A method and system for real-time monitoring and self-repair of vehicle software performance

Abstract

This invention relates to the field of data processing technology, and particularly to a method and system for real-time monitoring and self-repair of vehicle software performance. The method involves monitoring vehicle software memory to obtain monitored memory information, dividing and obtaining key memory information and service software memory information, and performing memory stress analysis on key software to obtain an initial software stress rate. Road condition information of the road surface where the vehicle is located is collected and weighted to obtain repair weights and service weights. A first repair scheme for memory cleanup and repair is randomly set. Based on the repair weights, service weights, and initial software stress rate, a first repair fitness is analyzed and calculated, and iterative memory repair optimization is performed to obtain an optimal repair scheme. The optimization strategy is set based on road condition information. The vehicle software memory is cleaned and repaired according to the optimal repair scheme. By implementing this invention, intelligent and dynamic management of vehicle software memory can be achieved.

Description

Technical Field

[0001] This invention relates to the field of data processing technology, and in particular to a method and system for real-time monitoring and self-repair of vehicle software performance. Background Technology

[0002] Current in-vehicle memory management systems mostly employ fixed threshold cleanup or manual cleanup modes, only monitoring memory usage without prioritizing critical software versus service software, and without considering differentiated management based on driving scenarios. Problems with existing technologies include: memory cleanup lacks targeting, failing to adapt to varying road conditions and critical software needs, lacking quantitative analysis and iterative optimization mechanisms, and inaccurate assessment of critical software stress levels, easily leading to over- or under-fixing, or impacting user experience. Summary of the Invention

[0003] This invention addresses the problems in existing technologies, such as the lack of targeted memory cleanup, inability to adapt to the different needs of critical software under different road conditions, and the lack of quantitative analysis and iterative optimization mechanisms. It provides a method and system for real-time monitoring and self-repair of vehicle software performance.

[0004] The technical solution of the present invention to solve the above-mentioned technical problems is as follows:

[0005] In a first aspect, the present invention provides a method for real-time monitoring and self-repair of vehicle software performance, comprising: monitoring vehicle software memory to obtain monitored memory information, dividing and obtaining key memory information and service software memory information, and performing memory stress analysis of key software to obtain an initial software stress rate; collecting road condition information of the road surface where the vehicle is located, performing weight allocation to obtain repair weights and service weights; randomly setting a first repair scheme for memory cleaning and repair, analyzing and calculating a first repair fitness based on the repair weights, service weights, and initial software stress rate, and performing iterative memory repair optimization to obtain an optimal repair scheme, wherein an optimization strategy is set based on the road condition information; and cleaning and repairing the vehicle software memory according to the optimal repair scheme.

[0006] Optionally, vehicle software memory monitoring is performed to obtain monitored memory information, key memory information and service software memory information are divided, and memory stress analysis of key software is performed to obtain an initial software stress rate. This includes: during the operation of vehicle software, vehicle software memory monitoring is performed to obtain monitored memory information, wherein the monitored memory information includes the running software and the memory used by each running software; the monitored memory information is divided according to preset key software and service software categories to obtain key memory information and service software memory information; based on the key memory information and service software memory information, memory stress analysis of key software is performed to obtain an initial software stress rate.

[0007] The process includes: performing memory stress analysis on key software based on the key memory information and service software memory information to obtain an initial software stress rate; acquiring a key software stress analyzer trained using machine learning, wherein the key software stress analyzer is obtained by convergence of a sample set of key memory information, a sample set of service software memory information, and sample software stress rate supervised training values; each sample software stress rate includes different sample key memory information and the percentage of times the key software response time exceeds a response time threshold under the sample software memory information; inputting the key memory information and service software memory information into the key software stress analyzer, and outputting the initial software stress rate.

[0008] Optionally, road condition information of the road surface where the vehicle is located is collected, and weighted allocation is performed to obtain repair weight and service weight. This includes: collecting road condition information of the road surface where the vehicle is located, wherein the road condition information includes vehicle density; and performing weighted allocation based on the road condition information to obtain repair weight and service weight.

[0009] The process of allocating weights based on the road condition information to obtain repair weights and service weights includes: obtaining extreme road condition information; calculating repair weights based on the road condition information and the extreme road condition information; and calculating service weights based on the repair weights.

[0010] Optionally, a first repair scheme for memory cleanup and repair is randomly set. Based on the repair weight, service weight, and initial software stress rate, a first repair fitness is calculated. This includes: constructing a memory repair space based on the service software memory information; randomly setting a first repair scheme within the memory repair space, wherein the first repair scheme includes the proportion of service software memory cleaned and repaired; obtaining first repair service software memory information under the first repair scheme based on the service software memory information; performing memory stress analysis based on the first repair service software memory information and the key memory information to obtain a first software stress rate, and calculating the first repair fitness in conjunction with the repair weight and service weight.

[0011] The process includes: performing memory stress analysis based on the memory information of the first repair service software and the key memory information to obtain a first software stress rate; and calculating a first repair fitness by combining the repair weight and service weight. This includes: inputting the memory information of the first repair service software and the service software memory information into a key software stress analyzer to output the first software stress rate; calculating a first service attenuation rate based on the first repair service software memory information and the service software memory information; calculating a first stress attenuation rate based on the first software stress rate and the initial software stress rate; and weighting the first stress attenuation rate and the first service attenuation rate according to the repair weight and service weight to obtain the first repair fitness.

[0012] Optionally, iterative memory repair optimization is performed to obtain the optimal repair scheme, including: based on the first stress decay rate and the first service decay rate analyzed and processed according to the first repair scheme, and combined with the road condition information, the optimization step size is configured to obtain the adjustment step size; the first repair scheme is adjusted using the adjustment step size to obtain the second repair scheme, and the second repair fitness is obtained; the adjustment step size is configured and iterative optimization is performed until convergence, and the optimal repair scheme with the maximum repair fitness is obtained.

[0013] Specifically, based on the first stress attenuation rate and the first service attenuation rate analyzed and processed according to the first repair scheme, and combined with the road condition information, an optimization step size configuration is performed to obtain an adjustment step size, including: obtaining a preset step size; calculating a first step size adjustment coefficient based on the first stress attenuation rate and the first service attenuation rate; calculating a second step size adjustment coefficient based on the road condition information and average road condition information, and calculating a fusion step size adjustment coefficient based on the first step size adjustment coefficient; and using the fusion step size adjustment coefficient to adjust the preset step size to obtain the adjustment step size.

