Simulation model optimization method, braking system calibration method, device, equipment and medium

Abstract

The invention discloses a simulation model optimization method, a braking system calibration method, a device, equipment and a medium. The simulation model optimization method comprises the steps that a training sample set corresponding to a calibration function is acquired, the training sample set comprises a plurality of training samples, and each training sample comprises non-functional training data, functional training data and preset brake build-up pressure; processing the non-functional training data and the functional training data in the first training sample by adopting a braking system simulation model, and determining first brake build-up pressure; on the basis of the preset brake build-up pressure and the first brake build-up pressure, updating model parameters of the brake system simulation model, and determining an initial function simulation model corresponding to the calibration function; and adopting the second training sample to test the initial function simulation model corresponding to the calibration function, and determining a target function simulation model corresponding to the calibration function. The method can shorten the whole vehicle development period, saves cost for whole vehicle project development, and is relatively convenient and rapid.

Description

Technical Field

[0001] The present invention relates to the technical field of automobile brake systems, and in particular to a simulation model optimization method, a brake system calibration method, a device, equipment and a medium. Background Art

[0002] Braking system control function calibration is primarily conducted on a real vehicle under real-world conditions. Calibration typically requires testing on a real vehicle under specific road conditions. This calibration method is often extremely sensitive to factors such as vehicle condition, geographical conditions, environmental factors, and weather, which in turn impacts vehicle development cycles and increases development costs. Therefore, how to calibrate a vehicle's braking system to reduce both the development cycle and costs is a pressing technical challenge. Summary of the Invention

[0003] The embodiments of the present invention provide a simulation model optimization method, a brake system calibration method, an apparatus, a device and a medium to solve the problem of how to calibrate the brake system of an automobile to reduce the cycle and development cost of the entire vehicle.

[0004] A simulation model optimization method, comprising: Obtaining a training sample set corresponding to a calibration function, the training sample set comprising a plurality of training samples, each of the training samples comprising non-functional training data, functional training data corresponding to the calibration function, and a preset brake build-up pressure; Processing the non-functional training data and the functional training data in a first training sample using a brake system simulation model to determine a first brake build-up pressure output by the brake system simulation model, wherein the first training sample is a training sample used for model training; Based on the preset brake build-up pressure and the first brake build-up pressure, updating model parameters of a brake system simulation model, and determining an initial function simulation model corresponding to the calibration function; The initial function simulation model corresponding to the calibration function is tested using a second training sample to determine a target function simulation model corresponding to the calibration function, wherein the second training sample is a training sample other than the first training sample.

[0005] Preferably, the calibration function includes at least one of a hydraulic brake assist function, a hill-start assist control function, a hill-down assist control function, a traction control function, a vehicle stability control function, a brake energy recovery function, an electronic anti-roll function, and an electronic parking function; The function training data corresponding to the hydraulic brake assist function includes a brake pedal travel signal and an accelerator pedal travel signal when the hydraulic brake assist system is working; The function training data corresponding to the hill-start assist control function includes a brake pedal travel signal, an accelerator pedal travel signal, an acceleration sensor signal, and a wheel speed signal when the hill-start assist control system is working; The function training data corresponding to the downhill assist control function includes a brake pedal travel signal, an accelerator pedal travel signal, an acceleration sensor signal, and a wheel speed signal when the downhill assist control system is working; The function training data corresponding to the traction control function includes a brake pedal travel signal, an accelerator pedal travel signal, an acceleration sensor signal, a wheel end speed signal, and an inclination angle signal when the traction control system is working; The function training data corresponding to the vehicle stability control function includes a brake pedal travel signal, an accelerator pedal travel signal, an acceleration sensor signal, a wheel end speed signal and an inclination angle signal when the vehicle stability control system is working; The functional training data corresponding to the braking energy recovery function includes a brake pedal travel signal, an accelerator pedal travel signal, an acceleration sensor signal, and a wheel end speed signal when the braking energy recovery system is working; The functional training data corresponding to the electronic rollover prevention function includes a brake pedal travel signal, an accelerator pedal travel signal, an acceleration sensor signal, and a wheel speed signal when the electronic rollover prevention system is working; The function training data corresponding to the electronic parking function includes acceleration sensor signals, wheel end speed signals and inclination angle signals when the electronic parking system is working.

[0006] Preferably, updating the model parameters of the brake system simulation model based on the preset brake build-up pressure and the first brake build-up pressure, and determining the initial function simulation model corresponding to the calibration function, includes: determining an initial cost function value corresponding to each of the first training samples, wherein the initial cost function value is determined based on a difference between a first brake build-up pressure corresponding to the first training sample and a preset brake build-up pressure; Determining a target cost function value, where the target cost function value is determined based on initial cost function values corresponding to all of the first training samples; Based on the target cost function value, the model parameters of the braking system simulation model are updated, and the initial function simulation model corresponding to the calibration function is determined.

[0007] Preferably, the preset brake build-up pressure includes preset brake build-up pressures corresponding to four brakes; The first brake build-up pressure includes the first brake build-up pressure corresponding to four brakes; The initial cost function is the sum of target differences corresponding to the four brakes, and the target difference corresponding to each brake is the square of the difference between the preset brake pressure building pressure and the first brake pressure building pressure.

[0008] Preferably, updating the model parameters of the braking system simulation model based on the target cost function value and determining the initial function simulation model corresponding to the calibration function includes: If the target cost function value reaches a minimum value, the braking system simulation model is determined as the initial function simulation model corresponding to the calibration function; If the target cost function value does not reach the minimum value, updating the model parameters of the braking system simulation model; Wherein, updating the model parameters of the braking system simulation model includes: Determining a gradient function value corresponding to a braking system simulation model based on the target cost function value; Based on the gradient function value, updating the weight value and the bias value in the braking system simulation model, determining the updated weight value and the bias value, and determining the first optimized simulation model based on the updated weight value and the bias value; When the updated weight value is less than or equal to the preset weight value, at least one of the model input, node, and connection relationship corresponding to the updated weight value in the first optimization simulation model is deleted, a second optimization simulation model is determined, the second optimization simulation model is updated to the braking system simulation model, and the acquisition of the training sample set corresponding to the calibration function is repeated; When the updated weight value is greater than the preset weight value, the first optimization simulation model is updated to the braking system simulation model, and the acquisition of the training sample set corresponding to the calibration function is repeatedly executed.

[0009] Preferably, the using the second training sample to test the initial function simulation model corresponding to the calibration function to determine the target function simulation model corresponding to the calibration function includes: Processing the non-functional training data and the functional training data in the second training sample using the initial functional simulation model corresponding to the calibration function to determine the second brake build-up pressure output by the braking system simulation model; determining a test accuracy rate based on the second brake build-up pressure and the preset brake build-up pressure, wherein the test accuracy rate is determined based on a difference between the second brake build-up pressure and the preset brake build-up pressure corresponding to the second training sample; If the test accuracy reaches a preset accuracy, the initial function simulation model corresponding to the calibration function is determined as the target function simulation model of the calibration function.

[0010] A brake system calibration method, comprising: Acquiring data to be calibrated, wherein the data to be calibrated includes functional calibration data and non-functional calibration data corresponding to the calibration function; Inputting the data to be calibrated into the target function simulation model corresponding to the calibration function, and outputting the calibration brake build-up pressure corresponding to the calibration function; The target function simulation model corresponding to the calibration function is determined based on the above-mentioned simulation model optimization method.

[0011] A simulation model optimization device, comprising: A training sample set acquisition module is used to acquire a training sample set corresponding to a calibration function, wherein the training sample set includes a plurality of training samples, each of which includes non-functional training data, functional training data corresponding to the calibration function, and a preset brake build-up pressure; a first brake build-up pressure determination module, configured to process the non-functional training data and the functional training data in a first training sample using a brake system simulation model to determine a first brake build-up pressure output by the brake system simulation model, wherein the first training sample is a training sample used for model training; an initial function simulation model determining module, which updates model parameters of a brake system simulation model based on the preset brake build-up pressure and the first brake build-up pressure, and determines an initial function simulation model corresponding to the calibration function; The target function simulation model determination module is used to test the initial function simulation model corresponding to the calibration function using a second training sample to determine the target function simulation model corresponding to the calibration function, where the second training sample is a training sample other than the first training sample.

