Suspension KC characteristic data generation method, device, equipment, medium and product
By normalizing the initial suspension parameters and inputting them into a pre-trained neural network model, suspension K&C characteristic data is generated, solving the problem of low efficiency in existing technologies and achieving efficient data generation.
Patent Information
- Application Number
- CN202511230368.1
- Authority / Receiving Office
- CN · China
- Patent Type
- Applications(China)
- Current Assignee / Owner
- Filing Date
- 2025-08-29
- Publication Date
- 2025-12-09
AI Technical Summary
In existing technologies, generating suspension K&C characteristic data requires simulation and testing, resulting in low efficiency.
By acquiring initial suspension parameters, normalizing them, and then inputting them into a pre-trained neural network model, data generation is performed to generate suspension K&C characteristic data.
Suspension K&C characteristic data can be generated without simulation and testing, improving generation efficiency and reducing costs.
Smart Images
Figure CN121093484A_ABST
Abstract
Description
Technical Field
[0001] This application relates to the field of vehicle technology, and in particular to a method, apparatus, device, medium and product for generating suspension K&C characteristic data. Background Technology
[0002] The vehicle suspension is the mechanism connecting the wheels and the vehicle body, playing a role in supporting the body and damping shocks, directly affecting the vehicle's comfort, handling, and reliability. To analyze the quality of a suspension design, it's necessary to analyze its Kinematic and Compliance (K&C) characteristics. K&C characteristics include K-characteristics and C-characteristics. K-characteristics describe the changes in wheel alignment parameters caused by the geometric motion of the suspension, while C-characteristics describe the changes in wheel alignment parameters caused by the deformation of the suspension's elastic elements due to tire stress.
[0003] In existing technologies, the method for generating suspension K&C characteristic data is usually to perform suspension simulation through a simulation system based on suspension parameters, and then test to obtain K&C characteristic data.
[0004] In summary, the existing technology requires simulation and testing to generate suspension K&C characteristic data, resulting in low efficiency in generating such data. Summary of the Invention
[0005] The suspension K&C characteristic data generation method, apparatus, equipment, medium and product provided in this application are used to solve the problem that the prior art requires simulation and testing to generate K&C characteristic data, resulting in low efficiency in generating suspension K&C characteristic data.
[0006] In a first aspect, embodiments of this application provide a method for generating suspension K&C characteristic data, including:
[0007] Obtain initial suspension parameters, which include hard point parameters, bushing parameters, tire parameters, and steering system parameters;
[0008] The initial suspension parameters are normalized to obtain the first suspension parameters;
[0009] The first suspension parameters are input into the characteristic data generation model to obtain suspension K&C characteristic data. The characteristic data generation model is a pre-trained neural network model used to generate K&C characteristic data based on the suspension parameters. The characteristic data generation model includes a fully connected layer in the form of vector equations.
[0010] In one possible implementation, normalizing the initial suspension parameters to obtain the first suspension parameters includes:
[0011] For each sub-parameter in the initial suspension parameters, the normalized value of the sub-parameter is calculated based on the preset upper limit value and preset lower limit value corresponding to the sub-parameter;
[0012] The first suspension parameters are generated based on the normalized value of each sub-parameter.
[0013] In one possible implementation, the method further includes:
[0014] Obtain target suspension K&C characteristic data;
[0015] Based on the aforementioned characteristic data, a model is generated, and the equations for generating suspension parameters are determined.
[0016] Calculate the second suspension parameters based on the target suspension K&C characteristic data and the suspension parameter generation function;
[0017] The second suspension parameters are inversely normalized to obtain the target suspension parameters.
[0018] In one possible implementation, determining the suspension parameter generation equation based on the characteristic data generation model includes:
[0019] Extract the characteristic data to generate the fully connected layer in the model;
[0020] Generate K&C characteristic data generation equations for each fully connected layer;
[0021] Based on the K&C characteristic data, the equation for generating the suspension parameters is determined.
[0022] In one possible implementation, prior to obtaining the initial suspension parameters, the method further includes:
[0023] Obtain multiple first training suspension parameters and the corresponding real K&C characteristic curves for each first training suspension parameter;
[0024] For each first training suspension parameter, the first training suspension parameter is normalized to obtain the second training suspension parameter corresponding to the first training suspension parameter.
[0025] For each first training suspension parameter, the real K&C characteristic curve corresponding to the first training suspension parameter is sampled to obtain the real K&C characteristic data corresponding to the first training suspension parameter.
[0026] For each first training suspension parameter, establish a correspondence between the second training suspension parameter corresponding to the first training suspension parameter and the real K&C characteristic data corresponding to the first training suspension parameter;
[0027] The initial neural network model is trained based on each second training suspension parameter and the corresponding real K&C characteristic data to obtain the characteristic data generation model.
[0028] In one possible implementation, training the initial neural network model based on each second training suspension parameter and the corresponding real K&C characteristic data to obtain the characteristic data generation model includes:
[0029] Select one second training suspension parameter from all the second training suspension parameters;
[0030] The second trained suspension parameters are input into the initial neural network model to obtain predicted K&C characteristic data;
[0031] The loss value is calculated based on the predicted K&C characteristic data and the actual K&C characteristic data corresponding to the second trained suspension parameters;
[0032] The initial neural network model is updated based on the loss value to obtain the trained network model;
[0033] If the loss value is less than a preset loss value threshold, then the trained network model is used as the feature data generation model.
