Road roughness identification method, model training method and related equipment

Abstract

The present application relates to the field of vehicle engineering technology, and discloses a road surface roughness identification method, a model training method, and related equipment. The model training method includes: generating a sample tire vertical acceleration according to the vertical dynamic relationship between the tire and the unpaved road surface; constructing a training sample using the sample vehicle speed and the sample tire vertical acceleration; inputting the training sample into a neural network model to obtain a second sample road surface roughness feature; determining the model loss information according to the first sample road surface roughness feature and the second sample road surface roughness feature; training the neural network model according to the model loss information to obtain a road surface roughness identification model. The embodiment of the present application can improve the recognition accuracy of the road surface roughness grade of unpaved roads.

Description

Technical Field

[0001] The present application relates to the field of vehicle engineering technology, and in particular to a road surface roughness recognition method, a model training method and related equipment. Background Art

[0002] Road roughness has a great impact on vehicle driving safety and riding comfort. Mastering road roughness information is of great significance for vehicle dynamics control, especially suspension control.

[0003] In related technologies, the realization of road roughness recognition requires the use of dedicated sensors to perceive and measure the road surface, or the use of a data set containing images of dry roads, wet roads, snowy roads, and muddy roads, and the use of a neural network model in deep learning for end-to-end identification.

[0004] However, unpaved roads are easy to deform, and the coupled deformation of tires and roads has a significant impact on vehicle stability analysis and dynamic control on unpaved roads, resulting in the model being unable to accurately simulate the tire center acceleration response on unpaved roads, reducing recognition accuracy. Summary of the invention

[0005] The purpose of this application is to provide a road surface roughness recognition method, a model training method and related equipment, aiming to improve the recognition accuracy of road surface roughness grades of unpaved roads.

[0006] The present application embodiment provides a model training method, including:

[0007] Generate a sample tire vertical acceleration according to the vertical dynamic relationship between the tire and the unpaved road surface; the sample tire vertical acceleration is the vertical acceleration of the tire when a vehicle with corresponding load and suspension parameters travels on a road surface with corresponding first sample road surface roughness characteristics at a corresponding sample vehicle speed;

[0008] Using the sample vehicle speed and the sample tire vertical acceleration, construct a training sample;

[0009] Inputting the training sample into a neural network model to obtain a second sample road surface roughness feature;

[0010] Determining model loss information according to the first sample road surface roughness feature and the second sample road surface roughness feature; the model loss information is a deviation between the first sample road surface roughness feature and the second sample road surface roughness feature;

[0011] The neural network model is trained according to the model loss information to obtain a road roughness recognition model.

[0012] In some embodiments, generating a sample tire vertical acceleration according to a vertical dynamic relationship between the tire and the unpaved road surface includes:

[0013] Acquiring road surface spectral density information of the road surface having the first sample road surface roughness characteristic;

[0014] Simulating the road elevation information of the tire according to the road spectrum density information;

[0015] coupling the radial deformation of the tire, the road subsidence and the road elevation information to obtain coupled road elevation information;

[0016] The coupled road elevation information is input into the tire vertical dynamics model, and the vertical acceleration of the unsprung mass is calculated to be used as the sample tire vertical acceleration; the tire vertical dynamics model is a dynamics model that describes the dynamic relationship between the vertical force of the tire and the coupled road elevation information.

[0017] In some embodiments, the calculation expression of the road elevation information is:

[0018] ,

[0019] The calculation expression of the coupled road elevation information is:

[0020] ,

[0021] The expression of the tire vertical dynamics model is:

[0022] ,

[0023] in, For The road elevation information at the moment, is the reference spatial frequency, is the lower cutoff frequency, is the road spectral density information at the reference spatial frequency, is the vehicle speed, is a uniformly distributed white noise with a mean of 0 and an intensity of 1. is the changing speed of the road elevation information at time t, , , and They are the coupled road elevation information of the left front tire, right front tire, left rear tire and right rear tire respectively. , , and are the radial deformations of the left front tire, right front tire, left rear tire, and right rear tire, respectively. , , and They are the road elevation information of the left front tire, right front tire, left rear tire, and right rear tire. , , and are the road subsidence of the left front tire, right front tire, left rear tire, and right rear tire, respectively. and are the unsprung masses of 1 / 4 of the front and rear suspensions, , , and are the vertical accelerations of the unsprung masses of the left front tire, right front tire, left rear tire, and right rear tire, respectively, , , and The vertical reaction forces from the road surface are respectively the left front tire, the right front tire, the left rear tire, and the right rear tire. , , and They are the suspension forces of the left front oil-gas suspension, the right front oil-gas suspension, the left rear oil-gas suspension and the right rear oil-gas suspension. is the tire stiffness, is tire damping, , , and are the vertical displacement velocities of the left front tire, right front tire, left rear tire, and right rear tire, respectively. , , and are the changing speeds of the coupled road elevation information of the left front tire, right front tire, left rear tire, and right rear tire respectively.

[0024] In some embodiments, the calculation formula of the road surface sinking amount of the tire is:

[0025] ,

[0026] ，

[0027] The calculation formula of the radial deformation of the tire is:

[0028] ,

[0029] ,

[0030] in, is the road subsidence of the tire, is the initial road subsidence, is the tire dynamic load, is the soil cohesion coefficient, is the tire width, is the soil friction coefficient, is the tire diameter, is the soil subsidence index, is the tire empirical parameter, is the tire pressure, is the tire stiffness, is the tire width after pressure deformation, is the radial deformation of the tire, is a constant related to the tire structure, is an empirical parameter.

