Model training method and device, use method and device, equipment and storage medium

Abstract

The invention relates to a model training method and device, a use method and device, equipment and a storage medium, and the method comprises the steps: inputting a first physical parameter and first performance data into a first model, and obtaining a plurality of association relationships between the first physical parameter and the first performance data; the first physical parameter is input into a value taking module to obtain a second physical parameter, the value taking module calculates a target value taking set based on the first physical parameter, and the second physical parameter is selected from the target value taking set; inputting the second physical parameters into the first model to obtain a plurality of pieces of second performance data; and updating the first model based on the multiple pieces of second performance data to obtain a second model. In this way, the technical problems of low precision and efficiency of model training, large resource consumption and the like in the prior art can be solved.

Description

Technical Field

[0001] The present disclosure relates to the field of computer technology, and in particular to a model training method, a use method, an apparatus, a device, and a storage medium. Background Art

[0002] With the rapid development of technologies such as artificial intelligence, big data and cloud computing, vehicle optimization design has gradually become intelligent. It uses intelligent algorithms, simulation technologies and other means to achieve intelligent optimization of the vehicle design process to improve vehicle design efficiency and product performance. Among them, vehicle performance prediction is one of the important technologies in vehicle optimization design. It mainly provides a scientific basis for vehicle design, manufacturing and maintenance by predicting vehicle performance.

[0003] At present, most of the existing vehicle performance prediction solutions are based on proxy models. During the implementation process, a large number of design variables and performance data need to be input into the machine learning algorithm, and it is necessary to combine manual experience and actual conditions for analysis, select a suitable proxy model, and then train it to obtain a prediction model. During the training process, users need to manually intervene to adjust parameters, resulting in low accuracy and efficiency of model training, and high resource consumption. Summary of the invention

[0004] In order to overcome the problems existing in the related technologies, the present disclosure provides a model training method, usage method, device, equipment and storage medium to solve the technical problems existing in the above-mentioned related technologies such as low accuracy and efficiency of model training and high resource consumption.

[0005] According to a first aspect of an embodiment of the present disclosure, a model training method is provided, including:

[0006] Inputting a first physical parameter and a first performance data into a first model to obtain a plurality of association relationships between the first physical parameter and the first performance data;

[0007] Inputting the first physical parameter into a value obtaining module to obtain a second physical parameter, wherein the value obtaining module calculates a target value set based on the first physical parameter, and selects the second physical parameter from the target value set;

[0008] Inputting the second physical parameter into the first model to obtain a plurality of second performance data;

[0009] Based on the plurality of second performance data, the first model is updated to obtain a second model.

[0010] In some embodiments, inputting the first physical parameter and the first performance data into the first model to obtain multiple associations between the first physical parameter and the first performance data includes:

[0011] The first physical parameter and the first performance data are generated by a third model, and the third model is used to perform simulation analysis on a vehicle structure.

[0012] In some embodiments, the step of inputting the first physical parameter into a value obtaining module to obtain a second physical parameter, the value obtaining module being used to calculate a target value set based on the first physical parameter, and selecting the second physical parameter from the target value set includes:

[0013] The value taking module calculates the target value set according to the first physical parameter and the preset value range, and performs DoE sampling within the target value set to obtain a second preprocessing physical parameter;

[0014] The second preprocessed physical parameters are input into a third model for simulation analysis to generate the second physical parameters.

[0015] In some embodiments, the first physical parameter is input into a value obtaining module to obtain a second physical parameter, the value obtaining module calculates a target value set based on the first physical parameter, and before selecting the second physical parameter from the target value set, the method further includes:

[0016] The first physical parameter is preprocessed, and the preprocessing includes at least one of data cleaning, standardization, data conversion, and outlier processing.

[0017] In some embodiments, the method further comprises:

[0018] Generate a third physical parameter and a third performance data through the third model, input the third physical parameter into the second model to obtain fourth performance data, perform a difference calculation between the fourth performance data and the third performance data, and output multiple fourth performance data that meet the difference threshold.

[0019] In some embodiments, the method further comprises:

[0020] The association relationships corresponding to the plurality of fourth performance data are post-processed to obtain an optimal solution to the association relationships, wherein the post-processing includes at least one of a voting method, an averaging method, and a ranking method.

[0021] According to a second aspect of an embodiment of the present disclosure, a model using method is provided, including:

[0022] The physical parameters are input into the second model for performance prediction to obtain predicted performance data corresponding to the physical parameters to be processed; wherein the second model is obtained by training using the model training method described in any one of claims 1 to 6 above.

[0023] According to a third aspect of an embodiment of the present disclosure, a model training device is provided, including:

[0024] a processing module configured to input a first physical parameter and a first performance data into a first model to obtain a plurality of associations between the first physical parameter and the first performance data;

[0025] The processing module is further configured to input the first physical parameter into a value obtaining module to obtain a second physical parameter, the value obtaining module calculates a target value set based on the first physical parameter, and selects the second physical parameter from the target value set;

[0026] The processing module is further configured to input the second physical parameter into the first model to obtain a plurality of second performance data;

[0027] The processing module is further configured to update the first model based on the plurality of second performance data to obtain a second model.

[0028] According to a fourth aspect of an embodiment of the present disclosure, there is provided a model using device, including:

[0029] The processing module is configured to input the physical parameters into the second model for performance prediction to obtain predicted performance data corresponding to the physical parameters; wherein the second model is obtained by training using the model training method described in any one of claims 1 to 6 above.

