Pedestrian head simulation optimization method, device and equipment based on ensemble learning

Abstract

The present application discloses a pedestrian head simulation optimization method, device and equipment based on ensemble learning, which relates to the technical field of automobile safety performance design, development and optimization. The method comprises: dividing a head test area in a vehicle simulation model into a plurality of grid points, wherein the head test area comprises at least a hood area and a non-hood area; acquiring characteristic data of the vehicle simulation model from the head test area; inputting the characteristic data into a hood grid point prediction model to obtain a hood head injury classification result corresponding to each grid point in the hood area, and performing a simulation experiment on the non-hood area to obtain a non-hood head injury classification result for each grid point in the non-hood area; and obtaining a total head injury score for the head test area based on the non-hood head injury classification result and the hood head injury classification result.

Description

Technical Field

[0001] The present application relates to the technical field of automobile safety performance design, development and optimization, and in particular to a pedestrian head simulation optimization method, device and equipment based on ensemble learning. Background Art

[0002] Pedestrian head protection performance is one of the important evaluation items of NCAP at home and abroad. The head injury index (HIC) is used to measure the degree of pedestrian head injury caused by vehicle collision. It is required that vehicles can reduce the impact force on pedestrians' heads and reduce the risk of pedestrian injuries in collisions within a certain speed range.

[0003] The simulation calculation method plays a predictive role before the test, so the development usually adopts a combination of a large number of CAE simulation optimization analysis points and a small number of test points. The simulation stage includes the establishment of a CAE simulation model of the whole vehicle, and then the meshing of the whole vehicle model, the placement of the head impactor, the setting of the working conditions and the submission of the calculation. During the simulation optimization process, more than a dozen versions of simulation model optimization iterations and calculations are usually performed to verify the solution. There are more than 100 grid points in the pedestrian protection head shape area. The HIC simulation calculation time for each grid point is more than hours. The simulation development stage generally takes several months to complete, which greatly delays the efficiency of vehicle simulation design and development. Summary of the invention

[0004] The purpose of this application is to provide a pedestrian head simulation optimization method, device and equipment based on ensemble learning.

[0005] In a first aspect, the present application provides a pedestrian head simulation optimization method based on ensemble learning, comprising:

[0006] Dividing a head test area in the vehicle simulation model into a plurality of grid points, wherein the head test area at least includes: a hood area and a non-hood area;

[0007] Acquire characteristic data of the vehicle simulation model from the head test area;

[0008] Inputting the characteristic data into a hood grid point prediction model to obtain a hood head injury classification result corresponding to each grid point in the hood area, and performing a simulation experiment on the non-hood area to obtain a non-hood head injury classification result for each grid point in the non-hood area;

[0009] Based on the non-hood head injury classification result and the hood head injury classification result, a total head injury score of the head experimental area is obtained.

[0010] Optionally, the hood grid point prediction model is trained by the following steps:

[0011] Acquire sample characteristic data of the head experimental area in the sample simulation model;

[0012] Inputting the sample feature data into an initial hood grid point prediction model for training to obtain a prediction result;

[0013] Obtaining a loss value of the prediction result;

[0014] When the loss value reaches the training requirement, it is confirmed that the training of the hood grid point prediction model is completed.

[0015] Optionally, the hood grid point prediction model includes at least two different classification models;

[0016] The step of inputting the sample feature data into the initial hood grid point prediction model for training to obtain a prediction result comprises:

[0017] Inputting the sample feature data into each of the classification models respectively to obtain the prediction results corresponding to each of the classification models;

[0018] The plurality of prediction results are combined to obtain the prediction result of the hood grid point prediction model.

[0019] Optionally, the feature data includes: first feature data and second feature data;

[0020] The step of acquiring characteristic data of the vehicle simulation model from the head experimental area comprises:

[0021] Reading or measuring the head experimental area of ​​the simulated vehicle model to obtain first characteristic data;

[0022] Receive input feature data for the simulated vehicle model to obtain second feature data, wherein the input feature data includes at least one of inner panel hollowing ratio, total number of inner panel openings, number of openings with an area greater than a fixed value 1, number of openings with an area less than a fixed value 2, ratio of irregular polygonal openings, ratio of rectangular openings, ratio of triangular openings, number of inner panel reinforcement ribs, and number of connection points of inner panel reinforcement ribs.

[0023] Optionally, the first feature data includes: local feature data and overall feature data;

[0024] The step of reading or measuring the head experimental area of ​​the simulated vehicle model to obtain the first characteristic data includes:

[0025] The grid points are measured to obtain local feature data, and the hood area is measured to obtain overall feature data.

[0026] Optionally, the step of obtaining a total score of head injury in the head experimental area based on the non-hood head injury classification result and the hood head injury classification result comprises:

[0027] Based on the proportion of the hood head injury classification results and the non-hood head injury classification results of each grid point in the head test area, the total head injury score of the head test area is calculated.

