DT model construction method and trajectory tracking control method of mine intelligent chassis

Abstract

The invention belongs to the field of digital twinning of driverless cars, and provides a DT model construction method and a trajectory tracking control method of a mine intelligent chassis in order to solve the problems that a mechanism model of a current mine intelligent chassis is limited in practical application and poor in path tracking capability and stability. Obtaining a whole vehicle test data set of the mine intelligent chassis under variable working conditions; a digital twinning mechanism model of the mine intelligent chassis is established by considering the longitudinal force effect of tire slippage on tires; based on the whole vehicle test data set and the digital twinning mechanism model, establishing a digital twinning data driving model of the mine intelligent chassis; and the digital twinning data driving model serves as an error compensator and is connected with the digital twinning mechanism model in parallel, and the mine intelligent chassis DT model based on data-mechanism hybrid driving is obtained. According to the method, the simulation precision of the model can be remarkably improved, and the mechanism-data hybrid driven DT model is used as the trajectory prediction model, so that the trajectory tracking precision and effect are greatly improved.

Description

Technical Field

[0001] The present invention belongs to the field of digital twins of driverless vehicles, and specifically relates to a DT model construction method and a trajectory tracking control method of a mine intelligent chassis. Background Art

[0002] At present, China is vigorously promoting the intelligent development of coal mines, and the mine transportation system is gradually moving towards unmanned operation. For underground coal mines, in addition to coal, the main means of transportation for other materials and operators is the mine auxiliary transportation system, which is crucial in the entire coal production system. Its technical level and operating efficiency directly affect the goal of reducing personnel and improving efficiency in coal mine production. At present, traditional mine auxiliary transportation mainly relies on driver driving and manual scheduling, with low automation and backward information level, which is prone to safety accidents. At the same time, these systems usually rely on mechanical and hydraulic transmission, with low transmission efficiency and poor reliability. In response to these problems, a mine intelligent chassis based on unmanned driving technology is proposed to realize unmanned transportation of underground materials.

[0003] At present, the trajectory tracking control of the intelligent chassis in mines is achieved by building a vehicle dynamics model to realize unmanned driving trajectory tracking control. Most related technologies also build a vehicle mechanism model around this method and perform simulation calculations to simulate the vehicle's driving conditions.

[0004] However, although the vehicle dynamics model derived from mathematical formulas can show the physical characteristics of the vehicle, such as its geometric features and spatial position, this mechanism model has great limitations. On the one hand, since it is difficult to obtain the detailed parameters of the whole vehicle, the whole vehicle modeling is often simplified; on the other hand, the influence of various working conditions and other factors on the whole vehicle results is difficult to be represented by the mechanism model, resulting in the mechanism model being unable to accurately calculate the real dynamic response of the vehicle during driving, that is, there is an error between the predicted results and the actual results, which reduces the simulation accuracy of the model. In addition, the trajectory tracking control method based on the model is usually affected by model parameter perturbations, uncertainty interference, time lag, and actuator saturation constraints, which makes some mathematical assumptions no longer valid, causing the vehicle to lose path tracking ability and stability. Therefore, considering the real-time and fidelity of the model calculation at the same time, establishing a digital twin (DT) model of the mine intelligent chassis, and building a trajectory prediction model based on the mine intelligent chassis DT model to realize the trajectory tracking control of the mine intelligent chassis has become an important issue that needs to be solved urgently. Summary of the invention

[0005] In order to solve at least one of the above technical problems existing in the prior art, the present invention provides a DT model construction method and a trajectory tracking control method for a mine intelligent chassis.

[0006] According to a first aspect, the present invention provides a mine intelligent chassis DT model, comprising the following steps: S1: Obtain the vehicle test data set under variable working conditions of the mine intelligent chassis; S2: Considering the longitudinal force of tire slip on the tire, a digital twin mechanism model of the mine intelligent chassis is established; S3: Based on the vehicle test data set and digital twin mechanism model, a digital twin data-driven model of the mine intelligent chassis is established; S4: The digital twin data-driven model is connected in parallel with the digital twin mechanism model as an error compensator to obtain the mine intelligent chassis DT model based on data-mechanism hybrid drive.