[0030] According to a fifth aspect of an embodiment of the present disclosure, a computer device is provided, comprising: a processor; a memory for storing processor executable instructions; wherein the processor is configured to execute the executable instructions to implement the steps of the above-mentioned model training method; or the processor is configured to execute the executable instructions to implement the steps of the above-mentioned model usage method.

[0031] According to the sixth aspect of an embodiment of the present disclosure, a computer-readable storage medium is provided, on which computer program instructions are stored, and when the program instructions are executed by a processor, the steps of the model training method provided in the first aspect of the present disclosure are implemented; or, when the program instructions are executed by a processor, the steps of the model usage method provided in the second aspect of the present disclosure are implemented.

[0032] The technical solution provided by the embodiments of the present disclosure may include the following beneficial effects: the computer device inputs the first physical parameter and the first performance data into the first model to obtain multiple associations between the first physical parameter and the first performance data; the first physical parameter is input into the value module to obtain the second physical parameter, the value module calculates the target value set based on the first physical parameter, and selects the second physical parameter from the target value set; the second physical parameter is input into the first model to obtain multiple second performance data; based on the multiple second performance data, the first model is updated to obtain the second model. It can be seen that the computer device can give multiple associations between the first physical parameter and the first performance data based on the first model, perform data expansion on the limited first physical parameter based on the value module to obtain the second physical parameter, and then train and update the above-mentioned first model based on the second physical parameter and its corresponding multiple second performance data to obtain a second model with higher accuracy. Compared with the above-mentioned related technologies, this embodiment does not need to input a large amount of learning data, saves the time of simulation to generate learning data, and does not require human participation. The model training is fully automated, which can not only solve the technical problems of low accuracy and efficiency of model training and large resource consumption in related technologies, but also improve the intelligence and reliability of model training, and at the same time improve the accuracy and efficiency of model training.

[0033] It is to be understood that the foregoing general description and the following detailed description are exemplary and explanatory only and are not restrictive of the present disclosure. BRIEF DESCRIPTION OF THE DRAWINGS

[0034] The accompanying drawings, which are incorporated in and constitute a part of this specification, illustrate embodiments consistent with the present disclosure and, together with the description, serve to explain the principles of the present disclosure.

[0035] Figure 1 It is a flowchart of a model training method according to an exemplary embodiment.

[0036] Figure 2 The figure is a flowchart of a method for using a model according to an exemplary embodiment.

[0037] Figure 3 It is a structural schematic diagram of a model training device according to an exemplary embodiment.

[0038] Figure 4 The figure is a schematic diagram of the structure of a model using device according to an exemplary embodiment.

[0039] Figure 5 The diagram is a schematic structural diagram of a computer device according to an exemplary embodiment.

[0040] Figure 6The figure is a schematic diagram showing the structure of a chip according to an exemplary embodiment. DETAILED DESCRIPTION

[0041] Exemplary embodiments will be described in detail herein, examples of which are shown in the accompanying drawings. When the following description refers to the drawings, the same numbers in different drawings represent the same or similar elements unless otherwise indicated. The embodiments described in the following exemplary embodiments do not represent all embodiments consistent with the present disclosure. Instead, they are merely examples of devices and methods consistent with some aspects of the present disclosure as detailed in the appended claims.

[0042] It should be noted that all actions of acquiring signals, information or data in the present disclosure are carried out in compliance with the relevant data protection laws and policies of the country where the device is located and with the authorization given by the owner of the corresponding device.

[0043] Vehicle optimization design refers to improving the vehicle's design and manufacturing process to improve the vehicle's performance, safety, fuel economy, and environmental performance. Among them, vehicle performance prediction is an important technology in vehicle optimization design. Currently, there are two existing vehicle performance prediction schemes.

[0044] The first is vehicle performance prediction based on simulation. This solution mainly uses the finite element method to simulate and analyze the vehicle and its components to obtain vehicle-related performance data. However, this solution takes too long to simulate once, and it is impossible to obtain corresponding performance data immediately / quickly due to changes in vehicle-related design variables.

[0045] The second method is to predict vehicle performance based on proxy models. This method requires manual analysis based on experience and actual conditions, selecting a suitable proxy model, and then training the selected proxy model. During the training process, manual intervention is required to adjust parameters, resulting in low efficiency of performance prediction and high resource consumption.

[0046] To solve the above problems, the present disclosure proposes a model training method, a method of use, an apparatus, a device, a storage medium and a chip. Figure 1 FIG. 1 is a flow chart of a model training method according to an exemplary embodiment. Figure 1 The method shown is applied to a computer device, and the method includes the following implementation steps:

[0047] S101. Input a first physical parameter and first performance data into a first model to obtain a plurality of association relationships between the first physical parameter and the first performance data.

[0048] The above-mentioned first physical parameter of the present disclosure may refer to a parameter or multiple parameters (also referred to as a set of parameters) used to describe the vehicle or vehicle performance, etc. For example, it may include the size data, weight data, material data, thickness data or other parameter data of one or more components in the vehicle. The above-mentioned first performance data refers to relevant data used to describe vehicle performance, such as vehicle modal performance data, vehicle stiffness performance data, collision safety performance data or other vehicle performance data. The above-mentioned first model is a meta-model customized by a computer device according to actual needs, which may include but is not limited to, for example, a linear regression model, a logistic regression model, a neural network model, a Bayesian model, a Gaussian process regression model, an extreme gradient boosting (XGBoost) model or other customized mathematical models, etc., which are not limited in the present disclosure.

