Axial flux motor multi-modal adaptive torque control method and system

Abstract

This invention discloses a multi-modal adaptive torque control method and system for an axial flux motor, relating to the field of motor control. The method includes: establishing bus communication between an onboard sensor array, an onboard AI chip, and a control center; performing multi-sensor synchronous acquisition of external motion and the motor's magnetic space to determine first macroscopic scale information and a second magnetic field state; constructing a torque control model and deploying it on the onboard AI chip; performing torque mode matching based on the first macroscopic scale information; performing trajectory decision-making under magnetic field topology phase transformation; determining the motor shaft torque parameters by comparing with the second magnetic field state and making differential control decisions; generating torque control commands and executing control drive management. This application solves the technical problems of insufficient multi-condition torque adaptation capability and low matching accuracy between magnetic field state and torque output in existing torque control systems, achieving the technical effect of improving multi-condition torque adaptation capability and increasing the matching accuracy between magnetic field state and torque output.

Description

Technical Field

[0001] This application relates to the field of motor control, and in particular to a multi-modal adaptive torque control method and system for axial flux motors. Background Technology

[0002] Precise torque control of axial flux motors plays a crucial role in improving the power performance, energy efficiency, and driving stability of new energy vehicles, directly impacting the vehicle's range and driving experience. Currently, the mainstream technical methods for solving the torque control problem of axial flux motors mostly employ a torque lookup table control strategy based on a pre-set MAP (Motor Mapping Map). Specifically, this involves calibrating the relationship between motor speed, current, and torque under different operating conditions to form a fixed MAP data table. During control, the data in the table is retrieved based on real-time collected speed and current parameters, and the corresponding torque control command is output. However, existing methods rely on statically calibrated MAP data, which cannot dynamically adapt to complex and changing actual driving conditions and motor magnetic field states. This results in a single torque mode, making it difficult to meet the needs of multiple scenarios. Furthermore, the low matching degree between the real-time magnetic field state and the target torque can easily lead to problems such as excessive torque pulsation and increased motor losses.

[0003] At present, the torque control of axial flux motors has technical problems such as insufficient torque adaptability under multiple working conditions and low matching accuracy between magnetic field state and torque output. Summary of the Invention

[0004] This application provides a multi-modal adaptive torque control method and system for axial flux motors. It employs a bus communication mechanism between an onboard sensor array, an onboard AI chip, and a control center. The sensor array collects external motion and motor magnetic space data at the same frequency to obtain first macroscopic information and second magnetic field state. A torque control model is constructed and deployed on the onboard AI chip. Based on the first macroscopic information, torque modes are matched, and trajectories under magnetic field topology phase transitions are planned. Differential control decisions are made in conjunction with the second magnetic field state to determine the motor shaft torque parameters. Control commands are generated based on the torque parameters to manage the axial flux motor's control and drive. This approach solves the technical problems of insufficient multi-condition torque adaptation capability and low matching accuracy between magnetic field state and torque output in existing axial flux motor torque control systems, achieving the technical effect of improving multi-condition torque adaptation capability and increasing the matching accuracy between magnetic field state and torque output.

[0005] This application provides a multi-modal adaptive torque control method for an axial flux motor, comprising: establishing bus communication between an on-board sensor array, an on-board AI chip, and a control center; using the on-board sensor array to perform multi-sensor synchronous acquisition of external motion and the motor's magnetic space to determine a first macroscopic scale information and a second magnetic field state; constructing a torque control model and embedding it in the on-board AI chip, performing torque mode matching based on the first macroscopic scale information, performing trajectory decision-making under magnetic field topology phase transformation, and determining the motor shaft torque parameters by comparing with the second magnetic field state and performing differential control decision-making; generating torque control commands based on the motor shaft torque parameters, and executing control drive management of the axial flux motor.

[0006] In a possible implementation, a torque control model is constructed and embedded in an onboard AI chip, performing the following processing: A top-level processing layer is deployed based on macroscopic information as input and torque mode combinations as the decision objective. This macroscopic information includes vehicle dynamics data and road condition data. A mid-level processing layer is deployed using inverse process analysis based on mapping relationships, with the torque mode decomposition into a target distribution trajectory of magnetic field topological invariants as the decision objective. A bottom-level processing layer is deployed based on control counter-differential analysis of the target distribution trajectory under real-time magnetic field conditions. A torque control model is built based on the top-level, mid-level, and bottom-level processing layers, and this model is then deployed in the onboard AI chip.

[0007] In possible implementations, the following processing is performed: establishing a mapping relationship between the magnetic field topological phase and the torque mode; wherein the mapping relationship includes: if the magnetic field topological phase is a trivial phase with uniformly parallel magnetic field lines, the torque mode is a smooth torque mode; if the magnetic field topological phase is a vortex lattice phase with regularly arranged magnetic field lines in a vortex array, the torque mode exhibits a high-frequency micro-amplitude pulsating mode; if the magnetic field topological phase is a topological insulator phase with a flat bulk magnetic field and magnetic field line channels at the edges, the torque mode exhibits a low-loss stable mode; if the magnetic field topological phase is a quantum spin liquid phase with highly entangled and disordered magnetic field lines and long-range entanglement, the torque mode exhibits an anti-interference mode.

[0008] In a possible implementation, torque mode matching is performed based on the first macroscopic scale information, and the following processing is performed: the first macroscopic scale information is input into the top-level processing layer, macroscopic features are extracted, and rule matching is performed with torque mode features to determine torque mode instructions. The torque mode instructions include torque mode type, macroscopic torque target value, and key mode parameters. The matching result is a single torque mode or a combination of torque modes. If it is a combination of torque modes, each torque mode identifier has a relative proportion.

[0009] In a possible implementation, trajectory decision-making under magnetic field topology phase transition is performed. By comparing with the second magnetic field state and making differential control decisions, the following processing is performed: the torque mode command is input into the middle processing layer, and the target magnetic field topology phase is determined according to the mapping relationship; for the target magnetic field topology phase, the target distribution trajectory of magnetic flux density in the air gap and stator teeth is measured, wherein the target distribution trajectory includes the spatial distribution target of air gap magnetic flux density, the magnetic flux density saturation target of stator teeth, and the upper limit of allowable local vortex density.

[0010] In a possible implementation, the motor shaft torque parameters are determined, and the following processing is performed: the target distribution trajectory is input into the bottom processing layer, and the second magnetic field state is read; by comparing the second magnetic field state with the target distribution trajectory, the multiphase winding current, rotor position and speed of the axial flux motor are used as control objects, and differential control analysis and optimization are performed to determine the motor shaft torque parameters.