[0007] Preferably, the whole vehicle test data set acquisition specifically includes: Acquire vehicle test data of the mine intelligent chassis under different road conditions, where the different road conditions include at least a dry asphalt road and a slippery muddy road; Data processing is performed based on the whole vehicle experimental data, and actual mine intelligent chassis dynamics data is extracted according to the processing results. The actual mine intelligent chassis dynamics data includes the mine intelligent chassis state quantity and control quantity, and a whole vehicle test data set is constructed.

[0008] Preferably, the construction process of the digital twin mechanism model includes the following steps: S201: Establish vehicle dynamics model; S202: Add center of mass sideslip angle constraints, vehicle adhesion condition constraints and tire sideslip angle constraints to the vehicle dynamics model; S203: Establish dynamic equations, add vehicle parameters of the mine intelligent chassis, and obtain the digital twin mechanism model.

[0009] Preferably, the construction process of the digital twin data-driven model includes the following steps: S301: Construct a convolutional recurrent neural network combined with SE attention mechanism; S302: Obtaining a residual value of the digital twin mechanism model based on a comparison of the simulation results of the vehicle test data set and the digital twin mechanism model; S303: The vehicle test data set is used as the input of the convolutional recurrent neural network combined with the SE attention mechanism, and the residual value of the digital twin mechanism model is used as the output of the convolutional recurrent neural network combined with the SE attention mechanism to train and obtain the digital twin data-driven model.

[0010] Preferably, the residual value of the digital twin mechanism model is the residual value between the predicted value of the digital twin mechanism model and the actual operation result of the physical entity of the mine intelligent chassis.

[0011] Preferably, the residual of the digital twin mechanism model is compensated and corrected by the digital twin data-driven model, and the compensation formula is: In the formula, is the predicted value of the DT model of the mine intelligent chassis after compensation, is the predicted value of the digital twin mechanism model, Predict values ​​for digital twin data-driven models.

[0012] According to a second aspect, the present invention provides a trajectory tracking control method for a mine intelligent chassis, which can execute the DT model construction method for a mine intelligent chassis according to the first aspect and any optional embodiment, comprising the following steps: Linearize the digital twin mechanism model to obtain a linear time-varying equation; The linear time-varying equation is discretized using a first-order difference quotient method to obtain a discrete state space expression; Based on the discrete state space expression and the model prediction control strategy, a trajectory prediction model is constructed based on the mine intelligent chassis DT model, and a trajectory tracking optimization problem is determined at the same time, and the trajectory tracking optimization problem is solved according to the trajectory prediction model; The trajectory control of the current control cycle is implemented according to the solution result of the trajectory tracking optimization problem.

[0013] Preferably, before determining the trajectory tracking optimization problem, the method further includes: A relaxation factor is added, a trajectory control function is constructed based on the discrete state space expression and the relaxation factor, and a trajectory tracking optimization problem is determined based on the trajectory control function and the model predictive control strategy.

[0014] Preferably, it also includes: According to the solution result of the trajectory tracking optimization problem, the control input increment and relaxation factor of the current control cycle are obtained, and a control sequence is constructed to generate a control input sequence in the future prediction time domain. The first control input increment of the control sequence is applied to the system to realize the movement of the mine intelligent chassis under the action of the new control input, and the state information is further updated to enter the next control cycle. This cycle is repeated to realize the trajectory tracking control of the mine intelligent chassis.

[0015] Compared with the prior art, the present invention has the following beneficial effects: 1. This paper proposes a mechanism-data hybrid driven DT model for the mine intelligent chassis, which solves the problem of the singleness of most current vehicle dynamics models. The model combines the interpretability of the mechanism analysis model with the nonlinear characteristics of the neural network, which can supplement the influence of some unmodeled effects and make the model output more consistent with the actual characteristics of high-degree-of-freedom vehicles. This comprehensive approach significantly improves the simulation accuracy of the model and has the ability to implicitly understand different road adhesion conditions.

[0016] 2. The present invention uses the mechanism-data hybrid driven DT model as the trajectory prediction model to achieve more accurate trajectory tracking control. This method provides more accurate state prediction by integrating the advantages of the mechanism model and the data-driven model, enabling the trajectory prediction model to formulate a more optimized control strategy, greatly improving the accuracy and effect of trajectory tracking. BRIEF DESCRIPTION OF THE DRAWINGS

[0017] In order to more clearly illustrate the embodiments of the present invention or the technical solutions in the prior art, the drawings required for use in the embodiments will be briefly introduced below. Obviously, the drawings described below are only some embodiments of the present invention. For ordinary technicians in this field, other drawings can be obtained based on these drawings without paying creative work.