[0034] If the loss value is greater than or equal to the preset loss value threshold, the trained network model is used as a new initial neural network model, and the above steps are repeated until the loss value is less than the preset loss value threshold. Then, the trained network model is used as the feature data generation model.
[0035] Secondly, embodiments of this application provide a suspension K&C characteristic data generation device, comprising:
[0036] The acquisition module is used to acquire initial suspension parameters, which include hard point parameters, bushing parameters, tire parameters, and steering system parameters;
[0037] The processing module is used to normalize the initial suspension parameters to obtain the first suspension parameters;
[0038] The generation module is used to input the first suspension parameters into the characteristic data generation model to obtain suspension K&C characteristic data. The characteristic data generation model is a pre-trained neural network model for generating K&C characteristic data based on the suspension parameters. The characteristic data generation model includes a fully connected layer in the form of vector equations.
[0039] Thirdly, embodiments of this application provide an electronic device, including:
[0040] Processor, memory, communication interface;
[0041] The memory is used to store the executable instructions of the processor;
[0042] The processor is configured to execute the suspension K&C characteristic data generation method according to any one of the first aspects by executing the executable instructions.
[0043] Fourthly, embodiments of this application provide a readable storage medium having a computer program stored thereon, wherein the computer program, when executed by a processor, implements the suspension K&C characteristic data generation method described in any of the first aspects.
[0044] Fifthly, embodiments of this application provide a computer program product, including a computer program, which, when executed by a processor, is used to implement the suspension K&C characteristic data generation method described in any of the first aspects.
[0045] The suspension K&C characteristic data generation method, apparatus, device, medium, and product provided in this application obtain initial suspension parameters, normalize these parameters to obtain first suspension parameters, and then input these first suspension parameters into a characteristic data generation model to obtain suspension K&C characteristic data. The characteristic data generation model is a pre-trained neural network model used to generate K&C characteristic data based on the suspension parameters, and includes fully connected layers in the form of vector equations. This solution obtains suspension K&C characteristic data by inputting the suspension parameters into the characteristic data generation model, eliminating the need for simulation and testing, thus improving the efficiency of suspension K&C characteristic data generation. Attached Figure Description
[0046] The accompanying drawings, which are incorporated in and form part of this specification, illustrate embodiments consistent with this application and, together with the description, serve to explain the principles of this application.
[0047] Figure 1 This is a schematic diagram of the suspension architecture provided in this application;
[0048] Figure 2 A flowchart illustrating an embodiment of the suspension K&C characteristic data generation method provided in this application;
[0049] Figure 3 A schematic diagram of the architecture of the feature data generation model provided in this application;
[0050] Figure 4 Flowchart of the feature data generation model provided in this application;
[0051] Figure 5 A flowchart illustrating Embodiment 2 of the suspension K&C characteristic data generation method provided in this application;
[0052] Figure 6 A flowchart illustrating Embodiment 3 of the suspension K&C characteristic data generation method provided in this application;
[0053] Figure 7 This is a schematic diagram of the actual K&C characteristic curves provided in this application;
[0054] Figure 8 A flowchart illustrating the training process of the feature data generation model provided in this application;
[0055] Figure 9 A graph showing the relationship between the number of training iterations and the test K&C characteristic data and the actual K&C characteristic data provided for this application;
[0056] Figure 10 A schematic diagram of an embodiment of the suspension K&C characteristic data generation device provided in this application;
[0057] Figure 11 This is a schematic diagram of the structure of an electronic device provided in this application.
[0058] The accompanying drawings have illustrated specific embodiments of this application, which will be described in more detail below. These drawings and descriptions are not intended to limit the scope of the concept in any way, but rather to illustrate the concept of this application to those skilled in the art through reference to specific embodiments. Detailed Implementation
[0059] Exemplary embodiments will now be described in detail, examples of which are illustrated in the accompanying drawings. When the following description relates to the drawings, unless otherwise indicated, the same numbers in different drawings denote the same or similar elements. The embodiments described in the following exemplary embodiments do not represent all embodiments consistent with this application. Rather, they are merely examples of apparatuses and methods consistent with some aspects of this application as detailed in the appended claims.
[0060] The terms “first,” “second,” “third,” “fourth,” etc. (if present) in the specification, claims, and accompanying drawings of this application are used to distinguish similar objects and are not necessarily used to describe a particular order or sequence. It should be understood that such data can be interchanged where appropriate so that the embodiments of this application described herein can be implemented in orders other than those illustrated or described herein. Furthermore, the terms “comprising” and “having,” and any variations thereof, are intended to cover a non-exclusive inclusion; for example, a process, method, system, product, or apparatus that comprises a series of steps or units is not necessarily limited to those steps or units explicitly listed, but may include other steps or units not explicitly listed or inherent to such processes, methods, products, or apparatus.
[0061] With the continuous development of technology, vehicle suspension, as the mechanism connecting the wheels and the vehicle body, has become increasingly sophisticated in design. Vehicle suspension plays a crucial role in supporting the vehicle body and damping shocks, directly impacting the vehicle's comfort, handling, and reliability. To analyze the quality of suspension design, it is necessary to analyze the suspension's Kinematic and Compliance characteristics. K&C characteristics include K-characteristics and C-characteristics. K-characteristics describe the changes in wheel alignment parameters caused by the geometrical motion of the suspension, while C-characteristics describe the changes in wheel alignment parameters caused by the deformation of the suspension's elastic elements due to tire stress.