[0011] In a possible implementation, after determining the motor shaft torque parameters, the following processing is performed: by decoupling the left and right halves of the motor shaft torque parameters, the first parameter part and the second parameter part are determined; for the same motion condition, the yaw torque compensation amount is determined by real-time yaw angle acquisition; based on the yaw torque compensation amount, an asymmetric fine-tuning command is generated, wherein the torque mode is not changed as a constraint; based on the asymmetric fine-tuning command, differential adjustment compensation is performed on the first parameter part and the second parameter part.

[0012] In a possible implementation, the following process is performed: generating a torque control command based on the motor shaft torque parameters; and issuing the torque control command based on the communication interaction between the on-board AI chip and the control center of the axial flux motor to execute the control drive of the axial flux motor.

[0013] In a possible implementation, after determining the motor shaft torque parameters, the following processing is performed: a unit amplitude constraint is determined based on the parameter change rate, thermodynamic limit, and magnetic field reconstruction speed; it is determined whether the motor shaft torque parameters meet the unit amplitude constraint; if not, the motor shaft torque parameters are processed in stages under the unit amplitude constraint to determine the motor shaft torque parameter sequence.

[0014] This application also provides a multi-modal adaptive torque control system for an axial flux motor, comprising: a bus communication establishment module for establishing bus communication between the vehicle-mounted sensor array, the vehicle-mounted AI chip, and the control center; a sensor frequency acquisition module for performing multi-sensor frequency acquisition of external motion and motor magnetic space through the vehicle-mounted sensor array to determine the first macroscopic scale information and the second magnetic field state; a motor shaft torque parameter determination module for constructing a torque control model and embedding it in the vehicle-mounted AI chip, performing torque mode matching based on the first macroscopic scale information, performing trajectory decision-making under magnetic field topology phase transformation, and determining the motor shaft torque parameters by comparing with the second magnetic field state and making differential control decisions; and a motor control drive management module for generating torque control commands based on the motor shaft torque parameters and performing control drive management of the axial flux motor.

[0015] The proposed method and system for multi-modal adaptive torque control of an axial flux motor first establishes bus communication between an onboard sensor array, an onboard AI chip, and a control center. Then, through the onboard sensor array, it performs multi-sensor synchronous acquisition of external motion and the motor's magnetic space to determine a first macroscopic scale information and a second magnetic field state. Next, by constructing a torque control model and embedding it into the onboard AI chip, torque mode matching is performed based on the first macroscopic scale information, and trajectory decision-making is executed under magnetic field topology phase transformation. By comparing and differentially controlling the results with the second magnetic field state, the motor shaft torque parameters are determined. Finally, torque control commands are generated based on the motor shaft torque parameters to execute the control and drive management of the axial flux motor. Through the above process, the proposed method and system achieve the technical effects of improving multi-condition torque adaptability and enhancing the matching accuracy between magnetic field state and torque output. Attached Figure Description

[0016] To more clearly illustrate the technical solutions of the embodiments of the present invention, the accompanying drawings of the embodiments of the present invention will be briefly described below. Flowcharts are used in this application to illustrate the operations performed by the system according to the embodiments of the present application. It should be understood that the preceding or following operations are not necessarily performed precisely in sequence. Instead, various steps can be processed in reverse order or simultaneously as needed. Furthermore, other operations can be added to these processes, or one or more steps can be removed from these processes.

[0017] Figure 1 This is a flowchart illustrating the multi-mode adaptive torque control method for an axial flux motor provided in an embodiment of this application.

[0018] Figure 2 A schematic diagram of the structure of the axial flux motor multimodal adaptive torque control system provided in the embodiments of this application.

[0019] Explanation of reference numerals in the attached diagram: Bus communication establishment module 10, sensor frequency acquisition module 20, motor shaft torque parameter determination module 30, motor control and drive management module 40. Detailed Implementation

[0020] To further illustrate the technical means and effects of the present invention in achieving its intended purpose, the following detailed description of the specific implementation methods, structures, features, and effects of the present invention, in conjunction with the accompanying drawings and preferred embodiments, is provided below.

[0021] This application provides a multi-mode adaptive torque control method for axial flux motors, such as... Figure 1 As shown, the method includes: Step S100: Establish bus communication between the vehicle-mounted sensor array, the vehicle-mounted AI chip, and the control center.

[0022] Specifically, vehicle Ethernet is selected as the bus communication carrier, and communication parameters are configured based on a deterministic communication protocol to ensure that the communication latency between the sensor array, AI chip and control center is less than 1 millisecond and the data transmission packet loss rate is 0.

[0023] Step S200: Through the vehicle-mounted sensor array, perform multi-sensor synchronous acquisition of external motion and motor magnetic space to determine the first macroscopic scale information and the second magnetic field state.

[0024] Specifically, the sensor array is divided into an external motion sensing unit and a motor magnetic space sensing unit. For example, the external motion sensing unit incorporates a six-axis inertial measurement unit (IMU), wheel speed sensors, tire pressure sensors, millimeter-wave radar, and a camera. The IMU collects the vehicle's longitudinal acceleration, lateral acceleration, and yaw rate; the wheel speed sensors collect the rotational speeds of the four wheels; the millimeter-wave radar collects the relative distance and relative speed to the vehicle in front; and the camera uses image recognition to determine the road surface type, such as asphalt, concrete, or icy / snowy surfaces. The motor magnetic space sensing unit deploys Hall effect sensors and fluxgate sensors at the stator teeth and air gap locations. The Hall effect sensors collect the radial magnetic flux density of the air gap magnetic field, and the fluxgate sensors collect the core magnetic flux density of the stator teeth. A high-precision clock synchronization module is used to achieve synchronous acquisition for both types of units, ensuring the acquisition frequency matches the motor's electrical frequency. All sensor data are synchronously transmitted to the onboard AI chip via the vehicle Ethernet. The first macroscopic information is vehicle dynamics and road condition data collected by the external motion sensing unit, including the vehicle's longitudinal acceleration, yaw rate, wheel speed, road surface type, and relative distance to the vehicle in front. The second magnetic field state is the internal magnetic field parameters of the motor collected by the motor magnetic space sensing unit, including air gap radial magnetic flux density distribution data and stator tooth core magnetic flux density data.

[0025] Step S300: By constructing a torque control model and embedding it into the vehicle AI chip, torque mode matching is performed based on the first macroscopic scale information, trajectory decision is executed under magnetic field topology phase transition, and the motor shaft torque parameters are determined by comparing with the second magnetic field state and making differential control decisions.

