Gear shifting control method, device and equipment for electric mining truck and storage medium

Abstract

The invention discloses a mining electric truck gear shifting control method, device and equipment and a storage medium. Comprising the steps of constructing a system dynamics model based on a vehicle electric drive assembly structure; training a physical information neural network by taking the system dynamics model as a physical constraint in an active synchronization stage of a gear shifting process; embedding the trained physical information neural network as a prediction model into a model prediction control framework, and performing rolling optimization solution based on a preset multi-objective optimization function to obtain an optimal control sequence; and performing control based on the optimal control sequence. According to the method, accurate coordination control over the gear shifting process is achieved according to the nonlinear characteristic of the gear shifting process of the mining truck under the extreme working condition, especially accurate control over the active synchronization stage; the problems of torque interruption, impact vibration, low energy efficiency and the like of the pure electric mining truck in the gear shifting process under the extreme working conditions of heavy load, low speed, large gradient, uneven road surfaces and the like are effectively solved.

Description

Technical Field

[0001] This application relates to the field of vehicle control technology, and more specifically, to a method, device, equipment, and storage medium for shifting control of an electric mining truck. Background Technology

[0002] With the rapid development of electrification and intelligentization technologies in mining, pure electric mining trucks, as a highly efficient and environmentally friendly heavy-duty transportation equipment, have been widely used in open-pit mines and other scenarios. However, their complex power transmission systems face multiple challenges during frequent gear shifts. Mining trucks operate under extreme conditions such as heavy loads, low speeds, steep gradients, and uneven road surfaces, making it difficult for traditional shift control strategies to simultaneously achieve smoothness, responsiveness, and energy efficiency. Especially under high load conditions, torque interruption and impact vibration during gear shifts can accelerate the wear of transmission components, reduce system reliability, and affect the overall vehicle operating efficiency.

[0003] In existing technologies, the shift control of mining trucks mainly adopts rule-based logic control or traditional PID control strategies. These methods are insufficient in modeling the nonlinear dynamic characteristics of the system and are difficult to adapt to changes in operating conditions. Summary of the Invention

[0004] This application provides a method, apparatus, device, and storage medium for controlling the gear shifting of an electric mining truck, in order to at least solve the problem of precise coordinated control of the gear shifting process of mining trucks in related technologies.

[0005] According to one aspect of the embodiments of this application, a shift control method for an electric mining truck is provided, comprising: A system dynamics model is constructed based on the vehicle electric drive assembly structure; During the active synchronization phase of the gear shifting process, a physical information neural network is trained using the system dynamics model as a physical constraint. The trained physical information neural network is embedded into the model predictive control framework as a prediction model, and the optimal control sequence is obtained by rolling optimization based on a preset multi-objective optimization function. Control is performed based on the optimal control sequence.

[0006] In one implementation, a system dynamics model is constructed based on the vehicle's electric drive assembly structure, including: Analysis of the structural parameters of the vehicle electric drive assembly; Based on the structural parameters of the electric drive assembly, a longitudinal dynamics model of the vehicle is established. The longitudinal dynamics model of the vehicle integrates the output torque of multiple drive motors, transmission ratio, rolling resistance, air resistance, total vehicle mass, wheel radius, slope resistance and braking torque parameters. Based on the vehicle longitudinal dynamics model, the system dynamics model for the active synchronization phase is determined.

[0007] In one implementation, before training the physical information neural network using the system dynamics model as a physical constraint during the active synchronization phase of the gear shifting process, the method further includes: The gear shifting process is divided into the torque unloading stage, the disengagement stage, the active synchronization stage, the gear engagement stage, and the torque recovery stage. The status of the gear shifting process is monitored in real time to determine whether the vehicle has entered the active synchronization phase.

[0008] In one implementation, training a physical information neural network using the system dynamics model as a physical constraint includes: Based on the system dynamics model, state decoupling is performed to establish system state equations that include state variables, control variables, and outputs; Discretize the system state equations to obtain discrete state transition equations; A physical information neural network is trained based on the discrete state transition equations as physical constraints.

[0009] In one implementation, training a physical information neural network based on the discrete state transition equation as a physical constraint includes: Substitute the predicted output value of the physical information neural network into the discrete state transition equation to calculate the physical loss; Based on a dataset containing system state, control input, and measured output, calculate the data loss; Based on the physical loss and data loss, a composite loss function is constructed; The composite loss function is optimized using the backpropagation algorithm, and the network weight parameters are updated to obtain the trained physical information neural network.

[0010] In one implementation, before performing rolling optimization based on a preset multi-objective optimization function, the method further includes: Determine the control objectives of the model predictive control framework, including speed synchronization accuracy, angle synchronization accuracy, shift smoothness, and minimum transient torsional vibration. The multi-objective optimization function is obtained by weighting and summing the various control objectives based on the quadratic cost function and preset weight coefficients.

