missile computer AI test and maintenance system based on big data learning proposed by the present invention; The figure shows: AD stands for analog-to-digital conversion; DA stands for digital to analog conversion; DC stands for direct current. DETAILED DESCRIPTION

[0026] The present invention will be described in detail below with reference to specific embodiments. The following examples will help those skilled in the art to further understand the present invention, but are not intended to limit the present invention in any form. It should be noted that, for those skilled in the art, several changes and improvements can be made without departing from the scope of the present invention. These all fall within the scope of protection of the present invention.

[0027] The present invention provides an AI test and maintenance system for missile-mounted computers based on big data learning, including: a hardware data acquisition module, a software data test and analysis fault repair module; the software data test and analysis fault repair module, as shown in the attached Figure 3 As shown, the missile data collected by the hardware data acquisition module can be input in parallel, and then the missile data can be analyzed, interpreted, and summarized, and preliminary repairs can be performed on unqualified products; The hardware data acquisition module is used to test the onboard computer product and collect onboard data; The hardware data acquisition module includes: a power supply control unit, a timing control unit, and a data acquisition unit; The hardware data acquisition module is written in C# language and can realize the full function of collecting various data indicators of the onboard computer; The power supply control unit is used to supply power to the missile computer intelligent AI test and maintenance system; The power supply control unit includes: a ripple suppression component, which can effectively reduce the impact of ripple and noise on subsequent accuracy data and rate data acquisition, while improving the stability of the system's long-term operation; The power supply control unit adopts a two-stage buffer overcurrent protection design. Specifically, when the load current exceeds the first level warning line, the voltage regulator (LDO) in the device continues to operate, and the onboard computer data processing system enters a half-load working mode. When the load current exceeds the second level current warning line, the voltage regulator (LDO) in the device is interrupted, and the onboard computer data processing system stops working. The power supply control unit, as shown in the attached Figure 2 As shown in the figure, the plug-in slide design is adopted to ensure balanced insertion and removal forces, thus avoiding distortion, shrinkage and breakage of the connector pins due to uneven insertion and removal forces or shaking and twisting of insertion and removal; The timing control unit is capable of processing cross-clock domains of 8 channels; The timing control unit adopts multiple measures to solve the problems of metastability, clock offset and jitter, including asynchronous FIFO buffering (first-in-first-out mechanism), bilateral handshake protocol confirmation, and clock domain crossing (CDC). The data acquisition unit is capable of performing analog-to-digital (AD) data sampling, digital-to-analog (DA) data sampling, and temperature and humidity data sampling; The data acquisition unit includes: a noise suppression component, a digital processing component, and a temperature drift compensation component. The data acquisition unit can effectively improve the reliability of the collected data; Preferably, as shown in the attached Figure 1 As shown, under the control of the data acquisition unit, analog digital (AD) data, digital analog (DA) data, and temperature and humidity data are transmitted to the noise suppression component respectively. After the digital processing component receives the data from the noise suppression component, it summarizes and sends it to the temperature drift compensation component; The software data testing and analysis fault repair module is used to analyze, interpret, and summarize the sampled data and perform preliminary repairs on unqualified products; The software data testing and analysis fault troubleshooting module includes: a main control program component, a data testing and analysis component, and a data fault troubleshooting AI component; Preferably, as shown in the attached Figure 1 As shown, after receiving the parallel input data, the main control program component sends the processed product data to the data test and analysis component, and sends it to the data fault repair AI component to analyze the cause of the fault and provide preliminary repair suggestions.

[0028] After receiving and processing product data from the main control program component, the data test and analysis component returns correct data to the main control program component, and at the same time sends correct product data to the data fault repair AI component to feed AI, and obtains incorrect product data from the data fault repair AI component.

[0029] The data fault repair AI component receives and analyzes the cause of the fault from the main control program component and gives preliminary repair suggestions, receives correct product data from the data test and analysis component and feeds it to the AI, and sends incorrect product data to the data test and analysis component.