[0014] Secondly, the present invention provides a real-time monitoring and self-repair system for vehicle software performance, comprising:

[0015] The memory stress analysis module is used to monitor the memory of in-vehicle software, obtain monitored memory information, classify and obtain key memory information and service software memory information, and perform memory stress analysis on key software to obtain the initial software stress rate.

[0016] The weight allocation module is used to collect road condition information of the road surface where the vehicle is located, perform weight allocation, and obtain repair weight and service weight.

[0017] The optimal repair solution acquisition module is used to randomly set a first repair solution for memory cleanup and repair, analyze and calculate the first repair fitness based on the repair weight, service weight, and initial software stress rate, and perform iterative memory repair optimization to obtain the optimal repair solution, wherein the optimization strategy is set based on the road condition information.

[0018] The cleaning and repair module is used to clean and repair the vehicle software memory according to the optimal repair scheme.

[0019] By implementing this invention, it is possible to monitor vehicle software memory, obtain monitored memory information, classify and obtain key memory information and service software memory information, perform memory stress analysis on key software, obtain the initial software stress rate, accurately grasp the current status of memory resource allocation, clearly define the degree of stress of key software on service software memory occupation, provide quantitative basis for subsequent memory repair, avoid blindly cleaning memory, and ensure that repair actions are highly targeted.

[0020] By implementing this invention, road condition information of the road surface where the vehicle is located can be collected, weighted, and repair weight and service weight can be obtained. Memory repair is bound to the actual driving scenario. When the road conditions are complex, the repair weight can be increased to prioritize the operation of critical software. When the road conditions are flat, the service weight can be appropriately increased to take into account the user experience, making the weight allocation more in line with driving needs and improving the rationality of the repair strategy.

[0021] By implementing this invention, a first repair scheme for randomly setting memory cleanup and repair can be achieved. Based on the repair weight, service weight, and initial software stress rate, a first repair fitness is calculated and analyzed. Iterative memory repair optimization is then performed to obtain the optimal repair scheme. In this process, an optimization strategy is set based on the road condition information. Through random scheme generation and iterative optimization, getting stuck in local optima is avoided. Combined with dynamic adjustment of the optimization strategy based on road conditions, the repair scheme can minimize the stress rate of critical software and minimize the degradation of service software functions, thus achieving a balance between repair effectiveness and service retention.

[0022] By implementing this invention, the vehicle software memory can be cleaned and repaired according to the optimal repair scheme, avoiding invalid memory cleanup operations. While ensuring the response speed of critical software, the impact on the user experience of service software is minimized. The repair process is efficient and accurate, and no manual intervention from the user is required.

[0023] In summary, by implementing this invention, intelligent and dynamic management of in-vehicle software memory can be achieved. It can identify memory stress risks in critical software in real time and dynamically adjust memory repair strategies based on driving conditions. While prioritizing the stable operation of core driving functions such as navigation, it also maximizes the consideration of user needs for service software, effectively solving the problem of slow response of critical software due to insufficient memory when multiple services are running in the vehicle. This improves the stability, security, and user experience of the in-vehicle software system. Attached Figure Description

[0024] Figure 1 A flowchart illustrating a method for real-time monitoring and self-repair of vehicle software performance provided by the present invention;

[0025] Figure 2 This is a schematic diagram of the structure of a real-time monitoring and self-repair system for vehicle software performance provided by the present invention.

[0026] In the attached diagram, the components represented by each number are as follows:

[0027] Memory stress analysis module 11, weight allocation module 12, optimal repair solution acquisition module 13, and cleanup and repair module 14. Detailed Implementation

[0028] The technical solutions of the embodiments of the present invention will be clearly and completely described below with reference to the accompanying drawings. Obviously, the described embodiments are only some embodiments of the present invention, and not all embodiments. Based on the embodiments of the present invention, all other embodiments obtained by those skilled in the art without creative effort are within the scope of protection of the present invention.

[0029] In the description of this invention, the terms "first" and "second" are used for descriptive purposes only and should not be construed as indicating or implying relative importance or implicitly specifying the number of indicated technical features. Thus, a feature defined as "first" or "second" may explicitly or implicitly include one or more features. In the description of this invention, "a plurality of" means two or more, unless otherwise explicitly specified.

[0030] In the description of this invention, the term "for example" is used to mean "used as an example, illustration, or description." Any embodiment described as "for example" in this invention is not necessarily to be construed as being more preferred or advantageous than other embodiments. The following description is provided to enable any person skilled in the art to make and use the invention. Details are set forth in the following description for purposes of explanation. It should be understood that those skilled in the art will recognize that the invention can be made without using these specific details. In other instances, well-known structures and processes will not be described in detail to avoid obscuring the description of the invention with unnecessary detail. Therefore, the invention is not intended to be limited to the embodiments shown, but is consistent with the broadest scope of the principles and features disclosed herein.

[0031] Example 1, as Figure 1 As shown, this embodiment of the invention provides a method for real-time monitoring and self-repair of vehicle software performance, including:

[0032] S100: Perform onboard software memory monitoring, obtain monitored memory information, classify and obtain key memory information and service software memory information, and perform memory stress analysis of key software to obtain the initial software stress rate;

[0033] S200: Collects road condition information of the road surface where the vehicle is located, performs weight allocation, and obtains repair weight and service weight;

[0034] S300: Randomly set a first repair scheme for memory cleanup and repair, analyze and calculate the first repair fitness based on the repair weight, service weight, and initial software stress rate, and perform iterative memory repair optimization to obtain the optimal repair scheme, wherein the optimization strategy is set according to the road condition information;

[0035] S400: Clean and repair the vehicle software memory according to the optimal repair solution.

[0036] In step S100 of this embodiment, vehicle software memory monitoring is performed to obtain monitored memory information, key memory information and service software memory information are obtained by segmentation, and memory stress analysis of key software is performed to obtain an initial software stress rate, including:

[0037] During the operation of the vehicle software, the vehicle software memory is monitored to obtain the monitored memory information, which includes the running software and the memory used by each running software.

[0038] The monitored memory information is divided according to the preset categories of key software and service software to obtain key memory information and service software memory information;

[0039] Based on the key memory information and service software memory information, a memory stress analysis of the key software is performed to obtain the initial software stress rate.