[0018] Figure 1 It is a flow chart of a method for constructing a DT model of a mine intelligent chassis provided by an embodiment of the present invention; Figure 2 It is a schematic diagram of the data collection process of the intelligent chassis of a mine provided by an embodiment of the present invention; Figure 3 It is a simplified diagram of the structure of a convolutional recurrent neural network combined with the SE attention mechanism provided in an embodiment of the present invention; Figure 4 It is a schematic diagram of the digital twin data-driven model provided in an embodiment of the present invention being connected in parallel with the digital twin mechanism model as an error compensator; Figure 5 It is a flow chart of a trajectory tracking control method for a mine intelligent chassis provided in an embodiment of the present invention. DETAILED DESCRIPTION

[0019] In conjunction with the drawings in the embodiments of the present invention, the technical solutions in the embodiments of the present invention are clearly and completely described. Obviously, the described embodiments are only part of the embodiments of the present invention, not all of the embodiments. Based on the embodiments of the present invention, all other implementation methods obtained by ordinary technicians in this field without making creative work are within the scope of protection of the present invention.

[0020] It should be noted that the structures, proportions, sizes, etc. illustrated in the drawings of this specification are only used to match the contents disclosed in the specification so that people familiar with this technology can understand and read them. They are not used to limit the conditions under which the present invention can be implemented, and therefore have no substantive technical significance. Any structural modification, change in proportional relationship or adjustment of size should fall within the scope of the technical contents disclosed in the present invention without affecting the effects and purposes that can be achieved by the present invention. It should be noted that in this specification, relational terms such as first and second are only used to distinguish one entity from several other entities, and do not necessarily require or imply any actual relationship or order between these entities.

[0021] As an important supporting technology for realizing autonomous driving, trajectory tracking control mainly uses control algorithms to adjust the vehicle tire angles so that the lateral position error and heading deviation of the vehicle are as small as possible while driving along the reference trajectory, while ensuring that the vehicle has a certain degree of stability and driving safety. At present, most control methods describe the operation of the vehicle by calculating physical quantities such as the vehicle yaw rate based on the vehicle dynamic model derived from physics, and then design a feedback control system for tracking. However, the currently constructed model cannot accurately calculate the real dynamic response of the vehicle during driving, and there is an error with the actual result. When operating underground, the path tracking capability and stability of the mine intelligent chassis cannot be guaranteed.

[0022] Therefore, in order to solve the above problems, a DT model construction method for a mine intelligent chassis is provided in this embodiment. Figure 1 : is a flow chart of constructing a DT model of a mine intelligent chassis according to an embodiment of the present invention, such as Figure 1 As shown, the process includes the following steps: S1: Obtain the vehicle test data set under variable working conditions of the mine intelligent chassis.

[0023] Optionally, the acquisition of the whole vehicle test data set specifically includes: acquiring the whole vehicle test data of the mine intelligent chassis under different road conditions, the different road surfaces at least including dry asphalt roads and wet muddy roads; performing data processing based on the whole vehicle test data, and extracting actual mine intelligent chassis dynamics data according to the processing results, the actual mine intelligent chassis dynamics data including mine intelligent chassis state quantities and control quantities, to construct a whole vehicle test data set.

[0024] In this embodiment, the surrounding environment data is collected by the data acquisition device installed on the mine intelligent chassis, and the feedback data from the vehicle encoder, speed sensor and other devices are received through the CAN bus. The mine intelligent chassis is controlled to perform vehicle tests and collect vehicle test data under different road conditions. Figure 2The data collection process of the mine intelligent chassis shown in the figure is as follows: the surrounding environment data is collected through laser radar, combined inertial navigation, RGB-D front camera and other equipment, and the mine intelligent chassis is operated to carry out various vehicle tests on dry asphalt roads and slippery muddy roads underground, such as sinusoidal swept frequency steering tests, quasi-static straight-line acceleration and deceleration tests, serpentine tests and other vehicle tests, to obtain the actual mine intelligent chassis dynamics data under high, medium and low road adhesion coefficients, and then the actual mine intelligent chassis dynamics data is processed to obtain the mine intelligent chassis state quantity and control quantity, thereby constructing the mine intelligent chassis data set.