[0012] A brake system calibration device, comprising: A module for acquiring data to be calibrated, used for acquiring data to be calibrated, wherein the data to be calibrated includes functional calibration data and non-functional calibration data corresponding to the calibration function; a calibration module, configured to input the data to be calibrated into a target function simulation model corresponding to the calibration function, and output a calibration brake build-up pressure corresponding to the calibration function; The target function simulation model corresponding to the calibration function is determined based on the above-mentioned simulation model optimization method.

[0013] A computer device comprises a memory, a processor, and a computer program stored in the memory and executable on the processor, wherein the processor implements the above-mentioned simulation model optimization method when executing the computer program, or implements the above-mentioned brake system calibration method when executing the computer program.

[0014] A computer-readable storage medium stores a computer program, wherein the computer program implements the above-mentioned simulation model optimization method when executed by a processor, or implements the above-mentioned brake system calibration method when executed by a processor.

[0015] The above-mentioned simulation model optimization method, brake system calibration method, device, equipment and medium update and test the brake system simulation model through multiple first training samples and multiple second training samples corresponding to the calibration function to obtain a target function simulation model corresponding to the calibration function. This method can obtain a target function simulation model for calibrating different calibration functions. The target function simulation model can quickly calibrate different calibration functions in the brake system without a real vehicle, regardless of region or season, which can shorten the vehicle development cycle and save costs for vehicle project development. It is relatively convenient and quick. BRIEF DESCRIPTION OF THE DRAWINGS

[0016] In order to more clearly illustrate the technical solutions of the embodiments of the present invention, the following briefly introduces the drawings required for use in the description of the embodiments of the present invention. Obviously, the drawings described below are only some embodiments of the present invention. For ordinary technicians in this field, other drawings can be obtained based on these drawings without paying any creative labor.

[0017] Figure 1 is a flow chart of a simulation model optimization method according to an embodiment of the present invention; Figure 2 is a schematic diagram of a simulation model optimization device according to an embodiment of the present invention; Figure 3 is a schematic structural diagram of a braking system simulation model according to an embodiment of the present invention; Figure 4 1 is a schematic diagram of a training optimization process of a braking system simulation model according to an embodiment of the present invention; Figure 5 is a schematic diagram of a process for collecting training samples corresponding to a hydraulic brake assist function in one embodiment of the present invention; Figure 6 This is a flow chart of collecting training samples corresponding to the uphill and downhill assist control function, the braking energy recovery function, and the electronic anti-rollover function in one embodiment of the present invention; Figure 7 is a schematic diagram of a process for collecting training samples corresponding to the traction control function and the vehicle stability control function in one embodiment of the present invention; Figure 8 1 is a flow chart of collecting training samples corresponding to the electronic parking function in one embodiment of the present invention. DETAILED DESCRIPTION

[0018] The following will clearly and completely describe the technical solutions in the embodiments of the present invention in conjunction with the accompanying drawings. Obviously, the described embodiments are only part of the embodiments of the present invention, not all of them. All other embodiments obtained by ordinary technicians in this field based on the embodiments of the present invention without making any creative efforts shall fall within the scope of protection of the present invention.

[0019] The simulation model optimization method provided in an embodiment of the present invention is applied in a computer device to calibrate the braking system of an automobile, thereby reducing the cycle and development cost of the entire vehicle.

[0020] In one embodiment, if Figure 1 As shown, a simulation model optimization method is provided, which is described by taking the application of the method in a computer device as an example, and includes the following steps: S101: Acquire a training sample set corresponding to a calibration function, where the training sample set includes multiple training samples, each training sample includes non-functional training data, functional training data corresponding to the calibration function, and a preset brake build-up pressure; S102: Processing non-functional training data and functional training data in a first training sample using a brake system simulation model to determine a first brake build-up pressure output by the brake system simulation model, where the first training sample is a training sample used for model training; S103: Based on the preset brake build-up pressure and the first brake build-up pressure, updating the model parameters of the brake system simulation model, and determining an initial function simulation model corresponding to the calibration function; S104: using a second training sample to test the initial function simulation model corresponding to the calibration function to determine a target function simulation model corresponding to the calibration function, wherein the second training sample is a training sample other than the first training sample.

[0021] The calibration function refers to any of the various control functions of the vehicle's braking system that requires calibration of the brake build-up pressure. The calibration function includes at least one of the following: hydraulic brake assist, hill-start assist, downhill assist, traction control, vehicle stability control, brake energy recovery, electronic rollover mitigation, and electronic parking brake.

[0022] The first training sample is a training sample used for model training. The second training sample is a training sample other than the first training sample, specifically a training sample used for model testing.

[0023] Non-functional training data refers to training data that is not affected by the calibration function. In this example, non-functional training data includes vehicle parameters, brake system angular assembly parameters, environmental monitoring signals, and other parameters that affect brake system calibration. Vehicle parameters include wheelbase, track width, curb weight, fully loaded mass, front and rear axle loads, fully loaded axle loads, center of mass height, tire specifications, tire rolling radius, vehicle speed, driven angular assembly, and non-driven angular assembly. Angular assembly parameters include caliper drag torque and wheel hub bearing torque. Here, angular assemblies include but are not limited to calipers, brake discs, friction pads, wheel hub bearings, etc. Environmental monitoring signals include temperature, humidity, altitude, and atmospheric pressure. In this example, since the driven angular assembly and non-driven angular assembly configurations are not specific values, for ease of processing, the driven angular assembly can be set to 1 and the non-driven angular assembly can be set to -1.

[0024] The functional training data refers to training data related to the calibration function. It is understandable that for different calibration functions, different functional training data are required to train and optimize the brake system simulation model to obtain target functional simulation models corresponding to different calibration functions.

[0025] The preset brake build-up pressure refers to the label data used for model training, which can be the pressure labeled by the user or determined according to actual conditions.

[0026] As an example, in step S101, the computer device obtains a training sample set corresponding to the calibration function, and divides all training samples in the training sample set into multiple first training samples and multiple second training samples for training and optimizing the braking system simulation model corresponding to the calibration function. The first training sample and the second training sample both include multiple non-functional training data, multiple functional training data, and preset brake pressure build-up pressures corresponding to the functional training data. For example, each training sample set contains N training samples, which can be divided into first training samples and second training samples based on a preset division ratio. For example, when N=10000 and the division ratio is 9:1, it is determined that the number of first training samples is 9000, and the number of second training samples is 1000. As Figure 4 As shown, 100 first training samples can be used for model training each time.

[0027] The first brake build-up pressure refers to the brake build-up pressure output by the brake system simulation model after processing the first training sample.

[0028] As an example, in step S102, the computer device uses a preset brake system simulation model to process the non-functional training data and functional training data in a plurality of first training samples in sequence, and outputs the first brake pressure build-up pressure corresponding to each first training sample. Figure 3 In the structural diagram of the braking system simulation model shown in FIG, , , , It represents the input of the braking system simulation model. The input of the braking system simulation model is the non-functional training data in the first training sample corresponding to the calibration function and the functional training data corresponding to the calibration function. and is the hidden layer, and etc. The symbol represents the weight value, 、 、 and is the output of the braking system simulation model, namely 、 、 and represents the first brake build-up pressure, where Indicates the first brake build-up pressure corresponding to the left front brake in the braking system. Indicates the first brake build-up pressure corresponding to the right front brake in the braking system. Indicates the first brake build-up pressure corresponding to the left rear brake in the braking system. Indicates the first brake build-up pressure corresponding to the right rear brake in the braking system. Figure 4 As shown in the figure, it is a schematic diagram of the training optimization process of the braking system simulation model. Figure 4 It can be seen that the computer device obtains a training sample set and obtains 100 first training samples for each training of the brake system simulation model. The computer device uses a preset brake system simulation model to process the non-functional training data and functional training data in the 100 first training samples in turn, and outputs the first brake pressure build-up pressure corresponding to each first training sample. 、 、 and .