[0062] For example, Figure 1 The suspension architecture diagram provided in this application is as follows: Figure 1 As shown, the vehicle suspension connects the wheels to the body, and can play a role in supporting the body and buffering shocks.
[0063] In existing technologies, the common method for generating suspension K&C characteristic data is to perform suspension simulation using a simulation system based on suspension parameters, and then test to obtain the K&C characteristic data. Because testing and simulation are required, the efficiency of generating suspension K&C characteristic data is relatively low.
[0064] To address the problems existing in the prior art, the inventors, during their research on suspension K&C characteristic data generation methods, discovered that to improve the generation efficiency of suspension K&C characteristic data, a neural network model can be trained based on suspension parameters and real K&C characteristic data to obtain a characteristic data generation model. When suspension K&C characteristic data needs to be generated, the initial suspension parameters are normalized and then input into the characteristic data generation model to obtain the suspension K&C characteristic data. Based on the above inventive concept, the suspension K&C characteristic data generation scheme of this application was designed.
[0065] The execution subject of the suspension K&C characteristic data generation method in this application can be a computer, or a server, terminal device, vehicle terminal, etc. This application does not limit it. The following description uses a computer as an example.
[0066] The following provides an example illustrating the application scenarios of the suspension K&C characteristic data generation method provided in this application.
[0067] For example, in this application scenario, the staff designed a vehicle suspension. In order to determine the quality of the suspension design, the staff used a terminal device to send the initial suspension parameters to the computer.
[0068] After the computer obtains the initial suspension parameters, it normalizes the initial suspension parameters to obtain the first suspension parameters, and then inputs the first suspension parameters into the characteristic data generation model to obtain the suspension K&C characteristic data.
[0069] The characteristic data generation model is a pre-trained neural network model used to generate K&C characteristic data based on suspension parameters.
[0070] The computer displays suspension K&C characteristic data, allowing staff to determine the quality of the designed vehicle suspension. If adjustments are needed, the computer can still be used to generate a model from the characteristic data to obtain the suspension K&C characteristic data after the adjustments are made.
[0071] The computer can also record and display the suspension K&C characteristic data generated each time, as well as generate visual charts, allowing staff to intuitively view the changes in the suspension K&C characteristic data.
[0072] It should be noted that the above scenario is only an example of an application scenario provided by the embodiments of this application. The embodiments of this application do not limit the actual form of the various devices included in the scenario, nor do they limit the interaction method between devices. In the specific application of the solution, it can be set according to actual needs.
[0073] The technical solution of this application will now be described in detail through specific embodiments. It should be noted that the following specific embodiments can be combined with each other, and the same or similar concepts or processes may not be described again in some embodiments.
[0074] Figure 2 This is a flowchart illustrating an embodiment of the suspension K&C characteristic data generation method provided in this application. This embodiment describes how a computer obtains suspension K&C characteristic data through a characteristic data generation model. The method in this embodiment can be implemented through software, hardware, or a combination of both. Figure 2 As shown, the method for generating suspension K&C characteristic data specifically includes the following steps:
[0075] S201: Obtain initial suspension parameters.
[0076] In this step, in order to generate suspension K&C characteristic data, the computer first obtains the initial suspension parameters.
[0077] The initial suspension parameters include hardpoint parameters, bushing parameters, tire parameters, and steering system parameters. Each of these parameters includes sub-parameters.
[0078] For example, the sub-parameters included in the hard point parameters may be the coordinates of the connection point between the upper control arm and the vehicle body, the coordinates of the ball joint center point connecting the lower control arm and the steering knuckle, the coordinates of the connection point between the steering tie rod and the steering gear, the coordinates of the connection point between the top of the shock absorber and the vehicle body, the coordinates of the connection point between the bottom of the shock absorber and the steering knuckle, etc.
[0079] The bushing parameters may include sub-parameters such as radial stiffness, torsional stiffness, and hysteresis damping.
[0080] Tire parameters can include sub-parameters such as tire vertical stiffness, tire lateral stiffness, and tire rolling radius.
[0081] The parameters of the steering system may include sub-parameters such as steering gear ratio, steering column stiffness, steering tie rod stiffness, and Ackermann ratio.
[0082] The embodiments of this application do not limit the hard point parameters, bushing parameters, tire parameters, and steering system parameters, which can be determined according to the actual situation.
[0083] It should be noted that the computer can obtain the initial suspension parameters in several ways: First, the operator can send the initial suspension parameters to the computer using a terminal device, and the computer can then obtain the initial suspension parameters. Second, the operator can input the initial suspension parameters into the computer using an input device, and the computer can then obtain the initial suspension parameters. Third, the operator can send the initial suspension parameters to a server using a terminal device, and the computer can obtain the initial suspension parameters from the server. This application does not limit the method by which the computer obtains the initial suspension parameters; it can be determined according to the actual situation.
[0084] It should be noted that the embodiments of this application do not limit the type of suspension, which may be MacPherson strut suspension, multi-link suspension, double wishbone suspension, etc.
[0085] S202: Normalize the initial suspension parameters to obtain the first suspension parameters.
[0086] In this step, after the computer obtains the initial suspension parameters, it normalizes the initial suspension parameters to obtain the first suspension parameters in order to obtain accurate suspension K&C characteristic data.
[0087] Since the input data for the subsequent characteristic data generation model is in the format of normalized data, the initial suspension parameters need to be normalized.
[0088] Specifically, for each sub-parameter in the initial suspension parameters, the normalized value of the sub-parameter is calculated based on the preset upper limit and preset lower limit values corresponding to the sub-parameter.