[0025] S2: Considering the longitudinal force of tire slip on the tire, a digital twin mechanism model of the mine intelligent chassis is established.

[0026] In this embodiment, when constructing the mechanism model of the mine intelligent chassis, only the pure side deviation tire characteristics are considered, and the lateral traction, angular momentum of the front and rear tires, and lateral displacement speed and longitudinal displacement speed of the mine intelligent chassis during operation are calculated, thereby establishing a single-track dynamic mine intelligent chassis mechanism model.

[0027] The lateral traction force of the mine intelligent chassis during operation is: In the formula, For the overall quality of the mine intelligent chassis, is the acceleration of lateral movement, is the speed of longitudinal movement, is the heading angular velocity, is the longitudinal lateral stiffness of the front wheel of the mine intelligent chassis, is the front wheel turning angle, is the lateral movement speed, is the distance from the front axle to the center of mass of the mine intelligent chassis, is the longitudinal lateral stiffness of the rear wheel of the mine intelligent chassis, It is the distance from the rear axle to the center of mass of the mine intelligent chassis.

[0028] The angular momentum of the front and rear tires is: In the formula, is the moment of inertia, is the heading angular acceleration.

[0029] The horizontal displacement velocity under the geodetic coordinate axis is: In the formula, is the heading angle.

[0030] The longitudinal displacement velocity under the geodetic coordinate axis is: .

[0031] Optionally, the construction process of the digital twin mechanism model includes the following steps: S201: Establish vehicle dynamics model; S202: Add center of mass sideslip angle constraints, vehicle adhesion condition constraints and tire sideslip angle constraints to the vehicle dynamics model; S203: Establish dynamic equations, add vehicle parameters of the mine intelligent chassis, and obtain the digital twin mechanism model.

[0032] In this embodiment, the state information and control information of the intelligent chassis of the vehicle dynamics model are used as the input features of the digital twin mechanism model. On this basis, vehicle dynamics constraints are added, including center of mass side slip angle constraints and vehicle adhesion condition constraints; at the same time, the side slip angle of the tire also needs to be restricted. The state information is , the control information is , where , They are the horizontal displacement and longitudinal displacement of the mine intelligent chassis under the geodetic coordinate axis. represents the transpose of a matrix, Indicates vehicle speed.

[0033] In this embodiment, the dynamic equations are written in MATLAB / Simulink software, and vehicle parameters are added, such as vehicle body length, width, height, vehicle mass, vehicle moment of inertia around the z-axis, and other parameter data.

[0034] S3: Based on the vehicle test data set and digital twin mechanism model, a digital twin data-driven model of the mine intelligent chassis is established.

[0035] Optionally, the construction process of the digital twin data-driven model includes the following steps: S301: Construct a convolutional recurrent neural network combined with SE attention mechanism; S302: Obtaining a residual value of the digital twin mechanism model based on a comparison of the simulation results of the vehicle test data set and the digital twin mechanism model; S303: The vehicle test data set is used as the input of the convolutional recurrent neural network combined with the SE attention mechanism, and the residual value of the digital twin mechanism model is used as the output of the convolutional recurrent neural network combined with the SE attention mechanism to train and obtain the digital twin data-driven model.

[0036] In this embodiment, the network structure diagram of the convolutional recurrent neural network (CNN-BiLSTM-SE) combined with the SE attention mechanism is as follows: Figure 3As shown in the figure, the local features of the input data are extracted by the convolutional neural network (CNN), and the long-term features are extracted by the recurrent neural network (RNN). Among them, CNN can be effectively applied to the prediction of time series. Its local perception and weight sharing can greatly reduce the number of parameters of the model, thereby improving the learning efficiency of the model. Each convolution layer contains multiple convolution kernels. CNN is used to obtain key local features from shorter fragments of the overall data set, and when the position of the feature in the data fragment is not highly correlated, CNN is very effective in extracting local features of the input.

[0037] In this embodiment, the output value after CNN convolution is: In the formula, is the output value after convolution, is the activation function, is the input vector, is the weight of the convolution kernel, is the bias of the convolution kernel.

