Deep learning based high smoothness automobile crane suspension design method and device

Abstract

The application discloses a high-ride comfort automobile crane suspension design method and device based on deep learning, and relates to the technical field of crane suspension design. The method comprises the following steps: acquiring a first training data set for each road spectrum; training a pre-constructed road vibration curve prediction model by using the first training data set corresponding to different road spectra, so as to obtain a trained road vibration curve prediction model corresponding to different road spectra; solving optimal suspension structure parameters corresponding to each road spectrum by using the trained road vibration curve prediction model, taking minimization of ride comfort indexes as an optimization target; constructing a second training data set according to the optimal suspension structure parameters corresponding to each road spectrum; training a pre-constructed optimal suspension structure parameter prediction model by using the second training data set; and performing suspension structure design by using the trained optimal suspension structure parameter prediction model. The application can realize rapid optimization design of a crane suspension structure with high ride comfort.

Description

TECHNICAL FIELD

[0001] The present application relates to the technical field of crane suspension design, and particularly relates to a high-ride comfort automobile crane suspension design method and device based on deep learning. BACKGROUND

[0002] In the process of operation and driving, the performance of the suspension system of the automobile crane directly affects the ride comfort, stability and maneuverability of the vehicle. The traditional crane suspension design method is based on a large number of tests and experience data for optimization design, and has the problems of long design cycle and poor adaptability. Especially when the vehicle operating conditions involve complex vibration spectrum, the traditional crane suspension design method is difficult to efficiently determine the optimal design parameters. At the same time, with the diversification of vehicle load and road conditions, the demand for fine and intelligent suspension design is increasing.

[0003] Therefore, there is an urgent need for an automobile crane suspension design method which can quickly respond to spectrum input and optimize design parameters. SUMMARY

[0004] To solve the above-mentioned technical problems in the prior art, the present application provides a high-ride comfort automobile crane suspension design method and device based on deep learning.

[0005] The technical scheme of the present application is as follows:

[0006] In a first aspect, a high-ride comfort automobile crane suspension design method based on deep learning is provided, comprising:

[0007] For each road spectrum, a first training data set is obtained, the first training data set comprising a plurality of first training data, the first training data comprising suspension structure parameters and corresponding cab acceleration vibration curves;

[0008] The first training data set corresponding to each pre-processed road spectrum is used to train a pre-constructed road vibration curve prediction model, to obtain a trained road vibration curve prediction model corresponding to each road spectrum, the input of the road vibration curve prediction model being the suspension structure parameters, and the output of the road vibration curve prediction model being the cab acceleration vibration curve;

[0009] The trained road vibration curve prediction model corresponding to each road spectrum is used to minimize the ride comfort index as the optimization objective, to solve the optimal suspension structure parameters corresponding to each road spectrum, the ride comfort being defined as the smoothness of the cab acceleration vibration curve, and the ride comfort index being defined as the root mean square value of the acceleration value of the cab acceleration vibration curve;

[0010] According to the optimal suspension structure parameters corresponding to each road spectrum, a second training data set is constructed, the second training data set comprising a plurality of second training data, the second training data comprising road spectrum characteristic parameters and corresponding optimal suspension structure parameters;

[0011] The pre-constructed optimal suspension structure parameter prediction model is trained by using the pre-processed second training data set, the input of the optimal suspension structure parameter prediction model being road spectrum characteristic parameters, and the output of the optimal suspension structure parameter prediction model being suspension structure parameters;

[0012] The trained optimal suspension structure parameter prediction model is used for suspension structure design.

[0013] In some optional embodiments, the first training data set is obtained by:

[0014] For a road spectrum, a plurality of sets of suspension structure parameters are set;

[0015] According to the plurality of sets of suspension structure parameters set, a suspension model corresponding to each set of suspension structure parameters is constructed;

[0016] Under the current road spectrum, the plurality of constructed suspension models are respectively subjected to dynamic simulation or test, the corresponding cab acceleration vibration curves of each suspension model are collected, a plurality of first training data comprising suspension structure parameters and corresponding cab acceleration vibration curves are obtained, and a first training data set corresponding to the current road spectrum is obtained.

[0017] In some optional embodiments, the suspension structure parameters comprise a plurality of suspension stiffness, sprung mass, unsprung mass, steel plate spring type, steel plate spring material, steel plate spring piece number, steel plate spring thickness, steel plate spring prestress, steel plate spring clamping distance, steel plate spring length, steel plate spring full load, damper damping force, damper stroke, and axle track.

[0018] In some optional embodiments, the pre-processing of the first training data set comprises normalizing the suspension structure parameters in the first training data;

[0019] The pre-processing of the second training data set comprises normalizing the road spectrum characteristic parameters in the second training data.