[0028] Optionally, the step of calculating the total score of the head injury in the head experimental area based on the proportion of the hood head injury classification results and the non-hood head injury classification results of each grid point in the head experimental area comprises:

[0029] Calculate the classification scores corresponding to the hood head injury classification results and the non-hood head injury classification results corresponding to each grid point in the head test area;

[0030] The head injury score of the head experimental area is determined based on the proportion of the different hood head injury classification results and the non-hood head injury classification results corresponding to all the grid points.

[0031] In a second aspect, the present application provides a pedestrian head simulation optimization device based on ensemble learning, comprising:

[0032] An acquisition module is used to divide a head test area in the vehicle simulation model into a plurality of grid points, wherein the head test area at least includes: a hood area and a non-hood area;

[0033] Acquire characteristic data of the vehicle simulation model from the head test area;

[0034] A prediction module, used for inputting the characteristic data into a hood grid point prediction model to obtain a hood head injury classification result corresponding to each grid point in the hood area, and performing a simulation experiment on the non-hood area to obtain a non-hood head injury classification result for each grid point in the non-hood area;

[0035] A processing module is used to obtain a total score of head injury in the head experimental area based on the non-hood head injury classification result and the hood head injury classification result.

[0036] Optionally, the prediction module is further used to:

[0037] Acquire sample characteristic data of the head experimental area in the sample simulation model;

[0038] Inputting the sample feature data into an initial hood grid point prediction model for training to obtain a prediction result;

[0039] Obtaining a loss value of the prediction result;

[0040] When the loss value reaches the training requirement, it is confirmed that the training of the hood grid point prediction model is completed.

[0041] Optionally, the hood grid point prediction model includes at least two different classification models;

[0042] The prediction module is further used for:

[0043] Inputting the sample feature data into each of the classification models respectively to obtain the prediction results corresponding to each of the classification models;

[0044] The plurality of prediction results are combined to obtain the prediction result of the hood grid point prediction model.

[0045] Optionally, the feature data includes: first feature data and second feature data;

[0046] The acquisition module is further used for:

[0047] Reading or measuring the head experimental area of ​​the simulated vehicle model to obtain first characteristic data;

[0048] Receive input feature data for the simulated vehicle model to obtain second feature data, wherein the input feature data includes at least one of inner panel hollowing ratio, total number of inner panel openings, number of openings with an area greater than a fixed value 1, number of openings with an area less than a fixed value 2, ratio of irregular polygonal openings, ratio of rectangular openings, ratio of triangular openings, number of inner panel reinforcement ribs, and number of connection points of inner panel reinforcement ribs.

[0049] Optionally, the first feature data includes: local feature data and overall feature data;

[0050] The acquisition module is further used for:

[0051] The grid points are measured to obtain local feature data, and the hood area is measured to obtain overall feature data.

[0052] Optionally, the processing module is further used to:

[0053] Based on the proportion of the hood head injury classification results and the non-hood head injury classification results of each grid point in the head test area, the total head injury score of the head test area is calculated.

[0054] Optionally, the processing module is further used to:

[0055] Calculate the classification scores corresponding to the hood head injury classification results and the non-hood head injury classification results corresponding to each grid point in the head test area;

[0056] The head injury score of the head experimental area is determined based on the proportion of the different hood head injury classification results and the non-hood head injury classification results corresponding to all the grid points.

[0057] In a third aspect, the present application provides a computer device comprising: a memory, a processor, and a computer program stored in the memory and executable on the processor, wherein the processor executes the computer program to implement the steps of any one of the above-mentioned pedestrian head simulation optimization methods based on integrated learning.

[0058] In a fourth aspect, the present application provides a computer-readable storage medium having a computer program stored thereon, which, when executed by a processor, implements the steps of any of the above-mentioned pedestrian head simulation optimization methods based on ensemble learning.

[0059] In a fifth aspect, the present application provides a computer program product, including a computer program, which, when executed by a processor, implements the steps of any of the above-mentioned pedestrian head simulation optimization methods based on ensemble learning.

[0060] According to the specific embodiments provided in this application, this application discloses the following technical effects:

[0061] The present application provides a pedestrian head simulation optimization method, device and equipment based on ensemble learning. The head test area in the vehicle simulation model is divided into a hood area and a non-hood area. The hood area is quickly predicted using a hood grid point prediction model, and the non-hood area is simulated to obtain results. The non-hood head injury classification results corresponding to each grid point in the hood area and the non-hood area are then combined with the hood head injury classification results to obtain a head injury score for the head test area. This not only makes the prediction more refined and accurate, but also enables faster verification and improvement of vehicle design solutions, thereby accelerating the development of vehicle products. BRIEF DESCRIPTION OF THE DRAWINGS

[0062] In order to more clearly illustrate the embodiments of the present application or the technical solutions in the prior art, the drawings required for use in the embodiments will be briefly introduced below. Obviously, the drawings described below are only some embodiments of the present application. For ordinary technicians in this field, other drawings can be obtained based on these drawings without paying creative work.