[0038] In this embodiment, BiLSTM is composed of a forward LSTM and a backward LSTM, so it has the characteristic of being extended according to time series and is widely used in time series analysis.

[0039] The core idea of ​​LSTM is to adjust the output of the state unit through the input gate, forget gate and output gate. By viewing Time input vector and Output of hidden state at each moment The value of determines whether to forget or retain The information stored in the state unit at the moment Similarly, the input gate This determines the input vector and Output at any time How much information needs to be passed to update the state unit information. Output gate Controls the flow of information from the current state unit information to the hidden state. The calculation formula of these gating units is as follows: Where: σ is the activation function Sigmoid; is the forget gate weight matrix; is the bias of the forget gate; is the input gate weight matrix; is the bias of the input gate; for Information stored in the moment state unit; A weight matrix that stores information for state units; Bias for storing information for state cells; is the output gate weight matrix; is the deviation of the output gate, for The output value of the hidden state at time t, Represents the product operation between neural networks.

[0040] The BiLSTM model is an improved model based on the LSTM network, which can read the relevant features of the input data in both forward and backward directions. The forward LSTM processes the information of the input data from left to right and its hidden state can be expressed as , for The hidden state of the forward LSTM at each moment; the backward LSTM processes information from right to left and its hidden state can be expressed as , for Reverse the hidden layer state of LSTM at every moment, and finally concatenate the forward and backward states to get the output of BiLSTM .

[0041] SEBlock (Squeeze-and-Excitation Block, SE) is a new structural unit that focuses on the channel dimension. It adds a channel attention mechanism to the model. This mechanism enhances or suppresses the corresponding channels for different tasks by adding weights of the importance of each feature channel, so as to extract useful features. The overall process is divided into three steps: first, the Squeeze compression operation compresses the features of the spatial dimension, keeping the number of feature channels unchanged, and then fuses the global information through global pooling to convert each two-dimensional feature channel into a real number. The second is the Excitation excitation operation, which first reduces the dimension of the feature and then increases the dimension. It consists of two fully connected layers and a Sigmoid function. Finally, the Reweight operation weights the previous input features channel by channel to complete the redistribution of the original features on each channel. The calculation formula is as follows: Where s is the output of the excitation operation; and are the corresponding parameters of the two fully connected layers; δ is the activation function ReLU; , refers to aggregation across spatial dimensions, i.e. is the height, is the width; is a vector; Represents a channel descriptor, for No. k elements.

[0042] Optionally, the residual value of the digital twin mechanism model is the residual value between the predicted value of the digital twin mechanism model and the actual operation result of the physical entity of the mine intelligent chassis.

[0043] In this embodiment, in order to build a digital twin data-driven model, it is necessary to operate the physical entity of the vehicle for testing under various working conditions, including movement on different road surfaces and different trajectories. The command input during its movement and the data fed back by the vehicle CAN bus are collected as the input and output of the BiLSTM network. The input end is consistent with the input of the digital twin mechanism model. The output end is the residual value between the predicted value of the digital twin mechanism model and the actual operation result of the physical entity.

[0044] In this embodiment, the detailed parameters for training the digital twin data-driven model are set as follows: dropoutLayer is 0.2, the preset maximum number of training times is 1000, the minimum batch size for each iteration is 10, the gradient threshold is 1, the initial value of the positive learning rate is 0.005, and it is updated according to the piecewise constant decay. The drop cycle is 150, the drop factor is 0.2, the verification frequency is once every 6 epochs, the training cutoff condition "the number of times the verification loss is allowed to be greater than or equal to the previous minimum loss" is 6, and the digital twin data-driven model is trained using the Adam solver.

[0045] S4: The digital twin data-driven model is connected in parallel with the digital twin mechanism model as an error compensator to obtain the mine intelligent chassis DT model based on data-mechanism hybrid drive.

[0046] In this embodiment, Figure 4 As shown in the figure, the digital twin data-driven model is connected in parallel with the digital twin mechanism model as an error compensator to compensate and correct the digital twin mechanism model, thereby improving the prediction accuracy of the mine intelligent chassis DT model based on data-mechanism hybrid drive. The specific compensation formula is: In the formula, is the predicted value of the DT model of the mine intelligent chassis after compensation, is the predicted value of the digital twin mechanism model, Predict values ​​for digital twin data-driven models.

