Safety performance detection method and system for new energy vehicle and storage medium

Abstract

The application relates to the technical field of machine learning, in particular to a safety performance detection method and system for a new energy vehicle and a storage medium, the method comprising the following steps: S1, determining a plurality of pending historical records from all historical records in a historical record combination, and performing modification processing on the plurality of pending historical records; S2, determining a historical record in the historical record combination after the modification processing, inputting historical performance data in the determined historical record into a detection model, and repeating the step in the case that there is an undetermined historical record; S3, judging whether the number of practice times is less than a threshold value of the number of practice times, stopping all steps in the case that the number of practice times is not less than the threshold value of the number of practice times, continuously judging whether a requirement for stopping practice is met in the case that the number of practice times is less than the threshold value of the number of practice times, jumping to S2 in the case that the requirement for stopping practice is not met, and stopping all steps in the case that the requirement for stopping practice is met; and S4, inputting performance data of a vehicle to be detected into the detection model after practice. The application can modify errors in the historical record combination.

Description

Technical Field

[0001] The present application relates to the field of machine learning technology, and in particular to a safety performance detection method, system, and storage medium for new energy vehicles. Background Art

[0002] With the development of computer application technology, the method of collecting performance data of new energy vehicles through performance testing equipment and obtaining analysis results about the performance of new energy vehicles by machine learning models has become more and more common.

[0003] The Chinese patent application with publication number CN119322980A discloses a method and system for analyzing automobile air-conditioning power performance based on machine learning. The system includes an automobile air-conditioning power influencing factor data collection module, a final random forest model construction module, a parameter tuning module, a first automobile air-conditioning power performance level determination module, and a second automobile air-conditioning power performance level determination module. The random forest model is trained by the collected automobile air-conditioning power influencing factor data set, and the automobile air-conditioning power influencing factor data to be tested is input into the final random forest model for classification to obtain the first automobile air-conditioning power performance level, and the cooling speed of the automobile air-conditioning to be tested is calculated, and the second automobile air-conditioning power performance level is obtained by analysis, and the final automobile air-conditioning power performance level to be tested is determined. A Chinese patent application with publication number CN118133657A discloses a method for detecting the deployment speed of an automobile airbag, which includes collecting collision data; storing the collision data to form a data set; expanding the data set through data enhancement technology; inputting the data as a training set into a neural network model to establish a collision model; inputting the collision data into the neural network model and adjusting the deployment speed of the airbag based on the prediction results. The collision model based on machine learning can learn and extract features from the collision data through a large amount of experimental data, and predict the severity of the collision and the deployment speed of the airbag.

[0004] However, both of the above two patent applications do not consider the problem that erroneous training data may exist in the training data set. Summary of the Invention

[0005] This application determines several pending historical records in a historical record combination, modifies the several pending historical records, uses the modified historical record combination to generate a detection model, and stops practicing the detection model at an appropriate time. This application aims to generate a detection model using the correct historical record combination.

[0006] This application provides a safety performance testing method for new energy vehicles, comprising the following steps:

[0007] S1. A preprocessing module obtains a historical record combination, wherein the historical record combination includes different historical records, each of which is composed of historical performance data and historical scores corresponding to the historical performance data. The preprocessing module determines a number of pending historical records from all the historical records in the historical record combination, and performs modification processing on the number of pending historical records.

[0008] S2. The training module determines a historical record from the modified historical record combination, inputs the historical performance data from the determined historical record into the detection model, adjusts the detection model based on the output score of the detection model, and further determines whether there is an undetermined historical record. If so, repeat this step; otherwise, proceed to the next step.

[0009] S3, the practice module determines whether the number of practice times is less than a preset practice number threshold. If not, all steps are stopped. If yes, the module continues to determine whether the requirement for stopping practice is met. If not, the module jumps to S2. If yes, the module stops all steps.

[0010] S4. The detection module obtains performance data of the vehicle to be detected, and inputs the performance data of the vehicle to be detected into the trained detection model, and the detection model outputs a score of the performance of the vehicle to be detected.

[0011] As a preferred technical solution of the present application, the exercise module determines whether the requirement for stopping exercise is met, including the following steps:

[0012] S31, the training module determines a historical record in the modified historical record combination, and inputs the historical performance data in the determined historical record into the current detection model, obtaining a first non-negative value of the difference between the output score of the current detection model and the historical score in the determined historical record, and a second non-negative value of the difference between the output score of the current detection model and the output score of the previous detection model with respect to the historical performance data in the determined historical record;

[0013] S32, the practice module determines whether the first non-negative value is less than a preset first threshold value. If so, it continues to determine whether there is a historical record that has not been determined. If so, it jumps to S31. If not, it is considered that the requirement for stopping practice is met and all steps are stopped. If not, it continues to the next step.