[0063] Figure 1 A flowchart of a pedestrian head simulation optimization method based on ensemble learning provided in one embodiment of the present application;

[0064] Figure 2A schematic diagram showing the effect of a method for dividing the head area of ​​a vehicle simulation model provided in one embodiment of the present application;

[0065] Figure 3 A schematic diagram of functional modules of a pedestrian head simulation optimization device based on ensemble learning provided in one embodiment of the present application;

[0066] Figure 4 A schematic diagram of the structure of a computer device provided in one embodiment of the present application. DETAILED DESCRIPTION

[0067] The following will be combined with the drawings in the embodiments of the present application to clearly and completely describe the technical solutions in the embodiments of the present application. Obviously, the described embodiments are only part of the embodiments of the present application, not all of the embodiments. Based on the embodiments in the present application, all other embodiments obtained by ordinary technicians in this field without creative work are within the scope of protection of this application.

[0068] like Figure 1 As shown, some embodiments of the present application provide a pedestrian head simulation optimization method based on ensemble learning, which includes the following steps 101 to 103. Among them:

[0069] Step 101: divide a head test area in a vehicle simulation model into a plurality of grid points, wherein the head test area at least includes a hood area and a non-hood area.

[0070] In the embodiments of the present application, the target vehicle refers to a vehicle that is in the pedestrian protection head target design stage in the safety performance simulation development. Usually, a vehicle needs to verify the feasibility of the design scheme through optimization iteration and calculation of multiple versions of simulation models. If the calculation verification process is often carried out in the form of simulation experiments, it will consume a lot of human resources and computing resources. However, the pedestrian head simulation optimization method based on ensemble learning provided by the embodiments of the present application can effectively reduce the resources required for the optimization iteration process of the vehicle simulation model in the pedestrian protection head target design stage, and can significantly improve the efficiency of the optimization iteration process. Therefore, the scenario targeted by the embodiments of the present application is the iterative optimization process of the human protection head target design in the safety performance simulation development, and the target vehicle refers to a vehicle for which a previous version of the vehicle simulation model already exists in the iterative optimization process. It is worth noting that the target vehicle is not necessarily a physical vehicle, but an existing vehicle design scheme, and a vehicle simulation model can be constructed according to the vehicle design scheme.

[0071] The vehicle whose head injury degree to pedestrians is to be predicted during a vehicle collision can generally be a vehicle that has undergone several rounds of simulation optimization, so that the predicted value of HIC (Head Injury Criteria) of the subsequent simulation optimization scheme of the target vehicle can be obtained through the pedestrian head injury prediction. Specifically, a high-precision CAE (Computer Aided Engineering) vehicle simulation model can be created for the target vehicle through simulation software to simulate and collect various image data of the target vehicle. The head test area is the area that may collide with the pedestrian's head. It is generally based on the requirements of various N-CAP or national standards and other regulations. It can be set according to actual needs and is not limited here.

[0072] The head test area in the vehicle simulation area can be finely divided into a plurality of grid points, which cover the entire head test area, including but not limited to the hood area and the non-hood area.

[0073] For example, refer to Figure 2 The head shape test area is divided into 18 equally divided areas, including the hood area, windshield area, fender area, A-pillar area, front bumper and headlight area, ventilation cover area, wiper area, etc. Except for the hood area, which is collectively referred to as the non-hood area, all grid points are divided into two categories, hood area grid points and non-hood area grid points. All grid points are numbered.

[0074] Step 102: Acquire characteristic data of the vehicle simulation model from the head test area.

[0075] In the embodiment of the present application, characteristic data are obtained from each grid point in the head test area and the surrounding vehicle structure such as the hood assembly. These characteristic data include both first characteristic data that can be directly read or measured from the simulation model, such as the three-dimensional coordinates of the grid point, the area position number, the head type, the head weight, the material thickness, the collision angle, the collision speed, etc., and second characteristic data that are further designed, measured and calculated based on the experience of engineers, such as the hollowing ratio of the inner panel, the proportion of the opening area, the number of reinforcing ribs, etc.

[0076] Step 103, input the characteristic data into the hood grid point prediction model to obtain the hood head injury classification results corresponding to each grid point in the hood area, and perform a simulation experiment on the non-hood area to obtain the non-hood head injury classification results of each grid point in the non-hood area.

[0077] In an embodiment of the present application, the hood grid point prediction model can be constructed based on an integrated learning algorithm, such as random forest, XGBoost, LightGBM, etc., which can quickly predict the classification results of head injuries that may be caused by each grid point in the hood area in a collision. The extracted feature data of the hood area is input into the pre-trained hood grid point prediction model to obtain the hood head injury classification result. For the non-hood area, since its structure and material properties are different from those of the hood area, the hood grid point prediction model is not directly used for prediction. A high-precision CAE simulation model can be used, such as input into LS-DYNA for calculation to obtain the non-hood head injury classification result, or an actual collision test can be used to obtain the head injury classification result of each grid point in the non-hood area.