[0026] Specifically, the torque control model adopts a three-layer processing architecture: top-level, middle-level, and bottom-level. It is based on a deep learning framework and a motor electromagnetic simulation tool for joint modeling. After training, the model size is compressed to less than 5MB to meet the embedded deployment requirements of automotive AI chips. The model inputs are first macroscopic information and second magnetic field state, and the output is the motor shaft torque parameters. The model first completes torque mode matching through the top-level processing layer, then completes the conversion from magnetic field topology phase to target magnetic field trajectory through the middle-level processing layer, and finally completes the differential analysis of the real-time magnetic field and target trajectory through the bottom-level processing layer, outputting the torque parameters.

[0027] In one possible implementation, a torque control model is constructed and embedded in an onboard AI chip. Step S300 further includes step S310, which involves deploying a top-level processing layer based on macroscopic information as input and torque mode combinations as the decision objective. The macroscopic information includes vehicle dynamics data and road condition data. Specifically, the top-level processing layer adopts an architecture combining rule-based reasoning and a deep learning classifier. First, a torque mode matching rule library is constructed, containing torque modes corresponding to different road conditions and vehicle dynamics characteristics. Then, a convolutional neural network classifier is built to extract features and classify modes from the macroscopic information. Rule-based reasoning serves as the basic decision-making module, and the deep learning classifier serves as the optimization and correction module. The decision results of both are weighted and fused to output the final torque mode command, with the weight coefficients dynamically adjusted based on the rule matching confidence level.

[0028] For example, the torque modal matching rule base defines: high-efficiency cruising condition (road condition is asphalt road, vehicle speed is 60 to 120 km / h, acceleration is 0 to 0.5 m / s²) 2 Corresponding to smooth torque mode; anti-skid condition for crushed and vibrating road surfaces (road conditions include icy and snowy surfaces, wheel speed difference greater than 10%, and acceleration less than 0.2 m / s²). 2 This corresponds to the high-frequency micro-amplitude pulsation mode; extreme impact conditions (road conditions include potholes and longitudinal acceleration greater than 5 m / s²). 2The first mode corresponds to low-loss stable mode; the second mode corresponds to complex off-road conditions (gravel road surface, yaw rate greater than 50° / s). The input layer of the convolutional neural network classifier has 12 dimensions, including longitudinal acceleration, lateral acceleration, yaw rate, four wheel speeds, road surface type encoding, and vehicle speed. The hidden layer contains two convolutional layers and two fully connected layers. The output layer is the probability distribution of four torque modes. When the macroscopic information is a vehicle speed of 80 km / h, asphalt road surface, and longitudinal acceleration of 0.3 m / s², the model is suitable for the following conditions: vehicle speed 80 km / h, asphalt road surface, and longitudinal acceleration 0.3 m / s². 2 At that time, the rule-based reasoning matched the smooth torque mode with a confidence level of 95%, and the deep learning classifier output the smooth torque mode with a probability of 98%. After weighted fusion, the final decision was the smooth torque mode, with the weight coefficients set to 0.6 for rule-based reasoning and 0.4 for classifier.

[0029] Step S320: Through inverse process analysis based on mapping relationships, the mid-level processing layer is deployed with the target distribution trajectory of the torque mode decomposition into magnetic field topological invariants as the decision objective. Specifically, the mid-level processing layer adopts an architecture combining a magnetic field topological phase-torque mode mapping model with electromagnetic field finite element simulation. First, a one-to-one mapping relationship between torque modes and magnetic field topological phases is established. Then, through electromagnetic field finite element simulation tools, simulation analysis is performed on each magnetic field topological phase to extract magnetic field topological invariants, such as magnetic flux density distribution, magnetic field line topology, and vortex density. Based on these invariants, a conversion model from the magnetic field topological phase to the target magnetic field distribution trajectory is constructed. This model uses polynomial fitting, taking the magnetic field topological invariants as input and outputting the target magnetic field distribution trajectory of the air gap and stator teeth.

[0030] For example, using the ANSYS Maxwell electromagnetic field finite element simulation tool, four magnetic field topological phases were simulated: the trivial phase corresponding to the smooth torque mode, the simulation showed a uniform distribution of air gap magnetic flux density of 0.8 Tesla, and a stator tooth magnetic flux density of 1.2 Tesla; the vortex lattice phase corresponding to the high-frequency micro-amplitude pulsation mode, the simulation showed that the air gap magnetic flux density was distributed in a vortex array, the vortex center spacing was 5 mm, and the vortex intensity was 0.2 Tesla; the topological insulator phase corresponding to the low-loss stable mode, the simulation showed that the volume magnetic field flux density was 0.5 Tesla, and the magnetic flux density of the edge magnetic field line channels was 1.0 Tesla; and the quantum spin liquid phase corresponding to the anti-interference mode, the simulation showed that the magnetic field lines were randomly entangled, and the long-range entanglement length was 10 mm.

[0031] Step S330: Using the control counter-differential analysis based on the target distribution trajectory under real-time magnetic field conditions as the underlying logic, a low-level processing layer is deployed. Specifically, the low-level processing layer adopts an architecture combining model predictive control and PID control. First, a magnetic field deviation calculation module is constructed to calculate the deviation value between the real-time magnetic field state and the target distribution trajectory. For example, the magnetic field deviation calculation module calculates the deviation value ΔB=B of the air gap magnetic flux density. target-B real B target For the target magnetic flux density, B real The real-time magnetic flux density is used. Then, using the magnetic field deviation as input, model predictive control (MMC) predicts the magnetic field state at multiple future time points and optimizes to obtain the optimal control sequence. MMC is based on the equivalent magnetic circuit model of an axial flux motor, which includes parameters such as stator resistance, inductance, and air gap reluctance. PID control, as a feedback correction module, fine-tunes the output of MMC, ultimately outputting the control parameters for the motor winding current. The prediction time domain of MMC is set to 10 steps, the control time domain to 5 steps, and the objective function is to minimize the sum of squares of the magnetic field deviation.

[0032] Step S340: Based on the top-level processing layer, middle-level processing layer, and bottom-level processing layer, a torque control model is built and deployed on the vehicle-mounted AI chip. Specifically, a modular programming method is used, encapsulating the top-level, middle-level, and bottom-level processing layers into independent functional modules. These modules interact with each other through standardized interfaces in JSON format. The overall model is then jointly debugged to verify the correctness of data transmission and logical connections between modules. Finally, a model deployment toolchain is used to convert the debugged model into an inference format supported by the vehicle-mounted AI chip, while configuring the chip's computing resources and allocating independent computing cores for torque control model inference.