[0014] S33, the practice module determines whether the second non-negative value is less than the preset second threshold value. If yes, continue to determine whether there is a historical record that has not been determined. If yes, jump to S31. If no, it is considered that the requirements for stopping practice are met and all steps are stopped. If no, it is considered that the requirements for stopping practice are not met and all steps are stopped.

[0015] As a preferred technical solution of the present application, the pre-processing module determines a number of pending historical records from all historical records of the historical record combination, and performs modification processing on the number of pending historical records, including the following steps:

[0016] S11, the preprocessing module generates a first verification model using all historical records, sequentially inputs historical performance data from all historical records into the generated first verification model, obtains output scores of the first verification model, and calculates a first ratio of the number of historical records whose corresponding output scores are consistent with the included historical scores to the total number of all historical records;

[0017] S12, the pre-processing module determines whether the first ratio is greater than a preset first ratio threshold, and if so, continues with subsequent processing; if not, continues with the next step;

[0018] S13. The preprocessing module determines whether the number of times the first test model is repeatedly generated and the number of times the first ratio is repeatedly calculated is greater than or equal to a preset number threshold. If not, several historical records whose corresponding output scores are inconsistent with the included historical scores are regarded as several pending historical records, and several pending historical records are modified, and the process jumps to S11. If so, several pending historical records are determined from all the historical records using a preset method, and several pending historical records are modified, and the process jumps to S11.

[0019] As a preferred technical solution of this application, the subsequent processing includes the following steps:

[0020] S121. The preprocessing module divides all historical records into a plurality of subcombinations, selects one subcombination from the plurality of subcombinations, generates a second verification model using all other subcombinations, sequentially inputs historical performance data from all historical records in the selected subcombination into the generated second verification model, obtains output scores of the second verification model, and calculates a second ratio of the number of historical records whose corresponding output scores are consistent with the included historical scores to the total number of all historical records in the selected subcombination;

[0021] S122, the pre-processing module determines whether the second ratio is greater than a preset second ratio threshold, and if so, stops all steps; if not, proceeds to the next step;

[0022] S123. The preprocessing module determines whether the number of times the second test model is repeatedly generated and the second ratio is repeatedly calculated is greater than or equal to a preset number threshold. If it is less than, several historical records whose corresponding output scores are inconsistent with the included historical scores are regarded as several pending historical records, and several pending historical records are modified, and the process jumps to S121. If it is greater than or equal to, a preset method is used to determine several pending historical records in all historical records, and several pending historical records are modified, and the process jumps to S121.

[0023] As a preferred technical solution of this application, the preset method includes the following steps:

[0024] S131: The preprocessing module generates a third verification model using all historical records, sequentially inputs historical performance data from all historical records into the generated third verification model, obtains output scores of the third verification model, and first stores historical records whose corresponding output scores are inconsistent with the included historical scores.

[0025] S132: The preprocessing module determines a plurality of historical records from all the historical records, generates a fourth verification model using all the other historical records, sequentially inputs the historical performance data from all the other historical records into the generated fourth verification model, obtains output scores of the fourth verification model, and stores the corresponding historical records whose output scores are consistent with the included historical scores for a second time.

[0026] S133, the preprocessing module determines whether there are historical records that have been stored twice. If not, continue to determine whether there are historical records that have not been determined. If so, jump to S132. If not, stop all steps. If so, treat the determined historical records as several pending historical records.

[0027] As a preferred technical solution of the present application, in the case of yes in S133, the following steps are further included:

[0028] S1331: The preprocessing module selects a number of historical records from all determined historical records, and uses all other historical records except the selected number of historical records from all historical records to generate a fifth verification model;

[0029] S1332: The preprocessing module sequentially inputs all historical performance data from other historical records into the generated fifth verification model, obtains output scores of the fifth verification model, and stores historical records corresponding to the output scores and the included historical scores for a third time.

[0030] S1333. The preprocessing module determines whether there are historical records that are stored both at the first and third storage times. If so, the selected historical records are regarded as pending historical records. If not, continue to determine whether there are historical records that have not been selected in all the determined historical records. If so, jump to S1331. If not, stop all steps.

[0031] This application also provides a safety performance testing system for new energy vehicles, including the following modules:

[0032] a preprocessing module for obtaining a historical record combination, wherein the historical record combination includes different historical records, the historical records consisting of historical performance data and historical scores corresponding to the historical performance data, determining a number of pending historical records from all the historical records in the historical record combination, and performing modification processing on the number of pending historical records;

[0033] a practice module, configured to determine a historical record in the modified historical record combination, input historical performance data in the determined historical record into the detection model, adjust the detection model based on an output score of the detection model, determine whether there is an undetermined historical record, and if so, repeat the current process, and determine whether the number of practice times is less than a preset practice time threshold, and if not, stop all processes, and if so, continue to determine whether a requirement for stopping practice is met, and if not, repeat the previous process, and if so, stop all processes;

[0034] The detection module is used to obtain performance data of the vehicle to be detected, input the performance data of the vehicle to be detected into a trained detection model, and the detection model outputs a score about the performance of the vehicle to be detected.