[0031] In some embodiments, constructing a training sample using the sample vehicle speed and the sample tire vertical acceleration includes:

[0032] Determining, according to the vertical dynamics relationship, the sample tire vertical accelerations corresponding to the plurality of sample vehicle speeds respectively under the influence of the corresponding first sample road surface roughness characteristics;

[0033] Using the Hilbert transform method, signal transformation is performed on the sample vehicle speed and the sample tire vertical acceleration to obtain envelope signals corresponding to the sample vehicle speed and the sample tire vertical acceleration;

[0034] The envelope signals corresponding to the sample vehicle speed and the sample tire vertical acceleration are associated, and the associated envelope signals are integrated to obtain the training sample.

[0035] In some embodiments, inputting the training sample into a neural network model to obtain a second sample road surface roughness feature includes:

[0036] Performing convolution encoding on the training sample to obtain an encoded feature vector;

[0037] Performing long short-term memory time series processing on the encoded feature vector, extracting feature information related to road surface roughness based on the correlation between the sample vehicle speed and the sample tire vertical acceleration, and obtaining a vector processed by long short-term memory time series;

[0038] Performing full connection operation and convolution decoding on the vector processed by long short-term memory time series to obtain the second sample road roughness feature.

[0039] In some embodiments, the neural network model is trained according to the model loss information to obtain a road roughness recognition model, including:

[0040] Determining whether the model loss information is within a loss threshold range;

[0041] If not, adjusting the weight parameters of the neural network model; returning to the step of inputting the training sample into the neural network model to obtain the second sample road roughness feature;

[0042] If yes, determine whether the number of sample parameter resets reaches the reset threshold;

[0043] If not reached, adjusting the vehicle load, the vehicle suspension parameters, the sample vehicle speed and / or the sampling duration; returning to the step of generating the sample tire vertical acceleration according to the vertical dynamic relationship between the tire and the unpaved road surface;

[0044] If it is reached, the training is terminated and the road roughness recognition model is obtained.

[0045] The present application also provides a method for identifying road roughness, including:

[0046] Obtaining the actual vehicle speed and the actual tire vertical acceleration of the vehicle;

[0047] Inputting the measured vehicle speed and the measured tire vertical acceleration into a road surface roughness recognition model to obtain road surface roughness characteristics; the road surface roughness recognition model is trained by the above-mentioned model training method;

[0048] The road surface roughness of the road surface on which the vehicle is traveling is determined according to the road surface roughness feature.

[0049] An embodiment of the present application further provides an electronic device, which includes a memory and a processor, wherein the memory stores a computer program, and the processor implements the above method when executing the computer program.

[0050] An embodiment of the present application further provides a computer-readable storage medium, wherein the computer-readable storage medium stores a computer program, and the computer program implements the above method when executed by a processor.

[0051] The beneficial effects of the present application are as follows: based on the vertical dynamic relationship between the tire and the unpaved road, the tire vertical acceleration corresponding to the load and suspension parameters of various vehicles, various vehicle speeds and various sample road surface roughness characteristics is simulated, and the road surface roughness is predicted based on the sample vehicle speed and the simulated sample tire vertical acceleration using a neural network model. The road surface roughness characteristics output by the neural network model are supervised by preset road surface roughness characteristics to determine the loss information between the two road surface roughness characteristics, and then the neural network model is trained. By constructing the vertical dynamic relationship between the tire and the unpaved road, simulating the tire vertical acceleration corresponding to the load and suspension parameters of various vehicles, various vehicle speeds and various sample road surface roughness characteristics, and taking the sample vehicle speeds and the simulated sample tire vertical acceleration as the materials for training the neural network model, the tire type variables and the road surface settlement amount when the vehicle is traveling on the unpaved road are fully considered. At the same time, the neural network model is trained with the load and suspension parameters of various vehicles, various vehicle speeds and various sample road surface roughness characteristics as preset conditions, so that the trained road surface roughness recognition model can adapt to the road surface roughness prediction of unpaved roads, thereby improving the recognition accuracy of the road surface roughness grade of unpaved roads. BRIEF DESCRIPTION OF THE DRAWINGS

[0052] Figure 1 It is a flow chart of the model training method provided in the embodiment of the present application.

[0053] Figure 2 It is a flowchart of the specific method of step S101 provided in an embodiment of the present application.

[0054] Figure 3 It is a flowchart of the specific method of step S102 provided in an embodiment of the present application.

[0055] Figure 4 It is a flowchart of the specific method of step S103 provided in an embodiment of the present application.

[0056] Figure 5 It is a flowchart of the specific method of step S105 provided in an embodiment of the present application.

[0057] Figure 6 It is a flow chart of the road roughness identification method provided in an embodiment of the present application.

[0058] Figure 7 It is a schematic diagram of the hardware structure of the electronic device provided in the embodiment of the present application. DETAILED DESCRIPTION

[0059] In order to make the purpose, technical solution and advantages of the present application more clearly understood, the present application is further described in detail below in conjunction with the accompanying drawings and embodiments. It should be understood that the specific embodiments described herein are only used to explain the present application and are not used to limit the present application.

[0060] It should be noted that, although the functional modules are divided in the device schematic diagram and the logical order is shown in the flowchart, in some cases, the steps shown can be performed in a different order than the module division in the device or the order in the flowchart. The terms "first", "second" and the like in the specification, claims and drawings are used to distinguish similar objects, and are not used to describe a specific order or sequence.