[0030] The main control program component is used for the overall operation of the system and the scheduling and coordination of various functions; The main control program component has the function of communicating and transmitting data with the onboard computer via the CAN bus, and can test the working status of receiving instructions to determine whether the instruction reception is normal; The main control program component is responsible for global resource mobilization and can monitor system resources in real time. When resources are surplus, it can perform data analysis with large computational load, which can effectively improve test efficiency. The data testing and analysis component can analyze, interpret and integrate the collected data in batches; The data test and analysis component can realize cascade calling of multiple software, including Matlab and Excel; The data test analysis component calls Excel through the C# main program, operates Excel, automatically fills in the qualification judgment basis of various indicators, and makes a preliminary interpretation of the test data, filters the data of unqualified products, and records the numbers; The data fault repair AI component can perform deep learning of test big data, conduct preliminary troubleshooting of data faults, and provide early, situation-based repair recommendations for faults that have not yet occurred. The data fault troubleshooting AI component adopts a supervised learning model; The data troubleshooting AI component uses an algorithm that combines logistic regression and decision trees, and is capable of self-iteration and continuous learning. The data troubleshooting AI component first optimizes the time it takes to retrieve the database through an algorithm and understands the user's query intent through natural language processing (NLP). The data troubleshooting AI component includes a set of query decision instructions. Preferably, as shown in the attached Figure 5 As shown, under the control of the data troubleshooting AI component, the input data undergoes natural language processing (NLP) and keyword extraction, and a judgment is made as to whether it is legal natural language. If it is judged to be legal natural language, a keyword search is performed and mapped to the database primary key, and a logic tree is established. If the judgment is passed, the process enters a recursive logic processing mode, in which iterative learning is performed to adjust the branch weights and update the decision instruction set, and then an action is taken, and then the judgment as to whether the action is correct is continued. If the action is correct, iterative learning is performed again to adjust the branch weights and update the decision instruction set. If the action is incorrect, the process again judges whether the query is correctly understood. If the query is judged to be incorrect before the recursive logic processing mode, the process will also be re-judged as to whether the query is correctly understood. If it is judged not to be legal natural language after the initial keyword extraction, the process ends directly. The data troubleshooting AI component first performs data integration, conversion, and anomaly detection through intelligent AI. It then analyzes abnormal data through regression algorithms and builds a fault tree including probability distribution to help users make maintenance decisions. The fault tree is as shown in the attached Figure 6 As shown, including: AD error is too large, AD device itself fails, AD device power supply is abnormal, power ripple suppression module fails, digital signal processing algorithm is abnormal, DC / DC is abnormal, power chip is abnormal, voltage regulator is abnormal; The data fault repair AI component can issue early warnings for abnormal and excessive data and provide maintenance suggestions based on the situation.

[0031] The present invention provides a method for testing and repairing missile-mounted computers using AI based on big data learning, comprising: Step S1: First, turn on the power switch, then make the device current self-check, and finally make the power supply control unit and the timing control unit start working; Step S2: First, start the main control software, then run the "one-button test" function, and finally the hardware data acquisition module starts working, transmitting the collected data to the main control program component through the controller local area network communication; Step S3: instructing the main control program component in step S2 to send the collected data to the software data test analysis and troubleshooting module, and call the data interpretation parameter setting; Step S4: The software data testing, analysis, and troubleshooting module analyzes and interprets the data based on preset parameter indicators, summarizes and outputs qualified product data, and draws a data envelopment diagram for the batch products. Finally, the qualified product data is fed to the AI for learning, and the unqualified products are sent to the data troubleshooting AI component. Step S5: The data fault repair AI component will analyze the fault cause corresponding to the fault phenomenon based on previous cases, automatically generate a fault tree, conduct troubleshooting according to the fault tree, and finally provide maintenance recommendations; Step S6: The data troubleshooting AI component analyzes the qualified product data and analyzes the data of products that are marked as "qualified" because they do not exceed the calibrated correct data range, but are significantly different from other products in the same batch. Based on previous cases, it provides case-by-case repair recommendations. Step S7: issuing a data test report based on the maintenance recommendations made in step S6.