[0040] In this embodiment of the application, the purpose of step S100 is to fully understand the memory usage of the vehicle software, accurately distinguish the memory usage of key software and service software, and then analyze the probability that the key software will experience a decrease in response speed due to memory resource occupation. This provides basic data support for the subsequent formulation of targeted memory cleaning and repair solutions, avoids the impact on the response speed of key software such as navigation and 360-degree imaging, and ensures driving safety.

[0041] To achieve the above objectives, it is first necessary to monitor the vehicle software memory during its operation to obtain the monitored memory information, which includes the running software and the memory used by each running software.

[0042] This involves monitoring the memory usage of in-vehicle software during operation and obtaining memory information. This information includes all currently running software and the memory usage of each individual application. For example, if the in-vehicle system is currently running navigation software, 360° camera software, music player software, and video software, monitoring might show that the navigation software uses 500MB of memory, the 360° camera software uses 800MB, the music player software uses 300MB, and the video software uses 600MB.

[0043] Next, the monitored memory information needs to be divided according to the preset key software and service software categories to obtain key memory information and service software memory information.

[0044] Based on preset categories of key software and service software, the monitored memory information is divided to obtain key memory information and service software memory information.

[0045] Key software may include navigation, 360-degree imaging and other driving safety-related software, while service software may include entertainment software that is not essential for driving, such as movie and music software.

[0046] Based on the examples above, the key memory information is that the navigation software uses 500MB of memory and the 360 ​​camera software uses 800MB of memory. The service software memory information is that the music player software uses 300MB of memory and the video software uses 600MB of memory.

[0047] Then, based on the key memory information and service software memory information, a memory stress analysis of the key software needs to be performed to obtain the initial software stress rate.

[0048] In step S100 of this embodiment, based on the key memory information and service software memory information, a memory stress analysis of the key software is performed to obtain an initial software stress rate, including:

[0049] A key software stress analyzer trained based on machine learning is obtained, wherein the key software stress analyzer is obtained by the convergence of sample key memory information set, sample service software memory information set and sample software stress rate supervised training values. Each sample software stress rate includes different sample key memory information and the proportion of times the key software response time is greater than the response time threshold under the sample software memory information.

[0050] The critical memory information and service software memory information are input into the critical software stress analyzer, and the initial software stress rate is output.

[0051] In this embodiment of the application, the purpose of the above-mentioned detailed steps in step S100 is to accurately quantify the probability of response delay of key software due to memory resources being occupied by service software, that is, the initial software stress rate, through professional analysis tools. This provides core data basis for subsequent judgment on whether memory repair is needed and for formulating repair plans, so as to avoid affecting the response speed of key software such as navigation and 360-degree imaging, and ensure the normal operation of driving-related functions.

[0052] To achieve the above steps, it is first necessary to obtain a key software stress analyzer trained based on machine learning. The key software stress analyzer is obtained by converging sample key memory information sets, sample service software memory information sets, and sample software stress rate supervised training values. Each sample software stress rate includes different sample key memory information and the percentage of times the key software response time exceeds the response time threshold under different sample software memory information.

[0053] Specifically, the set of key memory information samples is the core sample input for training the key software stress analyzer. It refers to the collection of memory usage data of multiple sets of key vehicle software. For example, the set includes memory usage values ​​of different key software such as navigation software (420MB memory usage), 360 camera software (780MB memory usage), and dashcam software (150MB memory usage). Each set of values ​​is independent key memory information samples, which are combined to form this set.

[0054] The set of sample service software memory information is another core sample input for training the critical software stress analyzer. It refers to the collection of memory usage data from multiple sets of in-vehicle service software. For example, this set includes memory usage values ​​for different service software such as music playback software (280MB memory usage), video playback software (620MB memory usage), and radio software (90MB memory usage). Each set of values ​​represents independent sample service software memory information, which are then combined to form this set.

[0055] The supervised training value of the sample software stress rate is the core supervised label value for training the critical software stress analyzer. It is a fixed reference judgment value and corresponds one-to-one with the two types of sample sets mentioned above. Each sample software stress rate supervised training value is the percentage of times the critical software response time exceeds the response time threshold under different sample critical memory information and sample service software memory information.

[0056] The response time threshold is the core benchmark parameter for determining the stress rate of sample software, representing the critical time threshold for the normal response of critical vehicle software. For example, setting the response time threshold to 0.6 seconds means that if the response time of a single operation of critical software exceeds 0.6 seconds, it is determined that the operation has timed out.

[0057] Critical software response time is a fundamental parameter for calculating the stress rate of sample software. It refers to the actual time taken for onboard critical software to complete loading and provide normal functionality after receiving a running command. For example, if navigation software takes 0.7 seconds to complete route planning and display after initiating a route planning command, this value is the critical software response time for that instance.

[0058] The initial software stress rate is the final output parameter of the critical software stress analyzer after completing its analysis. It is generated by inputting critical memory information and service software memory information into the analyzer. Essentially, it represents the percentage of times the critical software response time exceeds a threshold under the current memory conditions. For example, an output initial software stress rate of 35% means that, under the current memory conditions, the number of times the critical software response times out accounts for 35% of the total number of runs.

[0059] In building the critical software stress analyzer, the gradient boosting regression algorithm in supervised machine learning can be selected as the optimal method for building the critical software stress analyzer.

[0060] Optionally, in the parameter settings of the key software stress analyzer, the number of decision tree base learners is set to 300. The learning rate is set to 0.01. The maximum depth of the decision tree is set to 6 layers, and the minimum number of sample splits per decision tree is set to 20. The squared loss function is selected. The regularization coefficient is set to 0.02.

[0061] In training the critical software stress analyzer, the formal training rounds were set to 200 rounds. The training data consisted of the sample critical memory information set, the sample service software memory information set, and the sample software stress rate supervised training values, obtained through full memory monitoring and operational response data collection under real-world vehicle software operation scenarios. The model training loss value, calculated using the squared loss function, was considered converged if it remained stable below 0.001 for thirty consecutive training rounds, thus obtaining the critical software stress analyzer.

[0062] Next, the critical memory information and service software memory information need to be input into the critical software stress analyzer to output the initial software stress rate.

[0063] The critical memory information and service software memory information obtained from the aforementioned steps are then input into the pre-trained critical software stress analyzer. The critical software stress analyzer uses its internally trained model and algorithms to calculate and analyze the input memory information, and the final output is the initial software stress rate.