[0089] According to the formula Calculate the normalized value of the sub-parameter, where x1 represents the normalized value of the sub-parameter, x0 represents the sub-parameter, and x... min This represents the preset lower limit value corresponding to the sub-parameter, x. max This indicates the preset upper limit value corresponding to the sub-parameter.
[0090] For example, the sub-parameter is radial stiffness, with a preset lower limit of 800 N / mm and a preset upper limit of 17000 N / mm. The sub-parameter is tire lateral stiffness, with a preset lower limit of 2800 N / rad and a preset upper limit of 80000 N / rad. The sub-parameter is steering column stiffness, with a preset lower limit of 2200 Nm / rad and a preset upper limit of 2900 Nm / rad. This application does not limit the preset lower and upper limits of the sub-parameters; these values can be determined based on actual conditions.
[0091] Then, based on the normalized value of each sub-parameter, the first suspension parameters are generated. That is, the normalized values of each sub-parameter are combined into a column vector to obtain the first suspension parameters.
[0092] S203: Input the first suspension parameters into the characteristic data generation model to obtain the suspension K&C characteristic data.
[0093] In this step, after the computer obtains the first suspension parameters, it inputs the first suspension parameters into the characteristic data generation model to obtain the suspension K&C characteristic data.
[0094] Among them, the characteristic data generation model is a pre-trained neural network model used to generate K&C characteristic data based on suspension parameters. The characteristic data generation model includes fully connected layers in the form of vector equations.
[0095] For example, Figure 3 A schematic diagram of the architecture of the feature data generation model provided in this application is shown below. Figure 3 As shown, the characteristic data generation model includes an input layer, multiple fully connected layers, and an output layer. The computer inputs the first suspension parameters to the input layer, which then inputs these parameters to the first fully connected layer. The first fully connected layer processes the parameters and inputs them to the second fully connected layer for further processing. The output data of the preceding fully connected layer becomes the input data of the following fully connected layer. The last fully connected layer outputs the processed data to the output layer, which then outputs the suspension K&C characteristic data.
[0096] Each fully connected layer is in the form of a vector equation, so the input data can be substituted into the vector equation to obtain the output data.
[0097] For example, the characteristic data generation model has three fully connected layers. The vector equation of the first fully connected layer is X1 = W1X0 + b1, the vector equation of the second fully connected layer is X2 = W2X1 + b2, and the vector equation of the third fully connected layer is X3 = W3X2 + b3. Here, X0 represents the first suspension parameter, X1 represents the output data of the first fully connected layer, X2 represents the output data of the second fully connected layer, X3 represents the suspension K&C characteristic data, W1 represents the weight matrix of the first fully connected layer, b1 represents the bias matrix of the first fully connected layer, W2 represents the weight matrix of the second fully connected layer, b2 represents the bias matrix of the second fully connected layer, W3 represents the weight matrix of the third fully connected layer, and b3 represents the bias matrix of the third fully connected layer.
[0098] This application does not limit the number of fully connected layers and vector equations in the feature data generation model; these can be determined based on actual circumstances.
[0099] For example, Figure 4 The flowchart for using the feature data generation model provided in this application is as follows: Figure 4 As shown, the first suspension parameters are input into the characteristic data generation model to obtain the suspension K&C characteristic data.
[0100] The suspension K&C characteristic data generation method provided in this embodiment obtains initial suspension parameters, normalizes these parameters to obtain first suspension parameters, and then inputs these first suspension parameters into a characteristic data generation model to obtain suspension K&C characteristic data. The characteristic data generation model is a pre-trained neural network model used to generate K&C characteristic data based on the suspension parameters, and includes fully connected layers in the form of vector equations. Compared to the existing technology that requires simulation and testing to obtain suspension K&C characteristic data, this solution obtains suspension K&C characteristic data by inputting the suspension parameters into the characteristic data generation model, eliminating the need for simulation and testing, thus improving the generation efficiency of suspension K&C characteristic data and reducing costs.
[0101] Figure 5 This is a flowchart illustrating a second embodiment of the suspension K&C characteristic data generation method provided in this application. Based on the above embodiments, this application describes how a computer uses a characteristic data generation model to generate suspension parameters. For example... Figure 5 As shown, the method for generating suspension K&C characteristic data specifically includes the following steps:
[0102] S501: Obtain target suspension K&C characteristic data.
[0103] When you want to obtain the suspension parameters corresponding to specific suspension K&C characteristic data, you can input the target suspension K&C characteristic data into the computer to generate the suspension parameters.
[0104] In this step, in order to generate suspension parameters, the computer first acquires the target suspension K&C characteristic data.
[0105] S502: Generate a model based on the characteristic data and determine the equations for generating suspension parameters.
[0106] In this step, after the computer obtains the target suspension K&C characteristic data, it generates a model based on the characteristic data and determines the suspension parameter generation equation.
[0107] Specifically, we first extract the fully connected layers in the feature data generation model, that is, we extract the vector equations of each fully connected layer in the feature data generation model.
[0108] Then, based on each fully connected layer, a K&C characteristic data generation equation is generated.
[0109] For example, the feature data generation model has three fully connected layers. The vector equation for the first fully connected layer is X1 = W1X0 + b1, the vector equation for the second fully connected layer is X2 = W2X1 + b2, and the vector equation for the third fully connected layer is X3 = W3X2 + b3. Then, the relationship equation between X3 and X0 is calculated to obtain the K&C feature data generation equation. The K&C feature data generation equation is X3 = W3W2W1X0 + W3W2b1 + W3b2 + b3.