[0032] Wherein, the step S4 includes: Step S401: Execute data testing and analysis steps; Step S402: executing data integration step; Step S403: executing data output step; Furthermore, the step S401 specifically means that, assuming that the number of the first batch of weapons and equipment is 500 sets, first, the Excel file is opened through the C# main program, and the relevant initialization settings are completed; then, the built-in If function of Excel is used to set the implementation example to judge the legitimacy of the data; finally, 500 operations are executed through the For loop program, thereby realizing the interpretation of all the test data of these 500 sets of weapons and equipment, and further carrying out analysis and screening work.

[0033] Furthermore, the step S402 specifically refers to, first, opening the 500 Excel files of the 500 sets of products in step S401 one by one through the corresponding code, then copying and summarizing the key data therein into a data summary table, and finally analyzing to obtain the envelope and data distribution of the test data of the entire batch.

[0034] Furthermore, the step S403 specifically refers to drawing the data in the data summary table summarized in step S402 into an envelope diagram by setting C# code.

[0035] Preferably, as shown in the attached Figure 4 As shown in the figure, in a specific scenario, the cabinet power is first turned on to start the equipment, and then data collection begins. After collection is complete, the interpretation criteria for data indicators are entered by clicking the "Configure" button, which then leads to the data analysis phase. Clicking the "Analyze" button allows for in-depth data analysis using Matlab algorithms. The analyzed data is then evaluated for eligibility. If the data passes, the data integration phase proceeds directly. Conversely, if the data fails, the erroneous data is stored in a designated directory for subsequent troubleshooting and correction. After data integration is complete, clicking the "Summarize" button aggregates all data into a single table using Excel software, enabling centralized data management and display. Finally, a detailed data report and intuitive envelope chart are generated based on the aggregated data, providing a strong basis for decision-making, thus completing the entire process.

[0036] The present invention also provides an onboard computer AI testing and maintenance system based on big data learning. The onboard computer AI testing and maintenance system based on big data learning can be implemented by executing the process steps of the onboard computer AI testing and maintenance method based on big data learning, that is, those skilled in the art can understand the onboard computer AI testing and maintenance method based on big data learning as an optimal implementation of the onboard computer AI testing and maintenance system based on big data learning.

[0037] The above describes specific embodiments of the present invention. It should be understood that the present invention is not limited to the specific embodiments described above, and those skilled in the art may make various changes or modifications within the scope of the claims, which do not affect the essence of the present invention. The embodiments of this application and the features in the embodiments may be combined with each other in any manner unless there is a conflict.

Claims

1. A missile-mounted computer AI testing and maintenance system based on big data learning, characterized in that: include: Hardware data acquisition module, software data testing and analysis fault repair module; The software data testing and analysis fault repair module can input the missile data collected by the hardware data acquisition module in parallel, and then analyze, interpret and summarize the missile data, and perform preliminary repairs on unqualified products.

2. The AI testing and maintenance system for missile-mounted computers based on big data learning according to claim 1 is characterized in that: The hardware data acquisition module is used to test the onboard computer product and collect onboard data; The hardware data acquisition module includes: a power supply control unit, a timing control unit, and a data acquisition unit.