[0035] The present application also provides a storage medium storing program instructions, wherein when the program instructions are executed, the device where the storage medium is located is controlled to execute any one of the above methods.

[0036] Compared with the prior art, the present invention has at least the following advantages:

[0037] In the technical solution provided by the present application, first, a historical record combination is obtained, which includes different historical records. Several pending historical records are determined from all the historical records in the historical record combination, and the pending historical records are modified. Secondly, a historical record is determined in the modified historical record combination, and the historical performance data in the determined historical record is input into the detection model. The detection model is adjusted based on the output score of the detection model. It is also determined whether there are historical records that have not been determined. If so, the current process is repeated. If not, the next step is continued. Thirdly, it is determined whether the number of practice times is less than a preset practice times threshold. If not, all steps are stopped. If yes, it is continued to determine whether the requirements for stopping practice are met. If not, the previous process is jumped to. If yes, all steps are stopped. Finally, the performance data of the vehicle to be tested is obtained, and the performance data of the vehicle to be tested is input into the trained detection model. The detection model outputs a score for the performance of the vehicle to be tested. Through this application, not only can the pending historical records that may contain errors in the historical record combination be determined, and the pending historical records can be modified and processed, so as to generate a detection model through the correct historical record combination, which is beneficial to the accuracy of the detection model, but also the appropriate time to stop practicing the detection model can be determined, thereby avoiding waste of computing power. BRIEF DESCRIPTION OF THE DRAWINGS

[0038] In order to more clearly illustrate the technical solutions of the embodiments of the present invention, the following briefly introduces the drawings required for use in the description of the embodiments. Obviously, the drawings described below are some embodiments of the present invention. For ordinary technicians in this field, other drawings can be obtained based on these drawings without creative work.

[0039] Figure 1 This is a flow chart of a safety performance testing method for new energy vehicles in an embodiment of the present application;

[0040] Figure 2 A flowchart for determining several pending historical records in an embodiment of the present application;

[0041] Figure 3 This is a flowchart of the subsequent processing in the embodiment of this application;

[0042] Figure 4 This is a schematic diagram of a safety performance testing system for new energy vehicles in an embodiment of the present application. DETAILED DESCRIPTION

[0043] The embodiments of the present application provide a safety performance detection method, system and storage medium for new energy vehicles. The terms "first", "second", "third", "fourth", etc. (if any) in the specification and claims of this application and the above-mentioned drawings are used to distinguish similar objects and are not necessarily used to describe a specific order or sequence. It should be understood that the data used in this way can be interchangeable where appropriate so that the embodiments described herein can be implemented in an order other than that illustrated or described herein. In addition, the terms "including" or "having" and any variations thereof are intended to cover non-exclusive inclusions. For example, a process, method, system, product or device that includes a series of steps or units is not necessarily limited to those steps or units that are clearly listed, but may include other steps or units that are not clearly listed or that are inherent to these processes, methods, products or devices.

[0044] For ease of understanding, the specific process of the embodiment of the present application is described below. Figure 1 The safety performance testing method for new energy vehicles in the embodiment of the present application includes the following main steps:

[0045] S1. A preprocessing module obtains a historical record combination, wherein the historical record combination includes different historical records, each of which is composed of historical performance data and historical scores corresponding to the historical performance data. The preprocessing module determines a number of pending historical records from all the historical records in the historical record combination, and performs modification processing on the number of pending historical records.

[0046] S2. The training module determines a historical record from the modified historical record combination, inputs the historical performance data from the determined historical record into the detection model, adjusts the detection model based on the output score of the detection model, and further determines whether there is an undetermined historical record. If so, repeat this step; otherwise, proceed to the next step.

[0047] S3, the practice module determines whether the number of practice times is less than a preset practice number threshold. If not, all steps are stopped. If yes, the module continues to determine whether the requirement for stopping practice is met. If not, the module jumps to S2. If yes, the module stops all steps.

[0048] S4. The detection module obtains performance data of the vehicle to be detected, inputs the performance data of the vehicle to be detected into a trained detection model, and the detection model outputs a score of the performance of the vehicle to be detected.