[0110] This application does not limit the number of fully connected layers and vector equations in the feature data generation model; these can be determined based on actual circumstances.
[0111] Then, based on the K&C characteristic data, equations are generated to determine the suspension parameter generation equations.
[0112] Since the K&C characteristic data generation equation is the relationship equation between the K&C characteristic data and the suspension parameters, it can be transformed to obtain the relationship equation between the suspension parameters and the K&C characteristic data, which means determining the suspension parameter generation equation.
[0113] For example, based on the previous example, the determined suspension parameter generation equation is as follows:
[0114] S503: Calculate the second suspension parameters based on the target suspension K&C characteristic data and the suspension parameter generation function.
[0115] In this step, after obtaining the suspension parameter generation function, the computer calculates the second suspension parameters based on the target suspension K&C characteristic data and the suspension parameter generation function. In other words, the second suspension parameters are obtained by substituting the target suspension K&C characteristic data into the suspension parameter generation function.
[0116] S504: Perform inverse normalization on the second suspension parameters to obtain the target suspension parameters.
[0117] In this step, after the computer obtains the second suspension parameters, since the second suspension parameters are normalized data, in order to obtain the true data, it is necessary to perform inverse normalization on the second suspension parameters to obtain the target suspension parameters.
[0118] For each sub-parameter in the second suspension parameters, calculate the inverse normalized value of the sub-parameter based on the preset upper limit and preset lower limit values corresponding to the sub-parameter.
[0119] According to the formula x0=x1(x max -x min )+x min Calculate the inverse normalized value of the sub-parameter, where x1 represents the sub-parameter, x0 represents the inverse normalized value of the sub-parameter, and x... min This represents the preset lower limit value corresponding to the sub-parameter, x. max This indicates the preset upper limit value corresponding to the sub-parameter.
[0120] The inverse normalized value of each sub-parameter constitutes the target suspension parameters.
[0121] The suspension K&C characteristic data generation method provided in this embodiment determines the suspension parameter generation equation based on the characteristic data generation model, and then obtains the target suspension parameters by combining the target suspension K&C characteristic data. Compared with the prior art, which determines the target suspension parameters corresponding to the target suspension K&C characteristic data manually, this solution determines the target suspension parameters through the characteristic data generation model, which can improve the generation efficiency of suspension parameters.
[0122] Figure 6 This is a flowchart illustrating a third embodiment of the suspension K&C characteristic data generation method provided in this application. Based on the above embodiments, this application describes how a computer is trained to generate a characteristic data generation model. Figure 6 As shown, the method for generating suspension K&C characteristic data specifically includes the following steps:
[0123] S601: Obtain multiple first training suspension parameters and the corresponding real K&C characteristic curves for each first training suspension parameter.
[0124] In this step, in order to train the feature data generation model, the computer first obtains multiple first training suspension parameters and the real K&C feature curves corresponding to each first training suspension parameter.
[0125] It should be noted that there is at least one real K&C characteristic curve corresponding to each first training suspension parameter.
[0126] S602: For each first training suspension parameter, normalize the first training suspension parameter to obtain the second training suspension parameter corresponding to the first training suspension parameter.
[0127] In this step, after the computer obtains multiple first training suspension parameters and the actual K&C characteristic curves, in order to improve the accuracy of the model, it is necessary to normalize each first training suspension parameter to obtain the corresponding second training suspension parameter.
[0128] It should be noted that the normalization process is similar to step S202 in Example 1, and will not be described again here.
[0129] S603: For each first training suspension parameter, sample the real K&C characteristic curve corresponding to the first training suspension parameter to obtain the real K&C characteristic data corresponding to the first training suspension parameter.
[0130] In this step, after the computer obtains multiple first training suspension parameters and real K&C characteristic curves, for each first training suspension parameter, it samples the real K&C characteristic curve corresponding to that first training suspension parameter to obtain the real K&C characteristic data corresponding to that first training suspension parameter.
[0131] It should be noted that when the first training suspension parameter corresponds to multiple real K&C characteristic curves, it is necessary to sample each real K&C characteristic curve to obtain the real K&C characteristic data corresponding to the first training suspension parameter.
[0132] For example, Figure 7 This application provides a schematic diagram of the actual K&C characteristic curve sampling, such as... Figure 7 As shown in the figure, the curve is the true K&C characteristic curve, and the points in the figure are sampling points. The data of the sampling points constitute the true K&C characteristic data.
[0133] It should be noted that sampling can be performed according to a preset sampling number, and a preset sampling number of data can be collected from the real K&C characteristic curve. The preset sampling number can be 7, 11, 20, etc. This application embodiment does not limit the preset sampling number, and it can be determined according to the actual situation.
[0134] It should be noted that the sampling method can be equal arc length sampling, uniform sampling, adaptive sampling, random sampling, etc. This application embodiment does not limit the sampling method, and it can be determined according to the actual situation.
[0135] It should be noted that the execution order of steps S602 and S603 can be: step S602 is executed first, then step S603. Alternatively, step S603 is executed first, then step S602. Or, steps S602 and S603 are executed simultaneously. This embodiment does not limit the execution order of steps S602 and S603; it can be determined according to the actual situation.
[0136] S604: For each first training suspension parameter, establish a correspondence between the second training suspension parameter corresponding to the first training suspension parameter and the real K&C characteristic data corresponding to the first training suspension parameter.
