A motor torque fluctuation suppression method and system based on model predictive control

Abstract

The application discloses a motor torque fluctuation suppression method and system based on model predictive control, and belongs to the technical field of motor control. The method detects the motor operating state in real time, identifies key parameters such as stator resistance and inductance online, and uses a sliding mode observer to estimate multiple source disturbances such as load, tooth slot effect and power converter nonlinearity in real time. On this basis, an extended motor prediction model is constructed by fusing parameter adaptability and disturbance feedforward compensation, and the model is used to rollingly predict future torque and current. Taking tracking of an external torque instruction as a core target, multi-objective roll optimization is performed under the condition of taking into account current tracking accuracy and switch loss constraints, and an optimal voltage control instruction is solved, and finally the motor is driven to operate through space vector pulse width modulation. The application effectively solves the problems of poor parameter robustness, weak multi-disturbance suppression ability and difficult balance between dynamic and steady-state performance of the traditional method, significantly improves torque output stability, system adaptive ability and overall energy efficiency, and is suitable for various AC motor driving scenes with strict torque control performance requirements.

Description

Technical Field

[0001] This invention belongs to the field of motor control technology, and particularly relates to a method and system for suppressing motor torque fluctuations based on model predictive control. Background Technology

[0002] In high-performance motor drive applications, such as electric vehicles, industrial robots, and precision machine tools, the stability of the motor's output torque directly determines the overall system's operational accuracy, vibration and noise levels, and dynamic response capabilities. However, in actual operation, motor torque fluctuation is a common and difficult-to-avoid phenomenon. Its causes are complex and diverse, primarily including the inherent nonlinear characteristics of the motor itself, such as the non-ideal magnetic field distribution of permanent magnets in permanent magnet synchronous motors, the cogging effect between the stator and rotor, and magnetic circuit saturation caused by armature reaction; for asynchronous motors, there are factors such as stator winding magnetomotive force harmonics and rotor bar eddy current effects. Furthermore, sudden changes in external loads (such as road bumps and mechanical shocks), sensor measurement noise, and dead-zone effects during power converter switching can further exacerbate torque pulsation.

[0003] To suppress torque ripple, the industry commonly employs two main motor control strategies: Field-Oriented Control (FOC) and Direct Torque Control (DTC). Field-Oriented Control (FOC) achieves decoupled control of torque and flux linkage through coordinate transformation, exhibiting good steady-state performance. However, its control effectiveness heavily relies on an accurate mathematical model of the motor. In actual operation, motor parameters (such as stator resistance and inductance) drift with temperature and magnetic saturation, leading to model mismatch and consequently reducing torque control accuracy and weakening the ability to suppress ripple. Direct Torque Control, on the other hand, uses hysteresis comparators to directly adjust torque and flux linkage, providing rapid dynamic response. However, this method inherently suffers from drawbacks such as inconsistent switching frequency and significant torque ripple, particularly in the low-speed operating region where periodic torque ripple caused by cogging effect is especially prominent.

[0004] In summary, the existing technology has the following main limitations: High parameter sensitivity and insufficient robustness mean that both FOC and DTC performance rely on accurate motor parameters. Traditional methods lack effective online adjustment mechanisms to address unavoidable parameter variations during operation, leading to a decline in system anti-interference capabilities.

[0005] Dynamic performance and steady-state accuracy are difficult to balance. Increasing control damping to improve steady-state accuracy often sacrifices the system's dynamic response speed; conversely, pursuing a fast dynamic response may weaken the ability to suppress steady-state ripple. Existing control architectures have an inherent contradiction in balancing these two aspects.

[0006] The ability to suppress multi-source composite disturbances is limited. Traditional methods often use PID controllers based on linear models, whose parameters are usually tuned under specific operating conditions. When the motor faces multiple disturbances of different types and frequencies, such as cogging effect, load change, and dead zone effect, single linear feedback control is difficult to achieve accurate and coordinated compensation, resulting in poor suppression effect.

[0007] There is a trade-off between switching losses and torque performance. For example, in DTC, reducing torque ripple requires reducing hysteresis tolerance, but this leads to an increase in the switching frequency of power devices, increasing switching losses and heat dissipation burden. On the other hand, using a fixed switching frequency improvement scheme often comes at the cost of sacrificing some dynamic response characteristics.