[0061] Unless otherwise defined, all technical and scientific terms used herein have the same meaning as those commonly understood by those skilled in the art to which this application belongs. The terms used herein are only for the purpose of describing the embodiments of this application and are not intended to limit this application.

[0062] See also Figure 1 , Figure 1 is a flow chart of the model training method provided in the embodiments of the present application. In some embodiments of the present application, Figure 1 The method may specifically include but is not limited to steps S101 to S105. Figure 1 These five steps are introduced in detail.

[0063] Step S101, generating a sample tire vertical acceleration according to the vertical dynamic relationship between the tire and the unpaved road surface. The sample tire vertical acceleration is the vertical acceleration of the tire when a vehicle with corresponding load and suspension parameters travels at a corresponding sample vehicle speed on a road surface with a corresponding first sample road surface roughness characteristic.

[0064] In the embodiments of the present application, the unpaved road surface refers to a road surface that has not been paved, usually because it has not been paved or because the surface of the natural material is uneven.

[0065] In the embodiment of the present application, the road surface roughness is generally divided into 8 levels according to its unevenness, from level A (flattest) to level H (most rugged), and each level is determined by the specific value of its road surface roughness, reflecting the roughness of the road surface and the impact on vehicle driving, which helps to evaluate the driving conditions of different road surfaces, thereby affecting the design and driving strategy of the vehicle. For example, a high-level road surface may require a stronger suspension system and a more robust vehicle design to ensure driving comfort.

[0066] In step S101, a tire vertical dynamics model between the tire and the unpaved road surface may be constructed based on the consideration of the road surface settlement, tire deformation, and vehicle load and suspension parameters to describe the vertical dynamics relationship between the tire and the unpaved road surface. In the case of presetting the vehicle load and suspension parameters, a plurality of sample vehicle speeds are selected and input into the tire vertical dynamics model to obtain the sample tire vertical accelerations corresponding to the vehicle traveling on roads with various road surface roughness at the plurality of sample vehicle speeds. The construction of the tire vertical dynamics model between the tire and the unpaved road surface may be to express the action relationship between the tire and the unpaved road surface according to the Bekker-Wong model, obtain the tire deformation and road surface settlement through the simplified Bekker-Wong model, and after combining the vehicle load and suspension parameters, obtain the tire vertical dynamics model between the tire and the unpaved road surface through a series of dynamics relationship transformations to describe the vertical dynamics relationship between the tire and the unpaved road surface.

[0067] Step S102: construct training samples using sample vehicle speeds and sample tire vertical accelerations.

[0068] In step S102, a plurality of first sample road surface roughness characteristics and a plurality of sample vehicle speeds may be set with the vehicle load and suspension parameters pre-set, and the road surface roughness corresponding to the first sample road surface roughness characteristics may be respectively classified into different road surface roughness grades, and training samples may be constructed by associating sample vehicle speeds and sample tire vertical accelerations generated according to the vertical dynamic relationship between the tire and the unpaved road surface, that is, associating sample vehicle speeds and sample tire vertical accelerations when the vehicle is traveling on a road surface having the corresponding first sample road surface roughness characteristics.

[0069] Step S103: input the training sample into the neural network model to obtain the second sample road roughness feature.

[0070] In step S103, the sample vehicle speed and the sample tire vertical acceleration in the training sample are input into the neural network model in pairs to perform road roughness prediction to obtain a second sample road roughness feature.

[0071] In an embodiment of the present application, the neural network model is a pre-trained model, which can be directly applied to downstream tasks, or used as an initial model for downstream tasks to obtain better performance through fine-tuning. The neural network model can be used to perform parameter identification tasks of vehicle dynamics.

[0072] In the embodiment of the present application, the neural network model is used to predict the road surface roughness characteristics of the road surface on which the vehicle is traveling based on the sample vehicle speed and the sample tire vertical acceleration. That is, the neural network model is trained to learn the correlation information between the sample vehicle speed and the sample tire vertical acceleration.

[0073] In the embodiment of the present application, the second sample road surface roughness feature is used to indicate the road surface roughness of the road surface on which the vehicle is traveling. Further, the second sample road surface roughness feature includes road surface roughness sub-features obtained by dynamic fitting calculation of both the sample vehicle speed and the sample tire vertical acceleration.

[0074] Step S104: Determine model loss information based on the first sample road surface roughness feature and the second sample road surface roughness feature. The model loss information is the deviation between the first sample road surface roughness feature and the second sample road surface roughness feature.

[0075] In the embodiment of the present application, the model loss information represents the difference between the road surface roughness characteristics predicted by the neural network model and the actually set road surface roughness characteristics.

[0076] Step S105: training the neural network model according to the model loss information to obtain a road roughness recognition model.

[0077] In the embodiment of the present application, the sample vehicle speed, the first sample road surface roughness characteristics, and the vehicle load and suspension parameters are all multiple.

[0078] In step S105, when the load and suspension parameters of a vehicle are determined, multiple sample tire vertical accelerations are generated based on multiple sample vehicle speeds and multiple first sample road roughness characteristics to iteratively train the neural network model until the model loss information meets the training end conditions under the load and suspension parameters of each vehicle. After the training, a road roughness recognition model is obtained.