[0047] The embodiment of the present invention constructs a mechanism-data hybrid driven DT model of the mine intelligent chassis by connecting the digital twin data-driven model as an error compensator in parallel with the digital twin mechanism model, solving the problem of the singleness of most current vehicle dynamics models. The model combines the interpretability of the mechanism analysis model and the nonlinear characteristics of the neural network, and can supplement the influence of some unmodeled effects, making the model output more consistent with the actual characteristics of high-degree-of-freedom vehicles. This comprehensive method significantly improves the simulation accuracy of the model and has the ability to implicitly understand different road adhesion conditions.

[0048] The present invention also provides a trajectory tracking control method, which is applicable to the DT model of the mine intelligent chassis constructed in the above embodiment, and includes the following steps: Linearize the digital twin mechanism model to obtain a linear time-varying equation; The linear time-varying equation is discretized using a first-order difference quotient method to obtain a discrete state space expression; Based on the discrete state space expression and the model prediction control strategy, a trajectory prediction model is constructed based on the mine intelligent chassis DT model, and a trajectory tracking optimization problem is determined at the same time, and the trajectory tracking optimization problem is solved according to the trajectory prediction model; The trajectory control of the current control cycle is implemented according to the solution result of the trajectory tracking optimization problem.

[0049] Optionally, before determining the trajectory tracking optimization problem, the following steps are also included: A relaxation factor is added, a trajectory control function is constructed based on the discrete state space expression and the relaxation factor, and a trajectory tracking optimization problem is determined based on the trajectory control function and the model predictive control strategy.

[0050] Optionally, the control input increment and relaxation factor of the current control cycle are obtained according to the solution result of the trajectory tracking optimization problem, and a control sequence is constructed to generate a control input sequence in the future prediction time domain. The first control input increment of the control sequence is applied to the system to realize the movement of the mine intelligent chassis under the action of the new control input, and the state information is further updated to enter the next control cycle. This cycle is repeated to realize the trajectory tracking control of the mine intelligent chassis.

[0051] In this embodiment, Figure 5As shown in the figure, according to the mine intelligent chassis, a mine intelligent chassis DT model based on data-mechanism hybrid drive is constructed, and the mine intelligent chassis DT model is used as the prediction model of MPC to realize trajectory tracking control. First, the dynamic equation of the intelligent chassis is linearized to obtain a linear time-varying equation. Secondly, the linear time-varying equation is discretized using the first-order difference quotient method to obtain a discrete state space expression, and then the objective function is designed. In view of the high complexity of the vehicle dynamics model and more constraints, the controller is likely to fail to find the optimal solution within the specified time during execution. At this time, the vehicle steering wheel is in a dangerous state of out-of-control. In order to avoid this situation, a relaxation factor is added to the objective function, and the output of the constraint range is dynamically adjusted through the relaxation factor.

[0052] In this embodiment, the objective function is: In the formula, , is the relaxation factor; is status information; To control information; For the prediction time domain; To control the time domain; To control the output prediction value; To control the output reference value; Indicates based on The sampling information at the moment is used to predict The value at the moment, ; for The input increment is controlled at all times, where ; is the system output weight system matrix; is the system matrix of control increment and control amount weight.

[0053] After completing the solution of each control cycle, a series of control input increments and relaxation factors in the control time domain are obtained. The first element in the control sequence is used as the actual control input increment to act on the system, and then the next control cycle is entered. The above process is repeated, and the cycle is repeated to achieve tracking control of the desired trajectory.

[0054] The embodiment of the present invention uses the mechanism-data hybrid driven DT model as the trajectory prediction model to achieve more accurate trajectory tracking control. This method provides more accurate state prediction by integrating the advantages of the mechanism model and the data driven model, enabling the trajectory prediction model to formulate a more optimized control strategy, greatly improving the accuracy and effect of trajectory tracking.

[0055] The above is only a preferred specific embodiment of the present invention, but the protection scope of the present invention is not limited thereto. Any changes or substitutions that can be easily thought of by a person skilled in the art within the technical scope disclosed by the present invention should be included in the protection scope of the present invention. Therefore, the protection scope of the present invention shall be based on the protection scope of the claims.