[0008] Therefore, in modern motor drive applications that require high precision, high dynamics, and high energy efficiency, there is an urgent need for a new type of motor torque control method that can adapt to parameter changes online, actively compensate for multi-source disturbances, and achieve performance optimization under multiple constraints. Summary of the Invention

[0009] This invention aims to overcome the shortcomings of existing technologies and provide a motor torque ripple suppression scheme with superior comprehensive performance. Its core lies in constructing a collaborative control mechanism that integrates a model predictive control (MPC) framework, online parameter identification, and multi-source disturbance feedforward compensation.

[0010] To solve the above-mentioned technical problems, the specific technical solution of the present invention is as follows: In some embodiments of this application, a method for suppressing motor torque ripple based on model predictive control is provided, comprising the following steps: The operating status parameters of the motor are acquired in real time, including stator current, rotor position and speed; Based on the operating status parameters, the real-time parameters of the motor under the current operating conditions are identified online. The real-time parameters include stator resistance, quadrature-axis inductance, direct-axis inductance, and permanent magnet flux linkage. Based on the operating state quantity and the real-time parameters, the multi-source disturbances affecting the motor torque are observed and separated in real time. The multi-source disturbances include load torque disturbances, cogging effect disturbances, and equivalent voltage disturbances introduced by the nonlinearity of the power converter. Based on the real-time parameters and the multi-source disturbances, a motor extended prediction model that integrates parameter adaptability and disturbance feedforward compensation capability is constructed. Based on the motor extended prediction model, with the given torque command as the tracking target and minimizing the predicted torque deviation and current deviation as the main optimization objectives, while taking the switching loss of power devices as an optimization constraint, rolling optimization calculations are performed within the safe boundary of system operation to solve for the optimal voltage control command applied to the motor in the current control cycle. The optimal voltage control command is converted into a switching signal for the power device to drive the motor. The above steps are repeated in each control cycle to achieve closed-loop rolling optimization control.

[0011] In some embodiments of this application, the motor extended prediction model is constructed by embedding the real-time parameters and the equivalent voltage disturbance into the discretized voltage equation of the motor, while embedding the load torque disturbance and cogging effect disturbance into the torque equation of the motor in the form of compensation terms.

[0012] In some embodiments of this application, the online identification of real-time motor parameters is achieved using a recursive least squares estimation algorithm with a forgetting factor, which can continuously track the slow time-varying characteristics of motor parameters.

[0013] In some embodiments of this application, real-time observation of multi-source disturbances is achieved using a sliding mode observer. This observer constructs a sliding mode surface with the rotational speed observation error as the core and uses a continuous method to suppress output chattering in order to accurately estimate rapidly changing disturbance components.

[0014] In some embodiments of this application, the objective function in the rolling optimization calculation is a comprehensive performance index composed of the weighted sum of squares of the predicted torque deviation, the weighted sum of squares of the predicted current deviation, and the weighted sum of the changes in the switching state of adjacent control cycles.

[0015] In some embodiments of this application, the safety boundaries for system operation include an upper limit for the output voltage determined by the DC bus voltage, an upper limit for the stator current determined by the motor rated current, and a limit on the maximum number of changes in the switching states of power devices in adjacent cycles.

[0016] In some embodiments of this application, the rolling optimization calculation process discretizes the continuous voltage vector space into a finite set of voltage vector candidates, transforming the optimization solution into an enumeration search problem within the finite candidate set, in order to meet the real-time requirements of the digital control system.

[0017] In some embodiments of this application, a motor control system is disclosed, including: The status detection module is used to acquire the stator current, rotor position, and speed of the motor. The online parameter identification module is used to estimate the electrical parameters of the motor in real time based on the output of the state detection module; A multi-source disturbance observation module is used to observe disturbances caused by load, cogging effect and power converter nonlinearity in real time based on the output of the state detection module and the result of the online parameter identification module. An extended prediction model module is used to construct and update a dynamic model for predicting the behavior of the motor in multiple future cycles based on the outputs of the online parameter identification module and the multi-source disturbance observation module. The multi-objective rolling optimization module is used to solve for the optimal voltage control command based on the prediction results of the extended prediction model module, the given external commands, and the system constraints. A control signal generation module is used to convert the optimal voltage control command into a switching control signal for the power converter; A power converter for driving a motor according to the switch control signal; The online parameter identification module, multi-source disturbance observation module, extended prediction model module, and multi-objective rolling optimization module work together to achieve the motor torque fluctuation suppression method described in any one of the above-mentioned methods.