[0049] During the implementation process, the computer device can input the above-mentioned first physical parameter and the first performance data into the first model for processing, and output multiple associations between the above-mentioned first physical parameter and the above-mentioned first performance data, which are used to reflect the performance data prediction function between the physical parameter and the performance data. There can be multiple performance data prediction functions between a set of first physical parameters and the first performance data. For example, the first physical parameter is the thickness of a set of multiple parts with a connection relationship, and the first performance data is the body stiffness value. After the thickness of the multiple parts and the body stiffness value are input into the first model, multiple performance data prediction functions related to the multiple parts and performance data can be output. It can be understood that when the input first performance data is the same, but the combination of parts in the input first physical parameter is different (that is, the difference in the input first physical parameter is due to the different combination of parts), the output associations are different. For example, when the first input performance data is the body stiffness value, and the first input physical parameter includes the body roof thickness, engine hood outer panel thickness, door outer panel thickness and fender thickness, and when the first physical parameter includes the engine baffle thickness, floor thickness, engine hood inner panel thickness and door inner panel thickness, the output performance prediction function (i.e., the correlation relationship) is different; when the first input performance data is the same, but the physical parameters of the components in the first input physical parameters are different (i.e., the input component combination is the same but the physical parameters are different), the output correlation relationship is different. For example, when the first performance data input is the body stiffness value, and the first physical parameter input includes the body roof thickness, engine hood outer panel thickness, door outer panel thickness and fender thickness, and when the first physical data includes the engine baffle size, floor size, engine hood inner panel size and door inner panel size, the output performance prediction function (i.e., correlation relationship) is different; when the input first physical parameter is the same, but the input first performance data is different, the output correlation relationship is also different. For example, when the input first physical parameter is the body roof thickness, engine hood outer panel thickness, door outer panel thickness and fender thickness, but the input first performance data is the body stiffness value and when the first performance data is the body modal performance data, the output performance prediction function (i.e., correlation relationship) is different.

[0050] In one embodiment, the first physical parameter and the first performance data may be generated by a third model, and the third model may be used to perform simulation analysis on the vehicle structure. For example, the third model may generally be a vehicle analysis model, such as a vehicle finite element analysis model, etc. In other words, the present disclosure may use the third model to perform simulation analysis on the vehicle structure to be analyzed to obtain the first physical parameter and the first performance data.

[0051] S102: Input the first physical parameter into a value obtaining module to obtain a second physical parameter; the value obtaining module calculates a target value set based on the first physical parameter, and selects the second physical parameter from the target value set.

[0052] During implementation, the value module can perform data calculation based on the above-mentioned first physical parameter and its corresponding preset value range to obtain a target value set. The target value set may include the above-mentioned first physical parameter and some physical parameters calculated in the preset value range. Further, a preset sampling algorithm can be used to sample all physical parameters in the target value set to obtain a second pre-processed physical parameter. The sampling algorithm is a sampling algorithm customized by the device or the user, which may include but is not limited to, for example, a random sampling algorithm, a Latin hypercube sampling algorithm, an experimental design DoE sampling algorithm or other algorithms that support data sampling. Further, the present disclosure can input the above-mentioned second pre-processed physical parameter into the above-mentioned third model for simulation analysis. If the third model is simulated and analyzed, it is judged that the second pre-processed physical parameter and the corresponding performance data are both within a reasonable data range, the second pre-processed physical parameter is output as the second physical parameter. If the third model is simulated and analyzed, it is judged that the second pre-processed physical parameter and / or the corresponding performance data are not within a reasonable data range, the second pre-processed physical parameter is deleted, and an error prompt message is fed back to the value module to apply for adjustment of the preset value range.

[0053] In some optional embodiments, before the first physical parameter is input into the value taking module, the present disclosure may also pre-process the first physical parameter, and the pre-processing may include at least one of the following: data cleaning, standardization, data conversion, outlier processing, feature selection, feature extraction or other customized data processing. Among them, data cleaning may include but is not limited to operations such as cleaning noise data and processing missing values. Data conversion may include but is not limited to data type conversion, data standardization conversion, data scaling processing, etc. Standardization processing may include but is not limited to normalization processing, etc. Outlier processing refers to processing physical data with abnormalities, such as deleting physical data with missing values, or deleting physical data when the physical data exceeds the corresponding preset value range. Feature extraction and feature selection may be part of feature engineering processing, and the present disclosure does not make too many restrictions and details on this.

[0054] S103: Input the second physical parameter into the first model to obtain a plurality of second performance data.

[0055] The present disclosure can input the obtained second physical parameter into the first model for performance prediction to obtain corresponding multiple second performance data. The relevant introduction of the second physical parameter and the second performance data can refer to the relevant introduction of the first physical parameter and the first performance data, which will not be repeated here.

[0056] S104. Based on the plurality of second performance data, update the first model to obtain a second model.

[0057] The present disclosure can utilize the above-mentioned multiple second performance data to update the above-mentioned first model, thereby obtaining the final second model after training.

[0058] In practical applications, the present disclosure may use the second physical parameter and the second performance data as target sample data for training the first model, train and update the first model based on the target sample data, and obtain the trained second model.