[0079] The model training method provided in steps S101 to S105 of the embodiment of the present application simulates the tire vertical acceleration corresponding to the load and suspension parameters of various vehicles, various vehicle speeds and various sample road surface roughness characteristics according to the vertical dynamic relationship between the tire and the unpaved road surface, and uses the neural network model to predict the road surface roughness based on the sample vehicle speed and the simulated sample tire vertical acceleration. The road surface roughness characteristics output by the neural network model are supervised by preset road surface roughness characteristics to determine the loss information between the two road surface roughness characteristics, and then the neural network model is trained. By constructing the vertical dynamic relationship between the tire and the unpaved road, simulating the tire vertical acceleration corresponding to the load and suspension parameters of various vehicles, various vehicle speeds and various sample road surface roughness characteristics, and taking the sample vehicle speeds and the simulated sample tire vertical acceleration as the materials for training the neural network model, the tire type variables and the road surface settlement amount when the vehicle is traveling on the unpaved road are fully considered. At the same time, the neural network model is trained with the load and suspension parameters of various vehicles, various vehicle speeds and various sample road surface roughness characteristics as preset conditions, so that the trained road surface roughness recognition model can adapt to the road surface roughness prediction of unpaved roads, thereby improving the recognition accuracy of the road surface roughness grade of unpaved roads.

[0080] In a specific embodiment, see Figure 2 , Figure 2 is a flowchart of a specific method of step S101 provided in an embodiment of the present application. Step S101 may specifically include but is not limited to steps S201 to S204. Figure 2 These four steps are introduced in detail.

[0081] Step S201: obtaining road surface spectral density information of a road surface having a first sample road surface roughness characteristic.

[0082] Step S202: Simulate the road elevation information of the tire according to the road spectrum density information.

[0083] Step S203, coupling the radial deformation of the tire, the road surface settlement and the road surface elevation information to obtain coupled road surface elevation information.

[0084] Step S204: Input the coupled road elevation information into the tire vertical dynamics model, calculate the vertical acceleration of the unsprung mass, and use it as the sample tire vertical acceleration. The tire vertical dynamics model is a dynamics model that describes the dynamic relationship between the vertical force of the tire and the coupled road elevation information.

[0085] In step S201, the geometric mean of the road surface spectral density corresponding to each road surface roughness level is obtained as the road surface spectral density information of the road surface having the first sample road surface roughness characteristic. In the embodiment of the present application, the road surface spectral density information is an important indicator for measuring road surface roughness, which mainly uses the displacement power spectral density to describe the statistical characteristics of road surface roughness, and treats the time history of road surface roughness as a stationary random process for processing.

[0086] In step S202, the elevation characteristics are described by the road surface power spectrum density, and the road surface elevation information of the tire under different road surface roughness levels is simulated by grading the road surface spectrum density. The road surface roughness level is described by the road surface spectrum density recommended by the International Organization for Standardization document ISO 8608:2016.

[0087] In step S203, based on the tire road elevation information obtained by simulation, the tire radial deformation and road subsidence are introduced, and coupled road elevation information is obtained by coupling the tire radial deformation, road subsidence and road elevation information.

[0088] In step S204, a tire vertical dynamics model between the tire and the unpaved road is constructed based on the coupled road elevation information, taking into account the road subsidence, tire deformation, and vehicle load and suspension parameters. The tire vertical dynamics model describes the vertical dynamics relationship between the tire and the unpaved road, and obtains a complete dynamics output, which includes the vertical acceleration of the unsprung mass. In this embodiment, the vertical acceleration of the unsprung mass is used as the sample tire vertical acceleration.

[0089] In a specific embodiment, the calculation expression of the road elevation information is:

[0090] ,

[0091] The calculation expression of coupled road elevation information is:

[0092] ,

[0093] The expression of the tire vertical dynamics model is:

[0094] ,

[0095] in, For The road elevation information at the moment, is the reference spatial frequency, is the lower cutoff frequency, is the road spectral density information at the reference spatial frequency, is the vehicle speed, is a uniformly distributed white noise with a mean of 0 and an intensity of 1. is the changing speed of the road elevation information at time t, , , and They are the coupled road elevation information of the left front tire, right front tire, left rear tire and right rear tire respectively. , , and are the radial deformations of the left front tire, right front tire, left rear tire, and right rear tire, respectively. , , and They are the road elevation information of the left front tire, right front tire, left rear tire, and right rear tire. , , and are the road subsidence of the left front tire, right front tire, left rear tire, and right rear tire, respectively. and are the unsprung masses of 1 / 4 of the front and rear suspensions, , , and are the vertical accelerations of the unsprung masses of the left front tire, right front tire, left rear tire, and right rear tire, respectively, , , and The vertical reaction forces from the road surface are respectively the left front tire, the right front tire, the left rear tire, and the right rear tire. , , and They are the suspension forces of the left front oil-gas suspension, the right front oil-gas suspension, the left rear oil-gas suspension and the right rear oil-gas suspension. is the tire stiffness, is tire damping, , , and are the vertical displacement velocities of the left front tire, right front tire, left rear tire, and right rear tire, respectively. , , and are the changing speeds of the coupled road elevation information of the left front tire, right front tire, left rear tire, and right rear tire respectively.