Claims

1. A DT model construction method for a mine intelligent chassis, characterized in that: The steps include: S1: Obtain the vehicle test data set under variable working conditions of the mine intelligent chassis; S2: Considering the longitudinal force of tire slip on the tire, a digital twin mechanism model of the mine intelligent chassis is established; S3: Based on the vehicle test data set and digital twin mechanism model, a digital twin data-driven model of the mine intelligent chassis is established; S4: The digital twin data-driven model is connected in parallel with the digital twin mechanism model as an error compensator to obtain the mine intelligent chassis DT model based on data-mechanism hybrid drive.

2. The DT model construction method of the mine intelligent chassis according to claim 1, characterized in that: The acquisition of vehicle test data set specifically includes: Acquire vehicle test data of the mine intelligent chassis under different road conditions, where the different road conditions include at least a dry asphalt road and a slippery muddy road; Data processing is performed based on the whole vehicle experimental data, and actual mine intelligent chassis dynamics data is extracted according to the processing results. The actual mine intelligent chassis dynamics data includes the mine intelligent chassis state quantity and control quantity, and a whole vehicle test data set is constructed.

3. The DT model construction method of the mine intelligent chassis according to claim 1, characterized in that: The construction process of the digital twin mechanism model includes the following steps: S201: Establish vehicle dynamics model; S202: Add center of mass sideslip angle constraints, vehicle adhesion condition constraints and tire sideslip angle constraints to the vehicle dynamics model; S203: Establish dynamic equations, add vehicle parameters of the mine intelligent chassis, and obtain the digital twin mechanism model.

4. The DT model construction method of the mine intelligent chassis according to claim 1, characterized in that: The process of building a digital twin data-driven model includes the following steps: S301: Construct a convolutional recurrent neural network combined with SE attention mechanism; S302: Obtaining a residual value of the digital twin mechanism model based on a comparison of the simulation results of the vehicle test data set and the digital twin mechanism model; S303: The vehicle test data set is used as the input of the convolutional recurrent neural network combined with the SE attention mechanism, and the residual value of the digital twin mechanism model is used as the output of the convolutional recurrent neural network combined with the SE attention mechanism to train and obtain the digital twin data-driven model.

5. The DT model construction method of the mine intelligent chassis according to claim 4 is characterized in that: The residual value of the digital twin mechanism model is the residual value between the predicted value of the digital twin mechanism model and the actual operation result of the physical entity of the mine intelligent chassis.

6. The DT model construction method of the mine intelligent chassis according to claim 5, characterized in that: The residual of the digital twin mechanism model is compensated and corrected through the digital twin data-driven model. The compensation formula is: In the formula, is the predicted value of the DT model of the mine intelligent chassis after compensation, is the predicted value of the digital twin mechanism model, Predict values ​​for digital twin data-driven models.

7. A trajectory tracking control method for a mine intelligent chassis, based on the DT model construction method for a mine intelligent chassis according to any one of claims 1 to 6, characterized in that: The following steps are involved: Linearize the digital twin mechanism model to obtain a linear time-varying equation; The linear time-varying equation is discretized using a first-order difference quotient method to obtain a discrete state space expression; Based on the discrete state space expression and the model prediction control strategy, a trajectory prediction model is constructed based on the mine intelligent chassis DT model, and a trajectory tracking optimization problem is determined at the same time, and the trajectory tracking optimization problem is solved according to the trajectory prediction model; The trajectory control of the current control cycle is implemented according to the solution result of the trajectory tracking optimization problem.

8. The trajectory tracking control method of the mine intelligent chassis according to claim 7 is characterized in that: The determining of the trajectory tracking optimization problem also includes: A relaxation factor is added, a trajectory control function is constructed based on the discrete state space expression and the relaxation factor, and a trajectory tracking optimization problem is determined based on the trajectory control function and the model predictive control strategy.

9. The trajectory tracking control method of the mine intelligent chassis according to claim 7, characterized in that: Also includes: According to the solution result of the trajectory tracking optimization problem, the control input increment and relaxation factor of the current control cycle are obtained, and a control sequence is constructed to generate a control input sequence in the future prediction time domain. The first control input increment of the control sequence is applied to the system to realize the movement of the mine intelligent chassis under the action of the new control input, and the state information is further updated to enter the next control cycle. This cycle is repeated to realize the trajectory tracking control of the mine intelligent chassis.

Images