[0059] The present disclosure can use the above target sample data to optimize and adjust the hyperparameters of the above first model, and after multiple trainings, a trained second model can be obtained. The above first model involved in the present disclosure is a model to be trained that is customized and selected according to actual conditions, and can meet different vehicle performance prediction requirements. The present disclosure does not limit the type and quantity of the above first model, which can be determined according to actual conditions. For example, the first model may include n performance prediction models, where n is a positive integer set according to actual conditions.

[0060] It can be seen that the present disclosure can use automatic machine learning (AutoML) technology to quickly establish a mapping relationship between design variables (physical parameters) and objective functions (performance data) based on limited sample data (first physical parameters), such as establishing a correlation relationship between the first physical parameters and the first performance data. The purpose is that after a design variable (physical parameter) is given, the corresponding performance data can be quickly and accurately predicted by the trained second model. The present disclosure supports the use of different automatic machine learning technologies, and can automatically learn based on limited sample data without any manual adjustment of the model or hyperparameters. The workflow for establishing machine learning can include, but is not limited to, the above-mentioned data preprocessing, feature engineering, model customization selection, hyperparameter optimization, and model integration.

[0061] In an optional embodiment, the present disclosure can also generate third physical parameters and third performance data through the above-mentioned third model. The relevant introduction of the above-mentioned third physical parameters and the above-mentioned third performance data can refer to the relevant introduction of the above-mentioned first physical parameters and the first performance data, which will not be repeated here.

[0062] During the implementation process, when the third model performs simulation analysis on the vehicle structure to be analyzed, a certain number of physical parameters and performance data will be generated. A part of the physical parameters and performance data will be taken out as the first physical data and the first performance data to train the first model, and another part of the physical parameters and performance parameters will be taken out as the third physical parameters and the third performance data to verify the trained second model. Further, the present disclosure can input the above-mentioned third physical parameter into the above-mentioned second model after training to perform performance prediction to obtain the predicted fourth performance data. Then the above-mentioned fourth performance data (i.e., predicted performance data) and the third performance data (real performance data) are calculated by difference, and multiple fourth performance data that meet the difference threshold are output. The difference threshold is a threshold range that is customized according to actual conditions, and the present disclosure does not limit this. The present disclosure does not limit the specific implementation method of the above-mentioned output of multiple fourth performance data that meet the difference threshold. For example, the smallest fourth performance data in the difference calculation can be output as the optimal solution, and the fourth performance data that is within the difference threshold in the difference calculation can be averaged, and the performance data obtained after processing can be output as the optimal solution, etc. The present disclosure does not make too many restrictions and details on this.

[0063] In another optional embodiment, the present disclosure may post-process the associations corresponding to the above-mentioned multiple fourth performance data to obtain the optimal solution of the associations, where the associations may refer to one or more associations between the fourth performance data and the third physical parameters. The post-processing includes, but is not limited to, for example, a combination of at least one or more of the following: voting method, averaging method, sorting method or other customized processing methods. In other words, the present disclosure may post-process all or part of the associations between the multiple fourth performance data and the multiple third physical parameters, such as voting method, averaging method or sorting method, so as to obtain the associations that are finally recognized or considered to be the optimal solution.

[0064] It should be noted that the model training method involved in the present disclosure can be executed on the computer device (specifically, it can be a training device). The computer device can include, but is not limited to, any one or more of the following according to actual conditions: a terminal device, other devices different from the terminal device (such as a server, a cloud device, etc.), a cloud platform or other devices with model training capabilities, etc. After the computer device has trained the above-mentioned second model, it can be deployed in the end device (such as a terminal device) to facilitate the device to directly call the above-mentioned second model to perform performance prediction of the corresponding vehicle, which can be referred to above Figure 1 The relevant introduction in the embodiments will not be repeated here.

[0065] By implementing the embodiments of the present disclosure, the embodiments of the present disclosure can be combined with automatic machine learning technology to quickly establish a prediction model from design variables to different performance objective functions through a small amount of learning data, and can quickly predict vehicle performance based on the modification of design variables without the need for a complex simulation process. And the use of automatic machine learning technology can automatically train the entire process to obtain a second model with better prediction accuracy, without the need to manually select the type of model and manually adjust parameters, etc., greatly reducing the workload. In a specific implementation, the computer device inputs the first physical parameter and the first performance data into the first model to obtain multiple associations between the first physical parameter and the first performance data; the first physical parameter is input into a value module to obtain a second physical parameter, and the value module calculates a target value set based on the first physical parameter, and selects the second physical parameter from the target value set; the second physical parameter is input into the first model to obtain multiple second performance data; based on the multiple second performance data, the first model is updated to obtain a second model. It can be seen that the computer device can provide multiple correlations between the first physical parameter and the first performance data based on the first model, expand the data of the limited first physical parameter based on the value module, obtain the second physical parameter, and then train and update the above-mentioned first model based on the second physical parameter and its corresponding multiple second performance data to obtain a more accurate second model. Compared with the above-mentioned related technologies, this embodiment does not need to input a large amount of learning data, saves the time of simulating and generating learning data, and does not require human participation. The model training is fully automated, which can not only solve the technical problems of low accuracy and efficiency of model training and large resource consumption in related technologies, but also improve the intelligence and reliability of model training, and at the same time improve the accuracy and efficiency of model training.