[0049] Specifically, a large number of historical records are usually accumulated first, and then the large number of accumulated historical records are used to train the generated machine learning model, but if there are erroneous historical records in the large number of accumulated historical records, it will have a bad effect on the accuracy of the machine learning model, and in addition, when training the generated machine learning model, a suitable stopping time needs to be determined to avoid waste of computing power.

[0050] In S1, the preprocessing module obtains a historical record combination, the historical record combination including different historical records, wherein the historical record is composed of historical performance data and a historical score corresponding to the historical performance data, it should be noted that the method of obtaining the historical record is to set a historical score first, and then set the historical performance data corresponding to the historical score, so the historical score is considered to be accurate, the preprocessing module determines a plurality of pending historical records from all historical records in the historical record combination, the pending historical record being a historical record whose contained historical performance data may need to be modified, and the modification processing is performed on the plurality of pending historical records, it should be noted that the modification processing can be performed by artificial, and the pending historical record can not be modified when it is determined by artificial that it does not need to be modified, the detailed process will be described below. In S2, the training module determines a historical record in the historical record combination after the modification processing, inputs the historical performance data in the determined historical record into the detection model, and adjusts the detection model based on the output score of the detection model, the detection model can be a neural network model, so that the parameters of the neural network model can be adjusted based on the deviation value between the output score of the neural network model and the corresponding historical score, the training module determines whether there is an undetermined historical record, if there is, determines an undetermined historical record, and repeats the step, if not, considers that one training is ended, and continues to the next step. In S3, the training module judges whether the training times are less than the preset training times threshold, the training times threshold is set according to the actual application scene, if greater than or equal to, the whole step is stopped, if less than, it is judged whether the requirement of stopping training is met, the requirement of stopping training will be described below, if not met, it is jumped to S2, if met, the whole step is stopped. In S4, the detection module obtains the performance data of the to-be-detected vehicle from the new energy vehicle performance detection device, inputs the performance data of the to-be-detected vehicle into the trained detection model, and the detection model outputs the score of the performance of the to-be-detected vehicle.

[0051] Further, the training module judges whether the requirement of stopping training is met, including the following steps:

[0052] S31. The training module determines a historical record in the modified historical record combination, and inputs the historical performance data in the determined historical record into the current detection model, obtaining a first non-negative value of the difference between the output score of the current detection model and the historical score in the determined historical record, and a second non-negative value of the difference between the output score of the current detection model and the output score of the previous detection model with respect to the historical performance data in the determined historical record;

[0053] S32, the practice module determines whether the first non-negative value is less than a preset first threshold value. If so, it continues to determine whether there is a historical record that has not been determined. If so, it jumps to S31. If not, it is considered that the requirement for stopping practice is met and all steps are stopped. If not, it continues to the next step.

[0054] S33. The practice module determines whether the second non-negative value is less than the preset second threshold value. If yes, it continues to determine whether there is a historical record that has not been determined. If yes, it jumps to S31. If no, it is considered that the requirements for stopping practice are met and all steps are stopped. If no, it is considered that the requirements for stopping practice are not met and all steps are stopped.

[0055] Specifically, the process of the practice module judging whether the requirement of stopping practice is met is introduced. In S31, the practice module determines a historical record in the modified historical record combination, inputs the historical performance data in the determined historical record into the current detection model, and the current detection model has been practiced several times. The first non-negative value of the difference between the output score of the current detection model and the historical score in the determined historical record is obtained. At the same time, the second non-negative value of the difference between the output score of the current detection model and the output score of the previous detection model with respect to the historical performance data in the determined historical record is obtained. The previous detection model refers to the detection model at the end of the practice before the current practice. In S32, the practice module judges whether the first non-negative value is less than a preset first threshold value. The first threshold value is set according to the actual application scenario. If it is less than, continue to judge whether there is a historical record that has not been determined. If so, jump to S31. If not, it is considered that the requirement of stopping practice is met and stop executing the method from S31 to S33. If it is greater than or equal to, continue to the next step. In S33, the practice module determines whether the second non-negative value is less than a preset second threshold value, which is set according to the actual application scenario. If so, the module continues to determine whether there is a historical record that has not been determined. If so, the module jumps to S31. If not, the requirement for stopping practice is considered to be met, and the method from S31 to S33 is stopped. If it is greater than or equal to the threshold value, the requirement for stopping practice is considered not to be met, and the method from S31 to S33 is stopped. According to the above method, for all historical records, if the first non-negative values ​​are all less than the first threshold value, or the second non-negative values ​​are all less than the second threshold value, then the requirement for stopping practice is considered to be met.