3. The AI testing and maintenance system for missile-mounted computers based on big data learning according to claim 2 is characterized in that: The power supply control unit is used to supply power to the missile computer intelligent AI test and maintenance system; The power supply control unit includes: a ripple suppression component, which can effectively reduce the impact of ripple and noise on subsequent accuracy data and rate data acquisition, while improving the stability of the system's long-term operation; The power supply control unit adopts a plug-in slide design to balance the force during plugging and unplugging, thereby avoiding distortion, shrinkage and breakage of the connector core pins due to uneven plugging and unplugging force or shaking and twisting during plugging and unplugging; the power supply control unit adopts a two-level buffer overcurrent protection design. Specifically, when the load current exceeds the first warning line, the voltage stabilizer in the device continues to work, and the onboard computer data processing system enters a half-load working mode. When the load current exceeds the second current warning line, the voltage stabilizer in the device is interrupted, and the onboard computer data processing system stops working.

4. The AI testing and maintenance system for missile-mounted computers based on big data learning according to claim 2 is characterized in that: The timing control unit is capable of processing cross-clock domains of 8 channels; The timing control unit adopts multiple measures to solve the problems of metastability, clock offset and jitter, and the multiple measures include: asynchronous FIFO buffer mode, bilateral handshake protocol confirmation mode, and clock domain crossing mode.

5. The AI testing and maintenance system for missile-mounted computers based on big data learning according to claim 2 is characterized in that: The data acquisition unit is capable of performing analog digital data sampling, digital analog data sampling, and temperature and humidity data sampling; The data acquisition unit includes: a noise suppression component, a digital processing component, and a temperature drift compensation component. The data acquisition unit can effectively improve the reliability of the collected data; Under the control of the data acquisition unit, the analog digital data, digital analog data, and temperature and humidity data are transmitted to the noise suppression component respectively. After the digital processing component receives the data from the noise suppression component, it summarizes and sends it to the temperature drift compensation component.

6. The AI testing and maintenance system for missile-mounted computers based on big data learning according to claim 1 is characterized in that: The software data testing and analysis fault repair module is used to analyze, interpret, and summarize the sampled data, and perform preliminary repairs on unqualified products; the software data testing and analysis fault repair module includes: a main control program component, a data testing and analysis component, and a data fault repair AI component; after the main control program component receives the parallel input data, it sends the processed product data to the data testing and analysis component, and sends the analysis of the fault cause and preliminary repair suggestions to the data fault repair AI component; after the data testing and analysis component receives the processed product data from the main control program component, it returns the correct data to the main control program component, and at the same time sends the correct product data to the data fault repair AI component to feed the AI, and obtains incorrect product data from the data fault repair AI component; the data fault repair AI component receives the analysis of the fault cause and preliminary repair suggestions from the main control program component, receives the correct product data from the data testing and analysis component to feed the AI, and obtains incorrect product data from the data testing and analysis component; The main control program component is used for the overall operation of the system and the scheduling and coordination of various functions. The main control program component has the function of communicating and transmitting data with the onboard computer via the CAN bus, and can test the working status of received instructions to determine whether the instructions are received normally. The main control program component is responsible for global resource mobilization and can monitor system resources in real time. When resources are surplus, it can perform data analysis with high computational workload, which can effectively improve testing efficiency. The data test and analysis component can analyze, interpret and integrate the collected data in batches; the data test and analysis component can realize cascade calls to multiple software, including Matlab and Excel; The data test analysis component calls Excel through the C# main program, operates Excel, automatically fills in the qualification judgment basis of various indicators, and makes a preliminary interpretation of the test data, filters the data of unqualified products, and records the numbers; The data fault repair AI component is capable of deep learning test big data, preliminary troubleshooting of data faults, and early situation-based maintenance recommendations for faults that have not yet occurred. The data fault repair AI component adopts a supervised learning model. The data fault repair AI component uses an algorithm that combines logistic regression and decision trees, and is capable of self-iteration and continuous learning. The data fault repair AI component first optimizes the time for AI to search the database through an algorithm and understands the user's query intent through natural language processing. The data fault repair AI component includes a set of query decision instructions. The data fault repair AI component first performs data integration, conversion, and anomaly detection through intelligent AI, and then analyzes abnormal data through a regression algorithm to establish a fault tree including probability distribution to help users make maintenance decisions. The fault tree includes: excessive AD error, failure of the AD device itself, abnormal AD device power supply, failure of the power supply ripple suppression module, abnormal digital signal processing algorithm, DC / DC abnormality, power supply chip abnormality, and voltage regulator abnormality. The data fault repair AI component can issue early warnings for abnormal and excessive data and provide situation-based maintenance recommendations.

