Hierarchical parallel active disturbance rejection control method for permanent magnet synchronous motor to suppress compound multi-frequency disturbance

Abstract

The application provides a hierarchical parallel active disturbance rejection control method for suppressing composite multi-frequency disturbance of a permanent magnet synchronous motor. First, three-phase currents and a rotating speed of the motor are collected, and actual currents on dq axes are obtained through coordinate transformation. The rotating speed ring generates a q-axis expected current, and a d-axis expected current is set to zero. Parallel decoupling double-branch observers are configured on the d and q axes respectively. A cascade extended state observer is used for observing low-frequency non-periodic disturbance and compensating observation residual error, and a resonant harmonic observer is used for extracting 6th and 12th current harmonic disturbance. After superposition of two disturbance estimation paths, a total disturbance value is obtained, and is fed forward to a three-stage active disturbance rejection control law (including proportional feedback, disturbance compensation and reference differential feedforward) to generate a control voltage. The control voltage is subjected to inverse Park transformation and SVPWM modulation to drive an inverter. The method realizes separate observation of multiple types of disturbance, is simple to parameterize, can simultaneously suppress low-frequency disturbance and current harmonics, significantly reduces current distortion and rotating speed fluctuation, and improves control precision.

Description

Technical Field

[0001] This invention relates to the field of permanent magnet synchronous motor servo drive control technology, specifically to a hierarchical parallel active disturbance rejection control method for permanent magnet synchronous motors to suppress complex multi-frequency disturbances. It is applicable to new energy vehicles, precision machine tools, and high-precision servo systems in aerospace, and can simultaneously suppress non-periodic low-frequency disturbances and 6th / 12th order periodic harmonic disturbances during motor operation. Background Technology

[0002] Permanent magnet synchronous motors (PMSMs), with their advantages of high power density, high efficiency, and wide speed range, have become the core actuators of high-precision servo drive equipment. The industry commonly uses field-oriented control (FOC) to achieve decoupled speed regulation of the motor. Through Clark and Park coordinate transformations, FOC can decouple the strongly coupled, nonlinear time-varying model of a three-phase AC motor in the stationary coordinate system into two independent DC currents in the synchronous rotating coordinate system: the d-axis current representing the magnetic field component and the q-axis current representing the torque component.

[0003] However, there is a significant difference between the ideal control model and the complex actual operating environment. Under real-world conditions, the motor's d-axis and q-axis current loops are continuously affected by a combination of internal and external disturbances: firstly, non-periodic low-frequency disturbances such as model parameter drift, sudden load changes, frictional resistance, and ramp loads; and secondly, periodic harmonic disturbances of the 6th and 12th electrical frequencies caused by magnetic circuit asymmetry, cogging effect, and inverter dead-zone effect. These two types of combined disturbances cause current waveform distortion, torque fluctuations, and increased steady-state error, significantly reducing the accuracy of servo control.

[0004] Active disturbance rejection control (ADRC) relies on an extended state observer (ESO) to estimate and feedforward the total disturbance in real time, without requiring a precise model of the controlled object, making it the mainstream disturbance rejection scheme. However, the traditional linear extended state observer (LESO) has inherent drawbacks: limited observation bandwidth, amplitude attenuation and phase lag for high-frequency periodic harmonics, leading to steady-state observation errors. To reduce this error, theoretically, the bandwidth needs to be significantly increased, but increasing the observation bandwidth will directly amplify high-frequency noise in the current sampling, reduce the signal-to-noise ratio, and cause system oscillations, resulting in a "precision-noise" contradiction.

[0005] While existing improvement schemes such as bandwidth optimization, high-order / cascaded ESO, and resonant ESO can partially improve disturbance estimation capabilities, they generally suffer from defects such as disturbance processing path coupling, phase margin impairment, harmonic estimation distortion, complex parameter tuning, and poor frequency offset adaptability. They cannot simultaneously meet the high-precision and high-robust decoupling control requirements of aperiodic and periodic composite disturbances in the current loop of permanent magnet synchronous motors.

[0006] Therefore, proposing a hierarchical parallel active disturbance rejection control method for permanent magnet synchronous motors to suppress complex multi-frequency disturbances, in order to eliminate aperiodic and periodic disturbances generated during the control process and improve dynamic tracking capability and steady-state accuracy, has become an urgent problem to be solved. Summary of the Invention

[0007] In view of this, the present invention provides a hierarchical parallel active disturbance rejection control method for permanent magnet synchronous motors to suppress complex multi-frequency disturbances, so as to solve the problems existing in the prior art.