[0029] The initial functional simulation model refers to an updated model of the braking system simulation model. The model parameters are parameters constituting the initial functional simulation model, including but not limited to weight values and bias values.

[0030] As an example, in step S103, after the computer device obtains the preset brake build-up pressure corresponding to each first training sample and the first brake build-up pressure corresponding to the first training sample, it determines the difference between the preset brake build-up pressure corresponding to each first training sample and the first brake build-up pressure corresponding to the first training sample, and determines whether the braking system simulation model has converged based on the difference. If the braking system simulation model has not converged, the model parameters of the braking system simulation model are updated, and step S101 is repeated to continue updating the model parameters of the braking system simulation model until the braking system simulation model converges, thereby obtaining an initial functional simulation model corresponding to the calibration function. In this example, the model parameters of the braking system simulation model are updated using the preset brake build-up pressure corresponding to each first training sample and the first brake build-up pressure corresponding to the first training sample to obtain an initial functional simulation model that can accurately calibrate the calibration function of the braking system.

[0031] The target function simulation model refers to a simulation model that has been trained and optimized, and is used to calibrate the brake pressure under the calibration function of the automobile braking system.

[0032] As an example, in step S104, the computer device uses the second training sample to test the initial function simulation model corresponding to the calibration function to determine whether the initial function simulation model meets the preset test standard. If it is determined that the initial function simulation model meets the preset test standard, the initial function simulation model corresponding to the calibration function is updated to determine the target function simulation model corresponding to the calibration function. If it is determined that the initial function simulation model does not meet the preset test standard, the initial function simulation model is updated to the brake system simulation model, and step S101 is continued until the initial function simulation model meets the preset test standard, thereby obtaining the target function simulation model corresponding to the calibration function. In this example, the preset test standard can be the accuracy of the initial function simulation model. The computer device inputs the second training sample into the initial function simulation model to obtain the output of the initial function simulation model, obtains the difference between the output of the initial function simulation model and the preset brake build pressure corresponding to the second training sample, and determines whether the accuracy of the initial function simulation model is within a preset range based on the difference. If the accuracy is within the preset range, it is determined that the initial function simulation model meets the preset test standard, and the initial function simulation model corresponding to the calibration function is determined as the target function simulation model corresponding to the calibration function.

[0033] In this embodiment, the braking system simulation model is updated and tested by using multiple first training samples and multiple second training samples corresponding to the calibration function to obtain a target function simulation model corresponding to the calibration function. This method can obtain a target function simulation model for calibrating different calibration functions. The target function simulation model can quickly calibrate different calibration functions in the braking system without a real vehicle, regardless of region or season, which can shorten the vehicle development cycle, save costs for vehicle project development, and is relatively convenient and quick.

[0034] In one embodiment, the calibrated function includes at least one of a hydraulic brake assist function, a hill-start assist control function, a hill-down assist control function, a traction control function, a vehicle stability control function, a brake energy recovery function, an electronic rollover prevention function, and an electronic parking function; The functional training data corresponding to the hydraulic brake assist function includes the brake pedal travel signal and the accelerator pedal travel signal when the hydraulic brake assist system is working; The functional training data corresponding to the hill-start assist control function includes the brake pedal travel signal, accelerator pedal travel signal, acceleration sensor signal and wheel speed signal when the hill-start assist control system is working; The functional training data corresponding to the downhill assist control function includes the brake pedal travel signal, accelerator pedal travel signal, acceleration sensor signal and wheel end speed signal when the downhill assist control system is working; The functional training data corresponding to the traction control function includes the brake pedal travel signal, accelerator pedal travel signal, acceleration sensor signal, wheel end speed signal and inclination angle signal when the traction control system is working; The functional training data corresponding to the vehicle stability control function includes the brake pedal travel signal, accelerator pedal travel signal, acceleration sensor signal, wheel end speed signal and inclination angle signal when the vehicle stability control system is working; The functional training data corresponding to the braking energy recovery function includes the brake pedal travel signal, accelerator pedal travel signal, acceleration sensor signal and wheel end speed signal when the braking energy recovery system is working; The functional training data corresponding to the electronic rollover protection function includes the brake pedal travel signal, accelerator pedal travel signal, acceleration sensor signal and wheel speed signal when the electronic rollover protection system is working; The functional training data corresponding to the electronic parking function includes the acceleration sensor signal, wheel end speed signal and inclination angle signal when the electronic parking system is working.

[0035] As an example, the braking system includes at least one of a hydraulic brake assist system (HBA), a hill-start assist control system (HHC), a hill-descent assist control system (HDC), a traction control system (TCS), a vehicle stability control system (ESC), a brake energy recovery system (BERS), an electronic rollover mitigation system (RMI), and an electronic parking brake system (EPB), each of which corresponds to a calibrated function. In this example, the hydraulic brake assist system (HBA) corresponds to the hydraulic brake assist function, the hill-start assist control system (HHC) corresponds to the hill-start assist control function, the hill-descent assist control system (HDC) corresponds to the hill-descent assist control function, the traction control system (TCS) corresponds to the traction control function, the vehicle stability control system (ESC) corresponds to the vehicle stability control function, the brake energy recovery system (BERS) corresponds to the brake energy recovery function, the electronic rollover mitigation system (RMI) corresponds to the electronic rollover mitigation function, and the electronic parking brake system (EPB) corresponds to the electronic parking brake function. During vehicle development, different calibration functions require different brake build-up pressures. In this example, functional training data corresponding to different calibration functions is obtained so that target function simulation models corresponding to different calibration functions can be trained based on the functional training data, thereby achieving accurate calibration of the brake build-up pressures corresponding to different calibration functions.

[0036] In this example, in order to accurately train and optimize the brake system simulation models corresponding to different calibration functions, and obtain target function simulation models corresponding to different calibration functions, different calibration functions require different function training data when performing model optimization training. Figure 5 、 Figure 6 、 Figure 7 and Figure 8 As shown, the training samples include non-functional training data, functional training data corresponding to the calibration function and preset brake build pressure. The non-functional training data includes vehicle parameters, angular assembly parameters, environmental monitoring signals and other parameters. Among them, vehicle parameters include wheelbase, track width, curb weight, fully loaded weight, front and rear axle loads, fully loaded axle loads, center of mass height, tire specifications, tire rolling radius, vehicle speed, drive angular assembly and non-drive angular assembly. Angular assembly parameters include caliper drag torque and wheel hub bearing torque. Environmental monitoring signals include temperature signals, humidity signals, altitude signals and atmospheric pressure signals. Different calibration functions correspond to different functional training data and the preset brake build pressure corresponding to the left front brake corresponding to the functional training data. , the preset brake build pressure corresponding to the right front brake , the preset brake build pressure corresponding to the left rear brake Preset brake build-up pressure corresponding to the right rear brake .Depend on Figure 5It can be seen that the function training data corresponding to the hydraulic brake assist function includes the brake pedal stroke signal and the accelerator pedal stroke signal when the hydraulic brake assist system is working. Figure 6 It can be seen that the functional training data corresponding to the uphill assist control function includes the brake pedal stroke signal, throttle pedal stroke signal, acceleration sensor signal and wheel-end speed signal when the uphill assist control system is working; the functional training data corresponding to the downhill assist control function includes the brake pedal stroke signal, throttle pedal stroke signal, acceleration sensor signal and wheel-end speed signal when the downhill assist control system is working; the functional training data corresponding to the brake energy recovery function includes the brake pedal stroke signal, throttle pedal stroke signal, acceleration sensor signal and wheel-end speed signal when the brake energy recovery system is working; the functional training data corresponding to the electronic parking function includes the acceleration sensor signal, wheel-end speed signal and inclination signal when the electronic parking system is working. Figure 7 It can be seen that the functional training data corresponding to the traction control function includes the brake pedal travel signal, accelerator pedal travel signal, acceleration sensor signal, wheel end speed signal and tilt angle signal when the traction control system is working; the functional training data corresponding to the vehicle stability control function includes the brake pedal travel signal, accelerator pedal travel signal, acceleration sensor signal, wheel end speed signal and tilt angle signal when the vehicle stability control system is working; Figure 8 It can be seen that the function training data corresponding to the electronic anti-roll function includes the brake pedal travel signal, throttle pedal travel signal, acceleration sensor signal and wheel end speed signal when the electronic anti-roll system is working. Figures 5 to 8 It can be seen that the computer device collects function training data and environmental monitoring signals once every preset time (for example, 1ms) for each calibration function.