[0056] Furthermore, the pre-processing module determines a number of pending historical records from all historical records of the historical record combination, and performs modification processing on the number of pending historical records, including the following steps:

[0057] S11. The preprocessing module generates a first verification model using all historical records, sequentially inputs historical performance data from all historical records into the generated first verification model, obtains output scores of the first verification model, and calculates a first ratio of the number of historical records whose corresponding output scores are consistent with the included historical scores to the total number of all historical records;

[0058] S12, the pre-processing module determines whether the first ratio is greater than a preset first ratio threshold, and if so, continues with subsequent processing; if not, continues with the next step;

[0059] S13. The preprocessing module determines whether the number of times the first test model is repeatedly generated and the number of times the first ratio is repeatedly calculated is greater than or equal to a preset number threshold. If not, several historical records whose corresponding output scores are inconsistent with the included historical scores are regarded as several pending historical records, and several pending historical records are modified, and the process goes to S11. If so, several pending historical records are determined from all the historical records using a preset method, and several pending historical records are modified, and the process goes to S11.

[0060] Specifically, see Figure 2 , describes how the pre-processing module determines several pending historical records from all the historical records of the historical record combination and performs modification processing on the several pending historical records. In S11, the pre-processing module uses all the historical records to generate a first verification model. The first verification model has the same function as the detection model and can output performance scores based on the input historical performance data. After the first verification model is generated, the historical performance data in all the historical records are sequentially input into the generated first verification model to obtain the output scores of the first verification model. The pre-processing module calculates a first ratio of the number of historical records whose corresponding output scores are consistent with the included historical scores to the total number of all historical records. The corresponding output score and the included historical score can be consistent in that the two are the same or the difference between the two is less than a preset difference threshold. The difference threshold is set according to the actual application scenario. In S12, the pre-processing module determines whether the first ratio is greater than a preset first ratio threshold. The first ratio threshold is set according to the actual application scenario. If it is greater, the subsequent processing continues. The steps of the subsequent processing will be described below. If it is less than or equal to, the next step is continued. In S13, the preprocessing module determines whether the number of times the first test model is repeatedly generated and the number of times the first ratio is repeatedly calculated is greater than or equal to a preset number threshold. The number threshold is set according to the actual application scenario. If it is less than, several historical records whose corresponding output scores are inconsistent with the included historical scores are regarded as several pending historical records, and several pending historical records are modified, specifically the historical performance data in the several pending historical records are modified, and the process jumps to S11. If it is greater than or equal to, a preset method is used to determine several pending historical records in all historical records, and several pending historical records are modified, and the process jumps to S11. The steps of the preset method will be described below.

[0061] Furthermore, the subsequent processing includes the following steps:

[0062] S121. The preprocessing module divides all historical records into a plurality of subcombinations, selects one subcombination from the plurality of subcombinations, generates a second verification model using all other subcombinations, sequentially inputs historical performance data from all historical records in the selected subcombination into the generated second verification model, obtains output scores of the second verification model, and calculates a second ratio of the number of historical records whose corresponding output scores are consistent with the included historical scores to the total number of all historical records in the selected subcombination;

[0063] S122, the pre-processing module determines whether the second ratio is greater than a preset second ratio threshold value, and if so, stops all steps; if not, proceeds to the next step;

[0064] S123. The preprocessing module determines whether the number of times the second test model is repeatedly generated and the second ratio is repeatedly calculated is greater than or equal to a preset number threshold. If it is less than, several historical records whose corresponding output scores are inconsistent with the included historical scores are regarded as several pending historical records, and several pending historical records are modified, and the process jumps to S121. If it is greater than or equal to, a preset method is used to determine several pending historical records in all historical records, and several pending historical records are modified, and the process jumps to S121.

[0065] Specifically, see Figure 3, the steps of subsequent processing are introduced. In S121, the pre-processing module divides all historical records into several sub-combinations, selects a sub-combination from the several sub-combinations, and uses all other sub-combinations to generate a second test model. The second test model has the same function as the detection model, and can output performance scores based on the input historical performance data. After the second test model is generated, the historical performance data of all historical records in the selected sub-combination are sequentially input into the generated second test model, and the output scores of the second test model are obtained respectively. The pre-processing module calculates the second ratio of the number of historical records that are consistent with the corresponding output score and the historical score to the total number of all historical records in the selected sub-combination. The consistency here has the same meaning as the consistency mentioned above. In S122, the pre-processing module determines whether the second ratio is greater than a preset second ratio threshold value. The second ratio threshold value is set according to the actual application scenario. If it is greater than, all steps are stopped. If it is less than or equal to, proceed to the next step. In S123, the preprocessing module determines whether the number of times the second test model is repeatedly generated and the number of times the second ratio is repeatedly calculated is greater than or equal to a preset number threshold. If so, several historical records whose corresponding output scores are inconsistent with the included historical scores are regarded as several pending historical records, and several pending historical records are modified, specifically the historical performance data in the several pending historical records are modified, and the process jumps to S121. If it is greater than or equal to, several pending historical records are determined in all historical records using a preset method, and several pending historical records are modified, and the process jumps to S121. The steps of the preset method will be described below.