[0033] For example, modular programming using Python is employed. The top-level processing layer module is named Top_Layer.py, the middle-level processing layer module is named Middle_Layer.py, and the bottom-level processing layer module is named Bottom_Layer.py. The interface input and output data between modules includes macroscopic scale information, torque mode commands, target magnetic field trajectory, magnetic field deviation, torque parameters, etc. During joint debugging, 100,000 sets of test datasets are input to verify that the error between the torque parameters output by the model and the actual motor test data is less than 5%. The chip's ATC toolchain is used to convert the TensorFlow Lite format model into an OM format inference model. Two computing cores of the chip are configured for model inference, and 256MB of memory is allocated to ensure the real-time performance of model inference.

[0034] In one possible implementation, step S320 further includes step S321, establishing a mapping relationship between the magnetic field topological phase and the torque mode; wherein the mapping relationship includes: if the magnetic field topological phase is a trivial phase with uniformly parallel magnetic field lines, the torque mode is a smooth torque mode; if the magnetic field topological phase is a vortex lattice phase with regularly arranged magnetic field lines in a vortex array, the torque mode exhibits a high-frequency micro-amplitude pulsating mode; if the magnetic field topological phase is a topological insulator phase with a flat bulk magnetic field and magnetic field line channels at the edges, the torque mode exhibits a low-loss stable mode; if the magnetic field topological phase is a quantum spin liquid phase with highly entangled and disordered magnetic field lines and long-range entanglement, the torque mode exhibits an anti-interference mode.

[0035] Specifically, a mapping relationship was established by combining experimental measurement and simulation verification. First, an axial flux motor experimental platform was built. By changing the amplitude, frequency and phase of the motor winding current, the magnetic field topology inside the motor was adjusted. Then, the torque characteristics of the motor shaft end under different magnetic field topologies were measured by a torque sensor. Finally, the experimental data were fitted and analyzed by combining the electromagnetic field simulation results to determine the characteristic parameters of each magnetic field topology phase and the performance parameters of the corresponding torque mode, forming a quantitative mapping relationship table.

[0036] For example, the axial flux motor experimental platform includes the motor body, torque sensor, current source, and magnetic field measurement system. During the experiment, the current source is adjusted to output a sinusoidal current of constant amplitude, so that the magnetic field lines inside the motor are uniformly and parallelly distributed. The measured torque amplitude is stable at 300 N·m, and the torque pulsation amplitude is less than ±5 N·m. This magnetic field topological phase is defined as the trivial phase, corresponding to a smooth torque mode, suitable for efficient cruise conditions. The current source is adjusted to output a multi-phase superimposed pulsed current, so that the magnetic field lines inside the motor are distributed in a regular vortex array. The measured torque amplitude is 200 N·m, the pulsation frequency is 50 Hz, and the pulsation amplitude is ±20 N·m. This magnetic field topological phase is defined as the vortex lattice phase, corresponding to a high-frequency micro-amplitude pulsation mode, suitable for anti-skid conditions on crushed and vibrating road surfaces. By adjusting the current source output of a multiphase stepped wave current, the magnetic flux density of the internal magnetic field of the motor is maintained at a low stable value, forming concentrated magnetic field line channels in the edge region. The measured torque amplitude is 350 N·m, and the torque pulsation amplitude is less than ±3 N·m. The motor iron loss is reduced by 15%. This magnetic field topology is defined as a topological insulator phase, corresponding to a low-loss stable mode, suitable for extreme impact conditions. By adjusting the current source output of a multi-band random pulse current, the magnetic field lines inside the motor exhibit a highly entangled and disordered distribution with long-range entanglement characteristics. The measured torque amplitude is 280 N·m, and the torque pulsation amplitude is ±10 N·m. Under an external magnetic field interference intensity of 0.2 Tesla, the torque fluctuation rate is still less than 2%. This magnetic field topology is defined as a quantum spin liquid phase, corresponding to an anti-interference mode, suitable for complex off-road conditions. Through repeated verification through experiments and simulations, a mapping table is established, containing fields such as magnetic field topology phase type, magnetic field line characteristics, torque mode type, suitable scenarios, and torque performance parameters.

[0037] In one possible implementation, torque mode matching is performed based on the first macroscopic scale information. Step S300 further includes step S350, whereby the first macroscopic scale information is input into the top-level processing layer, and macroscopic features are extracted and matched with torque mode features according to rules to determine torque mode instructions. The torque mode instructions include torque mode type, macroscopic torque target value, and key mode parameters. The matching result is a single torque mode or a combination of torque modes. If it is a combination of torque modes, each torque mode identifier has a relative proportion.

[0038] Specifically, a macroscopic feature extraction module is constructed, employing principal component analysis to reduce the dimensionality of the first macroscopic scale information and extract key features such as road surface adhesion coefficient, vehicle power demand, and driving stability indicators. Then, a torque modal feature library is built, containing feature vectors for each torque mode. The similarity between the extracted macroscopic features and the feature vectors in the modal feature library is calculated using cosine similarity; rule matching is triggered when the similarity is greater than or equal to 90%. For multi-modal combinations, weighted voting is used to determine the proportion of each mode, with weights calculated based on feature similarity. The torque mode type includes one or a combination of smooth torque mode, high-frequency micro-amplitude pulsation mode, low-loss stable mode, and anti-interference mode; the macroscopic torque target value is the basic torque amplitude that the motor needs to output; and key modal parameters include the pulsation frequency and amplitude corresponding to the high-frequency micro-amplitude pulsation mode.

[0039] In one possible implementation, trajectory decision-making under magnetic field topology phase transition is performed. This involves comparison with the second magnetic field state and differential control decision-making. Step S300 further includes step S360, where the torque mode command is input to the middle processing layer, and the target magnetic field topology phase is determined based on the mapping relationship. Specifically, a mapping relationship query module is constructed to store the mapping relationship between torque modes and magnetic field topology phases as a key-value pair database table. Then, the mode type in the torque mode command is used as the query key to retrieve the corresponding magnetic field topology phase from the database table. For combinations of torque modes, magnetic field topology phase superposition is used. The target magnetic field topology phase after superposition is calculated based on the proportion of each mode, with the superposition rule based on a weighted summation of magnetic field energy.