[0137] In this step, the computer obtains the second training suspension parameters and the real K&C characteristic data. In order to ensure the smooth progress of subsequent model training, for each first training suspension parameter, a correspondence is established between the second training suspension parameter corresponding to the first training suspension parameter and the real K&C characteristic data corresponding to the first training suspension parameter.
[0138] S605: Based on each second training suspension parameter and the corresponding real K&C characteristic data, train the initial neural network model to obtain the characteristic data generation model.
[0139] In this step, after the computer establishes a correspondence between the second training suspension parameters and the real K&C characteristic data, it trains the initial neural network model based on each second training suspension parameter and the corresponding real K&C characteristic data to obtain the characteristic data generation model.
[0140] For example, Figure 8 The training flowchart for the feature data generation model provided in this application is as follows: Figure 8 As shown, training the initial neural network model to obtain the feature data generation model can be achieved through the following steps:
[0141] S801: Select a second training suspension parameter from all second training suspension parameters.
[0142] S802: Input the second training suspension parameters into the initial neural network model to obtain the predicted K&C characteristic data.
[0143] S803: Calculate the loss value based on the predicted K&C characteristic data and the actual K&C characteristic data corresponding to the second trained suspension parameters.
[0144] S804: Update the initial neural network model based on the loss value to obtain the trained network model.
[0145] In the above steps, after the computer establishes a correspondence between the second trained suspension parameters and the real K&C characteristic data, it selects one second trained suspension parameter from all available parameters and inputs it into the initial neural network model to obtain the predicted K&C characteristic data. Then, based on the predicted K&C characteristic data and the corresponding real K&C characteristic data of the second trained suspension parameter, a loss value is calculated. The initial neural network model is updated based on the loss value to obtain the trained network model.
[0146] It should be noted that the predicted K&C characteristic data and the actual K&C characteristic data can be substituted into the loss function to calculate the loss value. The loss function can be the mean squared error loss function, the mean absolute error loss function, the Huber loss function, etc. This application does not limit the loss function, and it can be determined according to the actual situation.
[0147] S805: Determine whether the loss value is less than the preset loss value threshold; if the loss value is less than the preset loss value threshold, proceed to step S806; if the loss value is greater than or equal to the preset loss value threshold, proceed to step S807.
[0148] In this step, after the computer obtains the trained network model, it needs to determine whether the loss value is less than a preset loss value threshold in order to decide whether to stop training.
[0149] It should be noted that the preset loss value threshold can be 0.01, 0.001, or 0.0001. This application embodiment does not limit the preset loss value threshold, which can be determined according to the actual situation.
[0150] S806: Use the trained network model as a feature data generation model.
[0151] In this step, if the computer determines that the loss value is less than the preset loss value threshold, it means that training can be stopped and the trained network model can be used as the feature data generation model.
[0152] S807: Use the trained network model as the new initial neural network model and return to step S801.
[0153] In this step, if the computer determines that the loss value is greater than or equal to the preset loss value threshold, it means that further training is needed. The trained network model is then used as the new initial neural network model, and the process returns to step S801. That is, the trained network model is used as the new initial neural network model, and the above steps are repeated until the loss value is less than the preset loss value threshold. Then, the trained network model is used as the feature data generation model.
[0154] The trained network model is used as the new initial neural network model. A new second training suspension parameter is selected from all the second training suspension parameters and input into the new initial neural network model to obtain new predicted K&C characteristic data. A new loss value is calculated based on the new predicted K&C characteristic data and the actual K&C characteristic data corresponding to the new second training suspension parameter. The new initial neural network model is updated based on the new loss value to obtain a new trained network model. If the new loss value is greater than or equal to a preset loss threshold, this process is repeated until the loss value is less than the preset loss threshold. The trained network model is then used as the characteristic data generation model.
[0155] In one implementation, after each training of the network model, the computer can test the trained network model based on the test data to obtain test K&C characteristic data, and record the test K&C characteristic data and the corresponding real K&C characteristic data, as well as the absolute value of the deviation between the two.
[0156] For example, Table 1 is a table showing the relationship between the number of training sessions and the test K&C characteristic data, the actual K&C characteristic data, and the absolute value of the deviation value provided in this application.
[0157] Table 1
[0158]
[0159]
[0160] As shown in Table 1, as the number of training iterations increases, the absolute value of the bias decreases and the accuracy of the model increases.
[0161] For example, based on Table 1, Figure 9 The relationship between the number of training iterations and the test K&C characteristic data and the actual K&C characteristic data provided in this application is shown in the following figure. Figure 9 As shown, the dashed line represents the actual K&C characteristic data, and the solid line represents the test K&C characteristic data. The test K&C characteristic data and the actual K&C characteristic data are quite similar, indicating that the model has good accuracy.
[0162] The suspension K&C characteristic data generation method provided in this embodiment obtains second training suspension parameters and real K&C characteristic data by processing the first training suspension parameters and the real K&C characteristic curve. Then, the initial neural network model is trained based on the second training suspension parameters and the real K&C characteristic data to obtain a characteristic data generation model, which can improve the accuracy of the characteristic data generation model.
[0163] The following are embodiments of the apparatus described in this application, which can be used to execute the embodiments of the method described in this application. For details not disclosed in the apparatus embodiments of this application, please refer to the embodiments of the method described in this application.