[0037] In this embodiment, functional training data corresponding to different calibration functions are obtained to train and optimize the braking system simulation model according to the functional training data corresponding to different calibration functions, and a target function simulation model corresponding to each calibration function is obtained, so that each calibration function can be accurately calibrated according to the target function simulation model corresponding to each calibration function.

[0038] In one embodiment, step S103, i.e., updating the model parameters of the brake system simulation model based on the preset brake build-up pressure and the first brake build-up pressure, and determining the initial function simulation model corresponding to the calibration function, includes: S201: Determine an initial cost function value corresponding to each first training sample, where the initial cost function value is determined based on a difference between a first brake build-up pressure corresponding to the first training sample and a preset brake build-up pressure; S202: Determine a target cost function value, where the target cost function value is determined based on initial cost function values corresponding to all first training samples; S203: Based on the target cost function value, update the model parameters of the braking system simulation model and determine the initial function simulation model corresponding to the calibration function.

[0039] The initial cost function value refers to a function value used to determine the difference between the first brake build-up pressure corresponding to each first training sample and the preset brake build-up pressure.

[0040] As an example, in step S201, the computer device performs difference processing on the first brake pressure build-up pressure and the preset brake pressure build-up pressure corresponding to the same first training sample, determines the difference between the two, and determines the initial cost function value corresponding to the first training sample based on the difference between the two. .For example, Figure 4 As shown, when the number of first training data z=100, for each first training data, the corresponding initial cost function value is determined to include , , , In this example, the initial cost function value is used to characterize the difference between the preset brake build-up pressure and the first brake build-up pressure in the same first training sample, and the initial cost function value corresponding to each first training sample is obtained.

[0041] The target cost function value refers to the function value used to characterize the training error of the braking system simulation model.

[0042] As an example, in step S202, the computer device performs statistical processing on the initial cost function values corresponding to all first training samples to determine the target cost function value. In this example, the computer device performs average processing on the initial cost function values corresponding to all first training samples, and the average value of the initial cost function values corresponding to all first training samples can be determined as the target cost function value, that is, the target cost function value for ,in, is the number of the first training samples, is the initial cost function value corresponding to the first training sample z. Figure 4 As shown, z=100, Understandably, since the brake build-up pressure output by the model is credible when the average of the initial cost function values corresponding to all first training samples is minimum, the average of the initial cost function values corresponding to all first training samples is determined as the target cost function value, so as to facilitate updating the brake system simulation model according to the target cost function value. In this example, the target cost function value can also be a statistical value such as the sum, median, and mode of the initial cost function values corresponding to all first training samples.

[0043] As an example, in step S203, the computer device updates the weight value and bias value of the braking system simulation model according to the target cost function value to obtain an updated braking system simulation model. When the updated braking system simulation model converges, the initial function simulation model corresponding to the calibration function is obtained. In this example, when the computer device determines that the updated braking system simulation model does not converge, it continues to execute step S101, updates the braking system simulation model, and obtains the initial function simulation model corresponding to the calibration function until the updated braking system simulation model converges. Figure 4 As shown, the computer device determines the target cost function value of the one hundred first training samples at the current moment to evaluate whether the braking system simulation model converges. If the braking system simulation model does not converge, the next one hundred first training samples are used to train and optimize the braking system simulation model.

[0044] In this embodiment, the computer device determines the initial cost function value corresponding to each first training sample based on the preset brake build-up pressure and the first brake build-up pressure corresponding to the same first training sample, and determines the target cost function value for updating the braking system simulation model based on the initial cost function values corresponding to all first training samples. Based on the target cost function value, the model parameters of the braking system simulation model are updated to obtain a converged initial functional simulation model, which is used to calibrate the calibration function of the braking system without a real vehicle, saving manpower and material resources, and saving development costs and development cycles.

[0045] In one embodiment, the preset brake building pressure includes the preset brake building pressure corresponding to the four brakes; the first brake building pressure includes the first brake building pressure corresponding to the four brakes; the initial cost function is the sum of the target differences corresponding to the four brakes; the target difference corresponding to each brake is the square of the difference between the preset brake building pressure and the first brake building pressure.

[0046] As an example, the computer device obtains the sum of squares of the differences between the preset brake pressure build-up pressure and the first brake pressure build-up pressure corresponding to the same first training sample, and determines the sum of squares as the initial cost function value corresponding to the group of first training samples. Figure 3 As shown, the first brake pressure build-up pressure includes the first brake pressure build-up pressure corresponding to the left front brake , the first brake pressure corresponding to the right front brake , the first brake pressure corresponding to the left rear brake , and the first brake build-up pressure corresponding to the right rear brake The first training sample corresponds to Corresponding preset brake build-up pressure 、 Corresponding preset brake build-up pressure 、 Corresponding preset brake build-up pressure and Corresponding preset brake build-up pressure In this example, the computer device performs the same first training sample on and , and , and , and Perform difference processing to obtain the difference processing result ( )、( )、( )and( ), square the difference processing result to get the target difference and , the sum of the target differences is determined as the initial cost function value corresponding to the first training sample , ,in, is the initial cost function value corresponding to the zth first training sample.

[0047] In this example, the initial cost function value is used to characterize the difference between the preset brake build-up pressure and the first brake build-up pressure in the same first training sample, and the initial cost function value for training each first training sample is obtained, so as to determine the target cost function value corresponding to the braking system simulation model based on the initial cost function values corresponding to multiple groups of first training samples.

[0048] In one embodiment, step S203, i.e., updating the model parameters of the brake system simulation model based on the target cost function value and determining the initial function simulation model corresponding to the calibration function, includes: S301: If the target cost function value reaches the minimum value, the braking system simulation model is determined as the initial function simulation model corresponding to the calibration function; S302: If the target cost function value does not reach the minimum value, updating the model parameters of the braking system simulation model; The model parameters of the braking system simulation model are updated, including: S3021: Determine a gradient function value corresponding to the braking system simulation model based on the target cost function value; S3022: updating weight values and bias values in the braking system simulation model based on the gradient function value, determining updated weight values and bias values, and determining a first optimized simulation model based on the updated weight values and bias values; S3023: When the updated weight value is less than or equal to the preset weight value, at least one of the model input, node, and connection relationship corresponding to the updated weight value in the first optimization simulation model is deleted, a second optimization simulation model is determined, the second optimization simulation model is updated to the braking system simulation model, and the training sample set corresponding to the calibration function is repeatedly obtained; S3024: When the updated weight value is greater than the preset weight value, the first optimization simulation model is updated to the braking system simulation model, and the training sample set corresponding to the calibration function is repeatedly obtained.

[0049] As an example, in step S301, when the computer device determines that the target cost function value is the minimum value, it determines that the difference between the first brake pressure build-up pressure output by the braking system simulation model at this time and the preset brake pressure build-up pressure in the same first training sample is small, and the output of the braking system simulation model is credible and can meet the model convergence condition. At this time, the braking system simulation model is determined as the initial function simulation model corresponding to the calibration function. Figure 4 As shown, whether the braking system simulation model converges is determined based on whether the target cost function value is the minimum value. When the target cost function value is the minimum value, the braking system simulation model converges and can be determined as the initial function simulation model corresponding to the calibration function.