[0066] The following are some examples of how to use the model. Figure 2 FIG. 1 is a flow chart showing a method for using a model according to an exemplary embodiment. Figure 2 The method shown is applied to a computer device, and the method includes the following implementation steps:

[0067] S201, input the physical parameters to be processed into the second model for performance prediction, and obtain the predicted performance data corresponding to the physical parameters to be processed. Figure 1 The trained model in the embodiment.

[0068] The computer device can input the physical parameters to be processed into the trained second model for performance prediction, thereby outputting corresponding predicted performance data. Figure 1The relevant introduction in the described embodiments will not be repeated here. The above-mentioned physical parameters to be processed may refer to a certain parameter or multiple parameters (also referred to as a set of parameters) used to describe the vehicle to be processed or the performance of the vehicle, for example, it may include the size data, weight data, material data, thickness data or other parameter data of one or more components in the vehicle. The above-mentioned predicted performance data refers to data related to some performances of the vehicle to be processed, such as vehicle modal performance data (also referred to as durability performance data), vehicle stiffness performance data (also referred to as noise vibration and acoustic roughness (NoiseVibrationHarshness, NVH) performance data), collision safety performance data or other vehicle performance data. For the above-mentioned physical parameters to be processed and the above-mentioned predicted performance data, please refer to the above-mentioned Figure 1 The relevant introduction of the first physical parameter and the first performance data in the embodiment will not be repeated here.

[0069] In some embodiments, the above-mentioned second model can also be applied to multi-objective optimization scenarios. That is to say, there can be multiple second models, and the present disclosure can perform multi-objective optimization on different types of predicted vehicle performance output by multiple second models. In a specific implementation, the present disclosure can determine the constraints of each of the multiple second models, and the constraints include but are not limited to, for example, the constraints of the corresponding physical parameters input into each of the above-mentioned second models, such as the constraints of the vehicle body material thickness data or other custom constraints. Furthermore, the present disclosure can perform multi-objective optimization calculations on the multiple second models based on the determined constraints, so as to obtain the optimization performance corresponding to each of the multiple second models. The present disclosure can adopt a preset multi-objective optimization algorithm to perform multi-objective optimization calculations on the multiple second models based on all the determined constraints, so as to solve the optimization performance corresponding to each of the multiple second models.

[0070] For example, the second model is used as a performance prediction model for predicting vehicle stiffness performance and vehicle modal performance, and the optimization of the vehicle's gross vehicle weight, vehicle stiffness performance, and vehicle modal performance is used as an example. In practical applications, the vehicle stiffness performance involved in the present disclosure may actually include performance such as vehicle torsional stiffness and / or vehicle bending stiffness, and the vehicle modal performance involved in the present disclosure may actually include performance such as vehicle torsional mode and / or vehicle bending mode.

[0071] In the actual process, the present disclosure can first determine the value range of the material thickness data of the S metal sheets constituting the body of the target vehicle, and the four performance prediction models (i.e., the four different second models mentioned above) used to predict the vehicle torsional stiffness, vehicle bending stiffness, vehicle torsional mode, and vehicle bending mode, respectively. Then, the vehicle's total body weight function is determined, for example, the basic weight of each component of the vehicle is determined according to the vehicle finite element model, and then the change ratio between it and the basic material thickness is calculated, the weight of each component is deduced, and then the total amount of each component is summed up, thereby obtaining the target vehicle's total body weight function, which is not limited or detailed in the present disclosure.

[0072] Next, according to the multiple objective functions to be optimized, here are the above four performance prediction models and the above vehicle body weight function, combined with the constraints corresponding to each function, such as the value range of the material thickness data of S metal sheets, etc., a multi-objective genetic algorithm is used to solve the above multiple objective functions, thereby achieving the purpose of multi-objective optimization. In practical applications, the applicable scenarios of multi-objective optimization can be determined according to actual conditions, for example, it may include but is not limited to the following scenarios:

[0073] (1) Taking lightweight body and optimal vehicle performance as the optimization goal, the vehicle stiffness performance and vehicle modal performance are required to be as high as possible. In this case, the optimization goal can be set as: minimizing the total weight function of the vehicle body, maximizing the vehicle torsional stiffness, vehicle bending stiffness, vehicle torsional mode and vehicle bending mode.

[0074] (2) Taking lightweight body and optimal vehicle performance as the optimization goals, the vehicle weight is required to be reduced by x kilograms, and the vehicle stiffness performance and vehicle modal performance are as high as possible. In this case, based on the first scenario above, the constraint of reducing the vehicle weight by x kilograms is added, and the other conditions remain the same.

[0075] (3) Taking lightweight body and optimal vehicle performance as the optimization goals, the vehicle stiffness performance and vehicle modal performance are required to be greater than the corresponding basic thresholds. In this case, based on the first scenario above, additional constraints are added: vehicle torsional stiffness, vehicle bending stiffness, vehicle torsional mode and vehicle bending mode are all greater than the corresponding thresholds, and the others remain the same.

[0076] It should be noted that the above examples of the present disclosure are only for illustrative purposes and do not constitute limitations.

[0077] By implementing the disclosed embodiment, the computer device inputs the physical parameters to be processed into the second model for performance prediction, and obtains the predicted performance data corresponding to the physical parameters to be processed. It can be seen that the computer device can directly use the second model to perform intelligent and rapid performance prediction on the physical parameters to be processed, so that the corresponding performance results can be immediately / quickly predicted when the physical data changes, which is conducive to improving the convenience and efficiency of vehicle performance prediction. At the same time, it also solves the technical problems existing in the existing related technologies, such as the simulation time consumption is too long and it is impossible to immediately / quickly obtain the corresponding performance data due to changes in vehicle-related design variables.