[0066] Furthermore, the preset method includes the following steps:

[0067] S131: The preprocessing module generates a third verification model using all historical records, sequentially inputs historical performance data from all historical records into the generated third verification model, obtains output scores of the third verification model, and first stores historical records whose corresponding output scores are inconsistent with the included historical scores.

[0068] S132: The preprocessing module determines a number of historical records from all the historical records, generates a fourth verification model using all the other historical records, sequentially inputs the historical performance data from all the other historical records into the generated fourth verification model, obtains output scores of the fourth verification model, and stores the corresponding historical records whose output scores are consistent with the included historical scores for a second time.

[0069] S133. The preprocessing module determines whether there are historical records that have been stored twice. If not, continue to determine whether there are historical records that have not been determined. If so, jump to S132. If not, stop all steps. If so, treat the determined historical records as several pending historical records.

[0070] Specifically, the steps of the preset method are introduced. In S131, the preprocessing module uses all the historical records to generate a third verification model. The third verification model has the same function as the detection model and can output performance scores based on the input historical performance data. After the third verification model is generated, the historical performance data in all the historical records are sequentially input into the generated third verification model to obtain the output scores of the third verification model. For the first time, multiple historical records whose corresponding output scores are inconsistent with the included historical scores are stored. The consistency here has the same meaning as the consistency mentioned above. In S132, the preprocessing module determines several historical records from all the historical records and uses all the other historical records to generate a fourth verification model. The fourth verification model has the same function as the detection model and can output performance scores based on the input historical performance data. After the fourth verification model is generated, the historical performance data in all the other historical records are sequentially input into the generated fourth verification model to obtain the output scores of the fourth verification model. For the second time, multiple historical records whose corresponding output scores are consistent with the included historical scores are stored. The consistency here has the same meaning as the consistency mentioned above. In S133, the preprocessing module determines whether there are historical records that have been stored twice. For example, a historical record is recorded both when executing S131 and when executing S132. If not, continue to determine whether there are historical records that have not been determined. If so, jump to S132, determine several historical records that have not been determined from all historical records and continue to execute subsequent steps. If not, stop all steps. If so, it is considered that the reason for this situation is that the determined historical records were not used when generating the fourth verification model in S132, and therefore the determined historical records are regarded as several pending historical records.

[0071] Furthermore, in S133, if yes, the following steps are further included:

[0072] S1331: The preprocessing module selects a number of historical records from all determined historical records, and uses all other historical records except the selected number of historical records from all historical records to generate a fifth verification model;

[0073] S1332: The preprocessing module sequentially inputs all historical performance data from other historical records into the generated fifth verification model, obtains output scores of the fifth verification model, and stores historical records corresponding to the output scores and the included historical scores for a third time.

[0074] S1333. The preprocessing module determines whether there are historical records stored both at the first and third storage times. If so, the selected historical records are regarded as pending historical records. If not, continue to determine whether there are historical records that have not been selected among all the determined historical records. If so, jump to S1331. If not, stop all steps.

[0075] Specifically, the steps of the preset method are continued. By executing S131 to S133, a number of pending historical records can be determined. However, the total number of the pending historical records may be relatively large, and it takes a long time to modify the pending historical records. Therefore, S1331 to S1333 are proposed. In S133, when the judgment result is that there are historical records that have been stored twice, it is executed. In S1331, the preprocessing module selects a number of historical records from the determined total historical records. Here, the determined total historical records refer to the number of historical records determined when S132 is executed. All other historical records in the total historical records except the selected number of historical records are used to generate a fifth verification model. The fifth verification model has the same function as the detection model and can output performance scores based on the input historical performance data. In S1332, after generating the fifth verification model, the preprocessing module inputs the historical performance data from all other historical records into the generated fifth verification model in sequence, obtains the output scores of the fifth verification model respectively, and stores a plurality of historical records whose corresponding output scores are consistent with the included historical scores for the third time. The consistency here has the same meaning as the consistency mentioned above. In S1333, the preprocessing module determines whether there are historical records that are stored both at the first storage and the third storage. If so, it is considered that the reason for this situation is that when generating the fifth verification model in S131, several historical records selected from all the determined historical records were not used, so the selected several historical records are regarded as several pending historical records. If not, continue to determine whether there are historical records that have not been selected in all the determined historical records. If so, jump to S1331, select several historical records that have not been selected from all the determined historical records and continue to execute subsequent steps. If not, stop all steps.