[0078] Step 104: obtaining a total head injury score of the head experimental area based on the non-hood head injury classification result and the hood head injury classification result.

[0079] In the embodiment of the present application, the head injury score is calculated based on the NCAP (New Car Assessment Program) regulations to be developed, such as China's C-NCAP, Europe's E-NCAP or the United States' A-NCAP. The NCAP regulations usually specify in detail the points corresponding to the injury classification results of each grid point or area, and how to summarize these points into the final score. Only based on this, the non-hood head injury classification results and the hood head injury classification results can be summarized based on the regulations as the head injury score of the head experimental area.

[0080] The embodiment of the present application divides the head test area in the vehicle simulation model into a hood area and a non-hood area, uses a hood grid point prediction model to perform rapid prediction on the hood area, and uses simulation tests to obtain results on the non-hood area, and then combines the non-hood head injury classification results corresponding to each grid point in the hood area and the non-hood area with the hood head injury classification results to obtain the head injury score of the head test area, which not only makes the prediction more refined and accurate, but also can verify and improve the vehicle design scheme more quickly, thereby accelerating the development of vehicle products.

[0081] Optionally, the hood grid point prediction model is trained by the following steps:

[0082] Step 201, obtaining sample feature data of a head experimental area in a sample simulation model.

[0083] Step 202: input the sample feature data into an initial hood grid point prediction model for training to obtain a prediction result.

[0084] Step 203, obtaining the loss value of the prediction result.

[0085] Step 204, when the loss value reaches the training requirement, confirm that the hood grid point prediction model training is completed.

[0086] In the embodiment of the present application, sample feature data of the head test area can be extracted from multiple versions of the simulation model optimization scheme of the vehicle. These data include but are not limited to basic features such as the three-dimensional coordinates of the grid points, the area position number, the head type, the head weight, the material thickness, the collision angle, and the collision speed, as well as the second features such as the hollowing ratio of the inner plate, the opening area ratio, and the number of reinforcing ribs obtained by further designing, measuring, and calculating based on the experience of engineers. The extracted sample feature data is input into the initial hood grid point prediction model.

[0087] During the training process, the model will learn the complex relationship between the grid points in the hood area and the head injury classification results based on the input sample feature data, and gradually adjust the internal parameters to improve the accuracy of the prediction. The prediction formula for the hood area can be as follows (1):

[0088] (1)

[0089] Among them, E(x) is the final prediction result of the integrated learning model for the input sample x

[0090] {ht(x)|t=1,…,T} is the set of prediction results of T basic learners for the input sample x

[0091] F is a combination function that generates a final prediction based on the predictions of multiple classifiers and possible additional parameters θ, which can be weights, model parameters, etc., depending on the type of ensemble learning algorithm.

[0092] After one or more rounds of training, the model will give a prediction result for the input sample feature data, that is, the head injury classification of the grid points in the hood area. However, these prediction results may deviate from the actual head injury classification results. In order to quantify this deviation, it is necessary to calculate the loss value between the prediction result and the actual result. The loss value is usually used to measure the prediction performance of the model. The smaller the loss value, the closer the model's prediction result is to the actual result, that is, the stronger the model's prediction ability is.

[0093] During the training process, a training requirement or target loss value is set. This requirement or target is set based on actual needs and acceptable error range. As the training progresses, the model's prediction ability will gradually improve and the loss value will gradually decrease. When the loss value reaches or is lower than the set training requirement, it can be considered that the model has sufficient prediction ability, and the hood grid point prediction model training can be confirmed. The parameter optimization formula can be as follows (2):

[0094] (2)

[0095] For example, the optimal parameter configuration of each prediction model can be obtained according to the minimum loss function. After the hood grid point prediction model is trained, it can be applied to the actual vehicle safety performance evaluation.

[0096] The embodiment of the present application can quickly predict the head injury classification results of the grid points in the hood area by inputting the feature data of the head experimental area of ​​the vehicle simulation model into the trained model, thereby greatly saving evaluation time and improving evaluation efficiency.

[0097] Optionally, the hood grid point prediction model includes at least two different classification models, and step 202 includes:

[0098] Step 2021, input the sample feature data into each of the classification models respectively to obtain the prediction results corresponding to each of the classification models.

[0099] Step 2022: Merge the multiple prediction results to obtain the prediction result of the hood grid point prediction model.

[0100] In an embodiment of the present application, the collected sample feature data (including various parameters and features of the hood area grid points extracted from the vehicle simulation model) are input into each classification model respectively. These feature data may include the three-dimensional coordinates of the grid points, material thickness, collision angle, etc., and may also include second features designed according to the experience of engineers, such as the hollowing ratio of the inner plate, the proportion of the opening area, etc. After receiving the input feature data, each classification model will make predictions based on its own algorithm and parameter settings, and output the prediction results of the model for the classification of head injuries at the hood grid points. Since different classification models may use different algorithms and optimization strategies, their prediction results may be different. After obtaining the prediction results of all classification models, it is necessary to use a certain merging strategy (also called a combination function or strategy) to merge these results to generate the final prediction results of the hood grid point prediction model. The choice of merging strategy depends on the specific algorithm type of ensemble learning (such as Bagging, Boosting, Stacking, etc.). For example, in the Bagging method, voting (such as majority voting) is usually used to merge the prediction results; while in the Boosting method, weighted merging may be performed based on the prediction accuracy and weight of each classification model.