[0064] For example, the key memory information is that the navigation software uses 500MB of memory and the 360 ​​imaging software uses 800MB of memory. The service software memory information is that the music software uses 300MB of memory and the video software uses 600MB of memory. After inputting this information into the key software stress analyzer, the key software stress analyzer outputs a result of 25%. The 25% is the initial software stress rate in the current scenario, which means that the number of times the key software response time exceeds the preset threshold accounts for 25% of the total number of runs.

[0065] In step S200 of this application embodiment, road condition information of the road surface where the vehicle is located is collected, weighted and allocated to obtain repair weight and service weight, including:

[0066] Collect road condition information of the road surface where the vehicle is located, including vehicle density;

[0067] Based on the road condition information, weights are allocated to obtain repair weights and service weights.

[0068] In step S200 of this application embodiment, the purpose of the above steps is to reasonably allocate repair weights and service weights based on the actual road conditions of the vehicle, so that the subsequently formulated memory repair plan can adapt to different road condition requirements. The more complex the road conditions, the more priority is given to ensuring the operation of critical software such as navigation and 360-degree imaging, while balancing the user experience of service software, avoiding the impact of excessive memory cleanup on the use of non-critical functions, and ultimately ensuring a balance between driving safety and user experience.

[0069] To achieve the above objectives, it is first necessary to collect road condition information of the road surface where the vehicle is located, including the key indicator of vehicle density.

[0070] For example, during the morning rush hour in a city, if there are 80 cars per kilometer on a road segment, this 80 cars per kilometer is the current vehicle density information, which is one of the collected road condition information.

[0071] Next, based on the road condition information, weights need to be allocated to obtain repair weights and service weights.

[0072] In step S200 of this embodiment, weight allocation is performed based on the road condition information to obtain repair weights and service weights, including:

[0073] Obtain information on extreme road conditions;

[0074] Based on the road condition information and extreme road condition information, the repair weight is calculated, and based on the repair weight, the service weight is calculated.

[0075] The extreme road condition information is a preset reference standard representing extremely complex road conditions. For example, the vehicle density under extreme road conditions is set at 120 vehicles per kilometer. This value is a clear example of extreme road condition information, and the vehicle density can be obtained through vehicle radar or navigation platforms.

[0076] Then, based on the collected road condition information and preset extreme road condition information, the repair weight is calculated. The repair weight is used to measure the priority of memory repair; the closer the road condition is to the extreme road condition, the higher the repair weight. For example, if the current vehicle density is 80 vehicles per kilometer and the extreme road condition vehicle density is 120 vehicles per kilometer, the repair weight is calculated using a preset algorithm. In the example above, the repair weight is calculated as 80 / 120≈0.67 based on the ratio of the current road condition parameters to the extreme road condition parameters.

[0077] Then, the service weight is calculated based on the repair weight. The sum of the repair weight and the service weight is usually 1, and the two have an inverse relationship.

[0078] Continuing with the example above, if the repair weight is 0.67, then the service weight is 1 minus 0.67, which is 0.33. This means that when formulating a memory repair plan, 67% of the priority is given to ensuring the memory requirements of critical software, while 33% of the priority is given to considering the user experience of service software.

[0079] In step S300 of this embodiment, a first repair scheme for memory cleanup and repair is randomly set. Based on the repair weight, service weight, and initial software stress rate, a first repair fitness is calculated and analyzed, including:

[0080] Based on the service software memory information, a memory repair space is constructed;

[0081] A first repair scheme is randomly set within the memory repair space, wherein the first repair scheme includes cleaning up and repairing the memory ratio of the service software;

[0082] Based on the service software memory information, obtain the first repair service software memory information under the first repair scheme;

[0083] Memory stress analysis is performed based on the memory information of the first repair service software and the key memory information to obtain the first software stress rate. Combined with the repair weight and service weight, the first repair fitness is calculated.

[0084] In this embodiment, the purpose of step S300 is to construct a reasonable memory repair space, randomly generate an initial memory cleanup and repair scheme, and then combine the repair weight, service weight, and memory stress analysis results to quantify the suitability of the scheme, i.e., the first repair fitness, laying the foundation for subsequent iterative optimization to find the optimal repair scheme. This ensures that the repair scheme can reduce memory stress on critical software and minimize the impact on the user experience of service software, making the scheme more in line with actual needs.

[0085] To achieve the above objectives, firstly, a memory repair space needs to be constructed based on the service software's memory information.

[0086] This involves constructing a memory repair space based on the service software's memory information. The memory repair space is defined as the boundary of the area where memory can be cleaned up, based on the total memory usage of all service software.

[0087] For example, if the service software's memory information shows that the music software uses 300MB of memory, the video software uses 600MB of memory, and the total memory of the service software is 900MB, then the memory repair space is the cleanable range corresponding to 0 to 900MB. This step clarifies the memory value range for subsequent cleanup and repair.

[0088] Then, a first repair scheme needs to be randomly set within the memory repair space, wherein the first repair scheme includes cleaning up and repairing the memory ratio of the service software.

[0089] This involves randomly setting a first repair plan within the memory repair space. The core of this first repair plan is to clean up and repair the memory usage of service software. For example, within the aforementioned 0 to 900 MB memory repair space, randomly determining that the memory usage of service software to be cleaned up and repaired is 30% is a specific first repair plan.

[0090] Next, based on the service software memory information, it is necessary to obtain the first repair service software memory information under the first repair scheme.

[0091] Continuing with the example above, the total memory usage of the service software is 900MB. The first repair plan cleans up 30% of that memory, so the amount of memory cleaned is 900MB × 30% = 270MB. Assuming we prioritize cleaning up the video software's memory, then the memory information for the first repair of the service software would be: music software using 300MB of memory, and video software using 600MB - 270MB = 330MB of memory.

[0092] In step S300 of this embodiment, memory stress analysis is performed based on the memory information of the first repair service software and the key memory information to obtain a first software stress rate. Combined with the repair weight and service weight, a first repair fitness is calculated, including:

[0093] Input the memory information of the first repair service software and the service software memory information into the critical software stress analyzer, and output the first software stress rate.

[0094] The first service attenuation rate is calculated based on the memory information of the first repair service software and the memory information of the service software.

[0095] The first stress attenuation rate is calculated based on the first software stress rate and the initial software stress rate.

[0096] Based on the repair weight and service weight, the first stress attenuation rate and the first service attenuation rate are weighted and calculated to obtain the first repair fitness.