[0078] Based on the above examples, see Figure 3 FIG. 1 is a schematic diagram showing the structure of a model training device according to an exemplary embodiment. Figure 3 The device shown can be applied to a computer device, and the device may include a processing module 301.

[0079] The processing module 301 is configured to input the first physical parameter and the first performance data into the first model to obtain multiple associations between the first physical parameter and the first performance data;

[0080] The processing module 301 is further configured to input the first physical parameter into a value obtaining module to obtain a second physical parameter, the value obtaining module calculates a target value set based on the first physical parameter, and selects the second physical parameter from the target value set;

[0081] The processing module 301 is further configured to input the second physical parameter into the first model to obtain a plurality of second performance data;

[0082] The processing module 301 is further configured to update the first model based on the plurality of second performance data to obtain a second model.

[0083] In some embodiments, the first physical parameter and the first performance data are generated by a third model, and the third model is used to perform simulation analysis on the vehicle structure.

[0084] In some embodiments, the value-taking module calculates the target value set based on the first physical parameter and the preset value range, and performs DoE sampling within the target value set to obtain a second preprocessed physical parameter; the second preprocessed physical parameter is input into a third model for simulation analysis to generate the second physical parameter.

[0085] In some embodiments, the processing module 301 is further configured to preprocess the first physical parameter, and the preprocessing includes at least one of data cleaning, standardization, data conversion, and outlier processing.

[0086] In some embodiments, the processing module 301 is also configured to generate a third physical parameter and a third performance data through the third model, input the third physical parameter into the second model to obtain fourth performance data, perform a difference calculation between the fourth performance data and the third performance data, and output multiple fourth performance data that meet the difference threshold.

[0087] In some embodiments, the processing module 301 is further configured to post-process the association relationships corresponding to the multiple fourth performance data to obtain an optimal solution for the association relationships, and the post-processing includes at least one of a voting method, an averaging method, and a sorting method.

[0088] Regarding the device in the above embodiment, the specific manner in which each module performs operations has been described in detail in the embodiment of the relevant model training method, and will not be elaborated here.

[0089] See also Figure 4 FIG. 1 is a schematic diagram showing a structure of a model using device according to an exemplary embodiment. Figure 4 The device shown can be applied to a computer device, and the device may include a processing module 401.

[0090] The processing module 401 is configured to input the physical parameters to be processed into the second model for performance prediction, and obtain predicted performance data corresponding to the physical parameters to be processed; wherein the second model is a model trained using the above model training method.

[0091] Regarding the device in the above embodiment, the specific manner in which each module performs operations has been described in detail in the embodiment of the model use method, and will not be elaborated here.

[0092] The present disclosure also provides a computer-readable storage medium having computer program instructions stored thereon, which, when executed by a processor, implement the steps of the model training method or model using method provided by the present disclosure.

[0093] Figure 5 1 is a schematic diagram of a computer device according to an exemplary embodiment. The computer device involved in the present disclosure may include, but is not limited to, a server, a mobile phone, a computer, a handheld computer, a digital broadcast terminal, a message transceiver device, a game console, a medical device, a fitness device, a personal digital assistant or other terminal devices. Figure 5 The computer device is taken as a terminal device (such as a handheld computer) as an example, but it does not constitute a limitation.

[0094] Reference Figure 5The terminal device 500 may include one or more of the following components: a processing component 502 , a memory 504 , a power component 506 , a multimedia component 508 , an audio component 510 , an input / output interface 512 , a sensor component 514 , and a communication component 516 .

[0095] The processing component 502 generally controls the overall operation of the device 500, such as operations associated with display, phone calls, data communications, camera operations, and recording operations. The processing component 502 may include one or more processors 520 to execute instructions to complete all or part of the steps of the above-mentioned model training method / model use method. In addition, the processing component 502 may include one or more modules to facilitate the interaction between the processing component 502 and other components. For example, the processing component 502 may include a multimedia module to facilitate the interaction between the multimedia component 508 and the processing component 502.

[0096] The memory 504 is configured to store various types of data to support operations on the device 500. Examples of such data include instructions for any application or method operating on the device 500, contact data, phone book data, messages, pictures, videos, etc. The memory 504 can be implemented by any type of volatile or non-volatile storage device or a combination thereof, such as static random access memory (SRAM), electrically erasable programmable read-only memory (EEPROM), erasable programmable read-only memory (EPROM), programmable read-only memory (PROM), read-only memory (ROM), magnetic memory, flash memory, magnetic disk or optical disk.

[0097] Power supply component 506 provides power to various components of device 500. Power supply component 506 may include a power management system, one or more power supplies, and other components associated with generating, managing, and distributing power to device 500.

[0098] The multimedia component 508 includes a screen that provides an output interface between the device 500 and the user. In some embodiments, the screen may include a liquid crystal display (LCD) and a touch panel (TP). If the screen includes a touch panel, the screen may be implemented as a touch screen to receive input signals from the user. The touch panel includes one or more touch sensors to sense touch, slide, and gestures on the touch panel. The touch sensor may not only sense the boundaries of the touch or slide action, but also detect the duration and pressure associated with the touch or slide operation. In some embodiments, the multimedia component 508 includes a front camera and / or a rear camera. When the terminal device 500 is in an operating mode, such as a shooting mode or a video mode, the front camera and / or the rear camera may receive external multimedia data. Each front camera and the rear camera may be a fixed optical lens system or have a focal length and optical zoom capability.