7. The AI testing and maintenance system for missile-mounted computers based on big data learning according to claim 6 is characterized in that: Under the control of the data troubleshooting AI component, the input data is processed through natural language processing and keyword extraction is performed to determine whether it is a legal natural language; if it is determined to be a legal natural language, keyword retrieval is performed and mapped to the database primary key while establishing a logic tree. The judgment will enter the recursive logic processing mode, in which iterative learning is performed to adjust the branch weights and update the decision instruction set, and an action is taken, and then it is determined whether the action is correct; if the action is correct, iterative learning is performed again to adjust the branch weights and update the decision instruction set. If the action is incorrect, it is determined again whether the query is understood correctly; if the query is determined to be incorrect before the recursive logic processing mode, it is re-determined whether the query is understood correctly; if it is determined not to be a legal natural language after the initial keyword extraction, it is terminated directly.

8. A method for AI testing and maintenance of missile-mounted computers based on big data learning, characterized in that: include: Step S1: First, turn on the power switch, then make the device current self-check, and finally make the power supply control unit and the timing control unit start working; Step S2: First, the main control software is started, then the one-key test function is run, and finally the hardware data acquisition module starts working and transmits the collected data to the main control program component through the controller local area network communication; Step S3: instructing the main control program component in step S2 to send the collected data to the software data test analysis and troubleshooting module, and call the data interpretation parameter setting; Step S4: The software data testing, analysis, and troubleshooting module analyzes and interprets the data based on preset parameter indicators, summarizes and outputs qualified product data, and draws a data envelopment diagram for the batch products. Finally, the qualified product data is fed to the AI for learning, and the unqualified products are sent to the data troubleshooting AI component. Step S5: The data fault repair AI component will analyze the fault cause corresponding to the fault phenomenon based on previous cases, automatically generate a fault tree, conduct troubleshooting according to the fault tree, and finally provide maintenance recommendations; Step S6: The data troubleshooting AI component analyzes the qualified product data and analyzes the data of products that are marked as qualified because they do not exceed the calibrated correct data range, but are significantly different from other products in the same batch. Based on previous cases, it provides repair recommendations based on the situation. Step S7: issuing a data test report based on the maintenance recommendations made in step S6.

9. The AI testing and maintenance method for missile-mounted computers based on big data learning according to claim 8 is characterized in that: The step S4 comprises: Step S401: Execute data testing and analysis steps; Step S402: executing data integration step; Step S403: executing data output step; Furthermore, step S401 specifically refers to, assuming that the number of the first batch of weapons and equipment is 500 sets, first, opening the Excel file through the C# main program and completing the relevant initialization settings; then, using the built-in If function of Excel to set the embodiment, thereby interpreting the legitimacy of the data; finally, executing the operation 500 times through the For loop program, thereby interpreting all the test data of the 500 sets of weapons and equipment, and carrying out analysis and screening work; Furthermore, step S402 specifically refers to, first, opening the 500 Excel files of the 500 sets of products in step S401 one by one through the corresponding code, then copying and summarizing the key data therein into a data summary table, and finally analyzing and obtaining the envelope and data distribution of the test data of the entire batch; Furthermore, the step S403 specifically refers to drawing the data in the data summary table summarized in step S402 into an envelope diagram by setting C# code.

10. A missile-mounted computer AI testing and maintenance equipment based on big data learning, characterized in that: By adopting the onboard computer AI testing and maintenance system based on big data learning as described in any one of claims 1 to 7, a significant improvement in testing efficiency and performance is achieved through onboard computer AI testing and maintenance based on big data learning.

Images