[0097] In step S300 of this embodiment, the calculation of the first repair fitness quantifies the improvement effect of the first repair scheme on the memory stress of critical software, as well as the degree of impact of the scheme on the user experience of the service software. The first repair fitness is obtained through weighted calculation, thereby judging the merits of the first repair scheme and providing a clear evaluation basis for subsequent iterative optimization. This ensures that the final memory repair scheme can effectively reduce the memory stress of critical software while minimizing the negative impact on the use of the service software.

[0098] To achieve the above objectives, the first step is to input the memory information of the first repair service software and the service software memory information into the critical software stress analyzer, and output the first software stress rate.

[0099] The memory information of the first repair service software and the service software is input into a pre-trained critical software stress analyzer. The critical software stress analyzer calculates and outputs the first software stress rate through its internal model. For example, if the service software memory information is 300MB for music software and 600MB for video software, and the first repair service software memory information is 300MB for music software and 330MB for video software, after inputting these two sets of information into the critical software stress analyzer, the output result is 15%, and this 15% is the first software stress rate.

[0100] Then, based on the memory information of the first repair service software and the memory information of the service software, the first service attenuation rate is calculated.

[0101] The first service attenuation rate is the proportion by which the memory information of the first repair service software is reduced relative to the memory information of the service software.

[0102] The calculation formula is: First service attenuation rate = (Total memory of service software - Total memory of first repair service software) / Total memory of service software × 100%.

[0103] Continuing with the example above, the total memory of the service software is 300MB + 600MB = 900MB, and the total memory of the first repair service software is 300MB + 330MB = 630MB. Substituting these values ​​into the formula above, we can calculate the first service attenuation rate as (900 - 630) / 900 × 100% = 30%. Here, 30% is the first service attenuation rate.

[0104] Next, the first stress attenuation rate is calculated based on the first software stress rate and the initial software stress rate.

[0105] The first stress attenuation rate is the ratio of the initial software stress rate minus the first software stress rate to the initial software stress rate.

[0106] The calculation formula is: First stress attenuation rate = (Initial software stress rate - First software stress rate) / Initial software stress rate × 100%.

[0107] Assuming the initial software stress rate is 25%, and combined with the first software stress rate of 15% obtained above, the first stress attenuation rate can be calculated by substituting into the formula: (25%-15%) / 25%×100%=40%. Here, 40% is the first stress attenuation rate.

[0108] Finally, based on the repair weight and service weight, the first stress attenuation rate and the first service attenuation rate are weighted and calculated to obtain the first repair fitness.

[0109] Specifically, the calculation formula is: First Repair Fitness = Repair Weight × First Stress Decay Rate - Service Weight × First Service Decay Rate.

[0110] Assuming the repair weight calculated using the aforementioned steps is 0.67 and the service weight is 0.33, substituting the relevant values, we can calculate the first repair fitness as follows: 0.67 × 40% - 0.33 × 30% = 0.268 - 0.099 = 0.169 = 16.9%. That is, the first repair fitness is 16.9% or 0.169.

[0111] In step S300 of this application embodiment, iterative memory repair optimization is performed to obtain the optimal repair solution, including:

[0112] Based on the first stress attenuation rate and the first service attenuation rate analyzed and processed by the first repair scheme, and combined with the road condition information, the optimization step size configuration is performed to obtain the adjustment step size;

[0113] The first repair scheme is adjusted using the adjustment step size to obtain a second repair scheme, and the second repair fitness is obtained through processing.

[0114] Continue to adjust the step size for iterative optimization until convergence, obtaining the optimal repair scheme with the maximum repair fitness.

[0115] In this embodiment, the purpose of step S300 is to continuously iterate and optimize the repair scheme, gradually improve the repair adaptability, and ultimately find the optimal repair scheme that maximizes the balance between improving the memory stress of critical software and the user experience of service software. This allows memory repair to effectively reduce the response latency risk of critical software such as navigation and 360-degree imaging, while minimizing excessive impact on service software such as movies and music, thus adapting to the needs of real-world driving scenarios.

[0116] To achieve the above steps, it is first necessary to analyze and process the first stress attenuation rate and the first service attenuation rate based on the first repair scheme, and then optimize the step size configuration based on the road condition information to obtain the adjustment step size.

[0117] In step S300 of this embodiment, based on the first stress attenuation rate and the first service attenuation rate analyzed and processed by the first repair scheme, and combined with the road condition information, an optimization step size configuration is performed to obtain the adjustment step size, including:

[0118] Get the preset step size;

[0119] The first step length adjustment coefficient is calculated based on the first stress attenuation rate and the first service attenuation rate.

[0120] Based on the road condition information and average road condition information, the second step length adjustment coefficient is calculated, and combined with the first step length adjustment coefficient, the fusion step length adjustment coefficient is calculated.

[0121] The preset step size is adjusted using the fusion step size adjustment coefficient to obtain the adjustment step size.

[0122] In step S300 of this embodiment, the optimization step size configuration is performed by dynamically and reasonably adjusting the step size based on the actual effect of the first repair scheme and the safety requirements of the current road conditions. This allows the iterative optimization process to adjust the optimization pace according to the improvement of critical software memory stress and the degree of impact on service software, while also adapting to the road safety level. The more complex the road conditions and the less adequate the critical software repair guarantee, the larger the step size should be, thereby improving optimization efficiency.

[0123] To achieve the above objectives, the preset step size must first be obtained;

[0124] The preset step size is the basic adjustment unit for iterative optimization, used to determine the initial magnitude of the adjustment to the repair plan. For example, setting the preset step size to 8% represents the basic standard change in the proportion of service software memory cleanup during each adjustment.

[0125] Then, the first step long adjustment coefficient needs to be calculated based on the first stress attenuation rate and the first service attenuation rate.

[0126] The first-step length adjustment coefficient is the ratio of the first service attenuation rate to the first stress attenuation rate. A smaller first stress attenuation rate indicates insufficient support for the operation and repair of critical software in the first round of optimization. In this case, the first stress attenuation rate will increase the step size to improve optimization efficiency. Assuming the first service attenuation rate is 30% and the first stress attenuation rate is 40%, the first-step length adjustment coefficient is calculated to be 30% ÷ 40% = 0.75. If the first stress attenuation rate is only 20% and the first service attenuation rate remains 30%, then the first-step length adjustment coefficient is 30% ÷ 20% = 1.5.