[0099] The audio component 510 is configured to output and / or input audio signals. For example, the audio component 510 includes a microphone (MIC), and when the device 500 is in an operating mode, such as a call mode, a recording mode, and a speech recognition mode, the microphone is configured to receive an external audio signal. The received audio signal can be further stored in the memory 504 or sent via the communication component 516. In some embodiments, the audio component 510 also includes a speaker for outputting audio signals.

[0100] The input / output interface 512 provides an interface between the processing component 502 and the peripheral interface modules, which may be keyboards, click wheels, buttons, etc. These buttons may include but are not limited to: a home button, a volume button, a start button, and a lock button.

[0101] The sensor assembly 514 includes one or more sensors for providing various aspects of status assessment for the device 500. For example, the sensor assembly 514 can detect the open / closed state of the device 500, the relative positioning of components, such as the display and keypad of the device 500, and the sensor assembly 514 can also detect the position change of the device 500 or a component of the device 500, the presence or absence of user contact with the device 500, the orientation or acceleration / deceleration of the device 500, and the temperature change of the device 500. The sensor assembly 514 may include a proximity sensor configured to detect the presence of nearby objects without any physical contact. The sensor assembly 514 may also include an optical sensor, such as a CMOS or CCD image sensor, for use in imaging applications. In some embodiments, the sensor assembly 514 may also include an acceleration sensor, a gyroscope sensor, a magnetic sensor, a pressure sensor, or a temperature sensor.

[0102] The communication component 516 is configured to facilitate wired or wireless communication between the device 500 and other devices. The device 500 can access a wireless network based on a communication standard, such as WiFi, 2G or 3G, or a combination thereof. In an exemplary embodiment, the communication component 516 receives a broadcast signal or broadcast-related information from an external broadcast management system via a broadcast channel. In an exemplary embodiment, the communication component 516 also includes a near field communication (NFC) module to facilitate short-range communication. For example, the NFC module can be implemented based on radio frequency identification (RFID) technology, infrared data association (IrDA) technology, ultra-wideband (UWB) technology, Bluetooth (BT) technology and other technologies.

[0103] In an exemplary embodiment, the device 500 can be implemented by one or more application-specific integrated circuits (ASICs), digital signal processors (DSPs), digital signal processing devices (DSPDs), programmable logic devices (PLDs), field programmable gate arrays (FPGAs), controllers, microcontrollers, microprocessors, or other electronic components to perform the above-mentioned model training method or model usage method.

[0104] In an exemplary embodiment, a non-transitory computer-readable storage medium including instructions is also provided, such as a memory 504 including instructions, and the instructions can be executed by the processor 520 of the device 500 to complete the above-mentioned model training method or model use method. For example, the non-transitory computer-readable storage medium can be a ROM, a random access memory (RAM), a CD-ROM, a magnetic tape, a floppy disk, an optical data storage device, etc.

[0105] In addition to being an independent electronic device, the above-mentioned device can also be a part of an independent electronic device. For example, in one embodiment, the device can be an integrated circuit (IC) or a chip, wherein the integrated circuit can be an IC or a collection of multiple ICs; the chip can include but is not limited to the following types: GPU (Graphics Processing Unit), CPU (Central Processing Unit), FPGA (Field Programmable Gate Array), DSP (Digital Signal Processor), ASIC (Application Specific Integrated Circuit), SOC (System on Chip, SoC, system on chip or system-level chip), etc. The above-mentioned integrated circuit or chip can be used to execute executable instructions (or codes) to implement the above-mentioned model training method or model use method. The executable instructions can be stored in the integrated circuit or chip, or can be obtained from other devices or equipment, such as the integrated circuit or chip including a processor, a memory, and an interface for communicating with other devices. The executable instruction can be stored in the memory, and when the executable instruction is executed by the processor, the above-mentioned model training method or model usage method is implemented; alternatively, the integrated circuit or chip can receive the executable instruction through the interface and transmit it to the processor for execution, so as to implement the above-mentioned model training method or model usage method.

[0106] In another exemplary embodiment, a computer program product is also provided, which includes a computer program that can be executed by a programmable device, and the computer program has a code portion for executing the above-mentioned model training method or model usage method when executed by the programmable device.

[0107] See also Figure 6 FIG. 1 is a schematic diagram showing the structure of a chip according to an exemplary embodiment. Figure 6 The chip 600 shown includes a processor 601 and an interface 602. Optionally, it may also include a memory 603. The number of the processor 601 may be one or more, and the number of the interface 602 may be more than one.

[0108] In one embodiment, for a case where a chip is used to implement the method embodiment of the present disclosure:

[0109] The interface 602 is used to receive or output signals;

[0110] The processor 601 is used to execute part or all of the content in the model training method or model use method embodiment.

[0111] Understandably, the processor in the embodiment of the present disclosure may be an integrated circuit chip having signal processing capabilities. In the implementation process, each step of the above method embodiment may be completed by an integrated logic circuit of hardware in the processor or an instruction in the form of software. The above processor may be a general-purpose processor, a digital signal processor (DSP), an application-specific integrated circuit (ASIC), a field programmable gate array (FPGA) or other programmable logic devices, discrete gates or transistor logic devices, discrete hardware components.