[0018] Some embodiments of this application disclose an electronic device including a memory, a processor, and a computer program stored in the memory, wherein when the processor executes the computer program, it implements the steps of the method as described in any of the preceding claims.

[0019] Some embodiments of this application disclose a computer-readable storage medium having a computer program stored thereon, which, when executed by a processor, implements the steps of the method as described in any of the above.

[0020] Compared with the prior art, the beneficial effects of the present invention are as follows: This invention constructs an extended prediction model that integrates real-time parameters and multi-source disturbance compensation, and employs a rolling optimization strategy to directly track and control torque, achieving proactive and precise suppression of torque fluctuations. It effectively mitigates periodic pulsations caused by the cogging effect of the motor itself, as well as random fluctuations caused by external disturbances such as load abrupt changes. This invention uses the state changes of switching devices as one of the optimization objectives. By adjusting the weight of this term, the switching frequency of the power converter can be actively and flexibly constrained while ensuring torque control performance. This allows the system to effectively limit unnecessary switching actions while suppressing torque fluctuations, reducing switching losses and improving the energy efficiency of the entire drive system. Furthermore, by transforming the continuous optimization problem into an enumeration search of a finite voltage vector set, the online computational complexity is significantly reduced, ensuring that the algorithm can be completed within a microsecond-level control cycle, meeting the real-time requirements of high-performance digital control. Each functional module (such as parameter identification, disturbance observation, prediction model, and optimizer) is relatively independent. For different types of motors (such as permanent magnet synchronous motors and asynchronous motors), only the prediction model module needs to be adjusted according to its mathematical model; the core control logic and framework do not need to be reconstructed. Furthermore, the objective function and constraints can be flexibly configured according to specific application scenarios (such as electric vehicles, precision machine tools, and industrial robots) to highlight different performance focuses. Therefore, this invention has good universality and engineering application extension value. Attached Figure Description

[0021] Various other advantages and benefits will become apparent to those skilled in the art upon reading the following detailed description of preferred embodiments. The accompanying drawings are for illustrative purposes only and are not intended to limit the invention. Furthermore, the same reference numerals denote the same parts throughout the drawings. In the drawings: Figure 1 This is a flowchart of the motor torque ripple suppression method based on model predictive control in the embodiments of this application; Figure 2 This is a schematic diagram of the motor torque ripple suppression system module in the embodiments of this application; Figure 3 This is a flowchart illustrating the core process of multi-objective rolling optimization in this application embodiment. Detailed Implementation

[0022] The specific embodiments of the present invention will be described in further detail below with reference to the examples. These examples are for illustrative purposes only and are not intended to limit the scope of the invention.

[0023] To better understand the purpose, structure, and function of this invention, the invention will be further described in detail below with reference to embodiments.

[0024] This application provides a method for suppressing motor torque ripple based on model predictive control, including the following steps: The operating status parameters of the motor are acquired in real time, including stator current, rotor position and speed; Based on the operating status parameters, the real-time parameters of the motor under the current operating conditions are identified online. The real-time parameters include stator resistance, quadrature-axis inductance, direct-axis inductance, and permanent magnet flux linkage. Based on the operating state quantity and the real-time parameters, the multi-source disturbances affecting the motor torque are observed and separated in real time. The multi-source disturbances include load torque disturbances, cogging effect disturbances, and equivalent voltage disturbances introduced by the nonlinearity of the power converter. Based on the real-time parameters and the multi-source disturbances, a motor extended prediction model that integrates parameter adaptability and disturbance feedforward compensation capability is constructed. Based on the motor extended prediction model, with the given torque command as the tracking target and minimizing the predicted torque deviation and current deviation as the main optimization objectives, while taking the switching loss of power devices as an optimization constraint, rolling optimization calculations are performed within the safe boundary of system operation to solve for the optimal voltage control command applied to the motor in the current control cycle. The optimal voltage control command is converted into a switching signal for the power device to drive the motor. The above steps are repeated in each control cycle to achieve closed-loop rolling optimization control.

[0025] The extended prediction model for the motor is constructed by embedding the real-time parameters and the equivalent voltage disturbance into the discretized voltage equation of the motor, and by embedding the load torque disturbance and the cogging effect disturbance into the torque equation of the motor in the form of compensation terms.

[0026] The online identification of real-time motor parameters is achieved using a recursive least squares estimation algorithm with a forgetting factor, which can continuously track the slow time-varying characteristics of motor parameters.