[0164] Figure 10 This is a schematic diagram of the structure of an embodiment of the suspension K&C characteristic data generation device provided in this application; as shown below. Figure 10 As shown, the suspension K&C characteristic data generation device 1000 includes:
[0165] The acquisition module 1001 is used to acquire initial suspension parameters, which include hard point parameters, bushing parameters, tire parameters, and steering system parameters;
[0166] Processing module 1002 is used to normalize the initial suspension parameters to obtain the first suspension parameters;
[0167] The generation module 1003 is used to input the first suspension parameters into the characteristic data generation model to obtain suspension K&C characteristic data. The characteristic data generation model is a pre-trained neural network model for generating K&C characteristic data based on the suspension parameters. The characteristic data generation model includes a fully connected layer in the form of vector equations.
[0168] Furthermore, the processing module 1002 is specifically used for:
[0169] For each sub-parameter in the initial suspension parameters, the normalized value of the sub-parameter is calculated based on the preset upper limit value and preset lower limit value corresponding to the sub-parameter;
[0170] The first suspension parameters are generated based on the normalized value of each sub-parameter.
[0171] Furthermore, the acquisition module 1001 is also used to acquire target suspension K&C characteristic data;
[0172] The processing module 1002 is further configured to:
[0173] Based on the aforementioned characteristic data, a model is generated, and the equations for generating suspension parameters are determined.
[0174] Calculate the second suspension parameters based on the target suspension K&C characteristic data and the suspension parameter generation function;
[0175] The second suspension parameters are inversely normalized to obtain the target suspension parameters.
[0176] Furthermore, the processing module 1002 is specifically used for:
[0177] Extract the characteristic data to generate the fully connected layer in the model;
[0178] Generate K&C characteristic data generation equations for each fully connected layer;
[0179] Based on the K&C characteristic data, the equation for generating the suspension parameters is determined.
[0180] Furthermore, the acquisition module 1001 is also used to acquire multiple first training suspension parameters and the real K&C characteristic curve corresponding to each first training suspension parameter;
[0181] The processing module 1002 is further configured to:
[0182] For each first training suspension parameter, the first training suspension parameter is normalized to obtain the second training suspension parameter corresponding to the first training suspension parameter.
[0183] For each first training suspension parameter, the real K&C characteristic curve corresponding to the first training suspension parameter is sampled to obtain the real K&C characteristic data corresponding to the first training suspension parameter.
[0184] For each first training suspension parameter, establish a correspondence between the second training suspension parameter corresponding to the first training suspension parameter and the real K&C characteristic data corresponding to the first training suspension parameter;
[0185] The initial neural network model is trained based on each second training suspension parameter and the corresponding real K&C characteristic data to obtain the characteristic data generation model.
[0186] Furthermore, the processing module 1002 is specifically used for:
[0187] Select one second training suspension parameter from all the second training suspension parameters;
[0188] The second trained suspension parameters are input into the initial neural network model to obtain predicted K&C characteristic data;
[0189] The loss value is calculated based on the predicted K&C characteristic data and the actual K&C characteristic data corresponding to the second trained suspension parameters;
[0190] The initial neural network model is updated based on the loss value to obtain the trained network model;
[0191] If the loss value is less than a preset loss value threshold, then the trained network model is used as the feature data generation model.
[0192] If the loss value is greater than or equal to the preset loss value threshold, the trained network model is used as a new initial neural network model, and the above steps are repeated until the loss value is less than the preset loss value threshold. Then, the trained network model is used as the feature data generation model.
[0193] The suspension K&C characteristic data generation device provided in this embodiment is used to execute the technical solution in any of the aforementioned method embodiments. Its implementation principle and technical effect are similar, and will not be described again here.
[0194] Figure 11 This is a schematic diagram of the structure of an electronic device provided in this application. Figure 11 As shown, the electronic device 1100 includes:
[0195] Processor 1101, memory 1102, and communication interface 1103;
[0196] The memory 1102 is used to store the executable instructions of the processor 1101;
[0197] The processor 1101 is configured to execute the technical solutions in any of the foregoing method embodiments by executing the executable instructions.
[0198] Optionally, the memory 1102 can be either standalone or integrated with the processor 1101.
[0199] Optionally, when the memory 1102 is a device independent of the processor 1101, the electronic device 1100 may further include:
[0200] Bus 1104, memory 1102 and communication interface 1103 are connected to processor 1101 through bus 1104 and complete communication with each other. Communication interface 1103 is used to communicate with other devices.
[0201] Optionally, the communication interface 1103 can be implemented using a transceiver. The communication interface is used to enable communication between the database access device and other devices (e.g., clients, read-write databases, and read-only databases). The memory may include random access memory (RAM) and may also include non-volatile memory, such as at least one disk drive.
[0202] Bus 1104 can be a Peripheral Component Interconnect (PCI) bus or an Extended Industry Standard Architecture (EISA) bus, etc. Buses can be categorized as address buses, data buses, control buses, etc. For ease of representation, only one thick line is used in the diagram, but this does not indicate that there is only one bus or one type of bus.
[0203] The processors mentioned above can be general-purpose processors, including central processing units (CPUs), network processors (NPs), etc.; they can also be digital signal processors (DSPs), application-specific integrated circuits (ASICs), field-programmable gate arrays (FPGAs), or other programmable logic devices, discrete gate or transistor logic devices, or discrete hardware components.
[0204] The electronic device is used to execute the technical solutions in any of the foregoing method embodiments. Its implementation principle and technical effect are similar, and will not be described again here.