[0040] For example, the key of the mapping relation database table is the torque mode type, and the value is the magnetic field topological phase type. The table stores: smooth torque mode - trivial phase, high-frequency micro-amplitude pulsating mode - vortex lattice phase, low-loss stable mode - topological insulator phase, and anti-interference mode - quantum spin liquid phase. When the torque mode command is a high-frequency micro-amplitude pulsating mode, the query module outputs the target magnetic field topological phase as a vortex lattice phase. When the torque mode command is a combination of anti-interference mode (60%) and low-loss stable mode (40%), the weighted summation method of magnetic field energy is used. The magnetic field energy of the quantum spin liquid phase corresponding to the anti-interference mode is E1, and the magnetic field energy of the topological insulator phase corresponding to the low-loss stable mode is E2. The energy of the target magnetic field topological phase after superposition is E = 0.6 × E1 + 0.4 × E2. Electromagnetic field simulation determines that the superimposed magnetic field topological structure is a mixed topological phase with quantum spin liquid phase as the main component and topological insulator phase as the auxiliary component.

[0041] Step S370: For the target magnetic field topological phase, measure the target distribution trajectory of magnetic flux density in the air gap and stator teeth. This target distribution trajectory includes the spatial distribution target of the air gap magnetic flux density, the magnetic flux density saturation target of the stator teeth, and the upper limit of the allowable local vortex density. Specifically, an electromagnetic field finite element simulation tool is used to simulate the target magnetic field topological phase, outputting a magnetic flux density distribution cloud map of the air gap and stator teeth. Based on the distribution cloud map, the spatial distribution function of the magnetic flux density is extracted, and a curve fitting method is used to obtain the functional expression of the air gap magnetic flux density as a function of radial and circumferential coordinates. Simultaneously, based on the material properties of the motor core, such as the magnetization curve of silicon steel sheets, the magnetic flux density saturation target of the stator teeth is determined, for example, set to 80% of the core saturation magnetic flux density to avoid increased losses due to core saturation. For the magnetic field topological phase with vortex structures, the local vortex density is calculated, and based on the principle of optimal motor efficiency, the upper limit of the allowable vortex density is determined.

[0042] For example, for the target magnetic field topological phase of a vortex lattice phase, ANSYS Maxwell simulation was used to obtain the distribution contour map of the air gap magnetic flux density. The contour map shows that the air gap magnetic flux density exhibits a periodic vortex distribution. The spatial distribution function of the air gap magnetic flux density was obtained through curve fitting. Where r is the radial coordinate and θ is the circumferential coordinate. The silicon steel sheet used in the stator teeth is model 50WW470, with a saturation magnetic flux density of 1.8 Tesla. The target magnetic flux density saturation of the stator teeth is set to 1.44 Tesla. The vortex density is calculated using a vortex identification method, counting the number of vortices per unit area. The upper limit of the allowable local vortex density is set to 10 vortices / cm². 2 This ensures that the vortex structure does not cause motor losses to exceed 5% of the rated losses.

[0043] In one possible implementation, after determining the motor shaft torque parameters, step S300 further includes step S380, inputting the target distribution trajectory into the underlying processing layer and reading the second magnetic field state. Specifically, a data input interface module is constructed, employing DMA (Direct Memory Access) technology to transfer the target distribution trajectory data, including the spatial distribution function, magnetic flux density saturation target, and vortex density upper limit, from the memory of the onboard AI chip to the buffer of the underlying processing layer. Then, a real-time magnetic field state reading module is constructed to acquire real-time data of the second magnetic field state from the sensor array via the onboard Ethernet. The input target distribution trajectory data and real-time magnetic field state data are standardized and uniformly converted into a floating-point data format.

[0044] Step S390: By comparing the second magnetic field state with the target distribution trajectory, and taking the multiphase winding current, rotor position, and speed of the axial flux motor as the control objects, differential control analysis and optimization are used to determine the motor shaft torque parameters. Specifically, a deviation analysis module is constructed to calculate various deviation indices between the real-time magnetic field state and the target distribution trajectory, including air gap flux density deviation, stator tooth flux density deviation, and vortex density deviation. Then, a differential control model is constructed, which is based on the mathematical model of the axial flux motor and includes voltage equations, flux linkage equations, and torque equations. Using the deviation indices as input, model predictive control is employed to optimize the amplitude, frequency, and phase of the multiphase winding current. Simultaneously, the control parameters are adjusted in real time by combining the rotor position and speed data collected by the rotor position sensor. Finally, the torque parameters at the motor shaft are calculated using the torque equations, including torque amplitude, torque pulsation frequency, and pulsation amplitude.

[0045] In one possible implementation, after determining the motor shaft torque parameters, the method further includes step S500, which involves decoupling the left and right half-motors based on the motor shaft torque parameters to determine a first parameter part and a second parameter part. Specifically, a torque decoupling model is constructed. This model is based on the structural characteristics of an axial flux motor, assuming that the electromagnetic parameters of the left and right half-motors are symmetrical, and the decoupling ratio is dynamically adjusted according to the vehicle stability requirements. The motor shaft torque parameters are decomposed into a first parameter part (left half-motor torque parameters) and a second parameter part (right half-motor torque parameters). The decomposition rule is based on the torque balance equation. When the vehicle has no yaw moment requirement, the torque parameters of the left and right half-motors are equal; when the vehicle has a yaw moment requirement, the torque difference between the left and right half-motors is adjusted according to the direction and magnitude of the yaw moment.

[0046] For example, the axial flux motor uses a dual-stator structure, with completely symmetrical electromagnetic parameters for the left and right halves of the motor. Both stator resistances are 0.1 ohms, and both inductances are 10 millihenries. When the motor shaft torque is 200 Nm and the pulsation frequency is 50 Hz, if the vehicle has no yaw torque requirement, the torque decoupling model outputs the first parameter (left half motor) as 100 Nm and 50 Hz, and the second parameter (right half motor) as 100 Nm and 50 Hz. If the vehicle requires positive yaw torque, the torque decoupling model adjusts the torque difference between the left and right half motors, outputting the first parameter as 120 Nm and 50 Hz, and the second parameter as 80 Nm and 50 Hz, thus generating positive yaw torque through the torque difference.

[0047] Step S600: For the same motion condition, the yaw moment compensation amount is determined by real-time yaw angle acquisition. Specifically, a six-axis inertial measurement unit is used to acquire the vehicle's real-time yaw angle and roll angle, and a steering angle sensor is used to acquire the steering wheel's steering angle. A yaw moment compensation model is constructed. This model is based on a two-degree-of-freedom model of vehicle dynamics, which includes the vehicle body's yaw motion and lateral motion. Using the yaw angle, roll angle, and steering angle as inputs, the deviation between the vehicle's ideal yaw rate and the actual yaw rate is calculated. The yaw moment compensation amount is determined based on the deviation value. The magnitude of the compensation amount is proportional to the deviation value, and the proportionality coefficient is determined based on the vehicle's wheelbase and center of gravity height.