[0011] In one implementation, the trained physical information neural network is embedded as a prediction model into a model predictive control framework, and rolling optimization is performed based on a preset multi-objective optimization function to obtain the optimal control sequence, including: The physical information neural network is used as a prediction model and embedded in the model prediction control framework. At each rolling optimization moment of model predictive control, the system state is predicted by the physical information neural network based on the current system state and the control input sequence in the future control time domain, resulting in a state prediction sequence; Based on the state prediction sequence, combined with the multi-objective optimization function and system constraints, the finite-time optimal control problem is solved online in each control cycle to obtain the optimal control sequence.

[0012] According to another aspect of the embodiments of this application, a shift control device for a mining electric truck is provided, comprising: The dynamics model building module is used to build system dynamics models based on the vehicle electric drive assembly structure. The neural network training module is used to train a physical information neural network with the system dynamics model as the physical constraint during the active synchronization phase of the gear shifting process. The optimization and solution module is used to embed the trained physical information neural network as a prediction model into the model predictive control framework, and perform rolling optimization based on a preset multi-objective optimization function to obtain the optimal control sequence. The shift control module is used to perform control based on the optimal control sequence.

[0013] According to another aspect of the embodiments of this application, an electronic device is also provided, including a memory and a processor, wherein the memory stores a computer program, and the processor is configured to execute the above-described electric mining truck shift control method through the computer program.

[0014] According to another aspect of the embodiments of this application, a computer-readable storage medium is also provided, wherein a computer program is stored in the computer program, and the computer program is configured to execute the above-described electric mining truck shift control method when running.

[0015] The technical solutions provided in this application embodiment may include the following beneficial effects: This application provides a shift control method for electric mining trucks. A system dynamics model is constructed based on the vehicle's electric drive assembly structure. During the active synchronization phase of the shifting process, a physical information neural network is trained using the system dynamics model as a physical constraint. The trained physical information neural network is then embedded into a model predictive control framework (MMC) for rolling optimization to obtain the optimal control sequence. This application integrates the physical information neural network with the MMC framework. Addressing the nonlinear characteristics of the shifting process in mining trucks under extreme conditions, particularly the precise control during the active synchronization phase, the dynamics model is used as a physical constraint to train the neural network. This allows the physical information neural network to utilize both measured data and prior physical knowledge, enabling it to accurately predict the nonlinear state evolution of the transmission system under heavy load conditions. It possesses stronger generalization ability and physical interpretability, making it particularly suitable for the shift control requirements of mining trucks under different loads, gradients, and road surface conditions. Combined with the MMC framework, this method coordinates the motor torque and the gearbox actuator in real time, suppressing shift shock and achieving precise coordinated control of the shifting process, significantly reducing shock and energy loss. Attached Figure Description

[0016] The accompanying drawings, which are included to provide a further understanding of this application and form part of this application, illustrate exemplary embodiments and are used to explain this application, but do not constitute an undue limitation of this application. In the drawings: Figure 1 This is a flowchart of a gear shifting control method for an electric mining truck according to an embodiment of this application; Figure 2 This is a simplified model diagram of a controlled object according to an embodiment of this application; Figure 3 This is a schematic diagram of a four-motor drive assembly structure according to an embodiment of this application; Figure 4 This is a schematic diagram of a five-stage gear shifting process according to an embodiment of this application; Figure 5 This is a schematic diagram of a gear shifting control method for a pure electric mining truck according to an embodiment of this application; Figure 6 This is a schematic diagram of a gear shifting control device for an electric mining truck according to an embodiment of this application; Figure 7 This is a schematic diagram of the structure of an electronic device according to an embodiment of this application. Detailed Implementation

[0017] To enable those skilled in the art to better understand the present application, the technical solutions in the embodiments of the present application will be clearly and completely described below with reference to the accompanying drawings. Obviously, the described embodiments are only some embodiments of the present application, and not all embodiments. Based on the embodiments in the present application, all other embodiments obtained by those of ordinary skill in the art without creative effort should fall within the scope of protection of the present application.

[0018] It should be noted that the terms "first," "second," etc., in the specification, claims, and accompanying drawings of this application are used to distinguish similar objects and are not necessarily used to describe a specific order or sequence. It should be understood that such data can be interchanged where appropriate so that the embodiments of this application described herein can be implemented in orders other than those illustrated or described herein. Furthermore, the terms "comprising" and "having," and any variations thereof, are intended to cover non-exclusive inclusion; for example, a process, method, system, product, or apparatus that comprises a series of steps or units is not necessarily limited to those steps or units explicitly listed, but may include other steps or units not explicitly listed or inherent to such processes, methods, products, or apparatus.

[0019] This application aims to address the shortcomings of existing technologies by proposing a shift control method for pure electric mining trucks based on physical information neural networks and model predictive control. Existing shift control strategies for mining trucks lack sufficient modeling of the system's nonlinear dynamic characteristics and struggle to adapt to changing operating conditions. Some advanced methods introduce model predictive control to improve shift quality, but these heavily rely on accurate system models and lack robustness under parameter perturbations and external disturbances. Particularly concerning is the critical active synchronization stage during shifting, where strong nonlinear characteristics and impact loads make it difficult for conventional control methods to accurately predict and suppress impact vibrations. This application addresses the precise control of the active synchronization stage, effectively solving problems such as torque interruption, impact vibrations, and low energy efficiency during shifting in pure electric mining trucks under extreme conditions such as heavy loads, low speeds, steep gradients, and uneven road surfaces.