[0076] According to another aspect of the embodiment of the present application, Figure 4As shown, the present application also provides a safety performance detection system for new energy vehicles, including a preprocessing module, a training module, and a detection module to implement the safety performance detection method for new energy vehicles described above.

[0077] The functions of each module are as follows:

[0078] a preprocessing module for obtaining a historical record combination, wherein the historical record combination includes different historical records, the historical records consisting of historical performance data and historical scores corresponding to the historical performance data, determining a number of pending historical records from all the historical records in the historical record combination, and performing modification processing on the number of pending historical records;

[0079] a practice module, configured to determine a historical record in the modified historical record combination, input historical performance data in the determined historical record into the detection model, adjust the detection model based on an output score of the detection model, determine whether there is an undetermined historical record, and if so, repeat the current process, and determine whether the number of practice times is less than a preset practice time threshold, and if not, stop all processes, and if so, continue to determine whether a requirement for stopping practice is met, and if not, repeat the previous process, and if so, stop all processes;

[0080] The detection module is used to obtain performance data of the vehicle to be detected, input the performance data of the vehicle to be detected into a trained detection model, and the detection model outputs a score about the performance of the vehicle to be detected.

[0081] According to another aspect of an embodiment of the present application, a storage medium is further provided, wherein the storage medium stores program instructions, wherein when the program instructions are executed, the device where the storage medium is located is controlled to execute any one of the above methods.

[0082] Those skilled in the art will clearly understand that, for the convenience and brevity of description, the specific working processes of the above-described systems, systems and units can refer to the corresponding processes in the aforementioned method embodiments and will not be repeated here.

[0083] If the integrated unit is implemented in the form of a software functional unit and sold or used as an independent product, it can be stored in a computer-readable storage medium. Based on this understanding, the technical solution of the present application, or the part that contributes to the prior art, or all or part of the technical solution can be embodied in the form of a software product. The computer software product is stored in a storage medium and includes several instructions for enabling a computer device (which can be a personal computer, server, or network device, etc.) to execute all or part of the steps of the method described in each embodiment of the present application. The aforementioned storage medium includes: U disk, mobile hard disk, read-only memory (ROM), random access memory (RAM), disk or optical disk, and other media that can store program code.

[0084] As described above, the above embodiments are only used to illustrate the technical solutions of the present application, rather than to limit them. Although the present application has been described in detail with reference to the above embodiments, those skilled in the art should understand that they can still modify the technical solutions described in the above embodiments, or make equivalent replacements for some of the technical features therein. However, these modifications or replacements do not deviate the essence of the corresponding technical solutions from the spirit and scope of the technical solutions of the embodiments of the present application.

Claims

1. A safety performance testing method for new energy vehicles, characterized in that: The method comprises the following steps: S1. A preprocessing module obtains a historical record combination, wherein the historical record combination includes different historical records, each of which is composed of historical performance data and historical scores corresponding to the historical performance data. The preprocessing module determines a number of pending historical records from all the historical records in the historical record combination, and performs modification processing on the number of pending historical records. S2. The training module determines a historical record from the modified historical record combination, inputs the historical performance data from the determined historical record into the detection model, adjusts the detection model based on the output score of the detection model, and further determines whether there is an undetermined historical record. If so, repeat this step; otherwise, proceed to the next step. S3, the practice module determines whether the number of practice times is less than a preset practice number threshold. If not, all steps are stopped. If yes, the module continues to determine whether the requirement for stopping practice is met. If not, the module jumps to S2. If yes, the module stops all steps. S4. The detection module obtains performance data of the vehicle to be detected, inputs the performance data of the vehicle to be detected into the trained detection model, and the detection model outputs a performance score of the vehicle to be detected; The pre-processing module determines a number of pending historical records from all historical records of the historical record combination, and performs modification processing on the number of pending historical records, including the following steps: S11, the preprocessing module generates a first verification model using all historical records, sequentially inputs historical performance data from all historical records into the generated first verification model, obtains output scores of the first verification model, and calculates a first ratio of the number of historical records whose corresponding output scores are consistent with the included historical scores to the total number of all historical records; S12, the pre-processing module determines whether the first ratio is greater than a preset first ratio threshold, and if so, continues with subsequent processing; if not, continues with the next step; S13. The preprocessing module determines whether the number of times the first test model is repeatedly generated and the number of times the first ratio is repeatedly calculated is greater than or equal to a preset number threshold. If not, several historical records whose corresponding output scores are inconsistent with the included historical scores are regarded as several pending historical records, and several pending historical records are modified, and the process jumps to S11. If so, several pending historical records are determined from all the historical records using a preset method, and several pending historical records are modified, and the process jumps to S11.