[0101] For example, the training process can be expressed as follows:

[0102] (3)

[0103] Among them, F is the final integrated learning model, is the t-th base model,

[0104] C is a function used to combine the prediction results of multiple base models into the final prediction result according to a certain combination strategy Strategy.

[0105] The hood grid point prediction model provided in the embodiment of the present application can fully utilize the advantages of different classification models, improve the accuracy and stability of the prediction, and thus provide more reliable results for vehicle safety performance evaluation.

[0106] Optionally, the feature data includes: first feature data and second feature data, and step 102 includes:

[0107] Step 1021, reading or measuring the head test area of ​​the simulated vehicle model to obtain first characteristic data.

[0108] Step 1022, receiving input feature data for the simulated vehicle model and obtaining second feature data, wherein the input feature data includes at least one of inner panel hollowing ratio, total number of inner panel openings, number of openings with an area greater than a fixed value 1, number of openings with an area less than a fixed value 2, ratio of irregular polygonal opening area, ratio of rectangular opening area, ratio of triangular opening area, number of inner panel reinforcement ribs, and number of connection points of inner panel reinforcement ribs.

[0109] In the embodiment of the present application, the first characteristic data is directly obtained by reading or measuring the head test area of ​​the simulated vehicle model. These data generally include basic information such as the three-dimensional coordinates of the grid points, the area position number, the head type, the head weight, the material thickness, the collision angle, and the collision speed. This information is directly given by the simulation model or can be obtained by simple measurement.

[0110] Exemplarily, the first characteristic data refers to data directly read or measured from the simulation model, such as the following Table 1, Table 2, Table 3, Table 4, Table 5, and Table 6, wherein Table 1 is local characteristic data of the grid points, and Table 2, Table 3, Table 4, Table 5, and Table 6 are overall characteristic data of the hood:

[0111] Table 1

[0112]

[0113] Table 2

[0114]

[0115] Table 3

[0116]

[0117] Table 4

[0118]

[0119] Table 5

[0120]

[0121] Table 6

[0122]

[0123] The second characteristic data is further designed, measured and calculated based on the experience and knowledge of engineers. These data are not read directly from the simulation model, but need to be obtained through additional analysis and processing. Specifically, the second characteristic data includes the hollowing ratio of the inner panel, the total number of inner panel openings, the proportion of opening areas in different ranges, such as the number greater than a fixed value 1, the number less than a fixed value 2, and the proportion of opening areas of different shapes, such as irregular polygons, rectangles, triangles, the number of inner panel reinforcement ribs and the number of connection points. These second characteristic data reflect the more detailed structural characteristics of the experimental area of ​​the head of the simulated vehicle model, and can more comprehensively describe the physical characteristics and mechanical properties of the area, thereby helping to improve the accuracy and reliability of the prediction model.

[0124] Exemplarily, the second characteristic data includes but is not limited to the inner panel hollowing ratio, the total number of inner panel openings, the number of openings with an area greater than a fixed value 1, the number of openings with an area less than a fixed value 2, the proportion of irregular polygonal openings, the proportion of rectangular openings, the proportion of triangular openings, the number of inner panel reinforcement ribs, and the number of inner panel reinforcement rib connection points.

[0125] The sample data set is input into the standardization processor for data preprocessing to obtain the input data set of the prediction model. The data preprocessing method is as follows:

[0126] Count all material types and convert them into numerical values ​​using one-hot encoding.

[0127] Type data: For the type data of "whether to set", 1 represents yes and 0 represents no.

[0128] The numerical data is standardized, and the standardized processor is trained by the previous 2-3 versions of the simulation model of the vehicle model and the obtained characteristic data.

[0129] Optionally, the first feature data includes: local feature data and overall feature data, and the step 1021 includes: measuring the grid points to obtain the local feature data, and measuring the hood area to obtain the overall feature data.

[0130] In the embodiment of the present application, the overall feature data is the feature data used in the hood model that affects the degree of head injury to pedestrians when a vehicle collides with a pedestrian, such as the feature data of the hinge reinforcement plate, the feature data of the lock reinforcement plate, the feature data of the outer plate reinforcement plate, etc., which can be set according to actual needs and are not limited here. The local feature data is the feature data used to reflect the grid points that affect the degree of head injury to pedestrians when a vehicle collides with a pedestrian.

[0131] Optionally, step 104 includes: calculating a total head injury score of the head experimental area based on a proportion of hood head injury classification results and non-hood head injury classification results of each grid point in the head experimental area.