[0020] This application addresses the nonlinear characteristics of gear shifting in mining trucks under extreme operating conditions, particularly the precise control of the active synchronization phase. First, the gear shifting process is divided into five stages: torque unloading, disengagement, active synchronization, engagement, and torque recovery. In the active synchronization phase, a physical information neural network is introduced for precise modeling. Unlike traditional data-driven methods, the physical information neural network embeds dynamic equations as soft constraints into the neural network training process, enabling the network to utilize both measured data and prior physical knowledge.

[0021] Specifically, based on the physical characteristics of the power transmission system of mining trucks, a physical model including the dynamic equations of the engagement sleeve and engagement gear ring is established, and a composite loss function is defined, which includes physical loss terms and data loss terms. This design enables the physical information neural network to maintain good predictive performance even in the harsh environment of mining areas, facing high sensor noise and low data reliability. After training, the physical information neural network model can accurately predict the dynamic changes of speed difference and angle difference during active synchronization, providing accurate state information for subsequent control. Compared with pure data-driven methods, the physical information neural network has stronger generalization ability and physical interpretability under sparse data conditions, making it particularly suitable for the shift control requirements of mining trucks under different loads, gradients, and road conditions.

[0022] This application combines a model predictive control framework to achieve real-time coordinated control of motor torque and gearbox actuators. Within the MPC (Model Predictive Control) framework, a PINN (Physics-Informed Neural Networks) model is used as the predictive model to construct a rolling optimization problem. The system expects the output to be a speed difference of 0 and an angle difference equal to the target value. The optimal control sequence is obtained by solving the optimization problem.

[0023] The advantages of this application are: (1) embedding a physical information neural network in the active synchronization stage, which takes into account both model accuracy and computational efficiency; (2) improving the generalization ability and interpretability of the neural network in data-scarce scenarios through the embedding of physical constraints, especially suitable for sensor data in harsh mining environments; (3) combining the model predictive control framework to achieve precise coordinated control of the shifting process, which significantly reduces the impact and energy loss.

[0024] The shift control method for electric mining trucks according to embodiments of this application will be described in detail below with reference to the accompanying drawings. Figure 1 As shown, the method mainly includes the following steps: S101 constructs a system dynamics model based on the vehicle electric drive assembly structure.

[0025] In one implementation, a longitudinal dynamics model of the vehicle is constructed based on the structure of the vehicle's electric drive assembly, including analyzing the structural parameters of the vehicle's electric drive assembly; and establishing a longitudinal dynamics model of the vehicle based on the structural parameters of the electric drive assembly, wherein the longitudinal dynamics model of the vehicle integrates the output torque of multiple drive motors, transmission ratio, rolling resistance, air resistance, total vehicle mass, wheel radius, slope resistance and braking torque parameters.

[0026] In an exemplary embodiment, a longitudinal dynamics model of the whole vehicle is established, taking the electric drive assembly of a new energy heavy-duty electric mining truck as an example.

[0027] This embodiment uses a new energy heavy-duty electric mining truck drive assembly as an example. The four-motor drive assembly adopts the following... Figure 3 The illustrated four-motor drive and dual AMT electric drive assembly includes four drive motors, an intermediate shaft, and a shifting mechanism. The shifting mechanism consists of two AMTs. Each pair of motors transmits power to a single AMT via intermediate gear shafts 1 and 2. The dual AMTs drive the vehicle through intermediate gear shaft 3. Each AMT has five gear selections (gears I, II, III, IV, and neutral), which are engaged and disengaged by two shifting motors. During shifting, alternating shifts and progressive upshifts are ensured. When one AMT needs to shift, the other AMT performs active torque compensation to ensure uninterrupted power during shifting.

[0028] according to Figure 3 The longitudinal dynamics model of the four-motor electric drive assembly shown is as shown in equation (1): (1) in, , , , This indicates the output torque of motor 1, motor 2, motor 3, and motor 4. Indicates braking torque, Let z1 and z2 be the transmission ratios of gear sets. , The gear ratios of AMT1 and AMT2 transmissions. The rolling resistance coefficient, For the wheel radius, This refers to the vehicle's drag coefficient. Let v be the total mass of the vehicle, v be the vehicle speed, and A be the vehicle's frontal area. This refers to the road slope.

[0029] In one implementation, before training the physical information neural network during the active synchronization phase of the gear shifting process, the gear shifting process is further divided into a torque unloading phase, a disengagement phase, an active synchronization phase, a gear engagement phase, and a torque recovery phase; the state of the gear shifting process is monitored in real time to determine when the vehicle enters the active synchronization phase.

[0030] like Figure 4 As shown, this application divides the gear shifting process into five stages: unloading torque stage, disengaging gear stage, active synchronization stage, engaging gear stage, and restoring torque stage. In an exemplary scenario, the shift from first to second gear will be used as an example for illustration.