[0020] In some optional embodiments, the road vibration curve prediction model adopts a multi-layer feedforward neural network, the activation function of the multi-layer feedforward neural network adopts a Leaky ReLU activation function, and a Dropout layer is arranged between the hidden layers of the multi-layer feedforward neural network.

[0021] In some alternative implementations, the root mean square value of the acceleration values ​​in the cab acceleration vibration curve is expressed as:

[0022]

[0023] Where RMS represents the root mean square value of the acceleration value of the cab acceleration vibration curve, T represents the length of the time series, t represents the time variable, and y(t) represents the cab acceleration value at time t.

[0024] In some alternative implementations, the optimal suspension structure parameters corresponding to the road surface spectrum are solved in the following manner:

[0025] Step 301: Set initial values ​​for suspension structure parameters for a given road surface spectrum;

[0026] Step 302: Input the suspension structure parameters into the trained road vibration curve prediction model corresponding to the current road surface spectrum to obtain the cab acceleration vibration curve output by the road vibration curve prediction model.

[0027] Step 303: Calculate the ride comfort index based on the cab acceleration vibration curve output by the road vibration curve prediction model, and determine whether the ride comfort index has reached the minimum value. If so, take the current suspension structure parameters as the optimal suspension structure parameters corresponding to the current road surface spectrum. If not, take the suspension structure parameters as the optimization variable, iteratively update the suspension structure parameters using the gradient descent method, and return to step 302.

[0028] In some optional implementations, the road surface spectrum characteristic parameters include: road surface spectrum and vibration intensity.

[0029] In some optional implementations, the optimal suspension structure parameter prediction model employs a multilayer perceptron, with the Leaky ReLU activation function used for the multilayer perceptron.

[0030] Secondly, a high-smoothness truck crane suspension design device based on deep learning is also provided, comprising:

[0031] The first training dataset acquisition unit is used to acquire the first training dataset corresponding to each road surface spectrum. The first training dataset includes multiple first training data, and the first training data includes suspension structure parameters and their corresponding cab acceleration vibration curves.

[0032] The road surface vibration curve prediction model generation unit is configured to train a pre-constructed road surface vibration curve prediction model by using the first training data set corresponding to the pre-processed different road surface spectra, to obtain a trained road surface vibration curve prediction model corresponding to the different road surface spectra, wherein the input of the road surface vibration curve prediction model is the suspension structure parameters, and the output of the road surface vibration curve prediction model is the cab acceleration vibration curve;

[0033] The optimal suspension structure parameter solving unit is configured to solve the optimal suspension structure parameters corresponding to the different road surface spectra by using the trained road surface vibration curve prediction model corresponding to each road surface spectrum, with minimization of the ride comfort index as the optimization target, wherein the ride comfort is defined as the smoothness of the cab acceleration vibration curve, and the ride comfort index is defined as the root mean square value of the acceleration value of the cab acceleration vibration curve.

[0034] The second training data set acquisition unit is configured to construct a second training data set according to the optimal suspension structure parameters corresponding to the different road surface spectra, wherein the second training data set comprises a plurality of second training data, and the second training data comprises the road surface spectrum characteristic parameters and the corresponding optimal suspension structure parameters.

[0035] The optimal suspension structure parameter prediction model generation unit is configured to train a pre-constructed optimal suspension structure parameter prediction model by using the pre-processed second training data set, wherein the input of the optimal suspension structure parameter prediction model is the road surface spectrum characteristic parameters, and the output of the optimal suspension structure parameter prediction model is the suspension structure parameters.

[0036] The suspension structure design unit is configured to perform suspension structure design by using the trained optimal suspension structure parameter prediction model.

[0037] The main advantages of the technical scheme of the present application are as follows:

[0038] The high-ride-comfort automobile crane suspension design method and device based on deep learning of the present application can determine the optimal suspension structure parameters corresponding to different road surface spectra by constructing and training a road surface vibration curve prediction model to fit the mapping relationship between the suspension structure parameters and the cab acceleration vibration curve, and train an optimal suspension structure parameter prediction model by using different road surface spectrum characteristic parameters and the corresponding optimal suspension structure parameters, so as to realize rapid prediction of the optimal suspension structure parameters under any road surface spectrum condition by using the optimal suspension structure parameter prediction model, ensure the accuracy and reliability of the predicted optimal suspension structure parameters, and realize rapid optimization design of the high-ride-comfort automobile crane suspension structure. BRIEF DESCRIPTION OF DRAWINGS

[0039] The accompanying drawings, which are included to provide a further understanding of the embodiments of the application and are incorporated in and constitute a part of this application, illustrate embodiments of the application and together with the description serve to explain the application. In the drawings:

[0040] Figure 1 A flowchart of a high-ride comfort automobile crane suspension design method based on deep learning provided by an embodiment of the application;

[0041] Figure 2 A structural schematic diagram of a high-ride comfort automobile crane suspension design device based on deep learning provided by an embodiment of the application. DETAILED DESCRIPTION

[0042] To make the objectives, technical solutions and advantages of the application clearer, the technical solutions of the application will be described below in detail with reference to the embodiments of the application and the corresponding drawings. Obviously, the described embodiments are only some of the embodiments of the application, but not all the embodiments of the application. Based on the embodiments of the application, all other embodiments obtained by those skilled in the art without creative work fall within the scope of protection of the application.