[0096] In a specific embodiment, the calculation formula of the road surface sinking amount of the tire is:

[0097] ,

[0098] ，

[0099] The calculation formula of the radial deformation of the tire is:

[0100] ,

[0101] ,

[0102] in, is the road subsidence of the tire, is the initial road subsidence, is the tire dynamic load, is the soil cohesion coefficient, is the tire width, is the soil friction coefficient, is the tire diameter, is the soil subsidence index, is the tire empirical parameter, is the tire pressure, is the tire stiffness, is the tire width after pressure deformation, is the radial deformation of the tire, is a constant related to the tire structure, is an empirical parameter.

[0103] In a specific embodiment, see Figure 3 , Figure 3 is a flowchart of a specific method of step S102 provided in an embodiment of the present application. Step S102 may specifically include but is not limited to steps S301 to S303. Figure 3 These three steps are introduced in detail.

[0104] Step S301, determining sample tire vertical accelerations corresponding to a plurality of sample vehicle speeds respectively under the influence of a corresponding first sample road surface roughness feature according to a vertical dynamics relationship.

[0105] Step S302, using the Hilbert transform method to perform signal transformation on the sample vehicle speed and the sample tire vertical acceleration to obtain envelope signals corresponding to the sample vehicle speed and the sample tire vertical acceleration;

[0106] Step S303 , correlating the envelope signals corresponding to the sample vehicle speed and the sample tire vertical acceleration, integrating the correlated envelope signals, and obtaining the training sample.

[0107] In step S301, when the vehicle load and suspension parameters are pre-set, multiple first sample road surface roughness characteristics and multiple sample vehicle speeds are set, and the sample vehicle speeds and the sample tire vertical accelerations generated according to the vertical dynamic relationship between the tire and the unpaved road surface are associated, that is, the sample vehicle speeds and the sample tire vertical accelerations when the vehicle is traveling on the road surface with the corresponding first sample road surface roughness characteristics are associated, and the sample tire vertical accelerations corresponding to the multiple sample vehicle speeds under the influence of the corresponding first sample road surface roughness characteristics are obtained. For example, four sample vehicle speeds are set, and the four sample vehicle speeds are 10km / h, 20km / h, 30km / h and 40km / h, respectively. When the vehicle load and suspension parameters are pre-set, the sample tire vertical accelerations of the four sample vehicle speeds under 8 sample road surface roughness levels are simulated respectively.

[0108] In this embodiment, the vehicle load and suspension parameters may be as shown in Table 1 below.

[0109] Table 1

[0110]

[0111] In step S302, the sample vehicle speed and the sample tire vertical acceleration are first converted from time domain signals to frequency domain signals by discrete Fourier transform, and then the positive and negative frequency components are multiplied by imaginary units of opposite signs to obtain analytical signals. Finally, the obtained analytical signals are converted into Hilbert transform signals by inverse discrete Fourier transform and then modulo-transformed to obtain the required upper envelope signal of the Hilbert transform, thereby obtaining the envelope signals corresponding to the sample vehicle speed and the sample tire vertical acceleration.

[0112] In step S303, based on the correlation between the sample vehicle speed and the sample tire vertical acceleration, the envelope signals corresponding to the sample vehicle speed and the sample tire vertical acceleration are correlated, that is, the sample vehicle speed and the sample tire vertical acceleration when the vehicle is traveling on a road surface having the corresponding first sample road surface roughness characteristics are correlated, thereby constructing a training sample.

[0113] In a specific embodiment, see Figure 4 , Figure 4 is a flowchart of a specific method of step S103 provided in an embodiment of the present application. Step S103 may specifically include but is not limited to steps S401 to S403. Figure 4 These three steps are introduced in detail.

[0114] Step S401, performing convolution encoding on the training sample to obtain an encoded feature vector.

[0115] Step S402, performing long short-term memory time series processing on the encoded feature vector, extracting feature information related to road surface roughness based on the correlation between the sample vehicle speed and the sample tire vertical acceleration, and obtaining a vector processed by long short-term memory time series.

[0116] Step S403, performing full connection operation and convolution decoding on the vector processed by the long short-term memory time series to obtain a second sample road roughness feature.

[0117] In this embodiment, the neural network model includes an input layer, an LSTM layer, a fully connected layer and an output layer.

[0118] In step S401, the sample vehicle speed and sample tire vertical acceleration in the training sample are input into the encoding layer, and the input sample vehicle speed and sample tire vertical acceleration are converted into a vector form representation that can be processed by the LSTM layer through a convolution operation to obtain a coded feature vector.

[0119] In step S402, the encoded feature vector is input into the LSTM layer, and the feature information related to the road roughness in the encoded feature vector is reorganized and filtered through the input gate and output gate of the long short-term memory gate, and unnecessary information is discarded through the forget gate to obtain a vector processed by the long short-term memory time series.

[0120] In step S403, the vector processed by the long short-term memory time series is input into the fully connected layer, and the vector processed by the long short-term memory time series is reduced in dimension through the fully connected operation to obtain the vector processed by the fully connected operation, and the vector processed by the fully connected operation is input into the output layer, and the vector input and output layer processed by the fully connected operation is converted into a corresponding second sample road surface roughness feature containing road surface roughness information through a convolution operation, and the road surface roughness level predicted according to the input sample vehicle speed and the sample tire vertical acceleration is determined through the second sample road surface roughness feature.

[0121] In a specific embodiment, see Figure 5 , Figure 5 is a flowchart of a specific method of step S105 provided in an embodiment of the present application. Step S105 may specifically include but is not limited to steps S501 to S505. Figure 5 These five steps are introduced in detail.

[0122] Step S501, determine whether the model loss information is within the loss threshold range. If not, execute step S502; if so, execute step S503.