2. The method according to claim 1, characterized in that The exercise module determines whether the requirement for stopping exercise is met, including the following steps: S31, the training module determines a historical record in the modified historical record combination, and inputs the historical performance data in the determined historical record into the current detection model, obtaining a first non-negative value of the difference between the output score of the current detection model and the historical score in the determined historical record, and a second non-negative value of the difference between the output score of the current detection model and the output score of the previous detection model with respect to the historical performance data in the determined historical record; S32, the practice module determines whether the first non-negative value is less than a preset first threshold value. If so, it continues to determine whether there is a historical record that has not been determined. If so, it jumps to S31. If not, it is considered that the requirement for stopping practice is met and all steps are stopped. If not, it continues to the next step. S33, the practice module determines whether the second non-negative value is less than the preset second threshold value. If yes, continue to determine whether there is a historical record that has not been determined. If yes, jump to S31. If no, it is considered that the requirements for stopping practice are met and all steps are stopped. If no, it is considered that the requirements for stopping practice are not met and all steps are stopped.

3. The method according to claim 1, characterized in that The subsequent processing includes the following steps: S121. The preprocessing module divides all historical records into a plurality of subcombinations, selects one subcombination from the plurality of subcombinations, generates a second verification model using all other subcombinations, sequentially inputs historical performance data from all historical records in the selected subcombination into the generated second verification model, obtains output scores of the second verification model, and calculates a second ratio of the number of historical records whose corresponding output scores are consistent with the included historical scores to the total number of all historical records in the selected subcombination; S122, the pre-processing module determines whether the second ratio is greater than a preset second ratio threshold, and if so, stops all steps; if not, proceeds to the next step; S123. The preprocessing module determines whether the number of times the second test model is repeatedly generated and the second ratio is repeatedly calculated is greater than or equal to a preset number threshold. If it is less than, several historical records whose corresponding output scores are inconsistent with the included historical scores are regarded as several pending historical records, and several pending historical records are modified, and the process jumps to S121. If it is greater than or equal to, a preset method is used to determine several pending historical records in all historical records, and several pending historical records are modified, and the process jumps to S121.

4. The method according to claim 3, characterized in that The preset method includes the following steps: S131: The preprocessing module generates a third verification model using all historical records, sequentially inputs historical performance data from all historical records into the generated third verification model, obtains output scores of the third verification model, and first stores historical records whose corresponding output scores are inconsistent with the included historical scores. S132: The preprocessing module determines a plurality of historical records from all the historical records, generates a fourth verification model using all the other historical records, sequentially inputs the historical performance data from all the other historical records into the generated fourth verification model, obtains output scores of the fourth verification model, and stores the corresponding historical records whose output scores are consistent with the included historical scores for a second time. S133, the preprocessing module determines whether there are historical records that have been stored twice. If not, continue to determine whether there are historical records that have not been determined. If so, jump to S132. If not, stop all steps. If so, treat the determined historical records as several pending historical records.

5. The method according to claim 4, characterized in that In the above S133, if yes, the following steps are further included: S1331: The preprocessing module selects a number of historical records from all determined historical records, and uses all other historical records except the selected number of historical records from all historical records to generate a fifth verification model; S1332: The preprocessing module sequentially inputs all historical performance data from other historical records into the generated fifth verification model, obtains output scores of the fifth verification model, and stores historical records corresponding to the output scores and the included historical scores for a third time. S1333. The preprocessing module determines whether there are historical records that are stored both at the first and third storage times. If so, the selected historical records are regarded as pending historical records. If not, continue to determine whether there are historical records that have not been selected in all the determined historical records. If so, jump to S1331. If not, stop all steps.

6. A safety performance testing system for new energy vehicles, used to implement the method according to any one of claims 1 to 5, characterized in that: Includes the following modules: a preprocessing module for obtaining a historical record combination, wherein the historical record combination includes different historical records, the historical records consisting of historical performance data and historical scores corresponding to the historical performance data, determining a number of pending historical records from all the historical records in the historical record combination, and performing modification processing on the number of pending historical records; a practice module, configured to determine a historical record in the modified historical record combination, input historical performance data in the determined historical record into the detection model, adjust the detection model based on an output score of the detection model, determine whether there is an undetermined historical record, and if so, repeat the current process, and determine whether the number of practice times is less than a preset practice time threshold, and if not, stop all processes, and if so, continue to determine whether a requirement for stopping practice is met, and if not, repeat the previous process, and if so, stop all processes; The detection module is used to obtain performance data of the vehicle to be detected, input the performance data of the vehicle to be detected into a trained detection model, and the detection model outputs a score about the performance of the vehicle to be detected.

7. A storage medium, characterized in that: The storage medium stores program instructions, wherein when the program instructions are executed, the device where the storage medium is located is controlled to execute the method according to any one of claims 1 to 5.

Images