[0205] This application also provides a readable storage medium storing a computer program thereon, which, when executed by a processor, implements the technical solutions provided in any of the foregoing method embodiments.
[0206] This application also provides a computer program product, including a computer program, which, when executed by a processor, is used to implement the technical solutions provided in any of the foregoing method embodiments.
[0207] Those skilled in the art will understand that all or part of the steps of the above-described method embodiments can be implemented by hardware related to program instructions. The aforementioned program can be stored in a computer-readable storage medium. When executed, the program performs the steps of the above-described method embodiments; and the aforementioned storage medium includes various media capable of storing program code, such as ROM, RAM, magnetic disks, or optical disks.
[0208] Finally, it should be noted that the above embodiments are only used to illustrate the technical solutions of this application, and are not intended to limit them. Although this application has been described in detail with reference to the foregoing embodiments, those skilled in the art should understand that modifications can still be made to the technical solutions described in the foregoing embodiments, or equivalent substitutions can be made to some or all of the technical features therein. Such modifications or substitutions do not cause the essence of the corresponding technical solutions to deviate from the scope of the technical solutions of the embodiments of this application.
Claims
1. A method for generating suspension K&C characteristic data, characterized in that, include: Obtain initial suspension parameters, which include hard point parameters, bushing parameters, tire parameters, and steering system parameters; The initial suspension parameters are normalized to obtain the first suspension parameters; The first suspension parameters are input into the characteristic data generation model to obtain suspension K&C characteristic data. The characteristic data generation model is a pre-trained neural network model used to generate K&C characteristic data based on the suspension parameters. The characteristic data generation model includes a fully connected layer in the form of vector equations.
2. The method according to claim 1, characterized in that, The normalization of the initial suspension parameters to obtain the first suspension parameters includes: For each sub-parameter in the initial suspension parameters, the normalized value of the sub-parameter is calculated based on the preset upper limit value and preset lower limit value corresponding to the sub-parameter; The first suspension parameters are generated based on the normalized value of each sub-parameter.
3. The method according to claim 1, characterized in that, The method further includes: Obtain target suspension K&C characteristic data; Based on the aforementioned characteristic data, a model is generated, and the equations for generating suspension parameters are determined. Calculate the second suspension parameters based on the target suspension K&C characteristic data and the suspension parameter generation function; The second suspension parameters are inversely normalized to obtain the target suspension parameters.
4. The method according to claim 3, characterized in that, The step of generating a model based on the characteristic data and determining the suspension parameter generation equation includes: Extract the characteristic data to generate the fully connected layer in the model; Generate K&C characteristic data generation equations for each fully connected layer; Based on the K&C characteristic data, the equation for generating the suspension parameters is determined.
5. The method according to any one of claims 1 to 4, characterized in that, Before obtaining the initial suspension parameters, the method further includes: Obtain multiple first training suspension parameters and the corresponding real K&C characteristic curves for each first training suspension parameter; For each first training suspension parameter, the first training suspension parameter is normalized to obtain the second training suspension parameter corresponding to the first training suspension parameter. For each first training suspension parameter, the real K&C characteristic curve corresponding to the first training suspension parameter is sampled to obtain the real K&C characteristic data corresponding to the first training suspension parameter. For each first training suspension parameter, establish a correspondence between the second training suspension parameter corresponding to the first training suspension parameter and the real K&C characteristic data corresponding to the first training suspension parameter; The initial neural network model is trained based on each second training suspension parameter and the corresponding real K&C characteristic data to obtain the characteristic data generation model.
6. The method according to claim 5, characterized in that, The step of training the initial neural network model based on each second training suspension parameter and the corresponding real K&C characteristic data to obtain the characteristic data generation model includes: Select one second training suspension parameter from all the second training suspension parameters; The second trained suspension parameters are input into the initial neural network model to obtain predicted K&C characteristic data; The loss value is calculated based on the predicted K&C characteristic data and the actual K&C characteristic data corresponding to the second trained suspension parameters; The initial neural network model is updated based on the loss value to obtain the trained network model; If the loss value is less than a preset loss value threshold, then the trained network model is used as the feature data generation model. If the loss value is greater than or equal to the preset loss value threshold, the trained network model is used as a new initial neural network model, and the above steps are repeated until the loss value is less than the preset loss value threshold. Then, the trained network model is used as the feature data generation model.
7. A suspension K&C characteristic data generation device, characterized in that, include: The acquisition module is used to acquire initial suspension parameters, which include hard point parameters, bushing parameters, tire parameters, and steering system parameters; The processing module is used to normalize the initial suspension parameters to obtain the first suspension parameters; The generation module is used to input the first suspension parameters into the characteristic data generation model to obtain suspension K&C characteristic data. The characteristic data generation model is a pre-trained neural network model for generating K&C characteristic data based on the suspension parameters. The characteristic data generation model includes a fully connected layer in the form of vector equations.
8. An electronic device, characterized in that, include: Processor, memory, communication interface; The memory is used to store the executable instructions of the processor; The processor is configured to execute the suspension K&C characteristic data generation method according to any one of claims 1 to 6 by executing the executable instructions.
9. A readable storage medium having a computer program stored thereon, characterized in that, When the computer program is executed by the processor, it implements the suspension K&C characteristic data generation method according to any one of claims 1 to 6.
10. A computer program product, characterized in that, Includes a computer program, which, when executed by a processor, is used to implement the suspension K&C characteristic data generation method according to any one of claims 1 to 6.