[0043] The technical solutions provided by the embodiments of the application will be described in detail below with reference to the drawings.

[0044] Reference Figure 1 In a first aspect, the embodiments of the application provide a high-ride comfort automobile crane suspension design method based on deep learning, which comprises the following steps:

[0045] Step 1: For each road spectrum, a first training data set is obtained respectively.

[0046] In the embodiments of the application, each road spectrum has a corresponding first training data set, and the first training data set comprises a plurality of first training data. The first training data comprises a suspension structure parameter and a corresponding cab acceleration vibration curve.

[0047] In the embodiments of the application, the types of road spectrum are set according to actual requirements.

[0048] Step 2: The pre-constructed road vibration curve prediction model is trained by using the first training data set corresponding to the different road spectrum after preprocessing, to obtain a trained road vibration curve prediction model corresponding to the different road spectrum.

[0049] In the embodiments of the application, the road vibration curve prediction model is constructed based on a neural network.

[0050] In the embodiment of the present application, the input of the road surface vibration curve prediction model is the suspension structure parameter, and the output of the road surface vibration curve prediction model is the cab acceleration vibration curve.

[0051] In the embodiment of the present application, each road surface spectrum corresponds to a trained road surface vibration curve prediction model.

[0052] Step 3, using the trained road surface vibration curve prediction model corresponding to each road surface spectrum, the optimal suspension structure parameter corresponding to each road surface spectrum is solved by taking the minimization of the ride comfort index as the optimization objective.

[0053] In the embodiment of the present application, the ride comfort is defined as the smoothness of the cab acceleration vibration curve, and the ride comfort index is defined as the root mean square value of the acceleration value of the cab acceleration vibration curve.

[0054] The smaller the ride comfort index is, the smoother the cab acceleration vibration curve is, and the higher the ride comfort is.

[0055] Step 4, constructing a second training data set according to the optimal suspension structure parameter corresponding to each road surface spectrum.

[0056] In the embodiment of the present application, the second training data set includes a plurality of second training data, and the second training data includes road surface spectrum characteristic parameters and the corresponding optimal suspension structure parameters.

[0057] In the embodiment of the present application, the second training data set is constructed according to the optimal suspension structure parameter corresponding to each road surface spectrum obtained in step 3. The optimal suspension structure parameter corresponding to one road surface spectrum can determine one second training data.

[0058] Step 5, using the preprocessed second training data set to train the pre-constructed optimal suspension structure parameter prediction model.

[0059] In the embodiment of the present application, the optimal suspension structure parameter prediction model is constructed based on a neural network.

[0060] In the embodiment of the present application, the input of the optimal suspension structure parameter prediction model is the road surface spectrum characteristic parameter, and the output of the optimal suspension structure parameter prediction model is the suspension structure parameter.

[0061] Step 6, using the trained optimal suspension structure parameter prediction model to design the suspension structure.

[0062] In the embodiment of the present application, the specified road spectrum characteristic parameter is input into the trained optimal suspension structure parameter prediction model to obtain the suspension structure parameter corresponding to the specified road spectrum characteristic parameter output by the optimal suspension structure parameter prediction model, and the suspension structure parameter output by the optimal suspension structure parameter prediction model is the optimal suspension structure parameter corresponding to the specified road spectrum characteristic parameter, and the obtained optimal suspension structure parameter is used for suspension structure design.

[0063] The deep learning-based high-ride comfort automobile crane suspension design method provided in the embodiment of the present application can quickly predict the optimal suspension structure parameter under the condition of any road spectrum by constructing and training a road vibration curve prediction model to fit the mapping relationship between the suspension structure parameter and the cab acceleration vibration curve, determining the optimal suspension structure parameter corresponding to different road spectra by using the road vibration curve prediction model, training an optimal suspension structure parameter prediction model by using different road spectrum characteristic parameters and the optimal suspension structure parameters corresponding thereto, and ensuring the accuracy and reliability of the predicted optimal suspension structure parameter, thereby realizing the rapid optimization design of the high-ride comfort automobile crane suspension structure.

[0064] Further, in the embodiment of the present application, the first training data set is obtained by the following method:

[0065] For a kind of road spectrum, multiple groups of suspension structure parameters are set;

[0066] According to the set multiple groups of suspension structure parameters, a suspension model corresponding to each group of suspension structure parameters is constructed;

[0067] Under the current road spectrum, the constructed multiple suspension models are respectively subjected to dynamic simulation or test, the cab acceleration vibration curve corresponding to each suspension model is collected, the first training data including the suspension structure parameter and the corresponding cab acceleration vibration curve is obtained, and the first training data set corresponding to the current road spectrum is obtained.