[0132] In the embodiment of the present application, after obtaining the head injury classification results and scores of all grid points, it is necessary to count the proportion of each score interval in the head experimental area. For example, the number of grid points with scores of 1, 2, 3, etc. can be counted, and their proportions of the total number of grid points in the head experimental area can be calculated.

[0133] According to the scoring rules and point allocation scheme specified in the NCAP regulations (such as C-NCAP, E-NCAP, etc.), the proportion of each scoring interval is converted into specific points. Specifically, the total head injury score of the head test area is calculated based on the proportion of each scoring interval and the points corresponding to the interval.

[0134] Optionally, the step 104 includes:

[0135] Step 1041, calculating the classification scores corresponding to the hood head injury classification results and the non-hood head injury classification results corresponding to each grid point in the head test area.

[0136] Step 1042, based on the proportion of the different hood head injury classification results and the non-hood head injury classification results corresponding to all grid points, determine the head injury score of the head experimental area.

[0137] In the embodiment of the present application, the head injury score percentage of the head test area is calculated based on the total score and the maximum score that can be obtained in the head test area, that is, the total score when all grid points are the highest score. This percentage is usually obtained by dividing the total score by the maximum score that can be obtained and then multiplying by 100. In some cases, in order to be consistent with NCAP regulations, this percentage may be converted into a final score (such as 0-10 points), which usually involves further conversion and rounding.

[0138] Exemplarily, assume that the maximum score of the head shape test area is 10 points and the minimum score is 0 points. The maximum score that can be obtained for each grid point or area in the head shape test area is 1.000, and the minimum score that can be obtained is 0.000. The head evaluation index is HIC15. Five intervals are set according to the head evaluation index HIC15 value. Each interval corresponds to a different score and is represented by a different color. The sum of the scores obtained by all grid points or areas in the head shape test area is divided by the maximum score that can be obtained for all grid points or areas to obtain the head shape test score percentage. Multiply the percentage by 10 to obtain the final score of the head shape test area, which is rounded to three decimal places.

[0139] Based on the same inventive concept, the embodiment of the present application also provides a pedestrian head simulation optimization device based on ensemble learning for implementing the pedestrian head simulation optimization method based on ensemble learning involved above. The implementation scheme for solving the problem provided by the device is similar to the implementation scheme recorded in the above method, so the specific limitations in one or more embodiments of the pedestrian head simulation optimization device based on ensemble learning provided below can refer to the limitations of the pedestrian head simulation optimization method based on ensemble learning above, and will not be repeated here.

[0140] In an exemplary embodiment, Figure 3 As shown, a pedestrian head simulation optimization device 30 based on ensemble learning is provided, comprising:

[0141] An acquisition module 301 is used to divide a head test area in a vehicle simulation model into a plurality of grid points, wherein the head test area at least includes: a hood area and a non-hood area;

[0142] Acquire characteristic data of the vehicle simulation model from the head test area;

[0143] A prediction module 302 is used to input the feature data into a hood grid point prediction model to obtain a hood head injury classification result corresponding to each grid point in the hood area, and to perform a simulation experiment on the non-hood area to obtain a non-hood head injury classification result for each grid point in the non-hood area;

[0144] The processing module 303 is used to obtain the total score of the head injury of the head experimental area based on the non-hood head injury classification result and the hood head injury classification result.

[0145] Optionally, the prediction module 302 is further configured to:

[0146] Acquire sample characteristic data of the head experimental area in the sample simulation model;

[0147] Inputting the sample feature data into an initial hood grid point prediction model for training to obtain a prediction result;

[0148] Obtaining a loss value of the prediction result;

[0149] When the loss value reaches the training requirement, it is confirmed that the training of the hood grid point prediction model is completed.

[0150] Optionally, the hood grid point prediction model includes at least two different classification models;

[0151] The prediction module 302 is further used for:

[0152] Inputting the sample feature data into each of the classification models respectively to obtain the prediction results corresponding to each of the classification models;

[0153] The plurality of prediction results are combined to obtain the prediction result of the hood grid point prediction model.

[0154] Optionally, the feature data includes: first feature data and second feature data;

[0155] The acquisition module 301 is further used for:

[0156] Reading or measuring the head experimental area of ​​the simulated vehicle model to obtain first characteristic data;

[0157] Receive input feature data for the simulated vehicle model to obtain second feature data, wherein the input feature data includes at least one of inner panel hollowing ratio, total number of inner panel openings, number of openings with an area greater than a fixed value 1, number of openings with an area less than a fixed value 2, ratio of irregular polygonal openings, ratio of rectangular openings, ratio of triangular openings, number of inner panel reinforcement ribs, and number of connection points of inner panel reinforcement ribs.

[0158] Optionally, the first feature data includes: local feature data and overall feature data;

[0159] The acquisition module 301 is further used for:

[0160] The grid points are measured to obtain local feature data, and the hood area is measured to obtain overall feature data.