[0027] Real-time observation of multi-source disturbances is achieved using a sliding mode observer. This observer constructs a sliding mode surface with the rotational speed observation error as the core and uses a continuous method to suppress output chattering in order to accurately estimate rapidly changing disturbance components.

[0028] The objective function in the rolling optimization calculation is a comprehensive performance index composed of the weighted sum of squares of the predicted torque deviation, the weighted sum of squares of the predicted current deviation, and the weighted sum of the changes in the switching state between adjacent control cycles.

[0029] The safety boundaries for system operation include the upper limit of output voltage determined by the DC bus voltage, the upper limit of stator current determined by the motor rated current, and the limit on the maximum number of changes in the switching state of power devices in adjacent cycles.

[0030] The rolling optimization calculation process discretizes the continuous voltage vector space into a finite set of voltage vector candidates, transforming the optimization solution into an enumeration search problem within the finite candidate set, in order to meet the real-time requirements of the digital control system.

[0031] In some embodiments of this application, a motor control system is disclosed, including: The status detection module is used to acquire the stator current, rotor position, and speed of the motor. The online parameter identification module is used to estimate the electrical parameters of the motor in real time based on the output of the state detection module; A multi-source disturbance observation module is used to observe disturbances caused by load, cogging effect and power converter nonlinearity in real time based on the output of the state detection module and the result of the online parameter identification module. An extended prediction model module is used to construct and update a dynamic model for predicting the behavior of the motor in multiple future cycles based on the outputs of the online parameter identification module and the multi-source disturbance observation module. The multi-objective rolling optimization module is used to solve for the optimal voltage control command based on the prediction results of the extended prediction model module, the given external commands, and the system constraints. A control signal generation module is used to convert the optimal voltage control command into a switching control signal for the power converter; A power converter for driving a motor according to the switch control signal; The online parameter identification module, multi-source disturbance observation module, extended prediction model module, and multi-objective rolling optimization module work together to achieve the motor torque fluctuation suppression method described in any one of the above descriptions.

[0032] Some embodiments of this application disclose an electronic device including a memory, a processor, and a computer program stored in the memory, wherein when the processor executes the computer program, it implements the steps of the method as described in any of the preceding claims.

[0033] Some embodiments of this application disclose a computer-readable storage medium having a computer program stored thereon, which, when executed by a processor, implements the steps of the method as described in any of the above.

[0034] The control system of this invention uses an embedded digital processor as its core, achieving the coordination of multiple functional modules at the software level. This system, together with the controlled motor and power converter, forms a closed loop. The power converter is typically a three-phase voltage source inverter, with its DC side connected to a DC power supply and its AC side connected to the motor stator windings. The inputs to the control system include three-phase stator current signals from current sensors, rotor position and speed signals from position sensors, and torque command signals from the host controller. Its output is a pulse-width modulated signal that controls the on / off switching of the transistors in the power converter.

[0035] The method is repeated with a fixed, microsecond-level control cycle. Within each control cycle, the following steps are executed sequentially: Motor operating status detection and signal preprocessing: Signal acquisition: The three-phase instantaneous current of the motor stator is acquired in real time through a high-precision Hall effect current sensor or sampling resistor. At the same time, the absolute or incremental position information of the rotor is obtained through position sensors such as photoelectric encoders and rotary transformers, and the mechanical speed of the rotor is calculated or directly measured.

[0036] Signal processing: The acquired raw three-phase current signals are subjected to hardware or digital filtering (such as low-pass filtering) to suppress switching noise and high-frequency interference. Then, the filtered current signals are subjected to coordinate transformation. First, the current in the three-phase stationary coordinate system is converted into current components in the two-phase stationary coordinate system through Clark transformation. Then, using the rotor electrical angle information obtained from the position sensor, the current in the two-phase stationary coordinate system is further converted into the two-phase rotating coordinate system that rotates synchronously with the rotor magnetic field through Park transformation, to obtain the direct-axis current component and the quadrature-axis current component. These two components correspond to the excitation current and torque current of the motor, respectively. At the same time, the mechanical speed is converted into electrical angular velocity.

[0037] Online real-time identification of key motor parameters: Model basis: This step is based on the motor voltage equation in a rotating coordinate system. The voltage equation describes the relationship between the applied voltage, current, speed and the internal parameters of the motor (resistance, inductance, flux linkage).