[0068] In the embodiment of the present application, the first training data set corresponding to each kind of road spectrum can be quickly obtained by the above-mentioned method, and the accuracy of the obtained first training data set is ensured.

[0069] In the embodiment of the present application, the type and value of the suspension structure parameter are specifically set according to the actual situation.

[0070] Specifically, for the leaf spring suspension structure, the suspension structure parameter includes multiple of the stiffness of the suspension, the mass on the spring, the mass under the spring, the type of the leaf spring, the material of the leaf spring, the number of the leaf spring, the thickness of the leaf spring, the pre-stress of the leaf spring, the clamping distance of the leaf spring, the length of the leaf spring, the full load of the leaf spring, the damping force of the shock absorber, the stroke of the shock absorber and the wheelbase.

[0071] Further, in the embodiment of the present application, in step 2, the preprocessing includes: performing normalization processing on the suspension structure parameters in the first training data.

[0072] In the embodiment of the present application, by performing normalization processing on the suspension structure parameters, the dimensional differences of different parameters can be eliminated, the parameter scales are unified, it is ensured that each parameter in the suspension structure parameters input into the model has a consistent data range, the influences of each parameter are balanced, and the model generalization capability is improved.

[0073] Further, in the embodiment of the present application, the road surface vibration curve prediction model adopts a multi-layer feedforward neural network (FNN), the activation function of the multi-layer feedforward neural network adopts a Leaky ReLU activation function, and a Dropout layer is arranged between the hidden layers of the multi-layer feedforward neural network.

[0074] In the embodiment of the present application, by constructing the road surface vibration curve prediction model by using the above structure, the complex nonlinear relationship between the suspension structure parameters and the acceleration vibration response can be fully captured by the road surface vibration curve prediction model; by using the Leaky ReLU activation function, the training effect of the model can be improved; and by arranging the Dropout layer, the model can be prevented from excessively relying on the training data, overfitting can be prevented, and the model generalization capability can be improved.

[0075] In the embodiment of the present application, the multi-layer feedforward neural network includes an input layer, an output layer, and a plurality of hidden layers arranged between the input layer and the output layer, and the output of each hidden layer of the multi-layer feedforward neural network can be expressed as:

[0076] h l =f(w l h l-1 +b l );

[0077] wherein h l represents the output of the lth layer of the multi-layer feedforward neural network, h l-1 represents the output of the (l-1)th layer of the multi-layer feedforward neural network, w l represents the weight matrix of the lth layer of the multi-layer feedforward neural network, b l represents the bias coefficient of the lth layer of the multi-layer feedforward neural network, and f represents the activation function of the multi-layer feedforward neural network.

[0078] wherein the activation function adopts a Leaky ReLU activation function.

[0079] Further, setting the dropout rate of the Dropout layer as p, the output of each layer after the Dropout layer processing can be represented as:

[0080]

[0081] wherein, represents the output of the lth layer after the Dropout layer processing.

[0082] In the embodiment of the present application, the Dropout layer enhances the generalization ability of the network by randomly discarding part of the neurons.

[0083] Further, in the embodiment of the present application, in step 2, the road vibration curve prediction model is trained by the following way:

[0084] The preprocessed suspension structure parameters are taken as the input of the road vibration curve prediction model, the cab acceleration vibration curve corresponding to the suspension structure parameters is taken as the output of the road vibration curve prediction model, and the road vibration curve prediction model is trained.

[0085] In the embodiment of the present application, the preprocessed suspension structure parameters are taken as the input of the road vibration curve prediction model, the cab acceleration vibration curve corresponding to the suspension structure parameters is taken as the output of the road vibration curve prediction model, and the road vibration curve prediction model is trained, further comprising the following steps:

[0086] Step 201, input the preprocessed suspension structure parameters into the road vibration curve prediction model to obtain the cab acceleration vibration curve prediction result output by the road vibration curve prediction model;

[0087] Step 202, according to the input suspension structure parameters corresponding to the cab acceleration vibration curve and the input suspension structure parameters corresponding to the cab acceleration vibration curve prediction result output by the road vibration curve prediction model, a first loss function is calculated;

[0088] Step 203, it is judged whether the preset training stop condition is reached, if yes, the current road vibration curve prediction model is taken as the road vibration curve prediction model completed training, if not, the parameters of the road vibration curve prediction model are updated by using the first loss function, and the step 201 is returned.

[0089] In the embodiment of the present application, before the training starts, the parameters of each layer of the road vibration curve prediction model are initialization parameters, and in the training process, the parameters of each layer of the road vibration curve prediction model are constantly updated.