[0161] Optionally, the processing module 303 is further configured to:

[0162] Based on the proportion of hood head injury classification results and non-hood head injury classification results of each grid point in the head test area, the total head injury score of the head test area is calculated.

[0163] Optionally, the processing module 303 is further configured to:

[0164] Calculate the classification scores corresponding to the hood head injury classification results and the non-hood head injury classification results corresponding to each grid point in the head test area;

[0165] The head injury score of the head experimental area is determined based on the proportion of the different hood head injury classification results and the non-hood head injury classification results corresponding to all the grid points.

[0166] The embodiment of the present application divides the head test area in the vehicle simulation model into a hood area and a non-hood area, uses a hood grid point prediction model to perform rapid prediction on the hood area, and uses simulation tests to obtain results on the non-hood area, and then combines the non-hood head injury classification results corresponding to each grid point in the hood area and the non-hood area with the hood head injury classification results to obtain the head injury score of the head test area, which not only makes the prediction more refined and accurate, but also can verify and improve the vehicle design scheme more quickly, thereby accelerating the development of vehicle products.

[0167] In an exemplary embodiment, a computer device is provided. The computer device may be a server or a terminal. The internal structure diagram thereof may be as follows: Figure 4 As shown. The computer device includes a processor, a memory, an input / output interface (Input / Output, referred to as I / O) and a communication interface. The processor, the memory and the input / output interface are connected through a system bus, and the communication interface is connected to the system bus through the input / output interface. The processor of the computer device is used to provide computing and control capabilities. The memory of the computer device includes a non-volatile storage medium and an internal memory. The non-volatile storage medium stores an operating system, a computer program and a database. The internal memory provides an environment for the operation of the operating system and the computer program in the non-volatile storage medium. The database of the computer device is used to store pedestrian head simulation optimization data based on integrated learning. The input / output interface of the computer device is used to exchange information between the processor and an external device. The communication interface of the computer device is used to communicate with an external terminal through a network connection. When the computer program is executed by the processor, a pedestrian head simulation optimization method based on integrated learning is implemented.

[0168] Those skilled in the art will understand that Figure 4 The structure shown in the figure is only a block diagram of a part of the structure related to the solution of the present application, and does not constitute a limitation on the computer device to which the solution of the present application is applied. The specific computer device may include more or fewer components than those shown in the figure, or combine certain components, or have a different arrangement of components.

[0169] In an exemplary embodiment, a computer device is further provided, including a memory and a processor, wherein a computer program is stored in the memory, and the processor implements the steps in the above-mentioned method embodiments when executing the computer program.

[0170] In an exemplary embodiment, a computer-readable storage medium is provided, storing a computer program, and when the computer program is executed by a processor, the steps in the above method embodiments are implemented.

[0171] In an exemplary embodiment, a computer program product is provided, including a computer program, and when the computer program is executed by a processor, the steps in the above method embodiments are implemented.

[0172] It should be noted that the user information (including but not limited to user device information, user personal information, etc.) and data (including but not limited to data used for analysis, stored data, displayed data, etc.) involved in this application are all information and data authorized by the user or fully authorized by all parties, and the collection, use and processing of relevant data must comply with relevant regulations.

[0173] Those skilled in the art can understand that all or part of the processes in the above-mentioned embodiment methods can be completed by instructing the relevant hardware through a computer program, and the computer program can be stored in a non-volatile computer-readable storage medium. When the computer program is executed, it can include the processes of the embodiments of the above-mentioned methods. Among them, any reference to the memory, database or other medium used in the embodiments provided in this application can include at least one of non-volatile and volatile memory. Non-volatile memory can include read-only memory (ROM), magnetic tape, floppy disk, flash memory, optical memory, high-density embedded non-volatile memory, resistive random access memory (ReRAM), magnetoresistive random access memory (MRAM), ferroelectric random access memory (FRAM), phase change memory (PCM), graphene memory, etc. Volatile memory can include random access memory (RAM) or external cache memory, etc. By way of illustration and not limitation, RAM may be in various forms, such as Static Random Access Memory (SRAM) or Dynamic Random Access Memory (DRAM).

[0174] The database involved in each embodiment provided in this application may include at least one of a relational database and a non-relational database. The non-relational database may include a distributed database based on blockchain, etc., but is not limited thereto. The processor involved in each embodiment provided in this application may be a general-purpose processor, a central processing unit, a graphics processor, a digital signal processor, a programmable logic device, a data processing logic device based on quantum computing, etc., but is not limited thereto.

[0175] The technical features of the above embodiments may be combined arbitrarily. To make the description concise, not all possible combinations of the technical features in the above embodiments are described. However, as long as there is no contradiction in the combination of these technical features, they should be considered to be within the scope of this specification.

[0176] This article uses specific examples to illustrate the principles and implementation methods of this application. The description of the above embodiments is only used to help understand the method and core ideas of this application. At the same time, for those skilled in the art, according to the ideas of this application, there will be changes in the specific implementation methods and application scope. In summary, the content of this specification should not be understood as limiting this application.