[0038] Algorithm implementation: A recursive estimation algorithm with the ability to update and forget old data online is adopted. The algorithm takes the direct-axis and quadrature-axis voltage commands output from the previous control cycle, the direct-axis and quadrature-axis currents measured and transformed in the current cycle, and the electrical angular velocity as inputs.

[0039] Identification Process: The algorithm runs once per control cycle, recursively calculating parameter estimates that best match the current actual operating state of the motor based on the latest voltage, current, and speed data. Key parameters identified online include: stator winding resistance, direct-axis synchronous inductance, quadrature-axis synchronous inductance, and the constant flux linkage amplitude generated by the permanent magnets. By continuously running this algorithm, the control system can automatically track the slow parameter drift caused by factors such as motor temperature changes and magnetic circuit saturation, ensuring that the internal model always remains consistent with the actual characteristics of the motor.

[0040] Observation and separation compensation of multi-source disturbances: Observer design: Construct a nonlinear state observer whose core purpose is to estimate various disturbances that cannot be directly measured and are not included in the basic motor model. The observer takes the actual electrical angular velocity of the motor and the instantaneous electromagnetic torque estimate calculated using the parameters identified in step 2 as the main inputs.

[0041] Disturbance estimation: The observer employs a special feedback design to enable its output to converge quickly and track the total disturbance torque acting on the motor shaft. This total disturbance is a composite quantity.

[0042] Disturbance Separation: After obtaining the total disturbance estimate, an internal algorithm decomposes it into several independent components with clear physical meaning. These components typically include: (a) load torque disturbance: caused by changes in external mechanical load; (b) cogging torque disturbance: periodic torque pulsations generated by the cogging structure of the motor stator and rotor cores; and (c) power converter nonlinear disturbance: nonlinear factors such as inverter switching dead zone and transistor voltage drop are equivalent to additional voltage disturbances acting on the direct and quadrature axes. These separated disturbance components will be used for subsequent accurate compensation.

[0043] Dynamic construction and rolling forecasting of extended forecasting models: Model Building: The core of this step is to integrate the results of the first two steps (real-time parameters and disturbance components) to dynamically generate a high-precision discrete-time prediction model. This model makes two key extensions to the standard motor model: First, in the voltage prediction equation, the equivalent voltage disturbance term observed in step 3 is explicitly added as compensation for the non-ideal characteristics of the inverter; second, in the electromagnetic torque prediction equation, the load torque disturbance and cogging torque disturbance terms observed in step 3 are directly subtracted.

[0044] Rolling Prediction: At the beginning of each control cycle, the current direct-axis and quadrature-axis current values ​​obtained in step 1 are used as the initial state of the prediction model. Then, assuming a series of candidate voltage vectors in the future, the extended model is used to progressively calculate (roll) forward to predict the possible trajectories of motor current and electromagnetic torque over several future control cycles (called the "prediction time domain"). The prediction results are two sets of sequences: one is the predicted current sequence for each future cycle, and the other is the predicted torque sequence.

[0045] Multi-objective rolling optimization and optimal instruction solution: Objective Function Design: A comprehensive evaluation function is designed to measure the control effect of different candidate voltage vectors in the future prediction time domain. This function typically includes three weighted terms: (a) Torque Tracking Term: This term measures the sum of squares of the deviations between the predicted torque sequence and the externally given torque command sequence. This term has the highest weight to ensure the primary control objective; (b) Current Tracking Term: This term measures the sum of squares of the deviations between the predicted current sequence and the desired current command (e.g., when using maximum torque-to-current ratio control, the desired direct-axis current is typically set to zero). This term is used to optimize the current waveform and suppress harmonics; (c) Switching Loss Term: This term measures the frequency of the switching state changes required to generate the candidate voltage vector. This is achieved by penalizing the changes in the switching signal between adjacent cycles, aiming to reasonably limit the switching frequency and losses of the power devices.

[0046] Constraint settings: The optimization solution must be performed within the physical limits of the system. The main constraints include: (i) voltage amplitude constraint, that is, the maximum output voltage capability of the inverter determined by the DC bus voltage; (ii) current amplitude constraint, that is, the maximum stator current allowed by the motor; (iii) switching action constraint, in order to reduce switching losses, the maximum number of switching transistors that can be changed in adjacent control cycles can be limited.