[0127] Next, based on the road condition information and average road condition information, the second step length adjustment coefficient needs to be calculated, and combined with the first step length adjustment coefficient, the fusion step length adjustment coefficient is calculated.

[0128] The second step adjustment factor is the ratio of the road condition information to the average road condition information. The more unsafe the road conditions, the larger the second step adjustment factor, and the larger the corresponding adjustment step, resulting in higher optimization efficiency. For example, if the average road condition information shows a vehicle density of 60 vehicles per kilometer, and the currently collected road condition information shows a vehicle density of 90 vehicles per kilometer, then the second step adjustment factor is 90 vehicles ÷ 60 vehicles = 1.5. If the current vehicle density is 40 vehicles per kilometer, indicating safer road conditions, then the second step adjustment factor is 40 vehicles ÷ 60 vehicles ≈ 0.67, and the step size will be relatively smaller.

[0129] The fusion step size adjustment coefficient is calculated by combining the first step size adjustment coefficient and the second step size adjustment coefficient. For example, both can be set to a weight of 0.5, and the weighted sum can be used for calculation. In the example above, the first step size adjustment coefficient is 0.75, and the second step size adjustment coefficient is 1.5. Then the fusion step size adjustment coefficient is 0.75×0.5+1.5×0.5=1.125.

[0130] Finally, the preset step size needs to be adjusted using the fusion step size adjustment coefficient to obtain the adjustment step size.

[0131] The preset step size is adjusted using a fusion step size adjustment coefficient to obtain the final adjustment step size. Continuing with the example above where the preset step size is 8% and the fusion step size adjustment coefficient is 1.125, the adjustment step size is calculated as preset step size × fusion step size adjustment coefficient, i.e., 8% × 1.125 = 9%. This 9% is the adjustment step size used in subsequent adjustments and repair schemes.

[0132] Furthermore, the first repair scheme needs to be adjusted using the aforementioned adjustment step size to obtain a second repair scheme, and then processed to obtain a second repair fitness.

[0133] The core of the first repair plan is the percentage of service software memory cleanup. Assuming the percentage of service software memory cleanup in the first repair plan is 30%, and the previously calculated adjustment step size is 9%, then considering the optimization requirements, if it is necessary to further improve the repair guarantee of critical software, the percentage of service software memory cleanup can be increased by the aforementioned adjustment step size based on the first repair plan, i.e., 30% + 9% = 39%. This plan, which includes a 39% percentage of service software memory cleanup, is the second repair plan.

[0134] Then, the second repair plan is processed according to the complete process of calculating the first repair fitness: First, the service software memory information after repair is calculated based on the total memory of the service software. Assuming the total memory of the service software is 900MB, after cleaning 39%, the remaining memory is 900MB × (1-39%) = 549MB. If the video software memory is cleaned first, the second repair service software memory information can be 300MB for music software and 249MB for video software. Next, this second repair service software memory information and the key memory information are input into the key software stress analyzer to obtain the second software stress rate. Assuming it's 12%; then calculating the second service attenuation rate as 39%, and the second stress attenuation rate as (initial software stress rate 25% - second software stress rate 12%) / initial software stress rate 25% × 100% = 52%; finally, combining the repair weight of 0.67 and the service weight of 0.33, using the formula Second Repair Fitness = Repair Weight × Second Stress Attenuation Rate - Service Weight × Second Service Attenuation Rate, we can obtain the second repair fitness as 0.67 × 52% - 0.33 × 39% = 0.3484 - 0.1287 = 0.2197, or 21.97%.

[0135] Continue to adjust the step size for iterative optimization until convergence, obtaining the optimal repair scheme with the maximum repair fitness.

[0136] Based on the second repair scheme, the aforementioned adjustment step size configuration process is repeated, and a new adjustment step size is recalculated by combining the second stress attenuation rate of 52%, the second service attenuation rate of 39%, and the current real-time traffic information of the second repair scheme.

[0137] Assuming the new adjustment step size is 7%, the 39% service software memory cleanup percentage of the second repair plan will be adjusted to 39% + 7% = 46%, resulting in the third repair plan. The fitness of the third repair will then be calculated using the same process, assumed to be 28.3%.

[0138] This process is repeated, with each round reconfiguring and adjusting the step size based on the decay rate of the previous round's repair plan and real-time traffic conditions, adjusting the service software memory cleanup ratio, and calculating the new repair fitness.

[0139] For example, in the fourth round of adjustments, the step size was 5%, and the service software memory cleanup ratio was adjusted to 46% + 5% = 51%, with a repair adaptability of 31.5%; in the fifth round of adjustments, the step size was 3%, and the service software memory cleanup ratio was adjusted to 51% + 3% = 54%, with a repair adaptability of 32.7%; in the sixth round of adjustments, the step size was 2%, and the service software memory cleanup ratio was adjusted to 54% + 2% = 56%, with a repair adaptability of 32.9%.

[0140] Optionally, a convergence threshold of 1% can be set. That is, when the change in fitness between two adjacent iterations is less than this threshold, the iteration is considered to have converged. For example, the change in fitness from 32.9% in the sixth iteration to 32.7% in the fifth iteration is 0.2%, which is less than 1%, so the iteration stops.

[0141] From all the repair solutions generated in the iterations, the solution with the highest repair fitness is selected as the optimal repair solution. In the above iteration process, the solution with the highest repair fitness of 32.9% in the sixth round, and the solution with a service software memory cleanup ratio of 56% is the optimal repair solution.

[0142] In step S400 of this application embodiment, the vehicle software memory is cleaned and repaired according to the optimal repair scheme.

[0143] For example, assuming the optimal repair plan determines that the service software memory cleanup accounts for 56%, then the cleanup is performed based on the service software memory information.

[0144] Assuming the total memory usage of the service software is 900MB, including 300MB for music software and 600MB for video software, and calculating based on a 56% memory cleanup ratio for service software, the amount of memory to be cleaned is 900MB × 56% = 504MB. Prioritize cleaning non-essential service software with high memory usage. For example, prioritize cleaning the video software, freeing up 504MB from its 600MB memory. After cleaning, the video software will have 96MB of memory remaining, while the music software's memory usage will remain unchanged at 300MB.

[0145] After the cleanup is complete, critical software will have more available memory, ensuring its responsiveness, while service software will still function normally, achieving the expected effect of memory repair.