[0048] Step S700: Based on the yaw moment compensation amount, an asymmetric fine-tuning command is generated, with the constraint of not changing the torque mode. Specifically, a fine-tuning command generation model is constructed, the constraint of which is that the torque modes of the left and right half-motors remain unchanged, i.e., the torque pulsation frequency and pulsation amplitude remain unchanged, only the difference in torque amplitude is adjusted. Based on the mapping relationship between the yaw moment compensation amount and the torque difference between the left and right half-motors, the torque amplitude fine-tuning amount of the left and right half-motors is calculated. The mapping relationship is based on the torque-torque conversion formula of vehicle dynamics, for example, M=ΔT×R, where M is the yaw moment compensation amount, ΔT is the torque difference between the left and right half-motors, and R is the wheel radius. Finally, an asymmetric fine-tuning command is generated, which includes the torque amplitude fine-tuning amount and adjustment direction of the left and right half-motors.

[0049] Step S800: According to the asymmetric fine-tuning instruction, differential adjustment compensation is performed on the first parameter portion and the second parameter portion. Specifically, a parameter adjustment module is constructed. This module receives the asymmetric fine-tuning instruction, extracts the torque amplitude fine-tuning amount of the left and right half-motors, and adjusts the torque amplitude of the first parameter portion (left half-motor) and the second parameter portion (right half-motor) according to the adjustment amount, while keeping the key parameters of the torque mode unchanged during the adjustment process. Finally, through differential control of the underlying processing layer, the adjusted torque parameters are converted into winding current control parameters of the left and right half-motors to achieve differential torque compensation.

[0050] In one possible implementation, after determining the motor shaft torque parameters, the method further includes step S900, which determines the unit amplitude adjustment constraint based on the parameter change rate, thermodynamic limit, and magnetic field reconstruction speed. Specifically, three constraint indicators are determined: first, the parameter change rate constraint, based on the mechanical characteristics of the motor, the torque amplitude change rate must not exceed 10% / second of the motor's rated torque to avoid sudden torque changes causing impact on the transmission system; second, the thermodynamic limit constraint, based on the motor's temperature rise characteristics, the adjustment of the torque amplitude must ensure that the motor winding temperature does not exceed the maximum temperature of the insulation class, and the torque adjustment amplitude is adjusted in real time by collecting the winding temperature through a temperature sensor; third, the magnetic field reconstruction speed constraint, based on the motor's electromagnetic characteristics, the reconstruction speed of the magnetic field topology phase must not exceed five times the motor's electrical frequency to avoid excessive torque pulsation caused by untimely magnetic field reconstruction. The most stringent of the three constraint indicators, i.e., the smallest value, is taken as the unit amplitude adjustment constraint.

[0051] Step S1000: Determine whether the motor shaft torque parameter satisfies the unit amplitude adjustment constraint. If not, perform phased processing on the motor shaft torque parameter under the unit amplitude adjustment constraint to determine the motor shaft torque parameter sequence. Specifically, a constraint judgment module is constructed to calculate the ratio of the initial motor shaft torque parameter adjustment amount to the unit amplitude adjustment constraint. If the ratio is less than or equal to 1, the constraint is satisfied, and the torque parameter is directly output; if the ratio is greater than 1, the constraint is not satisfied, and phased processing is required. The number of phases in the phased processing is equal to the initial torque adjustment amount divided by the unit amplitude adjustment constraint, and the result is rounded up. Based on the number of phases, the initial torque adjustment amount is evenly distributed to each phase to generate the motor shaft torque parameter sequence.

[0052] Step S400: Generate torque control command based on the motor shaft end torque parameters, and execute the control drive management of the axial flux motor.

[0053] Specifically, a command generation module is constructed to convert the motor shaft torque parameters into a command format supported by the motor control center. This command format includes parameters such as control mode, current amplitude, current frequency, and current phase. Then, a command sending module is constructed to send torque control commands to the motor control center via the vehicle's Ethernet network. Finally, a command feedback module is constructed to receive execution feedback data from the motor control center, verify the execution effect of the torque control commands, and if the error between the feedback data and the target torque parameters exceeds 5%, the torque parameters are readjusted and a new control command is generated.

[0054] In one possible implementation, step S400 further includes step S410, generating a torque control command based on the motor shaft torque parameters. Specifically, a mapping model between the torque parameters and control command parameters is established. This model is based on the electromagnetic-torque conversion relationship of the axial flux motor, converting the torque amplitude into the winding current amplitude, the torque pulsation frequency into the current frequency, and the torque pulsation amplitude into the current phase adjustment. The converted control parameters are then standardized to ensure they are within the allowable range of the motor control center. Finally, a torque control command containing the control mode, current parameters, and execution time is generated, using the standard format of the CAN bus protocol.

[0055] Step S420: Based on the communication interaction between the onboard AI chip and the control center of the axial flux motor, the torque control command is issued to execute the control drive of the axial flux motor. Specifically, the communication parameters between the onboard AI chip and the motor control center are configured, and the CAN bus communication protocol is used to ensure the real-time nature of the command issuance. A command issuance queue is constructed, and the torque control commands are sorted according to priority, which is set based on the vehicle's operating conditions. For example, commands for extreme impact conditions have the highest priority, followed by commands for efficient cruising conditions. After the command is issued, an execution monitoring module is constructed to collect the motor's operating parameters in real time and verify the execution effect of the command. If the error between the motor operating parameters and the target parameters exceeds a threshold, a command retransmission mechanism is triggered.

[0056] This application embodiment employs a bus communication mechanism between an onboard sensor array, an onboard AI chip, and a control center. It collects external motion and motor magnetic space data at the same frequency through the sensor array, acquiring first macroscopic information and second magnetic field state. A torque control model is constructed and deployed on the onboard AI chip. Based on the first macroscopic information, torque modes are matched, and trajectories under magnetic field topology phase transitions are planned. Differential control decisions are made in conjunction with the second magnetic field state to determine the motor shaft torque parameters. Control commands are generated based on the torque parameters, and control and drive management of the axial flux motor are performed. This approach solves the technical problems of insufficient multi-condition torque adaptation capability and low matching accuracy between magnetic field state and torque output in existing axial flux motor torque control systems, achieving the technical effect of improving multi-condition torque adaptation capability and increasing the matching accuracy between magnetic field state and torque output.

[0057] In the above text, refer to Figure 1 A multi-mode adaptive torque control method for an axial flux motor according to an embodiment of the present invention is described in detail. Next, reference will be made to... Figure 2 A multi-mode adaptive torque control system for an axial flux motor according to an embodiment of the present invention is described.