[0031] Phase 1: Unloading Torque. Before shifting begins, the output torque of the drive motor is transmitted to the transmission output shaft via the transmission input shaft, intermediate shaft, first gear, first gear engagement gear ring, and engagement sleeve. Therefore, there is a significant circumferential interaction force between the teeth of the engagement sleeve and engagement gear ring. To ensure smooth disengagement, after receiving the shift command, the transmission controller first adjusts the torque of the drive motor to a very small value, minimizing the friction between the engagement sleeve and engagement gear ring during disengagement.

[0032] Phase Two: Disengaging. After unloading the torque, the transmission controller sends a disengaging command to the shift motor controller. The shift motor then actuates, causing the engagement sleeve to separate from the engagement gear ring, returning to the neutral position. During disengagement, the shift actuator first needs to overcome the resistance of the shift fork self-locking mechanism. When the engagement sleeve moves near the target position, the locking effect of the self-locking mechanism on the shift fork ensures that the shift fork remains in the neutral position and does not move freely to either side.

[0033] Phase Three: Active Synchronization. After disengaging the gear, there is no power output from the transmission output end. If the vehicle is traveling on a straight road, the rotational speeds of the engagement sleeve and output shaft will gradually decrease with the vehicle speed. At this time, the rotational speeds of the first-gear engagement gear ring and the engagement sleeve are similar, while the rotational speed of the second-gear engagement gear ring is significantly higher than that of the engagement sleeve. Therefore, it is necessary to control the drive motor to adjust the rotational speed of the second-gear engagement gear ring to be close to the rotational speed of the engagement sleeve (i.e., "speed synchronization"), and simultaneously adjust the rotational angle of the second-gear engagement gear ring so that the tooth backlash of the gear ring aligns with the center line of the teeth of the engagement sleeve (i.e., "angle synchronization"). After both speed synchronization and angle synchronization are completed, active synchronization ends.

[0034] Phase Four: Gear Engagement. After active synchronization is complete, the transmission controller sends a gear engagement command to the shift motor controller. The shift motor actuates, causing the engagement sleeve to engage with the second-gear engagement gear ring. Similar to the disengagement process, the shift actuator must first overcome the resistance of the self-locking mechanism. Finally, under the force of the self-locking mechanism, the shift fork locks in the second-gear position, completing the gear engagement action.

[0035] Phase 5: Torque Restoration. After the gear shifting action is completed, the main shifting process is finished. At this point, the transmission controller sends a command to the drive motor controller to restore drive torque.

[0036] Furthermore, based on the vehicle's longitudinal dynamics model, a dynamic model for the active synchronization phase is determined. In the gear shifting process of a pure electric mining truck, the active synchronization phase is a crucial step in ensuring smooth engagement of the engagement sleeve and engagement gear ring. For example... Figure 2 The figure shows a simplified model of the controlled object in the active synchronization process. The following section derives the formulas for angle and speed synchronization to provide a basis for control.

[0037] For control systems, the electromagnetic torque response model is the starting point for control system design; it describes the TCU (Transmission Control Unit) that sends torque commands to the drive motor. Then, the dynamic response process of the actual electromagnetic torque is described. In practical systems, due to electrical and mechanical inertia, the motor cannot respond to commands instantaneously; therefore, a first-order delay is used to capture this hysteresis characteristic. This model simplifies the complex motor dynamics into linear equations, facilitating subsequent controller analysis and design.

[0038] (2) in, The rate of change of electromagnetic torque represents the speed at which the torque changes with time. This represents the actual electromagnetic torque, that is, the torque actually output by the motor; This represents the motor time constant, reflecting the delay in the motor's response to commands. A smaller value indicates a faster response, and the typical value is determined by the motor's characteristics. This indicates a torque command.

[0039] The engagement sleeve is a key component of the shifting mechanism, and its rotational speed directly determines the smoothness of the synchronization process. The dynamic equation of the engagement sleeve end satisfies: (3) in, This indicates the rate of change of the rotational speed of the coupling sleeve; and Indicates the transmission ratio, specifically the gear transmission ratio from the motor to the engagement sleeve; It represents the equivalent rotational inertia of the coupling sleeve end, including the equivalent inertia of components such as the motor rotor and drive shaft. The larger the value, the greater the system inertia.

[0040] The rotational speed of the engagement sleeve dynamically describes the behavior on the power input side, but shift synchronization also needs to consider the output side, namely the dynamics of the engagement gear ring. The engagement gear ring is affected by the vehicle load, and its model is more complex, involving multiple resistance factors.

[0041] The dynamic equations of the engaging gear ring describe the rotational speed dynamics on the load side. These equations are the core of the system load modeling, reflecting the forced response of the gear ring speed under actual operating conditions. The dynamic equations at the engaging gear ring end satisfy: (4) The results were: (5) in, This represents the rate of change of the rotational speed of the engaging gear ring; It represents the equivalent moment of inertia at the engagement gear ring end, including the inertia of components such as the gear ring and wheels; Indicates the road slope angle; positive for uphill sections. This indicates the vehicle's mass, affecting inertial force and drag. Represents gravitational acceleration; Indicates the vehicle's frontal area; Indicates the air drag coefficient; This represents the rolling resistance coefficient.