Claims

1. A pedestrian head simulation optimization method based on ensemble learning, characterized in that: The pedestrian head simulation optimization method based on ensemble learning includes: Dividing a head test area in the vehicle simulation model into a plurality of grid points, wherein the head test area at least includes: a hood area and a non-hood area; Acquire characteristic data of the vehicle simulation model from the head test area; Input the characteristic data into the hood grid point prediction model to obtain the hood head injury classification results corresponding to each grid point in the hood area, and perform simulation experiments on the non-hood area to obtain the non-hood head injury classification results of each grid point in the non-hood area, wherein the hood grid point prediction model is a model constructed based on an ensemble learning algorithm; Based on the non-hood head injury classification result and the hood head injury classification result, a total head injury score of the head experimental area is obtained.

2. The pedestrian head simulation optimization method based on ensemble learning according to claim 1 is characterized in that: The hood grid point prediction model is trained by the following steps: Acquire sample characteristic data of the head experimental area in the sample simulation model; Inputting the sample feature data into an initial hood grid point prediction model for training to obtain a prediction result; Obtaining a loss value of the prediction result; When the loss value reaches the training requirement, it is confirmed that the training of the hood grid point prediction model is completed.

3. The pedestrian head simulation optimization method based on ensemble learning according to claim 2 is characterized in that: The hood grid point prediction model includes at least two different classification models; The step of inputting the sample feature data into the initial hood grid point prediction model for training to obtain a prediction result comprises: Inputting the sample feature data into each of the classification models respectively to obtain the prediction results corresponding to each of the classification models; The plurality of prediction results are combined to obtain the prediction result of the hood grid point prediction model.

4. The pedestrian head simulation optimization method based on ensemble learning according to claim 1, characterized in that: The characteristic data includes: first characteristic data and second characteristic data; The step of acquiring characteristic data of the vehicle simulation model from the head experimental area comprises: Reading or measuring the head experimental area of ​​the simulated vehicle model to obtain first characteristic data; Receive input feature data for the simulated vehicle model to obtain second feature data, wherein the input feature data includes at least one of inner panel hollowing ratio, total number of inner panel openings, number of openings with an area greater than a fixed value 1, number of openings with an area less than a fixed value 2, ratio of irregular polygonal openings, ratio of rectangular openings, ratio of triangular openings, number of inner panel reinforcement ribs, and number of connection points of inner panel reinforcement ribs.

5. The pedestrian head simulation optimization method based on ensemble learning according to claim 4 is characterized in that: The first feature data includes: local feature data and overall feature data; The step of reading or measuring the head experimental area of ​​the simulated vehicle model to obtain the first characteristic data includes: The grid points are measured to obtain local feature data, and the hood area is measured to obtain overall feature data.

6. The pedestrian head simulation optimization method based on ensemble learning according to claim 1, characterized in that: The step of obtaining the total score of the head injury in the head experimental area based on the non-hood head injury classification result and the hood head injury classification result comprises: Based on the proportion of hood head injury classification results and non-hood head injury classification results of each grid point in the head test area, the total head injury score of the head test area is calculated.

7. The pedestrian head simulation optimization method based on ensemble learning according to claim 6, characterized in that: The step of calculating the total score of the head injury in the head experimental area based on the proportion of the hood head injury classification results and the non-hood head injury classification results of each grid point in the head experimental area comprises: Calculate the classification scores corresponding to the hood head injury classification results and the non-hood head injury classification results corresponding to each grid point in the head test area; The head injury score of the head experimental area is determined based on the proportion of the different hood head injury classification results and the non-hood head injury classification results corresponding to all the grid points.

8. A pedestrian head simulation optimization device based on ensemble learning, characterized in that: The pedestrian head simulation optimization device based on ensemble learning comprises: An acquisition module is used to divide a head test area in the vehicle simulation model into a plurality of grid points, wherein the head test area at least includes: a hood area and a non-hood area; Acquire characteristic data of the vehicle simulation model from the head test area; A prediction module, used for inputting the characteristic data into a hood grid point prediction model to obtain a hood head injury classification result corresponding to each grid point in the hood area, and performing a simulation experiment on the non-hood area to obtain a non-hood head injury classification result of each grid point in the non-hood area, wherein the hood grid point prediction model is a model constructed based on an ensemble learning algorithm; A processing module is used to obtain a total score of head injury in the head experimental area based on the non-hood head injury classification result and the hood head injury classification result.

9. A computer device comprising: A memory, a processor, and a computer program stored in the memory and executable on the processor, wherein the processor executes the computer program to implement the steps of the pedestrian head simulation optimization method based on ensemble learning described in any one of claims 1 to 7.

10. A computer-readable storage medium having a computer program stored thereon, characterized in that: When the computer program is executed by a processor, the steps of the pedestrian head simulation optimization method based on ensemble learning described in any one of claims 1 to 7 are implemented.

Images