[0050] As an example, in step S302, when the computer device determines that the target cost function value has not reached the minimum value, the difference between the first brake pressure building pressure corresponding to the first training sample and the preset brake pressure building pressure is large, and the model convergence condition is not met. It is necessary to update the model parameters of the braking system simulation model according to the target cost function to obtain the updated braking system simulation model, and continue to execute step S101 until the target cost function value is determined to be the minimum value, and the initial function simulation model corresponding to the calibration function is obtained.

[0051] In this example, in step S302, the model parameters of the braking system simulation model are updated according to the target cost function to obtain the initial function simulation model corresponding to the calibration function, which specifically includes: The gradient function value is a function value used to perform gradient update on the braking system simulation model.

[0052] As an example, in step S3021, the computer device processes the target cost function value and determines the gradient function value corresponding to the target cost function value. In this example, the gradient function value for: In this example, the target cost function value is processed to determine the gradient function value so that the braking system simulation model updates the model parameters according to the gradient function value. Figure 4 As shown, the target cost function value is determined according to the average cost function After that, the target cost function value is processed to determine the gradient function value corresponding to the target cost function value .

[0053] The first optimized simulation model refers to a model obtained by optimizing and updating the model parameters of the braking system simulation model.

[0054] As an example, in step S3022, the computer device calculates all weight values and bias values in the braking system simulation model according to the gradient function value. Update to obtain updated weight values and bias values, and determine the braking system simulation model containing the updated weight values and bias values as the first optimized simulation model. Figure 4 As shown, the computer device sorts all weight values and all bias values in the braking system simulation model according to the column vector W, and uses the gradient function value Perform difference processing on the column vector W to determine the optimized column vector , =W , column vector Contains all updated weight values and all bias values in the brake system simulation model.

[0055] For example, Figure 3 In the braking system simulation model shown in FIG, the specific process of obtaining and updating the column vector W includes: According to the i-dimensional column vector corresponding to the i nodes of the hidden layer (L-1) , determine the column vector corresponding to the input of the braking system simulation model and the first bias set B, =RELU ,in, = , = , = , =[ , ], is the node value of the i-th node, is the xth weight value corresponding to the i-th node in the hidden layer (L-1), For the x-th model input, is the bias value corresponding to the i-th node in the hidden layer (L-1), is the column vector corresponding to the bias values of all nodes in the hidden layer (L-1), is the matrix corresponding to all weights between the hidden layer (L-1) and the input layer in the braking system simulation model, is the column vector corresponding to the input of the braking system simulation model.

[0056] According to the i-dimensional column vector corresponding to the i nodes corresponding to the hidden layer L , determine the matrix corresponding to all weights between the hidden layer L and the hidden layer (L-1) in the braking system simulation model and the second bias set D, =RELU ,in, = , = , =[ , ], is the node value of the j-th node, is the i-th weight value corresponding to the j-th node in the hidden layer L, is the bias value corresponding to the jth node in the hidden layer L, is the column vector corresponding to the bias values of all nodes in the hidden layer L, is the matrix corresponding to all weights between the hidden layer L and the hidden layer (L-1) in the braking system simulation model.

[0057] According to the n-dimensional column vector corresponding to the n nodes corresponding to the output layer , determine the matrix corresponding to all weights between the hidden layer L and the output layer in the braking system simulation model and the third bias set E, =RELU ,in, = , = , =[ , ], is the nth model output, is the jth weight value corresponding to the nth output, is the bias value corresponding to the nth node, is a column vector formed by the bias values corresponding to the n nodes of the output layer, is the matrix corresponding to all weights between the hidden layer L and the output layer in the braking system simulation model, where RELU is the linear rectification function.

[0058] Computer equipment will 、 、 、 、 and , arranged according to the column vector, we get the column vector W, W=[ , , , , , ], using the gradient function value The column vector W is updated to update all weight values and bias values in the braking system simulation model, thereby updating the node values of all hidden layer nodes in the braking system simulation model, and obtaining a first optimized simulation model in which the weight values, bias values and node values are all updated.

[0059] The preset weight value refers to a preset value used to determine the size of the weight value.

[0060] As an example, in step S3023, the computer device determines the size of the updated weight value, and when it is determined that the updated weight value is less than or equal to the preset weight value r, deletes at least one of the model inputs, nodes, and connection relationships corresponding to the weight value that is less than or equal to the preset weight value r, obtains an updated second optimization simulation model, and updates the second optimization simulation model to a braking system simulation model, and repeats step S101, that is, repeatedly obtains the training sample set corresponding to the calibration function until the target cost function value reaches the minimum value, and obtains the initial function simulation model corresponding to the calibration function. In this example, when the computer device determines that an updated weight value is less than or equal to the preset weight value, it deletes the connection relationship corresponding to the weight value. Alternatively, when the computer determines that all updated weight values corresponding to a node are less than or equal to the preset weight value, it deletes the connection relationship corresponding to the node and all weight values that are less than or equal to the preset weight value. When the computer device determines that all updated weight values corresponding to a model input are less than or equal to the preset weight value, it deletes the model input and the connection relationship corresponding to the weight value that is less than or equal to the preset weight value. For example, Figure 3 As shown, if the updated weight value If it is less than or equal to the preset weight value r, the updated weight value will be Corresponding model input and the nodes in the hidden layer (L-1) If the nodes in the hidden layer (L-1) With all model inputs If the updated weight values between the two are less than or equal to the preset weight value r, the node and nodes With all model inputs The connection between them is deleted. If the model input If the updated weight values between all nodes in the hidden layer (L-1) are less than or equal to the preset weight value r, the model input And model input The connection relationship between the updated weight value and all nodes in the hidden layer (L-1) is deleted. Understandably, if the updated weight value is less than or equal to the preset weight value, it means that the weight value has little influence. Since the total number of all weight values and bias values in the model is T= ,in, is the number of nodes in the hidden layer L-1, The number of inputs to the model, For the number of nodes in the hidden layer L, at least one of the model inputs, nodes, and connection relationships that have less impact on the model is deleted, so as to reasonably reduce the total number T of weight values and bias values in the model and improve the running rate of the braking system simulation model. In this example, when the updated weight value is less than or equal to the preset weight value, the nodes and connection relationships corresponding to the updated weight value in the first optimized simulation model are deleted, and a second optimized simulation model is determined, which can delete the nodes and connection relationships that have less impact on the calibration function, thereby improving the running rate of the braking system simulation model. The second optimized simulation model with a reduced total number of weight values and bias values is updated to the braking system simulation model for model training optimization, thereby improving the model training optimization rate.

[0061] As an example, in step S3024, the computer device determines the size of the updated weight value. When it is determined that the updated weight value is greater than the preset weight value r, all updated weight values and bias values in the first optimization simulation model are retained, the first optimization simulation model is updated to the braking system simulation model, and step S101 is repeated, that is, the training sample set corresponding to the calibration function is repeatedly obtained until the target cost function value reaches the minimum value, and the initial function simulation model corresponding to the calibration function is obtained. It can be understood that if the updated weight value is greater than the preset weight value r, it indicates that all updated weight values and bias values in the first optimization simulation model have a significant impact on the calibration function. Therefore, all updated weight values and bias values in the first optimization simulation model are retained, the first optimization simulation model is updated to the braking system simulation model, and model training optimization is performed to obtain a more accurate initial function simulation model corresponding to the calibration function.

[0062] In this embodiment, when the updated weight value is less than or equal to the preset weight value, the nodes and connection relationships that have less impact on the calibration function will be deleted, the operation rate of the braking system simulation model will be improved, and the second optimized simulation model with a reduced total number of weight values and bias values will be updated to the braking system simulation model for model training optimization, thereby improving the model training optimization rate.