[0090] In the embodiment of the present application, the first loss function adopts the mean square error (MSE) loss function, which is specifically represented as:

[0091]

[0092] wherein, L1 represents a first loss function value, T represents a time series length, t represents a time variable, represents a cab acceleration prediction value at t time, y(t) represents a cab acceleration value at t time.

[0093] wherein, the cab acceleration prediction value is directly obtained from the cab acceleration vibration curve, and the cab acceleration value y(t) is directly obtained from the cab acceleration vibration curve.

[0094] In the embodiment of the present application, the training stop condition is specifically set according to actual conditions, for example, the training iteration number reaches a set number or the optimization index reaches a set threshold, wherein the optimization index can adopt the first loss function value described above.

[0095] In the embodiment of the present application, the gradient descent method is adopted to perform iterative updating of the parameters of the road vibration curve prediction model.

[0096] Further, in the embodiment of the present application, in step 3, the root mean square value of the acceleration value of the cab acceleration vibration curve is represented as:

[0097]

[0098] wherein, RMS represents the root mean square value of the acceleration value of the cab acceleration vibration curve, T represents a time series length, t represents a time variable, y(t) represents a cab acceleration value at t time, and the cab acceleration value y(t) is directly obtained from the cab acceleration vibration curve.

[0099] Further, in the embodiment of the present application, in step 3, the optimal suspension structure parameter corresponding to the road spectrum is solved by the following way:

[0100] Step 301, for a road spectrum, setting an initial value of a suspension structure parameter;

[0101] Step 302, inputting the suspension structure parameter into the trained road vibration curve prediction model corresponding to the current road spectrum to obtain a cab acceleration vibration curve output by the road vibration curve prediction model;

[0102] Step 303, calculating a ride comfort index according to the cab acceleration vibration curve output by the road vibration curve prediction model, judging whether the ride comfort index reaches a minimum value, if yes, taking the current suspension structure parameter as the optimal suspension structure parameter corresponding to the current road spectrum, if no, taking the suspension structure parameter as an optimization variable, iteratively updating the suspension structure parameter by the gradient descent method, and returning to step 302.

[0103] In the embodiment of the present application, for each road spectrum, the optimal suspension structure parameters corresponding to each road spectrum are determined by solving the optimal suspension structure parameters in the above manner respectively.

[0104] In the embodiment of the present application, the iterative update of the suspension structure parameters by the gradient descent method can be expressed as:

[0105]

[0106] wherein X t+1 represents the suspension structure parameters at the t+1th iteration, X t represents the suspension structure parameters at the tth iteration, and η represents the learning rate. represents the gradient of the ride comfort index with respect to the suspension structure parameters, and the learning rate is set according to actual requirements. The gradient of the ride comfort index with respect to the suspension structure parameters represents the rate of change of the ride comfort index with respect to the suspension structure parameters under the current suspension structure parameters.

[0107] In the embodiment of the present application, the optimal suspension structure parameters corresponding to each road spectrum are solved in the above manner, which can realize the fast solving of the optimal suspension structure parameters corresponding to each road spectrum and ensure the accuracy of the optimal suspension structure parameters obtained by solving.

[0108] Further, in the embodiment of the present application, the road spectrum characteristic parameters are set according to actual conditions.

[0109] Specifically, in the embodiment of the present application, the road spectrum characteristic parameters include: road spectrum, vibration intensity.

[0110] Further, in the embodiment of the present application, in step 5, the pretreatment includes: performing normalization processing on the road spectrum characteristic parameters in the second training data.

[0111] In the embodiment of the present application, by performing normalization processing on the road spectrum characteristic parameters, the dimensional differences of different parameters can be eliminated, the parameter scales are unified, it is ensured that each parameter in the road spectrum characteristic parameters input into the model has a consistent data range, the influences of each parameter are balanced, and the model generalization capability is improved.

[0112] Further, in the embodiment of the present application, the optimal suspension structure parameter prediction model adopts a multilayer perceptron (MLP), and the activation function of the multilayer perceptron adopts a Leaky ReLU activation function.

[0113] In the embodiment of the present application, the optimal suspension structure parameter prediction model is constructed by using the above structure, so that the complex relationship between the road spectrum characteristic parameters and the optimal suspension structure parameters can be fully captured by the optimal suspension structure parameter prediction model, and the training effect of the model can be improved by using the Leaky ReLU activation function.

[0114] In the embodiment of the present application, the multi-layer perception machine includes an input layer, an output layer, and a plurality of hidden layers arranged between the input layer and the output layer.