[0042] Furthermore, the speed difference between the engaging gear ring and the engaging sleeve satisfies the following relationship: (6) in, It includes various system load items, and the formula is as follows: (7) And the angle difference satisfies: (8) This application analyzes the system dynamics model, which can serve as a physical constraint for the system.

[0043] During the active synchronization phase of the gear shifting process, S102 uses the system dynamics model as the physical constraint to train a physical information neural network.

[0044] Furthermore, the shifting process is monitored in real time to determine whether the active synchronization stage has been reached. In the active synchronization stage, a physical information neural network model is introduced.

[0045] In one implementation, a physical information neural network is trained using a system dynamics model as a physical constraint. This includes decoupling the system based on the system dynamics model, establishing a system state equation that includes state variables, control variables, and outputs, discretizing the system state equation to obtain discrete state transition equations, and training the physical information neural network based on the discrete state transition equations as physical constraints.

[0046] During the active synchronization phase, the speed difference between the engagement sleeve and the engagement gear ring is adjusted by regulating the motor torque command Tcmd. It converges quickly to zero, while the angle difference is small. The target value is tracked to achieve smooth engagement. This system is a typical nonlinear dynamic system, and motor delay, load disturbance, and parameter uncertainties must be considered.

[0047] First, the system is decoupled, so that... ,get: (9) Decoupling yields the state variables as follows The control quantity is Output .

[0048] The system state equation is: (10) The system matrix is ​​defined as follows: (11) (12) (13) Meanwhile, system dynamics can be uniformly represented as: (14) Here, F is a Lipschitz continuous function, ensuring the existence and uniqueness of the solution.

[0049] Furthermore, to achieve digital control, the system needs to be discretized. Let the control step size be T, and the above equations be discretized as follows: (15) The discrete state transition equations can be obtained through numerical integration or approximation methods: (16) in, The mapping function is used. The control problem is that at each time step k, given the current state... Solving for the optimal control input , to make output Follow the reference trajectory while satisfying the constraints.

[0050] Furthermore, based on the discrete state transition equation as the physical constraint, a physical information neural network is trained, including substituting the predicted output value of the physical information neural network into the discrete state transition equation to calculate the physical loss; calculating the data loss based on a dataset containing system state, control input, and measured output; constructing a composite loss function based on the physical loss and data loss; optimizing the composite loss function through the backpropagation algorithm, updating the network weight parameters, and obtaining the trained physical information neural network.

[0051] Understandably, establishing accurate predictive models for complex systems is highly challenging. This application addresses the accuracy and generalization difficulties faced by traditional predictive control models in complex systems, such as electric drive systems for mining trucks, by proposing a model approximation method based on a physical information neural network. Due to parameter uncertainties and errors in real-world systems, traditional predictive controllers based on idealized nominal models struggle to accurately capture the dynamics of real systems. Therefore, this application embeds the system's physical equations, such as discrete state transition equations, as constraints into the neural network training process, constructing a composite loss function that includes both physical consistency loss and data fitting loss. This allows for the learning of a system predictive model that simultaneously conforms to physical laws and measured dynamics under limited experimental samples, significantly improving the model's generalization ability and predictive accuracy.

[0052] Specifically, the approximate solution operator and its derivative Substituting into the discrete state transition equation (16), the corresponding physical loss function can be obtained: (twenty three) Where p represents the physical dataset used for network training. Indicates the size of dataset p. This represents the pairwise points used to evaluate the physical loss during network training.

[0053] Relying solely on physical loss constraints will inevitably lead to inaccurate network modeling, further introducing data loss: (twenty four) Where d represents the real dataset of the test bench, which contains the uncertain perturbation components of the model, and the size of the dataset is given by Nd. Measured values ​​that represent system status, control inputs, and system outputs.

[0054] Furthermore, a composite loss function is constructed based on physical loss and data loss: (25) The physical information neural network model is trained based on a composite loss function. The network weight matrix w is updated through backpropagation until the weights under the minimum composite loss are found, thus obtaining the trained physical information neural network model.

[0055] S103 embeds the trained physical information neural network into the model predictive control framework as a predictive model, and performs rolling optimization based on a preset multi-objective optimization function to obtain the optimal control sequence.

[0056] In one implementation, before performing rolling optimization based on a preset multi-objective optimization function, the control objectives of the model predictive control framework are determined. The control objectives include speed synchronization accuracy, angle synchronization accuracy, shift smoothness, and minimum transient torsional vibration. The multi-objective optimization function is obtained by weighting and summing the various control objectives based on a quadratic cost function and preset weight coefficients.

[0057] Specifically, given the control system model in equation (16), the coordination control problem during gear shifting is defined as determining the optimal control input. Torque fluctuations are reduced through torque compensation based on the motor's fast response characteristics, ensuring that the system state remains consistent with a predefined reference performance standard.

[0058] This application is based on the Model Predictive Control (MPC) framework, and the control objective of the MPC cost function is as follows: Including speed synchronization accuracy, in order to reduce shift time and the impact during shifting, and improve shifting quality, this application takes speed synchronization accuracy as a control target.

[0059] (17) in, Let be the predicted speed difference in step i.