[0063] In one embodiment, step S104, i.e., testing the initial function simulation model corresponding to the calibration function using the second training sample to determine the target function simulation model corresponding to the calibration function, includes: S501: Processing non-functional training data and functional training data in a second training sample using an initial functional simulation model corresponding to the calibration function to determine a second brake build-up pressure output by a braking system simulation model; S502: Determine a test accuracy rate based on the second brake build-up pressure and the preset brake build-up pressure, where the test accuracy rate is determined based on a difference between the second brake build-up pressure and the preset brake build-up pressure corresponding to the second training sample; S503: If the test accuracy reaches the preset accuracy, the initial function simulation model corresponding to the calibration function is determined as the target function simulation model of the calibration function.

[0064] The second brake build-up pressure refers to the model output after the initial function simulation model corresponding to the specified function processes the second training sample.

[0065] As an example, in step S501, the computer device inputs the non-functional training data and functional training data from the second training samples corresponding to the calibration function into the initial functional simulation model corresponding to the calibration function, and outputs the second brake built-up pressure corresponding to the second training samples. In this example, the computer device may process a set of second training samples using the initial functional simulation model corresponding to the calibration function to obtain a set of second brake built-up pressures, or may process multiple sets of second training samples using the initial functional simulation model corresponding to the calibration function to obtain multiple sets of second brake built-up pressures.

[0066] The test accuracy refers to the accuracy of the output of the initial function simulation model determined based on the difference between the second brake build-up pressure and the preset brake build-up pressure.

[0067] As an example, in step S502, the computer device obtains the difference between the second brake pressure and the preset brake pressure, and determines the test accuracy based on the difference. Figure 3 As shown, if the test accuracy of the initial function simulation model is determined by the second brake pressure build-up pressure of group L (L=1), then its test accuracy can be C= ,in, is the second brake build-up pressure corresponding to the second training sample, is the preset brake build-up pressure corresponding to the same second training sample. Including the second brake pressure build-up pressure corresponding to the left front brake , the second brake pressure corresponding to the right front brake , the second brake pressure corresponding to the left rear brake The second brake pressure corresponding to the right rear brake The preset brake pressure build-up pressure corresponding to the same second training sample includes Corresponding , Corresponding , Corresponding , Corresponding When the test accuracy can also be C= ,in, It refers to the preset brake building pressure corresponding to the t-th second brake building pressure.

[0068] Alternatively, if the test accuracy of the initial function simulation model is determined by the L (L>1) group of second brake build-up pressures, the test accuracy can be C= ,like The first brake pressure in the L (L>1) group The second brake build-up pressure of each group. Including the second brake pressure build-up pressure corresponding to the left front brake , the second brake pressure corresponding to the right front brake , the second brake pressure corresponding to the left rear brake The second brake pressure corresponding to the right rear brake , each The corresponding preset brake build-up pressure is , each The corresponding preset brake build-up pressure is , each The corresponding preset brake build-up pressure is , each The corresponding preset brake build-up pressure is When the test accuracy can also be C= ,in, For the The t-th second brake build-up pressure in the group, For the The t-th preset brake build-up pressure in the group, t=1, 2, 3, and 4, represents the brakes in the four directions.

[0069] The preset accuracy rate refers to a preset value used to determine the accuracy rate of the test.

[0070] As an example, in step S503, the computer device obtains the test accuracy. If the test accuracy reaches a preset accuracy, the initial functional simulation model corresponding to the calibration function is determined as the target functional simulation model of the calibration function. If the preset accuracy is not reached, step S101 is repeated until the test accuracy reaches the preset accuracy, thereby obtaining the target functional simulation model of the calibration function. In this example, the preset accuracy is 0.8.

[0071] In this embodiment, non-functional training data and functional training data in the untrained second training sample are used to test the initial functional simulation model corresponding to the calibration function. When the test accuracy of the initial functional simulation model reaches the preset accuracy, a target functional simulation model with high accuracy and convergence corresponding to the calibration function is obtained, so as to facilitate functional calibration of the braking system according to the target functional simulation model.

[0072] In another embodiment, a brake system calibration method is provided, comprising: S601: Acquire data to be calibrated, where the data to be calibrated includes functional calibration data and non-functional calibration data corresponding to the calibration function; Input the data to be calibrated into the target function simulation model corresponding to the calibration function, and output the calibration brake build-up pressure corresponding to the calibration function; The target function simulation model corresponding to the calibration function is determined based on the above-mentioned simulation model optimization method.

[0073] The data to be calibrated refers to data used to calibrate the brake system's calibration functions. Functional calibration data refers to model input data corresponding to the target function simulation model associated with the calibration function. Non-functional calibration data refers to model input data corresponding to the target function simulation model that affects all calibration functions of the brake system.

[0074] As an example, in step S601, the computer device obtains data to be calibrated including functional calibration data and non-functional calibration data corresponding to the calibration function. In this example, the non-functional calibration data include vehicle parameters, angular assembly parameters of the braking system, and other parameters that have an impact on the calibration braking system, such as environmental monitoring signals. Vehicle parameters include wheelbase, track width, curb weight, fully loaded weight, front and rear axle loads, fully loaded axle loads, center of mass height, tire specifications, tire rolling radius, vehicle speed, drive angular assembly, and non-drive angular assembly. Angular assembly parameters include caliper drag torque and hub bearing torque. Environmental monitoring signals include temperature signals, humidity signals, altitude signals, and atmospheric pressure signals. In this example, if the calibration function is the hydraulic brake assist function, the function calibration data includes the brake pedal travel signal and the accelerator pedal travel signal; if the calibration function is the uphill assist control function, the function calibration data includes the brake pedal travel signal, the accelerator pedal travel signal, the acceleration sensor signal and the wheel end speed signal; if the calibration function is the downhill assist control function, the function calibration data includes the brake pedal travel signal, the accelerator pedal travel signal, the acceleration sensor signal and the wheel end speed signal; if the calibration function is the traction control function, the function calibration data includes the brake pedal travel signal, the accelerator pedal travel signal, the acceleration sensor signal, the wheel end speed signal and the inclination angle signal; if the calibration function is the vehicle stability control function, the function calibration data includes the brake pedal travel signal, the accelerator pedal travel signal, the acceleration sensor signal, the wheel end speed and the inclination angle signal; if the calibration function is the brake energy recovery function, the function calibration data includes the brake pedal travel signal, the accelerator pedal travel signal, the acceleration sensor signal and the wheel end speed signal; if the calibration function is the electronic anti-roll function, the function calibration data includes the brake pedal travel signal, the accelerator pedal travel signal, the acceleration sensor signal and the wheel end speed signal; if the calibration function is the electronic parking function, the function calibration data includes the acceleration sensor signal, the wheel end speed signal and the inclination angle signal.

[0075] The calibration brake build-up pressure refers to the calibration result of the calibration function of the brake system.

[0076] As an example, in step S602, the computer device inputs the data to be calibrated, including the functional calibration data and non-functional calibration data corresponding to the calibration function, into the target function simulation model corresponding to the calibration function, and outputs the calibration brake pressure corresponding to the calibration function. Figure 3 After optimizing the braking system simulation model corresponding to the calibration function shown in the figure, the input layer is obtained as , , , , the output layer is 、 、 and The target function simulation model of the calibration function is input into the target function simulation model corresponding to the calibration function, and the non-functional calibration data corresponding to the calibration function is output. 、 、 and , that is, the standardized brake build-up pressures corresponding to the left front brake, right front brake, left rear brake, and right rear brake of the braking system under the calibration function are obtained.

[0077] In this embodiment, the target function simulation model corresponding to the calibration function is used, and the functional calibration data and non-functional calibration data corresponding to the calibration function are processed to calibrate the brake building pressure corresponding to the brake in the braking system under the calibration function. This method does not need to consider the influence of factors such as vehicle condition, geographical conditions, environment and weather. It can complete the calibration task of the braking system under the calibration function without a real vehicle, regardless of region or season, saving costs for vehicle project development and shortening the vehicle development cycle. It is relatively convenient and quick.

[0078] It should be understood that the size of the serial numbers of the steps in the above embodiments does not mean the order of execution. The execution order of each process should be determined by its function and internal logic, and should not constitute any limitation on the implementation process of the embodiments of the present invention.