[0047] Optimization Solution: In each control cycle, the prediction model obtained in step 4, the current external commands (torque command, current command), and the aforementioned objective function and constraints are input into an optimization solver. The solver's task is to find the voltage vector sequence that minimizes the comprehensive evaluation function value from all feasible voltage vectors that satisfy the constraints. A "rolling time domain" strategy is adopted, using only the first optimal voltage vector (including direct-axis and quadrature-axis components) corresponding to the current moment in this sequence as the final output command for this control cycle.

[0048] Control signal generation and power execution: Inverse coordinate transformation: The optimal direct-axis and quadrature-axis voltage commands obtained in step 5 are transformed back into a two-phase stationary coordinate system through the Parker inverse transformation.

[0049] Pulse Width Modulation: Voltage commands in the stationary coordinate system are processed using a space vector pulse width modulation algorithm. This algorithm converts continuous voltage commands into specific six-channel pulse width modulation signals with specific duty cycles and timing sequences.

[0050] Power drive: These six PWM signals are sent to the drive circuit of the power converter to control the conduction and cutoff of the six power switching transistors respectively. Finally, the power converter generates a three-phase AC voltage corresponding to the optimal voltage command at its output terminal, which is applied to the stator winding of the motor to achieve precise control of the motor torque.

[0051] Periodic rolling iteration: After completing all calculations, outputs, and drives for the current control cycle, the system waits for the interruption of the next control cycle. Once the interruption is triggered, the time index is updated, and the entire process is repeated from step 1, forming a closed-loop, rolling forward optimization control. Through this continuous cycle of sensing, modeling, predicting, optimizing, and executing, the system can respond to parameter changes in real time and proactively compensate for various disturbances, thereby effectively suppressing torque fluctuations.

[0052] Optimal algorithm for online parameter identification module: It is recommended to use recursive least squares with a forgetting factor. The forgetting factor is a positive constant slightly less than 1 (usually selected between 0.95 and 0.99). Its function is to give higher weight to newly sampled data, thereby effectively "forgetting" old data and enhancing the tracking ability of time-varying parameters. During algorithm initialization, the parameter estimates can be set to the motor's nameplate parameters or typical values, and a large covariance matrix should be initialized to accelerate the initial convergence process.

[0053] Preferred structure of the multi-source disturbance observation module: A sliding mode variable structure theory-based observer is recommended. To overcome the inherent high-frequency chattering phenomenon of traditional sliding mode observers and improve the practicality and smoothness of the observed signal, a continuous saturation function or hyperbolic tangent function is used to replace the discontinuous sign switching function. By rationally designing the sliding surface gain and boundary layer thickness, this observer can output smooth and usable disturbance estimates while ensuring rapid response to abrupt disturbances, facilitating subsequent compensation.

[0054] Dynamic characteristics of the extended prediction model module: The predictive model of this invention is not static. In each control cycle, the resistance, inductance, and flux linkage parameters in the model are updated using the latest results identified online in step 2. Simultaneously, the voltage disturbance compensation term and torque disturbance compensation term in the model are also updated using the latest results observed in step 3. This dynamic updating mechanism ensures that the predictive model can accurately reflect the true dynamic behavior of the motor under the current operating conditions in real time, which is a key prerequisite for achieving high-performance predictive control.

[0055] Engineering implementation of the rolling optimization module: To accommodate the limited computing power of embedded controllers, a balance must be struck between control accuracy and computational burden. The prediction time domain length is typically chosen to be 2 to 5 steps. For optimization strategies, an efficient and practical approach is to use an "enumeration search" method. Specifically, based on the principle of space vector pulse width modulation, all possible effective voltage vectors (including basic and synthetic vectors) that can be physically generated by the inverter are predefined as a finite, controllable candidate set. In each control cycle, the rolling optimization module only needs to evaluate the effect of substituting each candidate voltage vector in this finite set into the prediction model and optimization function, and finally select the one with the best effect. This method transforms a complex continuous optimization problem into a fast discrete search problem, making it very suitable for completion within microsecond-level control cycles.

[0056] The various embodiments in this specification are described in a progressive manner, with each embodiment focusing on its differences from other embodiments. Similar or identical parts between embodiments can be referred to interchangeably. For the apparatus disclosed in the embodiments, since they correspond to the methods disclosed in the embodiments, the description is relatively simple; relevant parts can be referred to the method section.

[0057] The above description of the disclosed embodiments enables those skilled in the art to make or use the invention. Various modifications to these embodiments will be readily apparent to those skilled in the art, and the general principles defined herein may be implemented in other embodiments without departing from the spirit or scope of the invention. Therefore, the invention is not to be limited to the embodiments shown herein, but is to be accorded the widest scope consistent with the principles and novel features disclosed herein.