[0123] Step S502, adjust the weight parameters of the neural network model. Return to step S103.

[0124] Step S503, determine whether the number of sample parameter reset times reaches the reset threshold. If not, execute step S504; if reached, execute step S505.

[0125] Step S504, adjusting the vehicle load, the vehicle suspension parameters, the sample vehicle speed and / or the sampling time. Return to step S101.

[0126] Step S505, ending the training and obtaining a road roughness recognition model.

[0127] In step S501 to step S505, a loss threshold interval and a reset threshold are preset as training end conditions. When the model loss information is within the loss threshold interval and the number of sample parameter resets reaches the reset threshold, the training is terminated, and the neural network model of the last iteration is obtained, which is the road surface roughness recognition model. When the number of sample parameter resets does not reach the reset threshold, the weight parameters of the neural network model are adjusted according to the degree of deviation of the model loss information from the loss threshold interval, so that the model loss information gradually approaches the loss threshold interval during the iteration process and finally falls within the loss threshold interval, and then the vehicle load, vehicle suspension parameters, sample vehicle speed and / or sampling duration are adjusted to change the sample tire vertical acceleration. Even if the training sample changes, the neural network model is iteratively trained using the changed training sample until the model loss information is within the loss threshold interval. The above steps are repeated until the number of sample parameter resets reaches the reset threshold, and the training is terminated to obtain a road surface roughness recognition model, which can be adapted to a variety of vehicle loads, vehicle suspension parameters, sample vehicle speeds and / or sampling durations and accurately predict road surface roughness.

[0128] See also Figure 6 , Figure 6 is a flow chart of a road surface roughness identification method provided in an embodiment of the present application. In some embodiments of the present application, Figure 6 The method may specifically include but is not limited to steps S601 to S603. Figure 6 These three steps are introduced in detail.

[0129] Step S601, obtaining the actual vehicle speed and the actual tire vertical acceleration of the vehicle;

[0130] Step S602: Input the measured vehicle speed and the measured tire vertical acceleration into a road surface roughness recognition model to obtain road surface roughness characteristics.

[0131] The road roughness recognition model is obtained by training using the model training method described in the above embodiment.

[0132] Step S603, determining the road surface roughness of the road on which the vehicle is traveling according to the road surface roughness characteristics.

[0133] The road surface roughness identification method provided in steps S601 to S603 of the embodiment of the present application measures the vehicle speed and tire vertical acceleration during driving of the vehicle by corresponding measuring instruments to obtain the actual vehicle speed and the actual tire vertical acceleration, and uses a pre-trained road surface roughness identification model to predict the road surface roughness based on the actual vehicle speed and the actual tire vertical acceleration to obtain a road surface roughness feature, and determines the road surface roughness of the road surface on which the vehicle is driving according to the grade interval of the road surface roughness parameter corresponding to the road surface roughness feature, thereby achieving accurate prediction of the road surface roughness.

[0134] Figure 7 It is a block diagram of an electronic device according to an exemplary embodiment.

[0135] Refer to the following Figure 7 An electronic device 700 according to this embodiment of the present disclosure is described. Figure 7 The electronic device 700 shown is merely an example and should not bring any limitation to the functions and scope of use of the embodiments of the present disclosure.

[0136] like Figure 7 As shown, the electronic device 700 is in the form of a general computing device. The components of the electronic device 700 may include, but are not limited to: at least one processing unit 710, at least one storage unit 720, a bus 730 connecting different system components (including the storage unit 720 and the processing unit 710), a display unit 740, etc.

[0137] The storage unit stores program codes, and the program codes can be executed by the processing unit 710, so that the processing unit 710 executes the steps described in the above method part of this specification according to various exemplary embodiments of the present disclosure.

[0138] The storage unit 720 may include a readable medium in the form of a volatile storage unit, such as a random access storage unit (RAM) 7201 and / or a cache storage unit 7202 , and may further include a read-only storage unit (ROM) 7203 .

[0139] The storage unit 720 may also include a program / utility 7204 having a set (at least one) of program modules 7205, such program modules 7205 including but not limited to: an operating system, one or more application programs, other program modules, and program data, each of which or some combination may include an implementation of a network environment.

[0140] Bus 730 may represent one or more of several types of bus structures, including a memory unit bus or memory unit controller, a peripheral bus, an accelerated graphics port, a processing unit, or a local bus using any of a variety of bus architectures.

[0141] The electronic device 700 may also communicate with one or more external devices 700' (e.g., keyboards, pointing devices, Bluetooth devices, etc.), may also communicate with one or more devices that enable a user to interact with the electronic device 700, and / or communicate with any device that enables the electronic device 700 to communicate with one or more other computing devices (e.g., routers, modems, etc.). Such communication may be performed via an input / output (I / O) interface 750. Furthermore, the electronic device 700 may also communicate with one or more networks (e.g., a local area network (LAN), a wide area network (WAN), and / or a public network, such as the Internet) via a network adapter 760. The network adapter 760 may communicate with other modules of the electronic device 700 via a bus 730. It should be understood that, although not shown in the figure, other hardware and / or software modules may be used in conjunction with the electronic device 700, including but not limited to: microcode, device drivers, redundant processing units, external disk drive arrays, RAID systems, tape drives, and data backup storage systems.

[0142] An embodiment of the present application further provides a computer-readable storage medium, which stores a computer program, and the computer program implements the above method when executed by a processor.