[0115] Specifically, assuming that the multi-layer perception machine includes L hidden layers, the output of the L hidden layers of the multi-layer perception machine can be represented as:

[0116]

[0117] The output of the output layer of the multi-layer perception machine can be represented as:

[0118] X * =W out H L +B out ;

[0119] wherein H1 represents the output of the first hidden layer of the multi-layer perception machine, H2 represents the output of the second hidden layer of the multi-layer perception machine, H L-1 represents the output of the L-1th hidden layer of the multi-layer perception machine, H L represents the output of the Lth hidden layer of the multi-layer perception machine, S represents a vector containing road spectrum characteristic parameter information received by the input layer, W1 and B1 represent the weight matrix and the bias coefficient of the first hidden layer of the multi-layer perception machine respectively, W2 and B2 represent the weight matrix and the bias coefficient of the second hidden layer of the multi-layer perception machine respectively, W L and B L represent the weight matrix and the bias coefficient of the Lth hidden layer of the multi-layer perception machine respectively, and sigma represents the activation function of the multi-layer perception machine. * represents the output of the output layer of the multi-layer perception machine, and W out and B out represent the weight matrix and the bias coefficient of the output layer of the multi-layer perception machine respectively.

[0120] wherein the activation function is a Leaky ReLU activation function.

[0121] In the embodiment of the present application, the output X * of the output layer of the multi-layer perception machine is the predicted optimal suspension structure parameter.

[0122] Further, in the embodiment of the present application, in step 5, the optimal suspension structure parameter prediction model is trained in the following manner:

[0123] The preprocessed road spectrum characteristic parameters are taken as inputs of the optimal suspension structure parameter prediction model, and the optimal suspension structure parameters corresponding to the road spectrum characteristic parameters are taken as outputs of the optimal suspension structure parameter prediction model, and the optimal suspension structure parameter prediction model is trained.

[0124] In the embodiment of the present application, the preprocessed road spectrum characteristic parameters are taken as inputs of the optimal suspension structure parameter prediction model, and the optimal suspension structure parameters corresponding to the road spectrum characteristic parameters are taken as outputs of the optimal suspension structure parameter prediction model, and the optimal suspension structure parameter prediction model is trained, and the method further comprises the following steps:

[0125] In step 501, the road spectrum characteristic parameters in the preprocessed plurality of second training data are sequentially input into the optimal suspension structure parameter prediction model, and the optimal suspension structure parameter prediction results corresponding to each road spectrum characteristic parameter output by the optimal suspension structure parameter prediction model are obtained.

[0126] In step 502, the optimal suspension structure parameters corresponding to the input road spectrum characteristic parameters and the optimal suspension structure parameter prediction results corresponding to the input road spectrum characteristic parameters output by the optimal suspension structure parameter prediction model are used to calculate a second loss function.

[0127] In step 503, it is determined whether a preset training stop condition is reached, if yes, the current optimal suspension structure parameter prediction model is taken as the completed optimal suspension structure parameter prediction model, and if not, the parameters of the optimal suspension structure parameter prediction model are updated using the second loss function, and the step 501 is returned.

[0128] In the embodiment of the present application, before the training starts, the parameters of each layer of the optimal suspension structure parameter prediction model are initialization parameters, and in the training process, the parameters of each layer of the optimal suspension structure parameter prediction model are constantly updated.

[0129] In the embodiment of the present application, the second loss function adopts a mean square error (MSE) loss function, and is specifically represented as:

[0130]

[0131] Wherein, L2 represents the second loss function value, N represents the number of second training data used for training, X true,i represents the optimal suspension structure parameter in the i-th training data, represents the optimal suspension structure parameter prediction result output by the optimal suspension structure parameter prediction model corresponding to the road spectrum characteristic parameters in the i-th training data.

[0132] In the embodiment of the present application, the training stop condition is specifically set according to actual conditions, for example, the training iteration number reaches a set number or the optimization index reaches a set threshold, wherein the optimization index can adopt the second loss function value described above.

[0133] In the embodiment of the present application, the gradient descent method is adopted to perform iterative updating of the parameters of the optimal suspension structure parameter prediction model.

[0134] Specifically, the parameters of the optimal suspension structure parameter prediction model are updated by the following formula:

[0135]

[0136] wherein θ t+1 represents the parameters of the optimal suspension structure parameter prediction model at the t+1th iteration, θ t represents the parameters of the optimal suspension structure parameter prediction model at the tth iteration, η2 represents the learning rate, and θ represents the parameters of the optimal suspension structure parameter prediction model. The learning rate is set in advance and is used to control the speed of model parameter updating, and the parameters of the optimal suspension structure parameter prediction model include a weight matrix and a bias coefficient.

[0137] Reference Figure 2 , in a second aspect, the embodiment of the present application provides a high-ride comfort automobile crane suspension design device based on deep learning, which comprises:

[0138] A first training data set acquisition unit is configured to acquire a first training data set corresponding to each road spectrum, wherein the first training data set comprises a plurality of first training data, and each first training data comprises a suspension structure parameter and a corresponding cab acceleration vibration curve.