[0146] Example 2, as Figure 2 As shown, based on the same inventive concept as the vehicle software real-time performance monitoring and self-repair method provided in Embodiment 1, this embodiment of the invention also provides a vehicle software real-time performance monitoring and self-repair system, including:

[0147] The memory stress analysis module 11 is used to monitor the memory of the vehicle software, obtain monitored memory information, divide and obtain key memory information and service software memory information, and perform memory stress analysis on key software to obtain the initial software stress rate.

[0148] The weight allocation module 12 is used to collect road condition information of the road surface where the vehicle is located, perform weight allocation, and obtain repair weight and service weight.

[0149] The optimal repair solution acquisition module 13 is used to randomly set a first repair solution for memory cleanup and repair, analyze and calculate the first repair fitness based on the repair weight, service weight, and initial software stress rate, and perform iterative memory repair optimization to obtain the optimal repair solution, wherein the optimization strategy is set based on the road condition information.

[0150] The cleaning and repair module 14 is used to clean and repair the vehicle software memory according to the optimal repair scheme.

[0151] Furthermore, the memory stress analysis module 11 includes the following execution steps:

[0152] During the operation of the vehicle software, the vehicle software memory is monitored to obtain the monitored memory information, which includes the running software and the memory used by each running software.

[0153] The monitored memory information is divided according to the preset categories of key software and service software to obtain key memory information and service software memory information;

[0154] Based on the key memory information and service software memory information, a memory stress analysis of the key software is performed to obtain the initial software stress rate.

[0155] Specifically, based on the key memory information and service software memory information, a memory stress analysis of the key software is performed to obtain an initial software stress rate, including:

[0156] A key software stress analyzer trained based on machine learning is obtained, wherein the key software stress analyzer is obtained by the convergence of sample key memory information set, sample service software memory information set and sample software stress rate supervised training values. Each sample software stress rate includes different sample key memory information and the proportion of times the key software response time is greater than the response time threshold under the sample software memory information.

[0157] The critical memory information and service software memory information are input into the critical software stress analyzer, and the initial software stress rate is output.

[0158] Furthermore, the weight allocation module 12 includes the following execution steps:

[0159] Collect road condition information of the road surface where the vehicle is located, including vehicle density;

[0160] Based on the road condition information, weights are allocated to obtain repair weights and service weights.

[0161] The process involves assigning weights based on the road condition information to obtain repair weights and service weights, including:

[0162] Obtain information on extreme road conditions;

[0163] Based on the road condition information and extreme road condition information, the repair weight is calculated, and based on the repair weight, the service weight is calculated.

[0164] Furthermore, the optimal repair solution acquisition module 13 includes the following execution steps:

[0165] A first repair scheme is randomly set within the memory repair space, wherein the first repair scheme includes cleaning up and repairing the memory ratio of the service software;

[0166] Based on the service software memory information, obtain the first repair service software memory information under the first repair scheme;

[0167] Memory stress analysis is performed based on the memory information of the first repair service software and the key memory information to obtain the first software stress rate. Combined with the repair weight and service weight, the first repair fitness is calculated.

[0168] Specifically, memory stress analysis is performed based on the memory information of the first repair service software and the key memory information to obtain a first software stress rate. Combined with the repair weight and service weight, a first repair fitness is calculated, including:

[0169] Input the memory information of the first repair service software and the service software memory information into the critical software stress analyzer, and output the first software stress rate.

[0170] The first service attenuation rate is calculated based on the memory information of the first repair service software and the memory information of the service software.

[0171] The first stress attenuation rate is calculated based on the first software stress rate and the initial software stress rate.

[0172] Based on the repair weight and service weight, the first stress attenuation rate and the first service attenuation rate are weighted and calculated to obtain the first repair fitness.

[0173] Based on the first stress attenuation rate and the first service attenuation rate analyzed and processed by the first repair scheme, and combined with the road condition information, the optimization step size configuration is performed to obtain the adjustment step size;

[0174] The first repair scheme is adjusted using the adjustment step size to obtain a second repair scheme, and the second repair fitness is obtained through processing.

[0175] Continue to adjust the step size for iterative optimization until convergence, obtaining the optimal repair scheme with the maximum repair fitness.

[0176] Specifically, based on the first stress attenuation rate and first service attenuation rate analyzed and processed according to the first repair scheme, and combined with the road condition information, an optimization step size configuration is performed to obtain the adjustment step size, including:

[0177] Get the preset step size;

[0178] The first step length adjustment coefficient is calculated based on the first stress attenuation rate and the first service attenuation rate.

[0179] Based on the road condition information and average road condition information, the second step length adjustment coefficient is calculated, and combined with the first step length adjustment coefficient, the fusion step length adjustment coefficient is calculated.

[0180] The preset step size is adjusted using the fusion step size adjustment coefficient to obtain the adjustment step size.

[0181] It should be noted that the descriptions of each embodiment in the above embodiments have different focuses. For parts that are not described in detail in a certain embodiment, please refer to the relevant descriptions in other embodiments.

[0182] Those skilled in the art will understand that embodiments of the present invention can provide methods, systems, or computer program products. Therefore, the present invention can take the form of a completely hardware embodiment, a completely software embodiment, or an embodiment combining software and hardware aspects. Furthermore, the present invention can take the form of a computer program product embodied on one or more computer-usable storage media (including, but not limited to, disk storage, CD-ROM, optical storage, etc.) containing computer-usable program code.

[0183] This invention is described with reference to flowchart illustrations and / or block diagrams of methods, apparatus (systems), and computer program products according to embodiments of the invention. It will be understood that each block of the flowchart illustrations and / or block diagrams, as well as combinations of blocks in the flowchart illustrations and / or block diagrams, can be implemented by computer program instructions. These computer program instructions can be provided to a processor of a general-purpose computer, special-purpose computer, embedded computer, or other programmable data processing apparatus to produce a machine, such that the instructions, which execute via the processor of the computer or other programmable data processing apparatus, generate instructions for implementing the flowchart. Figure 1 One or more processes and / or boxes Figure 1 A device that provides the functions specified in one or more boxes.

[0184] These computer program instructions may also be stored in a computer-readable storage medium that can direct a computer or other programmable data processing device to function in a particular manner, such that the instructions stored in the computer-readable storage medium produce an article of manufacture including instruction means, which are implemented in a process Figure 1 One or more processes and / or boxes Figure 1The function specified in one or more boxes.