[0143] The road surface roughness identification method, model training method and related equipment provided in the embodiments of the present application simulate the tire vertical acceleration corresponding to the load and suspension parameters of various vehicles, various vehicle speeds and various sample road surface roughness characteristics according to the vertical dynamic relationship between the tire and the unpaved road surface, and use the neural network model to predict the road surface roughness based on the sample vehicle speed and the simulated sample tire vertical acceleration. The road surface roughness characteristics output by the neural network model are supervised by preset road surface roughness characteristics to determine the loss information between two road surface roughness characteristics, and then the neural network model is trained. By constructing the vertical dynamic relationship between the tire and the unpaved road, simulating the tire vertical acceleration corresponding to the load and suspension parameters of various vehicles, various vehicle speeds and various sample road surface roughness characteristics, and taking the sample vehicle speeds and the simulated sample tire vertical acceleration as the materials for training the neural network model, the tire type variables and the road surface settlement amount when the vehicle is traveling on the unpaved road are fully considered. At the same time, the neural network model is trained with the load and suspension parameters of various vehicles, various vehicle speeds and various sample road surface roughness characteristics as preset conditions, so that the trained road surface roughness recognition model can adapt to the road surface roughness prediction of unpaved roads, thereby improving the recognition accuracy of the road surface roughness grade of unpaved roads.

[0144] Through the description of the above implementation, it is easy for those skilled in the art to understand that the example implementation described here can be implemented by software, or by software combined with necessary hardware. Therefore, the technical solution according to the implementation of the present disclosure can be embodied in the form of a software product, which can be stored in a non-volatile storage medium (which can be a CD-ROM, a USB flash drive, a mobile hard disk, etc.) or on a network, including several instructions to enable a computing device (which can be a personal computer, a server, or a network device, etc.) to execute the above method according to the implementation of the present disclosure.

[0145] The program product may use any combination of one or more readable media. The readable medium may be a readable signal medium or a readable storage medium. The readable storage medium may be, for example, but not limited to, an electrical, magnetic, optical, electromagnetic, infrared, or semiconductor system, device or device, or any combination of the above. More specific examples of readable storage media (a non-exhaustive list) include: an electrical connection with one or more wires, a portable disk, a hard disk, a random access memory (RAM), a read-only memory (ROM), an erasable programmable read-only memory (EPROM or flash memory), an optical fiber, a portable compact disk read-only memory (CD-ROM), an optical storage device, a magnetic storage device, or any suitable combination of the above.

[0146] Computer readable storage media may include data signals propagated in baseband or as part of a carrier wave, wherein readable program codes are carried. Such propagated data signals may take various forms, including but not limited to electromagnetic signals, optical signals, or any suitable combination thereof. The readable storage medium may also be any readable medium other than a readable storage medium, which may send, propagate, or transmit a program for use by or in conjunction with an instruction execution system, device, or device. The program codes contained on the readable storage medium may be transmitted using any suitable medium, including but not limited to wireless, wired, optical cable, RF, etc., or any suitable combination thereof.

[0147] Program code for performing the operations of the present disclosure may be written in any combination of one or more programming languages, including object-oriented programming languages ​​such as Java, C++, etc., and conventional procedural programming languages ​​such as "C" or similar programming languages. The program code may be executed entirely on the user computing device, partially on the user device, as a separate software package, partially on the user computing device and partially on a remote computing device, or entirely on a remote computing device or server. In cases involving a remote computing device, the remote computing device may be connected to the user computing device through any type of network, including a local area network (LAN) or a wide area network (WAN), or may be connected to an external computing device (e.g., through the Internet using an Internet service provider).

[0148] Those skilled in the art will appreciate that the above modules can be distributed in the device according to the description of the embodiment, or can be changed accordingly and only used in one or more devices different from the embodiment. The modules of the above embodiments can be combined into one module, or further divided into multiple sub-modules.

[0149] Through the description of the above embodiments, it is easy for those skilled in the art to understand that the example embodiments described here can be implemented by software, or by software combined with necessary hardware. Therefore, the technical solution according to the embodiment of the present disclosure can be embodied in the form of a software product, which can be stored in a non-volatile storage medium (which can be a CD-ROM, a USB flash drive, a mobile hard disk, etc.) or on a network, and includes several instructions to enable a computing device (which can be a personal computer, a server, a mobile terminal, or a network device, etc.) to execute the method according to the embodiment of the present disclosure.

[0150] The exemplary embodiments of the present disclosure are specifically shown and described above. It should be understood that the present disclosure is not limited to the detailed structures, configurations or implementations described herein; on the contrary, the present disclosure is intended to cover various modifications and equivalent configurations included in the spirit and scope of the appended claims.