[0112] It is understandable that the memory in the embodiments of the present disclosure may be a volatile memory or a non-volatile memory, or may include both volatile and non-volatile memories. Among them, the non-volatile memory may be a read-only memory (ROM), a programmable read-only memory (PROM), an erasable programmable read-only memory (EPROM), an electrically erasable programmable read-only memory (EEPROM), or a flash memory. The volatile memory may be a random access memory (RAM), which is used as an external cache. By way of example and not limitation, many forms of RAM are available, such as static RAM (SRAM), dynamic RAM (DRAM), synchronous DRAM (SDRAM), double data rate SDRAM (DDRSDRAM), enhanced SDRAM (ESDRAM), synchronous link DRAM (SLDRAM), and direct RAM (DR RAM). It should be noted that the memory of the systems and methods described herein is intended to include, but is not limited to, these and any other suitable types of memory.

[0113] It should be noted here that the description of the above storage medium, device and chip embodiments is similar to the description of the above method embodiments, and has similar beneficial effects as the method embodiments. For technical details not disclosed in the storage medium, storage medium and device embodiments of the present disclosure, please refer to the description of the method embodiments of the present disclosure for understanding.

[0114] Those skilled in the art will readily appreciate other embodiments of the present disclosure after considering the specification and practicing the present disclosure. The present disclosure is intended to cover any variations, uses or adaptations of the present disclosure that follow the general principles of the present disclosure and include common knowledge or customary techniques in the art that are not disclosed in the present disclosure. The description and examples are to be considered as exemplary only, and the true scope and spirit of the present disclosure are indicated by the following claims.

[0115] It should be understood that the present disclosure is not limited to the exact structures that have been described above and shown in the drawings, and that various modifications and changes may be made without departing from the scope thereof. The scope of the present disclosure is limited only by the appended claims.

Claims

1. A model training method, characterized in that: The method comprises: Inputting a first physical parameter and a first performance data into a first model to obtain a plurality of association relationships between the first physical parameter and the first performance data; Inputting the first physical parameter into a value obtaining module to obtain a second physical parameter, wherein the value obtaining module calculates a target value set based on the first physical parameter, and selects the second physical parameter from the target value set; Inputting the second physical parameter into the first model to obtain a plurality of second performance data; Based on the plurality of second performance data, the first model is updated to obtain a second model.

2. The method according to claim 1, characterized in that The first physical parameter and the first performance data are generated by a third model, and the third model is used to perform simulation analysis on a vehicle structure.

3. The method according to claim 1, characterized in that The step of inputting the first physical parameter into a value obtaining module to obtain a second physical parameter, wherein the value obtaining module is used to calculate a target value set based on the first physical parameter, and selecting the second physical parameter from the target value set includes: The value taking module calculates the target value set according to the first physical parameter and the preset value range, and performs DoE sampling within the target value set to obtain a second preprocessing physical parameter; The second preprocessed physical parameters are input into a third model for simulation analysis to generate the second physical parameters.

4. The method according to claim 1, characterized in that: The step of inputting the first physical parameter into a value obtaining module to obtain a second physical parameter, wherein the value obtaining module calculates a target value set based on the first physical parameter, and before selecting the second physical parameter from the target value set, further comprises: The first physical parameter is preprocessed, and the preprocessing includes at least one of data cleaning, standardization, data conversion, and outlier processing.

5. The method according to claim 2, characterized in that: The method further comprises: Generate a third physical parameter and a third performance data through the third model, input the third physical parameter into the second model to obtain fourth performance data, perform a difference calculation between the fourth performance data and the third performance data, and output multiple fourth performance data that meet the difference threshold.

6. The method according to claim 5, characterized in that The method further comprises: The association relationships corresponding to the plurality of fourth performance data are post-processed to obtain an optimal solution to the association relationships, wherein the post-processing includes at least one of a voting method, an averaging method, and a ranking method.

7. A method for using a model, characterized in that: The method comprises: The physical parameters to be processed are input into the second model for performance prediction to obtain predicted performance data corresponding to the physical parameters to be processed; wherein the second model is obtained by training using the model training method described in any one of claims 1 to 6 above.

8. A model training device, characterized in that: The device comprises: a processing module configured to input a first physical parameter and a first performance data into a first model to obtain a plurality of associations between the first physical parameter and the first performance data; The processing module is further configured to input the first physical parameter into a value obtaining module to obtain a second physical parameter, the value obtaining module calculates a target value set based on the first physical parameter, and selects the second physical parameter from the target value set; The processing module is further configured to input the second physical parameter into the first model to obtain a plurality of second performance data; The processing module is further configured to update the first model based on the plurality of second performance data to obtain a second model.

9. A model using device, characterized in that: The device comprises: The processing module is configured to input the physical parameters into the second model for performance prediction to obtain predicted performance data corresponding to the physical parameters; wherein the second model is obtained by training using the model training method described in any one of claims 1 to 6 above.

10. A computer device, characterized in that: include: processor; a memory for storing processor-executable instructions; The processor is configured to execute the executable instructions to implement the steps of the method described in any one of claims 1 to 6; or the processor is configured to execute the executable instructions to implement the steps of the method described in claim 7.

11. A computer-readable storage medium having computer program instructions stored thereon, characterized in that: When the computer program instructions are executed by a processor, the steps of the method described in any one of claims 1 to 6 are implemented; or, when the computer program instructions are executed by a processor, the steps of the method described in claim 7 are implemented.

Images