[0185] These computer program instructions may also be loaded onto a computer or other programmable data processing equipment to cause a series of operational steps to be performed on the computer or other programmable equipment to produce a computer-implemented process, thereby providing instructions that execute on the computer or other programmable equipment for implementing the process. Figure 1 One or more processes and / or boxes Figure 1 The steps of the function specified in one or more boxes.

[0186] Although preferred embodiments of the invention have been described, those skilled in the art, upon learning the basic inventive concept, can make other changes and modifications to these embodiments. Clearly, those skilled in the art can make various alterations and variations to the invention without departing from its spirit and scope. Therefore, if these modifications and variations fall within the scope of the invention and its equivalents, the invention is also intended to include these modifications and variations.

Claims

1. A method for real-time monitoring and self-repair of vehicle software performance, characterized in that, The method includes: Perform onboard software memory monitoring to obtain monitored memory information, segment and obtain key memory information and service software memory information, and conduct memory stress analysis of key software to obtain the initial software stress rate; Collect road condition information of the road surface where the vehicle is located, assign weights, and obtain repair weights and service weights; A first repair scheme for memory cleanup and repair is randomly set. Based on the repair weight, service weight, and initial software stress rate, the first repair fitness is analyzed and calculated. Iterative memory repair optimization is then performed to obtain the optimal repair scheme. Optimization strategies are set based on the road condition information. Clean and repair the vehicle software memory according to the optimal repair solution described above; This includes monitoring in-vehicle software memory to obtain monitored memory information, identifying key memory information and service software memory information, and performing memory stress analysis on key software to obtain the initial software stress rate, including: During the operation of the vehicle software, the vehicle software memory is monitored to obtain the monitored memory information, which includes the running software and the memory used by each running software. The monitored memory information is divided according to the preset categories of key software and service software to obtain key memory information and service software memory information; Based on the key memory information and service software memory information, a memory stress analysis of the key software is performed to obtain the initial software stress rate; Specifically, based on the key memory information and service software memory information, a memory stress analysis of the key software is performed to obtain an initial software stress rate, including: A key software stress analyzer trained based on machine learning is obtained, wherein the key software stress analyzer is obtained by the convergence of sample key memory information set, sample service software memory information set and sample software stress rate supervised training values. Each sample software stress rate includes different sample key memory information and the proportion of times the key software response time is greater than the response time threshold under the sample software memory information. The critical memory information and service software memory information are input into the critical software stress analyzer, and the initial software stress rate is output. This includes iterative memory repair and optimization to obtain the optimal repair solution, including: Based on the first stress attenuation rate and the first service attenuation rate analyzed and processed by the first repair scheme, and combined with the road condition information, the optimization step size configuration is performed to obtain the adjustment step size; The first repair scheme is adjusted using the adjustment step size to obtain a second repair scheme, and the second repair fitness is obtained through processing. Continue to configure and adjust the step size for iterative optimization until convergence, and obtain the optimal repair scheme with the maximum repair fitness; Specifically, based on the first stress attenuation rate and first service attenuation rate analyzed and processed according to the first repair scheme, and combined with the road condition information, an optimization step size configuration is performed to obtain the adjustment step size, including: Get the preset step size; The first step length adjustment coefficient is calculated based on the first stress attenuation rate and the first service attenuation rate. Based on the road condition information and average road condition information, the second step length adjustment coefficient is calculated, and combined with the first step length adjustment coefficient, the fusion step length adjustment coefficient is calculated. The preset step size is adjusted using the fusion step size adjustment coefficient to obtain the adjustment step size.

2. The method for real-time monitoring and self-repair of vehicle software performance according to claim 1, characterized in that, The system collects road condition information about the road surface where the vehicle is located, assigns weights to determine repair weights and service weights, including: Collect road condition information of the road surface where the vehicle is located, including vehicle density; Based on the road condition information, weights are allocated to obtain repair weights and service weights.

3. The method for real-time monitoring and self-repair of vehicle software performance according to claim 2, characterized in that, Based on the road condition information, a weight allocation is performed to obtain repair weights and service weights, including: Obtain information on extreme road conditions; Based on the road condition information and extreme road condition information, the repair weight is calculated, and based on the repair weight, the service weight is calculated.

4. The method for real-time monitoring and self-repair of vehicle software performance according to claim 1, characterized in that, A first repair plan for random memory cleanup and repair is selected. Based on the repair weight, service weight, and initial software stress rate, the first repair fitness is analyzed and calculated, including: Based on the service software memory information, a memory repair space is constructed; A first repair scheme is randomly set within the memory repair space, wherein the first repair scheme includes cleaning up and repairing the memory ratio of the service software; Based on the service software memory information, obtain the first repair service software memory information under the first repair scheme; Memory stress analysis is performed based on the memory information of the first repair service software and the key memory information to obtain the first software stress rate. Combined with the repair weight and service weight, the first repair fitness is calculated.

5. The method for real-time monitoring and self-repair of vehicle software performance according to claim 4, characterized in that, Based on the memory information of the first repair service software and the key memory information, a memory stress analysis is performed to obtain a first software stress rate. Combined with the repair weight and service weight, a first repair fitness is calculated, including: Input the memory information of the first repair service software and the service software memory information into the critical software stress analyzer, and output the first software stress rate. The first service attenuation rate is calculated based on the memory information of the first repair service software and the memory information of the service software. The first stress attenuation rate is calculated based on the first software stress rate and the initial software stress rate. Based on the repair weight and service weight, the first stress attenuation rate and the first service attenuation rate are weighted and calculated to obtain the first repair fitness.

6. A real-time monitoring and self-repair system for vehicle software performance, characterized in that, The system is used to implement the real-time monitoring and self-repair method for vehicle software performance as described in any one of claims 1-5, and the system includes: The memory stress analysis module is used to monitor the memory of in-vehicle software, obtain monitored memory information, divide and obtain key memory information and service software memory information, and perform memory stress analysis on key software to obtain the initial software stress rate. The weight allocation module is used to collect road condition information of the road surface where the vehicle is located, perform weight allocation, and obtain repair weight and service weight. The optimal repair solution acquisition module is used to randomly set a first repair solution for memory cleanup and repair. Based on the repair weight, service weight, and initial software stress rate, it analyzes and calculates the first repair fitness and performs iterative memory repair optimization to obtain the optimal repair solution. The optimization strategy is set based on the road condition information. The cleaning and repair module is used to clean and repair the vehicle software memory according to the optimal repair scheme.

Images