[0058] The multi-mode adaptive torque control system for axial flux motors according to embodiments of the present invention addresses the technical problems of insufficient multi-condition torque adaptation capability and low matching accuracy between magnetic field state and torque output in existing axial flux motor torque control systems, achieving the technical effects of improving multi-condition torque adaptation capability and enhancing the matching accuracy between magnetic field state and torque output. The multi-mode adaptive torque control system for axial flux motors includes: a bus communication establishment module 10, a sensor frequency acquisition module 20, a motor shaft end torque parameter determination module 30, and a motor control and drive management module 40.

[0059] The bus communication establishment module 10 is used to establish bus communication between the vehicle-mounted sensor array, the vehicle-mounted AI chip, and the control center; the sensor frequency acquisition module 20 is used to perform multi-sensor frequency acquisition of external motion and motor magnetic space through the vehicle-mounted sensor array to determine the first macroscopic scale information and the second magnetic field state; the motor shaft torque parameter determination module 30 is used to construct a torque control model and embed it in the vehicle-mounted AI chip, perform torque mode matching based on the first macroscopic scale information, perform trajectory decision under magnetic field topology phase transformation, and determine the motor shaft torque parameter by comparing it with the second magnetic field state and making differential control decisions; the motor control drive management module 40 is used to generate torque control commands based on the motor shaft torque parameters and perform control drive management of the axial flux motor.

[0060] The detailed description of the specific configuration of the motor shaft torque parameter determination module 30 is explained as follows: As mentioned above, the motor shaft torque parameter determination module 30, which constructs a torque control model and embeds it into the vehicle AI chip, may further include: a top-level processing layer deployment unit for deploying the top-level processing layer based on macroscopic information as input and torque mode combination as the decision target, wherein the macroscopic information includes vehicle dynamics data and road condition data; a middle-level processing layer deployment unit for deploying the middle-level processing layer by performing inverse process analysis based on mapping relationship and using the target distribution trajectory of torque mode decomposition into magnetic field topological invariants as the decision target; a bottom-level processing layer deployment unit for deploying the bottom-level processing layer based on the control counter-differential analysis of the target distribution trajectory under real-time magnetic field conditions; and a torque control model building unit for building a torque control model based on the top-level processing layer, middle-level processing layer, and bottom-level processing layer, and deploying the torque control model into the vehicle AI chip.

[0061] The middle-layer processing unit may further include: a mapping relationship establishment subunit for establishing a mapping relationship based on the magnetic field topological phase and the torque mode; wherein the mapping relationship includes: if the magnetic field topological phase is a trivial phase with uniformly parallel magnetic field lines, the torque mode is a smooth torque mode; if the magnetic field topological phase is a vortex lattice phase with regularly arranged magnetic field lines in a vortex array, the torque mode exhibits a high-frequency micro-amplitude pulsating mode; if the magnetic field topological phase is a topological insulator phase with a flat bulk magnetic field and magnetic field line channels at the edges, the torque mode exhibits a low-loss stable mode; if the magnetic field topological phase is a quantum spin liquid phase with highly entangled and disordered magnetic field lines and long-range entanglement, the torque mode exhibits an anti-interference mode.

[0062] The torque mode matching module 30 based on the first macroscopic scale information can further include: a rule matching unit for inputting the first macroscopic scale information into the top-level processing layer, extracting macroscopic features, performing rule matching with torque mode features, and determining torque mode instructions. The torque mode instructions include torque mode type, macroscopic torque target value, and key mode parameters. The matching result is a single torque mode or a combination of torque modes. If it is a combination of torque modes, each torque mode identifier has a relative proportion.

[0063] The trajectory decision-making under the magnetic field topology phase transition, through comparison with the second magnetic field state and differential control decision, the motor shaft torque parameter determination module 30 may further include: a target magnetic field topology phase determination unit for inputting the torque mode command into the middle layer processing layer and determining the target magnetic field topology phase according to the mapping relationship; and a target distribution trajectory measurement unit for measuring the target distribution trajectory of magnetic flux density in the air gap and stator teeth for the target magnetic field topology phase, wherein the target distribution trajectory includes the spatial distribution target of air gap magnetic flux density, the magnetic flux density saturation target of stator teeth, and the upper limit of allowable local vortex density.

[0064] The motor shaft torque parameter determination module 30 may further include: a second magnetic field state reading unit for inputting the target distribution trajectory into the underlying processing layer and reading the second magnetic field state; and a differential control analysis unit for determining the motor shaft torque parameter by comparing the second magnetic field state with the target distribution trajectory, taking the multiphase winding current, rotor position and speed of the axial flux motor as the control object, and through differential control analysis and optimization.

[0065] After determining the motor shaft torque parameters, the system may further include: a left and right half-motor decoupling module for determining a first parameter portion and a second parameter portion by decoupling the motor shaft torque parameters from the left and right half-motors; a yaw torque compensation amount determination module for determining the yaw torque compensation amount by real-time yaw angle acquisition for the same motion condition; an asymmetric fine-tuning command generation module for generating an asymmetric fine-tuning command based on the yaw torque compensation amount, wherein the constraint condition is not to change the torque mode; and a differential adjustment compensation module for performing differential adjustment compensation on the first parameter portion and the second parameter portion based on the asymmetric fine-tuning command.

[0066] The specific configuration of the motor control drive management module 40 is described in detail below: As mentioned above, the motor control drive management module 40 may further include: a torque control command generation unit for generating torque control commands based on the motor shaft end torque parameters; and a control drive unit for issuing the torque control commands and executing the control drive of the axial flux motor based on the communication interaction between the vehicle-mounted AI chip and the control center of the axial flux motor.

[0067] After determining the motor shaft torque parameters, the system may further include: a unit amplitude constraint determination module for determining the unit amplitude constraint based on the parameter change rate, thermodynamic limit, and magnetic field reconstruction speed; and a motor shaft torque parameter sequence determination module for determining whether the motor shaft torque parameters meet the unit amplitude constraint. If not, the motor shaft torque parameters are processed in stages under the unit amplitude constraint to determine the motor shaft torque parameter sequence.

[0068] The axial flux motor multi-mode adaptive torque control system provided in the embodiments of the present invention can execute the axial flux motor multi-mode adaptive torque control method provided in any embodiment of the present invention, and has the corresponding functional modules and beneficial effects of the execution method.

[0069] Although this application makes various references to certain modules in the system according to the embodiments of this application, any number of different modules can be used and run on user terminals and / or servers. The various units and modules included are only divided according to functional logic, but are not limited to the above division, as long as the corresponding functions can be achieved; in addition, the specific names of each functional unit are only for easy distinction between each other and are not used to limit the scope of protection of this invention.