[0060] Including angular synchronization accuracy, in order to improve shifting quality and achieve alignment of the engagement sleeve and engagement gear ring, this application takes angular synchronization accuracy as the control target.

[0061] (18) in, This represents the angle difference predicted in step i. For reference angle.

[0062] Including shift smoothness, in order to reduce torque fluctuations during shifting and improve shift quality, this application takes shift smoothness as a control objective.

[0063] (19) in, This indicates the angular velocity of the wheel.

[0064] This includes minimizing transient torsional vibration. To optimize transient torsional vibration during gear shifting, the acceleration of the gear relative to the meshing direction is used as the optimization target.

[0065] (20) Based on the above control objectives, the coordinated optimal control problem during gear shifting can be transformed into a problem of tracking the speed difference and angle difference between the engagement sleeve and the engagement gear ring. Introducing a quadratic cost function, a multi-objective optimization function can be obtained: (twenty one) The first and second terms aim to reduce shift time, while the third and fourth terms aim to reduce torsional oscillations during shifting. Weighting coefficients Q1 to Q4 are introduced to balance these optimization objectives.

[0066] In model predictive control, a physical information neural network is used as the predictive model and embedded in the model predictive control framework. At each rolling optimization time of model predictive control, the system state is predicted by the physical information neural network based on the current system state and the control input sequence in the future control time domain, and the state prediction sequence is obtained. Based on the state prediction sequence, combined with the multi-objective optimization function and system constraints, the optimal control problem in the finite time domain is solved online in each control cycle to obtain the optimal control sequence.

[0067] Specifically, the system state is predicted by a physical information neural network to obtain a state prediction sequence.

[0068] For the prediction time domain Np, the state prediction is calculated iteratively through a physical information neural network. The state transition equations integrating the physical information neural network and the model predictive control framework are defined as follows: (26) In the rolling optimization of MPC, let the current time step be k, the prediction time domain be Np, and the control time domain be Nc, then the predicted sequence is: (27) For prediction steps exceeding the control time domain (i>Nc), since there are no more optimization variables, it is usually assumed that the control quantity remains unchanged at the last optimized value, i.e. .

[0069] Furthermore, based on the state prediction sequence, combined with the multi-objective optimization function and system constraints, the finite-time optimal control problem is solved online in each control cycle to obtain the optimal control sequence.

[0070] Based on the MPC principle with backtracking level control, within the prediction level [T, T+H] of the current time step T, the optimal control problem is transformed into a constrained moving optimization problem, which can be expressed as follows: (twenty two) Within the time interval [T, T+H], the optimal control sequence can be determined by solving equation (22). The solution methods include, but are not limited to, interior point optimization (IP-OPT), Hamilton-Jacobi Bellman solution, or sequential quadratic programming (SQP).

[0071] In each sampling time T of the model predictive control, the optimal control sequence obtained is determined by only the first control variable. The control is actually applied to the controlled object to complete the current control step. Afterwards, the actual output state of the system is measured in real time using sensors. Using this as a new initial condition, the prediction time domain and the optimization time domain are synchronously rolled forward by one step to enter the next sampling time T+1. The rolling optimization problem is reconstructed and solved in the new interval [T+1, T+H+1]. This process is repeated until the entire shifting process meets the termination condition, thereby realizing closed-loop optimization control based on real-time feedback.

[0072] S104 is controlled based on the optimal control sequence.

[0073] In this embodiment, control is based on an optimal control sequence, which is mainly used to precisely adjust the torque command Tcmd of the drive motor, thereby controlling the speed difference and angle difference between the engagement sleeve and the engagement gear ring. During the active synchronization phase, the MPC optimizes a finite-time-domain torque command sequence by predicting the system's dynamic response in the future time domain. The first control term is applied to the system, enabling the motor output torque to quickly and smoothly track the command, synchronizing the speed of the engagement sleeve with that of the engagement gear ring, while ensuring correct angle alignment, ultimately achieving a shock-free, high-precision shifting process. This control sequence effectively handles nonlinear factors such as system inertia and load disturbances, ensuring shifting quality.

[0074] To facilitate understanding of the methods in the embodiments of this application, the following description is provided in conjunction with the appendix. Figure 5 Further description.

[0075] like Figure 5 As shown, this paper presents an integrated framework for a gear shifting system based on physical information neural networks and model predictive control. Its operational logic begins with the powertrain system. After the control command Tcmd is input, it is transmitted to the vehicle via components such as the engagement sleeve and engagement gear ring under mechanical physical constraints, forming the controlled object. The core neural network module receives time, state, and control inputs, and calculates the loss function through automatic differentiation and physical equation residuals, achieving high-precision and interpretable modeling of system dynamics. The lower-level optimization and control system, based on multiple objectives such as driving comfort, minimum energy loss, and minimization of transient torsional vibration, continuously solves for the optimal control sequence within the model predictive control framework. This sequence is ultimately executed by the powertrain control unit, forming a closed-loop control system integrating hardware and software through electrical connections and energy flow. This achieves smooth, rapid, and low-impact coordinated control of the gear shifting process under complex operating conditions.