Claims

1. A method for suppressing motor torque ripple based on model predictive control, characterized in that, Includes the following steps: The operating status parameters of the motor are acquired in real time, including stator current, rotor position and speed; Based on the operating status parameters, the real-time parameters of the motor under the current operating conditions are identified online. The real-time parameters include stator resistance, quadrature-axis inductance, direct-axis inductance, and permanent magnet flux linkage. Based on the operating state quantity and the real-time parameters, the multi-source disturbances affecting the motor torque are observed and separated in real time. The multi-source disturbances include load torque disturbances, cogging effect disturbances, and equivalent voltage disturbances introduced by the nonlinearity of the power converter. Based on the real-time parameters and the multi-source disturbances, a motor extended prediction model that integrates parameter adaptability and disturbance feedforward compensation capability is constructed. Based on the motor extended prediction model, with the given torque command as the tracking target and minimizing the predicted torque deviation and current deviation as the main optimization objectives, while taking the switching loss of power devices as an optimization constraint, rolling optimization calculations are performed within the safe boundary of system operation to solve for the optimal voltage control command applied to the motor in the current control cycle. The optimal voltage control command is converted into a switching signal for the power device to drive the motor. The above steps are repeated in each control cycle to achieve closed-loop rolling optimization control.

2. The method according to claim 1, characterized in that, The motor extended prediction model is constructed by embedding the real-time parameters and the equivalent voltage disturbance into the discretized voltage equation of the motor, and by embedding the load torque disturbance and cogging effect disturbance into the torque equation of the motor in the form of compensation terms.

3. The method according to claim 1, characterized in that, The online identification of real-time motor parameters is achieved using a recursive least squares estimation algorithm with a forgetting factor, which can continuously track the slow time-varying characteristics of motor parameters.

4. The method according to claim 1, characterized in that, The real-time observation of multi-source disturbances is achieved using a sliding mode observer. This observer constructs a sliding mode surface with the rotational speed observation error as the core and uses a continuous method to suppress output chattering in order to accurately estimate rapidly changing disturbance components.

5. The method according to claim 1, characterized in that, The objective function in the rolling optimization calculation is a comprehensive performance index composed of the weighted sum of squares of the predicted torque deviation, the weighted sum of squares of the predicted current deviation, and the weighted sum of the changes in the switching state of adjacent control cycles.

6. The method according to claim 1 or 5, characterized in that, The safety boundaries for system operation include the upper limit of output voltage determined by the DC bus voltage, the upper limit of stator current determined by the rated current of the motor, and the limit on the maximum number of changes in the switching state of power devices in adjacent cycles.

7. The method according to claim 1, characterized in that, The rolling optimization calculation process discretizes the continuous voltage vector space into a finite set of voltage vector candidates, transforming the optimization solution into an enumeration search problem within the finite candidate set, in order to meet the real-time requirements of the digital control system.

8. A motor control system, characterized in that, include: The status detection module is used to acquire the stator current, rotor position, and speed of the motor. The online parameter identification module is used to estimate the electrical parameters of the motor in real time based on the output of the state detection module; A multi-source disturbance observation module is used to observe disturbances caused by load, cogging effect and power converter nonlinearity in real time based on the output of the state detection module and the result of the online parameter identification module. An extended prediction model module is used to construct and update a dynamic model for predicting the behavior of the motor in multiple future cycles based on the outputs of the online parameter identification module and the multi-source disturbance observation module. The multi-objective rolling optimization module is used to solve for the optimal voltage control command based on the prediction results of the extended prediction model module, the given external commands, and the system constraints. A control signal generation module is used to convert the optimal voltage control command into a switching control signal for the power converter; A power converter for driving a motor according to the switch control signal; The online parameter identification module, multi-source disturbance observation module, extended prediction model module, and multi-objective rolling optimization module work together to achieve the motor torque fluctuation suppression method as described in any one of claims 1 to 7.

9. An electronic device comprising a memory, a processor, and a computer program stored in the memory, characterized in that, When the processor executes the computer program, it implements the steps of the method as described in any one of claims 1 to 7.

10. A computer-readable storage medium having a computer program stored thereon, characterized in that, When the computer program is executed by a processor, it implements the steps of the method as described in any one of claims 1 to 7.

Images