Claims

1. A model training method, characterized in that: include: Generate sample tire vertical acceleration according to the vertical dynamic relationship between the tire and the unpaved road surface; The sample tire vertical acceleration is the vertical acceleration of the tire when a vehicle with corresponding load and suspension parameters travels on a road with corresponding first sample road surface roughness characteristics at a corresponding sample vehicle speed; constructing a training sample using the sample vehicle speed and the sample tire vertical acceleration; Inputting the training sample into a neural network model to obtain a second sample road surface roughness feature; Determining model loss information according to the first sample road surface roughness feature and the second sample road surface roughness feature; the model loss information is a deviation between the first sample road surface roughness feature and the second sample road surface roughness feature; Training the neural network model according to the model loss information to obtain a road roughness recognition model; The generating of the sample tire vertical acceleration according to the vertical dynamic relationship between the tire and the unpaved road surface comprises: Acquiring road surface spectral density information of a road surface having a first sample road surface roughness characteristic; Simulating the road elevation information of the tire according to the road spectrum density information; coupling the radial deformation of the tire, the road subsidence and the road elevation information to obtain coupled road elevation information; The coupled road elevation information is input into a tire vertical dynamics model to calculate the vertical acceleration of the unsprung mass as the sample tire vertical acceleration; the tire vertical dynamics model is a dynamics model that describes the dynamic relationship between the vertical force of the tire and the coupled road elevation information; The calculation expression of the road elevation information is: ， The calculation expression of the coupled road elevation information is: ， The expression of the tire vertical dynamics model is: ， in, For The road elevation information at the moment, is the reference spatial frequency, is the lower cutoff frequency, is the road spectral density information at the reference spatial frequency, is the vehicle speed, is a uniformly distributed white noise with a mean of 0 and an intensity of 1. is the changing speed of the road elevation information at time t, , , and They are the coupled road elevation information of the left front tire, right front tire, left rear tire and right rear tire respectively. , , and are the radial deformations of the left front tire, right front tire, left rear tire, and right rear tire, respectively. , , and They are the road elevation information of the left front tire, right front tire, left rear tire, and right rear tire. , , and are the road subsidence of the left front tire, right front tire, left rear tire, and right rear tire, respectively. and are the unsprung masses of 1 / 4 of the front and rear suspensions, , , and are the vertical accelerations of the unsprung masses of the left front tire, right front tire, left rear tire, and right rear tire, respectively, , , and The vertical reaction forces from the road surface are respectively the left front tire, the right front tire, the left rear tire, and the right rear tire. , , and They are the suspension forces of the left front oil-gas suspension, the right front oil-gas suspension, the left rear oil-gas suspension and the right rear oil-gas suspension. is the tire stiffness, is tire damping, , , and are the vertical displacement velocities of the left front tire, right front tire, left rear tire, and right rear tire, respectively. , , and are the changing speeds of the coupled road elevation information of the left front tire, right front tire, left rear tire, and right rear tire respectively.

2. The model training method according to claim 1, characterized in that: The calculation formula of the road subsidence of the tire is: ， ， The calculation formula of the radial deformation of the tire is: ， ， in, is the road subsidence of the tire, is the initial road subsidence, is the tire dynamic load, is the soil cohesion coefficient, is the tire width, is the soil friction coefficient, is the tire diameter, is the soil subsidence index, is the tire empirical parameter, is the tire pressure, is the tire stiffness, is the tire width after pressure deformation, is the radial deformation of the tire, is a constant related to the tire structure, is an empirical parameter.

3. The model training method according to claim 1, characterized in that: The constructing a training sample by using the sample vehicle speed and the sample tire vertical acceleration includes: Determining, according to the vertical dynamics relationship, the sample tire vertical accelerations corresponding to the plurality of sample vehicle speeds respectively under the influence of the corresponding first sample road surface roughness characteristics; Using the Hilbert transform method, signal transformation is performed on the sample vehicle speed and the sample tire vertical acceleration to obtain envelope signals corresponding to the sample vehicle speed and the sample tire vertical acceleration; The envelope signals corresponding to the sample vehicle speed and the sample tire vertical acceleration are associated, and the associated envelope signals are integrated to obtain the training sample.

4. The model training method according to claim 1, characterized in that: The step of inputting the training sample into a neural network model to obtain a second sample road surface roughness feature comprises: Performing convolution encoding on the training sample to obtain an encoded feature vector; Performing long short-term memory time series processing on the encoded feature vector, extracting feature information related to road surface roughness based on the correlation between the sample vehicle speed and the sample tire vertical acceleration, and obtaining a vector processed by long short-term memory time series; Performing full connection operation and convolution decoding on the vector processed by long short-term memory time series to obtain the second sample road roughness feature.

5. The model training method according to claim 1, characterized in that: The neural network model is trained according to the model loss information to obtain a road roughness recognition model, including: Determining whether the model loss information is within a loss threshold range; If not, adjusting the weight parameters of the neural network model; returning to the step of inputting the training sample into the neural network model to obtain the second sample road roughness feature; If yes, determine whether the number of sample parameter resets reaches the reset threshold; If not reached, adjusting the vehicle load, the vehicle suspension parameters, the sample vehicle speed and / or the sampling duration; returning to the step of generating the sample tire vertical acceleration according to the vertical dynamic relationship between the tire and the unpaved road surface; If it is reached, the training is terminated and the road roughness recognition model is obtained.

6. A method for identifying road roughness, characterized in that: include: Obtaining the actual vehicle speed and the actual tire vertical acceleration of the vehicle; Inputting the measured vehicle speed and the measured tire vertical acceleration into a road surface roughness recognition model to obtain road surface roughness characteristics; the road surface roughness recognition model is trained by the model training method according to any one of claims 1 to 5; The road surface roughness of the road surface on which the vehicle is traveling is determined according to the road surface roughness feature.

7. An electronic device, characterized in that: The electronic device comprises a memory and a processor, the memory stores a computer program, and the processor implements the method according to any one of claims 1 to 6 when executing the computer program.

8. A computer-readable storage medium storing a computer program, characterized in that: When the computer program is executed by a processor, the method according to any one of claims 1 to 6 is implemented.

Images