[0076] This application addresses the shortcomings of existing shift control strategies for mining trucks in modeling complex nonlinear dynamic characteristics and adaptability to various operating conditions. It introduces a physical information neural network for high-precision modeling in the core active synchronization stage of the shifting process. This model embeds physical equations as intrinsic constraints into the network training process, ensuring that the model output strictly follows the physical laws of the system. Compared to purely data-driven methods, the established physical information-driven model maintains superior predictive robustness even in harsh environments with high sensor noise and poor data quality in mining areas. It can accurately predict the dynamic differences in speed and angle during synchronization, providing precise state information for control decisions. Furthermore, it exhibits stronger generalization ability and physical interpretability under sparse data conditions, effectively adapting to the diverse shift control requirements of mining trucks under different loads, gradients, and complex road conditions.

[0077] According to another aspect of the embodiments of this application, an electric mining truck shift control device for implementing the above-described electric mining truck shift control method is also provided. For example... Figure 6 As shown, the device includes: The dynamics model construction module 601 is used to construct a longitudinal dynamics model of the whole vehicle based on the structure of the vehicle electric drive assembly. The neural network training module 602 is used to train a physical information neural network with the longitudinal dynamics model of the whole vehicle as the physical constraint during the active synchronization phase of the gear shifting process. The optimization and solution module 603 is used to embed the trained physical information neural network as a prediction model into the model predictive control framework, and perform rolling optimization based on a preset multi-objective optimization function to obtain the optimal control sequence. The shift control module 604 is used for shift control based on the optimal control sequence.

[0078] It should be noted that the electric mining truck shift control device provided in the above embodiments is only illustrated by the division of the above functional modules when executing the electric mining truck shift control method. In actual applications, the above functions can be assigned to different functional modules as needed, that is, the internal structure of the equipment can be divided into different functional modules to complete all or part of the functions described above. In addition, the electric mining truck shift control device and the electric mining truck shift control method embodiments provided in the above embodiments belong to the same concept, and the implementation process is detailed in the method embodiments, which will not be repeated here.

[0079] According to another aspect of the embodiments of this application, an electronic device corresponding to the electric mining truck shift control method provided in the foregoing embodiments is also provided to execute the electric mining truck shift control method described above.

[0080] Please refer to Figure 7 This illustrates a schematic diagram of an electronic device provided by some embodiments of this application. For example... Figure 7 As shown, the electronic device includes: a processor 700, a memory 701, a bus 702, and a communication interface 703. The processor 700, the communication interface 703, and the memory 701 are connected via the bus 702. The memory 701 stores a computer program that can run on the processor 700. When the processor 700 runs the computer program, it executes the electric mining truck shift control method provided in any of the foregoing embodiments of this application.

[0081] The memory 701 may include high-speed random access memory (RAM) or non-volatile memory, such as at least one disk storage device. Communication between this system network element and at least one other network element is achieved through at least one communication interface 703 (which can be wired or wireless), such as the Internet, wide area network, local area network, or metropolitan area network.

[0082] Bus 702 can be an ISA bus, PCI bus, or EISA bus, etc. Buses can be divided into address buses, data buses, control buses, etc. Memory 701 is used to store programs. After receiving execution instructions, processor 700 executes the programs. The electric mining truck shift control method disclosed in any of the aforementioned embodiments of this application can be applied to processor 700, or implemented by processor 700.

[0083] The processor 700 may be an integrated circuit chip with signal processing capabilities. In implementation, each step of the above method can be completed by the integrated logic circuitry in the hardware of the processor 700 or by instructions in software form. The processor 700 may be a general-purpose processor, including a central processing unit (CPU), a network processor (NP), etc.; it may also be a digital signal processor (DSP), an application-specific integrated circuit (ASIC), an off-the-shelf programmable gate array (FPGA), or other programmable logic devices, discrete gate or transistor logic devices, or discrete hardware components. It can implement or execute the methods, steps, and logic block diagrams disclosed in the embodiments of this application. The general-purpose processor may be a microprocessor or any conventional processor. The steps of the methods disclosed in the embodiments of this application can be directly embodied in the execution of a hardware decoding processor, or executed by a combination of hardware and software modules in the decoding processor. The software modules may reside in random access memory, flash memory, read-only memory, programmable read-only memory, electrically erasable programmable memory, registers, or other mature storage media in the art. The storage medium is located in memory 701. Processor 700 reads the information in memory 701 and, in conjunction with its hardware, completes the steps of the above method.

[0084] The electronic device provided in this application embodiment and the electric mining truck shift control method provided in this application embodiment are based on the same inventive concept and have the same beneficial effects as the methods they adopt, operate or implement.

[0085] According to another aspect of the embodiments of this application, a computer-readable storage medium corresponding to the electric mining truck shift control method provided in the foregoing embodiments is also provided, wherein a computer program (i.e., a program product) is stored thereon, and when the computer program is run by a processor, it executes the electric mining truck shift control method provided in any of the foregoing embodiments.

[0086] It should be noted that examples of computer-readable storage media may also include, but are not limited to, phase-change memory (PRAM), static random access memory (SRAM), dynamic random access memory (DRAM), other types of random access memory (RAM), read-only memory (ROM), electrically erasable programmable read-only memory (EEPROM), flash memory or other optical and magnetic storage media, which will not be elaborated here.