[0079] In one embodiment, a simulation model optimization device is provided, which corresponds one-to-one to the simulation model optimization method in the above embodiment. Figure 2 As shown, the simulation model optimization device includes a training sample set acquisition module 21, a first brake pressure building pressure determination module 22, an initial function simulation model determination module 23 and a target function simulation model determination module 24. The functional modules are described in detail as follows: A training sample set acquisition module 21 is used to acquire a training sample set corresponding to a calibration function, wherein the training sample set includes a plurality of training samples, each of which includes non-functional training data, functional training data corresponding to the calibration function, and a preset brake build-up pressure; a first brake build-up pressure determination module 22, configured to process non-functional training data and functional training data in a first training sample using a brake system simulation model to determine a first brake build-up pressure output by the brake system simulation model, wherein the first training sample is a training sample used for model training; The initial function simulation model determination module 23 updates the model parameters of the brake system simulation model based on the preset brake build-up pressure and the first brake build-up pressure, and determines the initial function simulation model corresponding to the calibration function; The target function simulation model determination module 24 is configured to test the initial function simulation model corresponding to the calibration function using a second training sample to determine the target function simulation model corresponding to the calibration function, wherein the second training sample is a training sample other than the first training sample.

[0080] In one embodiment, the initial function simulation model determination module 23 includes: an initial cost function value determination submodule, configured to determine an initial cost function value corresponding to each first training sample, the initial cost function value being determined based on a difference between a first brake build-up pressure corresponding to the first training sample and a preset brake build-up pressure; A target cost function value determination submodule is used to determine a target cost function value, where the target cost function value is determined based on the initial cost function values corresponding to all first training samples; The initial function simulation model determination submodule updates the model parameters of the braking system simulation model based on the target cost function value and determines the initial function simulation model corresponding to the calibration function.

[0081] In one embodiment, the initial function simulation model determination submodule includes: a first judgment unit, configured to determine the braking system simulation model as an initial function simulation model corresponding to the calibration function if the target cost function value reaches a minimum value; The second judgment unit is configured to update the model parameters of the braking system simulation model if the target cost function value does not reach the minimum value.

[0082] In one embodiment, the second determining unit includes: Determining a gradient function value corresponding to a braking system simulation model based on a target cost function value; A first model updating subunit updates weight values and bias values in the braking system simulation model based on the gradient function value, determines updated weight values and bias values, and determines a first optimized simulation model based on the updated weight values and bias values; The second model updating subunit is configured to, when the updated weight value is less than or equal to the preset weight value, delete at least one of the model inputs, nodes, and connection relationships corresponding to the updated weight value in the first optimization simulation model, determine a second optimization simulation model, update the second optimization simulation model to the braking system simulation model, and repeatedly execute the acquisition of a training sample set corresponding to the calibration function; The third model updating subunit is used to update the first optimization simulation model to the braking system simulation model when the updated weight value is greater than the preset weight value, and repeatedly execute the acquisition of the training sample set corresponding to the calibration function.

[0083] In one embodiment, the target function simulation model determination module 24 includes: a second brake build-up pressure determination submodule, configured to process the non-functional training data and the functional training data in the second training sample using the initial functional simulation model corresponding to the calibration function, and determine the second brake build-up pressure output by the braking system simulation model; a test accuracy determination submodule, configured to determine a test accuracy based on the second brake build-up pressure and a preset brake build-up pressure, wherein the test accuracy is determined based on a difference between the second brake build-up pressure and the preset brake build-up pressure corresponding to the second training sample; The target function simulation model determination submodule is used to determine the initial function simulation model corresponding to the calibration function as the target function simulation model of the calibration function if the test accuracy reaches a preset accuracy.

[0084] In another embodiment, a brake system calibration device is provided. The simulation model optimization device corresponds one-to-one with the brake system calibration method described in the above embodiment. The brake system calibration device includes a data acquisition module to be calibrated and a calibration module. The functional modules are described in detail as follows: The module for acquiring data to be calibrated is used to acquire data to be calibrated, where the data to be calibrated includes functional calibration data and non-functional calibration data corresponding to the calibration function; The calibration module is used to input the data to be calibrated into the target function simulation model corresponding to the calibration function, and output the calibration brake build-up pressure corresponding to the calibration function.

[0085] For the specific definitions of the simulation model optimization device and the brake system calibration device, please refer to the definitions of the simulation model optimization method and the brake system calibration method above, which will not be repeated here. The various modules in the above-mentioned simulation model optimization device and the brake system calibration device can be implemented in whole or in part by software, hardware, and a combination thereof. The above-mentioned modules can be embedded in or independent of the processor in the computer device in the form of hardware, or can be stored in the memory of the computer device in the form of software, so that the processor can call and execute the operations corresponding to the above modules.

[0086] In one embodiment, a computer device is provided, including a memory, a processor, and a computer program stored in the memory and executable on the processor. When the processor executes the computer program, the simulation model optimization method in the above embodiment is implemented, for example Figure 1 Alternatively, when the processor executes the computer program, the functions of each module / unit in the embodiment of the simulation model optimization device are realized, for example, Figure 2 The functions of the training sample set acquisition module 21 , the first brake build-up pressure determination module 22 , the initial function simulation model determination module 23 and the target function simulation model determination module 24 are not described here in detail to avoid repetition.

[0087] In another embodiment, a computer device is provided, including a memory, a processor, and a computer program stored in the memory and executable on the processor. When the processor executes the computer program, the brake system calibration method in the above embodiment is implemented. Alternatively, when the processor executes the computer program, the functions of the modules / units in the above brake system calibration device embodiment are implemented, such as the functions of the calibration data acquisition module and the calibration module. To avoid repetition, they are not described here.

[0088] In one embodiment, a computer-readable storage medium is provided, on which a computer program is stored. When the computer program is executed by a processor, the simulation model optimization method in the above embodiment is implemented, for example Figure 1 Alternatively, when the computer program is executed by a processor, the functions of each module / unit in the embodiment of the simulation model optimization device are realized, for example, Figure 2 The functions of the training sample set acquisition module 21 , the first brake build-up pressure determination module 22 , the initial function simulation model determination module 23 and the target function simulation model determination module 24 are not described here in detail to avoid repetition.

[0089] In another embodiment, a computer-readable storage medium is provided, on which a computer program is stored. When the computer program is executed by a processor, the brake system calibration method in the above embodiment is implemented. Alternatively, when the computer program is executed by a processor, the functions of each module / unit in the above brake system calibration device in this embodiment are implemented, such as the functions of the calibration data acquisition module and the calibration module. To avoid repetition, they are not repeated here.

[0090] Those skilled in the art will understand that all or part of the processes in the above-mentioned embodiments can be implemented by instructing the relevant hardware through a computer program. The computer program can be stored in a non-volatile computer-readable storage medium. When the computer program is executed, it can include the processes of the embodiments of the above-mentioned methods. Among them, any reference to memory, storage, database or other media used in the embodiments provided in this application may include non-volatile and / or volatile memory. Non-volatile memory may include read-only memory (ROM), programmable ROM (PROM), electrically programmable ROM (EPROM), electrically erasable programmable ROM (EEPROM) or flash memory. Volatile memory may include random access memory (RAM) or external cache memory. By way of illustration and not limitation, RAM is available in many forms such as static RAM (SRAM), dynamic RAM (DRAM), synchronous DRAM (SDRAM), double data rate SDRAM (DDRSDRAM), enhanced SDRAM (ESDRAM), Synchronous Link DRAM (SLDRAM), Rambus direct RAM (RDRAM), direct memory bus dynamic RAM (DRDRAM), and memory bus dynamic RAM (RDRAM), etc.

[0091] The embodiments described above are only used to illustrate the technical solutions of the present invention, rather than to limit the same. Although the present invention has been described in detail with reference to the aforementioned embodiments, those skilled in the art should understand that they can still modify the technical solutions described in the aforementioned embodiments, or make equivalent replacements for some of the technical features therein. These modifications or replacements do not deviate the essence of the corresponding technical solutions from the spirit and scope of the technical solutions of the various embodiments of the present invention, and should all be included in the scope of protection of the present invention.