[0139] A road vibration curve prediction model generation unit is configured to train a pre-constructed road vibration curve prediction model by using the first training data set corresponding to different road spectra after preprocessing, to obtain a trained road vibration curve prediction model corresponding to different road spectra, wherein the input of the road vibration curve prediction model is a suspension structure parameter, and the output of the road vibration curve prediction model is a cab acceleration vibration curve.

[0140] An optimal suspension structure parameter solving unit is configured to solve an optimal suspension structure parameter corresponding to each road spectrum by using the trained road vibration curve prediction model corresponding to each road spectrum, and taking minimization of a ride comfort index as an optimization target, wherein the ride comfort is defined as the smoothness of the cab acceleration vibration curve, and the ride comfort index is defined as the root mean square value of the acceleration value of the cab acceleration vibration curve.

[0141] The second training data set acquisition unit is configured to construct a second training data set according to the optimal suspension structure parameters corresponding to each road spectrum, and the second training data set includes a plurality of second training data, and each second training data includes road spectrum characteristic parameters and corresponding optimal suspension structure parameters.

[0142] The optimal suspension structure parameter prediction model generation unit is configured to train a pre-constructed optimal suspension structure parameter prediction model by using the preprocessed second training data set, and the input of the optimal suspension structure parameter prediction model is road spectrum characteristic parameters, and the output of the optimal suspension structure parameter prediction model is suspension structure parameters.

[0143] The suspension structure design unit is configured to perform suspension structure design by using the trained optimal suspension structure parameter prediction model.

[0144] The device for designing a high-ride comfort automobile crane suspension based on deep learning provided by the embodiment of the application includes units corresponding to the above method steps, and can realize all processes of the method for designing a high-ride comfort automobile crane suspension based on deep learning described in any of the above embodiments, and the specific working principles, roles and technical effects of each unit are the same as those of the method for designing a high-ride comfort automobile crane suspension based on deep learning described in the above embodiments, which will not be described here.

[0145] It should be noted that, in this document, relational terms such as“first” and“second”, and the like, are used solely to distinguish one entity or action from another entity or action, without necessarily requiring or implying any actual such relationship or order between such entities or actions. Moreover, the terms“comprises”,“comprising”, or any other variations thereof, are intended to cover a non-exclusive inclusion, such that a process, method, article, or apparatus that comprises a list of elements does not include only those elements but can include other elements not expressly listed or inherent to such process, method, article, or apparatus. Furthermore, the terms“front”,“rear”,“left”,“right”,“up”, and“down” are used in this document for the purpose of description and refer to the positions shown in the drawings.

[0146] Finally, it should be noted that: the above embodiments are only used to illustrate the technical solutions of the application, and not to limit them; although the application has been described in detail with reference to the foregoing embodiments, those skilled in the art should understand that: it can still modify the technical solutions recorded in the foregoing embodiments, or make equivalent replacement for part of the technical features; and these modifications or replacements do not make the essence of the corresponding technical solutions deviate from the spirit and scope of the technical solutions of the embodiments of the application.

Claims

1. A deep learning based high-ride comfort automobile crane suspension design method, characterized in that, The method comprises the following steps: For each road spectrum, a first training data set is obtained, which comprises a plurality of first training data, and each first training data comprises a suspension structure parameter and a corresponding cab acceleration vibration curve; A pre-constructed road vibration curve prediction model is trained using the first training data set corresponding to each road spectrum, to obtain a trained road vibration curve prediction model corresponding to each road spectrum, wherein the input of the road vibration curve prediction model is a suspension structure parameter, and the output of the road vibration curve prediction model is a cab acceleration vibration curve; The trained road vibration curve prediction model corresponding to each road spectrum is used to solve an optimal suspension structure parameter corresponding to each road spectrum, with minimization of a ride comfort index as an optimization objective, wherein the ride comfort is defined as the smoothness of the cab acceleration vibration curve, and the ride comfort index is defined as the root mean square value of the acceleration value of the cab acceleration vibration curve; A second training data set is constructed according to the optimal suspension structure parameter corresponding to each road spectrum, which comprises a plurality of second training data, and each second training data comprises a road spectrum feature parameter and a corresponding optimal suspension structure parameter; A pre-constructed optimal suspension structure parameter prediction model is trained using the pre-processed second training data set, wherein the input of the optimal suspension structure parameter prediction model is a road spectrum feature parameter, and the output of the optimal suspension structure parameter prediction model is a suspension structure parameter; The trained optimal suspension structure parameter prediction model is used for suspension structure design.