[0087] The computer-readable storage medium provided in the above embodiments of this application and the electric mining truck shift control method provided in the embodiments of this application are based on the same inventive concept and have the same beneficial effects as the methods adopted, run or implemented by the application programs stored therein.

[0088] The technical features of the above embodiments can be combined in any way. For the sake of brevity, not all possible combinations of the technical features in the above embodiments are described. However, as long as there is no contradiction in the combination of these technical features, they should be considered to be within the scope of this specification.

[0089] The above embodiments merely illustrate several implementation methods of the present invention, and their descriptions are relatively specific and detailed, but they should not be construed as limiting the scope of the present invention. It should be noted that those skilled in the art can make various modifications and improvements without departing from the concept of the present invention, and these all fall within the protection scope of the present invention. Therefore, the protection scope of this patent should be determined by the appended claims.

Claims

1. A method for shifting gears in an electric mining truck, characterized in that, include: A system dynamics model is constructed based on the vehicle electric drive assembly structure; During the active synchronization phase of the gear shifting process, a physical information neural network is trained using the system dynamics model as a physical constraint. The trained physical information neural network is embedded into the model predictive control framework as a prediction model, and the optimal control sequence is obtained by rolling optimization based on a preset multi-objective optimization function. Control is performed based on the optimal control sequence.

2. The method according to claim 1, characterized in that, A system dynamics model is constructed based on the vehicle electric drive assembly structure, including: Analysis of the structural parameters of the vehicle electric drive assembly; Based on the structural parameters of the electric drive assembly, a longitudinal dynamics model of the vehicle is established. The longitudinal dynamics model of the vehicle integrates the output torque of multiple drive motors, transmission ratio, rolling resistance, air resistance, total vehicle mass, wheel radius, slope resistance and braking torque parameters. Based on the vehicle longitudinal dynamics model, the system dynamics model for the active synchronization phase is determined.

3. The method according to claim 1, characterized in that, In the active synchronization phase of the gear shifting process, before training the physical information neural network using the system dynamics model as a physical constraint, the following steps are also included: The gear shifting process is divided into the torque unloading stage, the disengagement stage, the active synchronization stage, the gear engagement stage, and the torque recovery stage. The status of the gear shifting process is monitored in real time to determine whether the vehicle has entered the active synchronization phase.

4. The method according to claim 1, characterized in that, Using the system dynamics model as physical constraints, a physical information neural network is trained, including: Based on the system dynamics model, state decoupling is performed to establish system state equations that include state variables, control variables, and outputs; Discretize the system state equations to obtain discrete state transition equations; A physical information neural network is trained based on the discrete state transition equations as physical constraints.

5. The method according to claim 4, characterized in that, Based on the discrete state transition equations as physical constraints, a physical information neural network is trained, including: Substitute the predicted output value of the physical information neural network into the discrete state transition equation to calculate the physical loss; Based on a dataset containing system state, control input, and measured output, calculate the data loss; Based on the physical loss and data loss, a composite loss function is constructed; The composite loss function is optimized using the backpropagation algorithm, and the network weight parameters are updated to obtain the trained physical information neural network.

6. The method according to claim 1, characterized in that, Before performing rolling optimization based on the preset multi-objective optimization function, the following steps are also included: Determine the control objectives of the model predictive control framework, including speed synchronization accuracy, angle synchronization accuracy, shift smoothness, and minimum transient torsional vibration. The multi-objective optimization function is obtained by weighting and summing the various control objectives based on the quadratic cost function and preset weight coefficients.

7. The method according to claim 1, characterized in that, The trained physical information neural network is embedded as a prediction model into the model predictive control framework. Rolling optimization is performed based on a preset multi-objective optimization function to obtain the optimal control sequence, including: The physical information neural network is used as a prediction model and embedded in the model prediction control framework. At each rolling optimization moment of model predictive control, the system state is predicted by the physical information neural network based on the current system state and the control input sequence in the future control time domain, resulting in a state prediction sequence; Based on the state prediction sequence, combined with the multi-objective optimization function and system constraints, the finite-time optimal control problem is solved online in each control cycle to obtain the optimal control sequence.

8. A gear shifting control device for an electric mining truck, characterized in that, include: The dynamics model building module is used to build system dynamics models based on the vehicle electric drive assembly structure. The neural network training module is used to train a physical information neural network with the system dynamics model as the physical constraint during the active synchronization phase of the gear shifting process. The optimization and solution module is used to embed the trained physical information neural network as a prediction model into the model predictive control framework, and perform rolling optimization based on a preset multi-objective optimization function to obtain the optimal control sequence. The shift control module is used to perform control based on the optimal control sequence.

9. An electronic device, characterized in that, It includes a processor and a memory storing program instructions, the processor being configured to execute the shift control method for mining electric trucks as described in any one of claims 1 to 7 when executing the program instructions.

10. A computer-readable medium, characterized in that, It stores computer-readable instructions that are executed by a processor to implement a shift control method for a mining electric truck as described in any one of claims 1 to 7.

Images