Claims

1. A simulation model optimization method, characterized in that: include: Obtaining a training sample set corresponding to a calibration function, the training sample set comprising a plurality of training samples, each of the training samples comprising non-functional training data, functional training data corresponding to the calibration function, and a preset brake build-up pressure; Processing the non-functional training data and the functional training data in a first training sample using a brake system simulation model to determine a first brake build-up pressure output by the brake system simulation model, wherein the first training sample is a training sample used for model training; Based on the preset brake build-up pressure and the first brake build-up pressure, updating model parameters of a brake system simulation model, and determining an initial function simulation model corresponding to the calibration function; The initial function simulation model corresponding to the calibration function is tested using a second training sample to determine a target function simulation model corresponding to the calibration function, wherein the second training sample is a training sample other than the first training sample.

2. The simulation model optimization method according to claim 1, characterized in that: The calibrated functions include at least one of a hydraulic brake assist function, a hill-start assist control function, a hill-down assist control function, a traction control function, a vehicle stability control function, a brake energy recovery function, an electronic rollover prevention function, and an electronic parking function; The function training data corresponding to the hydraulic brake assist function includes a brake pedal travel signal and an accelerator pedal travel signal when the hydraulic brake assist system is working; The function training data corresponding to the hill-start assist control function includes a brake pedal travel signal, an accelerator pedal travel signal, an acceleration sensor signal, and a wheel speed signal when the hill-start assist control system is working; The function training data corresponding to the downhill assist control function includes a brake pedal travel signal, an accelerator pedal travel signal, an acceleration sensor signal, and a wheel speed signal when the downhill assist control system is working; The function training data corresponding to the traction control function includes a brake pedal travel signal, an accelerator pedal travel signal, an acceleration sensor signal, a wheel end speed signal, and an inclination angle signal when the traction control system is working; The function training data corresponding to the vehicle stability control function includes a brake pedal travel signal, an accelerator pedal travel signal, an acceleration sensor signal, a wheel end speed signal and an inclination angle signal when the vehicle stability control system is working; The functional training data corresponding to the braking energy recovery function includes a brake pedal travel signal, an accelerator pedal travel signal, an acceleration sensor signal, and a wheel end speed signal when the braking energy recovery system is working; The functional training data corresponding to the electronic rollover prevention function includes a brake pedal travel signal, an accelerator pedal travel signal, an acceleration sensor signal, and a wheel speed signal when the electronic rollover prevention system is working; The function training data corresponding to the electronic parking function includes acceleration sensor signals, wheel end speed signals and inclination angle signals when the electronic parking system is working.

3. The simulation model optimization method according to claim 1, characterized in that: The updating of the model parameters of the brake system simulation model based on the preset brake build-up pressure and the first brake build-up pressure, and determining the initial function simulation model corresponding to the calibration function, includes: determining an initial cost function value corresponding to each of the first training samples, wherein the initial cost function value is determined based on a difference between a first brake build-up pressure corresponding to the first training sample and a preset brake build-up pressure; Determining a target cost function value, where the target cost function value is determined based on initial cost function values corresponding to all of the first training samples; Based on the target cost function value, the model parameters of the braking system simulation model are updated, and the initial function simulation model corresponding to the calibration function is determined.

4. The simulation model optimization method according to claim 3, characterized in that: The preset brake build-up pressure includes the preset brake build-up pressure corresponding to the four brakes; The first brake build-up pressure includes the first brake build-up pressure corresponding to four brakes; The initial cost function is the sum of target differences corresponding to the four brakes, and the target difference corresponding to each brake is the square of the difference between the preset brake pressure building pressure and the first brake pressure building pressure.

5. The simulation model optimization method according to claim 3, characterized in that: The updating of the model parameters of the braking system simulation model based on the target cost function value and determining the initial function simulation model corresponding to the calibration function include: If the target cost function value reaches a minimum value, the braking system simulation model is determined as the initial function simulation model corresponding to the calibration function; If the target cost function value does not reach the minimum value, updating the model parameters of the braking system simulation model; Wherein, updating the model parameters of the braking system simulation model includes: Determining a gradient function value corresponding to a braking system simulation model based on the target cost function value; Based on the gradient function value, updating the weight value and the bias value in the braking system simulation model, determining the updated weight value and the bias value, and determining the first optimized simulation model based on the updated weight value and the bias value; When the updated weight value is less than or equal to the preset weight value, at least one of the model input, node, and connection relationship corresponding to the updated weight value in the first optimization simulation model is deleted, a second optimization simulation model is determined, the second optimization simulation model is updated to the braking system simulation model, and the acquisition of the training sample set corresponding to the calibration function is repeated; When the updated weight value is greater than the preset weight value, the first optimization simulation model is updated to the braking system simulation model, and the acquisition of the training sample set corresponding to the calibration function is repeatedly executed.

6. The simulation model optimization method according to claim 1, characterized in that: The using the second training sample to test the initial function simulation model corresponding to the calibration function to determine the target function simulation model corresponding to the calibration function includes: Processing the non-functional training data and the functional training data in the second training sample using the initial functional simulation model corresponding to the calibration function to determine the second brake build-up pressure output by the braking system simulation model; determining a test accuracy rate based on the second brake build-up pressure and the preset brake build-up pressure, wherein the test accuracy rate is determined based on a difference between the second brake build-up pressure and the preset brake build-up pressure corresponding to the second training sample; If the test accuracy reaches a preset accuracy, the initial function simulation model corresponding to the calibration function is determined as the target function simulation model of the calibration function.

7. A brake system calibration method, characterized in that: include: Acquiring data to be calibrated, wherein the data to be calibrated includes functional calibration data and non-functional calibration data corresponding to the calibration function; Inputting the data to be calibrated into the target function simulation model corresponding to the calibration function, and outputting the calibration brake build-up pressure corresponding to the calibration function; Wherein, the target function simulation model corresponding to the calibration function is determined based on the simulation model optimization method according to any one of claims 1 to 6.

8. A simulation model optimization device, characterized in that: include: A training sample set acquisition module is used to acquire a training sample set corresponding to a calibration function, wherein the training sample set includes a plurality of training samples, each of which includes non-functional training data, functional training data corresponding to the calibration function, and a preset brake build-up pressure; a first brake build-up pressure determination module, configured to process the non-functional training data and the functional training data in a first training sample using a brake system simulation model to determine a first brake build-up pressure output by the brake system simulation model, wherein the first training sample is a training sample used for model training; an initial function simulation model determining module, which updates model parameters of a brake system simulation model based on the preset brake build-up pressure and the first brake build-up pressure, and determines an initial function simulation model corresponding to the calibration function; The target function simulation model determination module is used to test the initial function simulation model corresponding to the calibration function using a second training sample to determine the target function simulation model corresponding to the calibration function, where the second training sample is a training sample other than the first training sample.

9. A brake system calibration device, characterized in that: include: A module for acquiring data to be calibrated, used for acquiring data to be calibrated, wherein the data to be calibrated includes functional calibration data and non-functional calibration data corresponding to the calibration function; a calibration module, configured to input the data to be calibrated into a target function simulation model corresponding to the calibration function, and output a calibration brake build-up pressure corresponding to the calibration function; Wherein, the target function simulation model corresponding to the calibration function is determined based on the simulation model optimization method according to any one of claims 1 to 6.

10. A computer device comprising a memory, a processor, and a computer program stored in the memory and executable on the processor, wherein: When the processor executes the computer program, the simulation model optimization method according to any one of claims 1 to 6 is implemented, or when the processor executes the computer program, the brake system calibration method according to claim 7 is implemented.

11. A computer-readable storage medium storing a computer program, characterized in that: When the computer program is executed by a processor, the simulation model optimization method according to any one of claims 1 to 6 is implemented, or when the computer program is executed by a processor, the brake system calibration method according to claim 7 is implemented.

Images