2. The deep learning based high-ride comfort automotive crane suspension design method of claim 1, wherein, The first training data set is obtained by the following method: For a road spectrum, a plurality of groups of suspension structure parameters are set; A suspension model corresponding to each group of suspension structure parameters is constructed according to the set plurality of groups of suspension structure parameters; Dynamic simulation or test is respectively performed on the constructed plurality of suspension models under the current road spectrum, the cab acceleration vibration curve corresponding to each suspension model is collected, a plurality of first training data comprising a suspension structure parameter and a corresponding cab acceleration vibration curve are obtained, and a first training data set corresponding to the current road spectrum is obtained.

3. The deep learning based high-ride comfort automotive crane suspension design method of claim 1, wherein, The suspension structure parameter comprises a plurality of suspension stiffness, sprung mass, unsprung mass, type of leaf spring, material of leaf spring, number of leaf spring, thickness of leaf spring, pre-stress of leaf spring, clamping distance of leaf spring, length of leaf spring, full load of leaf spring, damping force of shock absorber, stroke of shock absorber and wheelbase.

4. The deep learning based high-ride comfort automotive crane suspension design method of claim 1, wherein, The pre-processing of the first training data set comprises normalization processing of the suspension structure parameters in the first training data; The pre-processing of the second training data set comprises normalization processing of the road spectrum feature parameters in the second training data.

5. The deep learning based high-ride comfort automotive crane suspension design method of claim 1, wherein, The road vibration curve prediction model adopts a multi-layer feedforward neural network, the activation function of the multi-layer feedforward neural network adopts a Leaky ReLU activation function, and a Dropout layer is arranged between the hidden layers of the multi-layer feedforward neural network.

6. The deep learning based high-ride comfort automotive crane suspension design method of claim 1, wherein, The root mean square value of the acceleration value of the cab acceleration vibration curve is represented as: Wherein, RMS represents the root mean square value of the acceleration value of the cab acceleration vibration curve, T represents the time series length, t represents the time variable, y(t) represents the cab acceleration value at t time.

7. The deep learning based high-ride comfort automotive crane suspension design method of claim 1, wherein, The optimal suspension structure parameter corresponding to the road surface spectrum is solved in the following manner: Step 301, for a kind of road surface spectrum, set the initial value of suspension structure parameter; Step 302, input the suspension structure parameter into the trained road surface vibration curve prediction model corresponding to the current road surface spectrum, to obtain the cab acceleration vibration curve output by the road surface vibration curve prediction model; Step 303, calculate the ride comfort index according to the cab acceleration vibration curve output by the road surface vibration curve prediction model, judge whether the ride comfort index reaches the minimum value, if yes, the current suspension structure parameter is taken as the optimal suspension structure parameter corresponding to the current road surface spectrum, if not, the suspension structure parameter is taken as the optimization variable, and the suspension structure parameter is updated iteratively by gradient descent method, and return to step 302.

8. The deep learning based high-ride comfort automotive crane suspension design method of claim 1, wherein, The road surface spectrum characteristic parameters include: road surface spectrum, vibration intensity.

9. The deep learning based high-ride comfort automotive crane suspension design method of claim 1, wherein, The optimal suspension structure parameter prediction model adopts multilayer perception machine, and the activation function of the multilayer perception machine adopts LeakyReLU activation function.

10. A deep learning based high-ride comfort automobile crane suspension design apparatus, characterized by, It includes: The first training data set acquisition unit is used for acquiring the first training data set corresponding to each road surface spectrum, and the first training data set includes a plurality of first training data, and the first training data includes suspension structure parameter and its corresponding cab acceleration vibration curve; The road surface vibration curve prediction model generation unit is used for training the pre-constructed road surface vibration curve prediction model by using the preprocessed first training data set corresponding to different road surface spectra, to obtain the trained road surface vibration curve prediction model corresponding to different road surface spectra, the input of the road surface vibration curve prediction model is the suspension structure parameter, and the output of the road surface vibration curve prediction model is the cab acceleration vibration curve; The optimal suspension structure parameter solving unit is used for solving the optimal suspension structure parameter corresponding to each road surface spectrum by using the trained road surface vibration curve prediction model corresponding to each road surface spectrum, taking minimizing the ride comfort index as the optimization target, the ride comfort is defined as the smooth degree of the cab acceleration vibration curve, and the ride comfort index is defined as the root mean square value of the acceleration value of the cab acceleration vibration curve; The second training data set acquisition unit is used for constructing the second training data set according to the optimal suspension structure parameter corresponding to each road surface spectrum, and the second training data set includes a plurality of second training data, and the second training data includes road surface spectrum characteristic parameter and its corresponding optimal suspension structure parameter; The optimal suspension structure parameter prediction model generation unit is used for training the pre-constructed optimal suspension structure parameter prediction model by using the preprocessed second training data set, the input of the optimal suspension structure parameter prediction model is the road surface spectrum characteristic parameter, and the output of the optimal suspension structure parameter prediction model is the suspension structure parameter; The suspension structure design unit is used for carrying out suspension structure design by using the trained optimal suspension structure parameter prediction model.

Images