[0070] The above description is merely a preferred embodiment of the present invention and is not intended to limit the present invention in any way. Although the present invention has been disclosed above with reference to preferred embodiments, it is not intended to limit the present invention. Any person skilled in the art can make some modifications or alterations to the above-disclosed technical content to create equivalent embodiments without departing from the scope of the present invention. Any modifications, equivalent changes, and alterations made to the above embodiments based on the technical essence of the present invention without departing from the scope of the present invention shall still fall within the scope of the present invention.

Claims

1. A multi-mode adaptive torque control method for an axial flux motor, characterized in that, The method includes: Establish bus communication between the vehicle-mounted sensor array, the vehicle-mounted AI chip, and the control center; By using an onboard sensor array, multi-sensor synchronous acquisition of external motion and motor magnetic space is performed to determine the first macroscopic scale information and the second magnetic field state. By constructing a torque control model and embedding it into an onboard AI chip, torque mode matching is performed based on the first macroscopic information, trajectory decision is executed under magnetic field topology phase transition, and the motor shaft torque parameters are determined by comparing with the second magnetic field state and making differential control decisions. Based on the torque parameters at the motor shaft end, a torque control command is generated to execute the control and drive management of the axial flux motor.

2. The multi-mode adaptive torque control method for an axial flux motor as described in claim 1, characterized in that, The torque control model is constructed and embedded in the automotive AI chip, including: Based on macroscopic information as input and torque mode combination as the decision objective, a top-level processing layer is deployed, wherein the macroscopic information includes vehicle dynamics data and road condition data; By using inverse process analysis based on mapping relationships, and taking the target distribution trajectory of torque mode decomposition into magnetic field topological invariants as the decision objective, a mid-level processing layer is deployed. Based on the control counter-differential analysis of the target distribution trajectory under real-time magnetic field conditions, the underlying processing layer is deployed. Based on the top-level processing layer, the middle-level processing layer, and the bottom-level processing layer, a torque control model is built, and the torque control model is deployed on the vehicle-mounted AI chip.

3. The multi-mode adaptive torque control method for an axial flux motor as described in claim 2, characterized in that, Establish a mapping relationship between magnetic field topological phase and torque mode; The mapping relationships include: If the topological phase of the magnetic field is a trivial phase with uniformly parallel magnetic field lines, the torque mode is a smooth torque mode. If the magnetic field topological phase is a vortex lattice phase in which the magnetic field lines are regularly arranged in a vortex array, the torque mode exhibits a high-frequency micro-amplitude pulsating mode; If the magnetic field topological phase is a topological insulator phase with a flat volume magnetic field and magnetic field line channels at the edges, the torque mode exhibits a low-loss stable mode; If the topological phase of the magnetic field is a quantum spin liquid phase with highly entangled and disordered magnetic field lines and long-range entanglement, the torque mode exhibits an anti-interference mode.

4. The multi-mode adaptive torque control method for an axial flux motor as described in claim 3, characterized in that, Torque mode matching is performed based on the first macroscopic scale information, including: The first macroscopic scale information is input into the top-level processing layer. By extracting macroscopic features and performing rule matching with torque mode features, torque mode instructions are determined. The torque mode instructions include torque mode type, macroscopic torque target value and key mode parameters. The matching result is either a single torque mode or a combination of torque modes. If it is a combination of torque modes, each torque mode identifier has a relative proportion.

5. The multi-mode adaptive torque control method for an axial flux motor as described in claim 4, characterized in that, Trajectory decision-making under magnetic field topological phase transition is performed, through comparison with the second magnetic field state and differential control decision-making, including: The torque mode command is input into the middle processing layer, and the target magnetic field topological phase is determined according to the mapping relationship. For the target magnetic field topological phase, the target distribution trajectory of magnetic flux density in the air gap and stator teeth is measured, wherein the target distribution trajectory includes the spatial distribution target of air gap magnetic flux density, the magnetic flux density saturation target of stator teeth, and the upper limit of allowable local vortex density.

6. The multi-mode adaptive torque control method for an axial flux motor as described in claim 5, characterized in that, Determine the motor shaft torque parameters, including: The target distribution trajectory is input into the bottom processing layer, and the second magnetic field state is read. By comparing the second magnetic field state with the target distribution trajectory, and taking the multiphase winding current, rotor position and speed of the axial flux motor as the control objects, the motor shaft torque parameters are determined through differential control analysis and optimization.

7. The multi-mode adaptive torque control method for an axial flux motor as described in claim 1, characterized in that, After determining the motor shaft torque parameters, the following steps are included: By decoupling the left and right halves of the motor shaft torque parameters, the first parameter part and the second parameter part are determined; For the same motion condition, the yaw moment compensation is determined by real-time yaw angle acquisition; Based on the yaw moment compensation amount, an asymmetric fine-tuning command is generated, wherein the constraint condition is that the torque mode is not changed. According to the asymmetric fine-tuning instruction, differential adjustment compensation is performed on the first parameter part and the second parameter part.

8. The multi-mode adaptive torque control method for an axial flux motor as described in claim 1, characterized in that, Based on the motor shaft torque parameters, a torque control command is generated; Based on the communication interaction between the vehicle-mounted AI chip and the control center of the axial flux motor, the torque control command is issued to execute the control drive of the axial flux motor.

9. The multi-mode adaptive torque control method for an axial flux motor as described in claim 1, characterized in that, After determining the motor shaft torque parameters, the following steps are included: The unit amplitude modulation constraint is determined by the parameter change rate, the thermodynamic limit, and the magnetic field reconstruction rate; Determine whether the motor shaft torque parameter meets the unit amplitude constraint. If not, perform phased processing on the motor shaft torque parameter under the unit amplitude constraint to determine the motor shaft torque parameter sequence.

10. A multi-mode adaptive torque control system for an axial flux motor, characterized in that, The system is used to implement the multi-mode adaptive torque control method for an axial flux motor according to any one of claims 1-9, and the system comprises: The bus communication establishment module is used to establish bus communication between the vehicle-mounted sensor array, the vehicle-mounted AI chip, and the control center. The synchronous sensing acquisition module is used to perform multi-sensor synchronous acquisition of external motion and motor magnetic space through the vehicle-mounted sensor array to determine the first macroscopic scale information and the second magnetic field state. The motor shaft torque parameter determination module is used to construct a torque control model and embed it in the vehicle AI chip, perform torque mode matching based on the first macroscopic scale information, execute trajectory decision under magnetic field topology phase transition, and determine the motor shaft torque parameter by comparing it with the second magnetic field state and making differential control decisions. The motor control and drive management module is used to generate torque control commands based on the motor shaft end torque parameters and to execute the control and drive management of the axial flux motor.

Images