[0008] This invention proposes a hierarchical parallel active disturbance rejection control method for permanent magnet synchronous motors to suppress complex multi-frequency disturbances, comprising:

[0009] S1: Collect the three-phase sub-currents of the permanent magnet synchronous motor, and obtain the actual d-axis current and q-axis current in the synchronous rotating dq coordinate system through Clark transformation and Park transformation; collect the actual speed of the permanent magnet synchronous motor; use the speed loop PI controller to combine the difference between the desired speed and the actual speed to obtain the desired q-axis current, and set the desired d-axis current to 0;

[0010] S2: Construct parallel decoupled dual-branch observers for the d-axis current loop and the q-axis current loop respectively, estimate the two disturbance components respectively, and superimpose them to obtain the total disturbance estimate of the corresponding current loop. The parallel decoupled dual-branch observers for the d-axis and q-axis current loops have the same structure. The dual-branch observer includes a cascaded extended state observer branch and a resonant harmonic observer branch with parallel architecture. The transfer functions of the two branches are completely decoupled, and the parameters are independently adjustable. The cascaded extended state observer branch is used to estimate low-frequency and aperiodic disturbances, employing a two-stage series structure. The first-stage observer is used to coarsely estimate the disturbance of the current loop system, and the second-stage observer is used to re-estimate and compensate for the estimation error of the first-stage observer. The resonant harmonic observer branch is used to estimate the 6th and 12th current harmonic disturbances in the current loop.

[0011] S3: The total disturbance estimate is compensated into the control law to obtain the control voltage, wherein the control law consists of three parts: a proportional feedback term, a disturbance compensation term, and a reference differential feedforward term;

[0012] S4: Input the control voltage and the real-time electrical angle of the rotor into the inverse Park transform module, and convert it into... Coordinate system voltage The signal is fed into the SVPWM (Space Vector Pulse Width Modulation) module for pulse width modulation, outputting six control signals. These six control signals are then input into the inverter to output three-phase control current. Drives the permanent magnet synchronous motor to rotate.

[0013] Preferably, in the cascaded extended state observer branch of the q-axis current loop, the mathematical model of the first-stage observer is as follows:

[0014] ;

[0015] In the formula, This is the estimated q-axis current output from the first-stage observer. For the estimated perturbation of the first-level observer, This represents the actual current along the q-axis. This is the output voltage of the q-axis current loop. , This is the proportional gain of the first-stage observer; For the gain of the controlled object in the current loop;

[0016] The mathematical model for the second-level observer is as follows:

[0017] ;

[0018] In the formula, This is the amount of compensation for the error in the first-stage current estimation. It is the compensation amount for the estimation error of the first-level disturbance. , This is the second-level observation gain;

[0019] The total perturbation estimate for the cascaded extended state observer is: .

[0020] Further optimization, , This is the observation bandwidth of the cascaded extended state observer.

[0021] Further preferably, the transfer function of the perturbation estimation error of the cascaded extended state observer is as follows:

[0022] ;

[0023] in, This is the observation bandwidth of the cascaded extended state observer.

[0024] Further preferably, the resonant harmonic observer branch includes a pre-low-frequency disturbance filtering module, a 6th resonant observer, and a 12th resonant observer. The pre-low-frequency disturbance filtering module filters out low-frequency disturbances and aperiodic disturbances in advance, retaining only the periodic disturbance component. The 6th and 12th resonant observers are connected in series after the pre-low-frequency disturbance filtering module to extract the 6th and 12th harmonic disturbance components in the current.

[0025] Further optimization involves using a pre-amplified low-frequency disturbance filtering module with a classic ESO filter unit, as shown in the following formula:

[0026] ;

[0027] In the formula, This is an estimated value for the q-axis current. The q-axis current observation error is defined as follows: , This represents the actual current along the q-axis. This is the output voltage of the q-axis current loop. This is the estimated total disturbance of the resonant harmonic observer branch; For fundamental / aperiodic disturbance components, This is the 6th harmonic disturbance component; It is the 12th harmonic disturbance component.

[0028] Further optimization involves constructing 6th and 12th resonant observers using a second-order generalized integrator structure, as shown in the following models:

[0029] ;

[0030] In the formula, These are the orthogonal auxiliary state variables of the 6th and 12th resonance observers, respectively; These are the resonant angular frequencies of the 6th and 12th harmonics, respectively. This is the resonant damping coefficient; is the observer gain coefficient.

[0031] Further optimization, ;

[0032] in, The expected convergence bandwidth represents the resonant channel. , This is the gain coefficient.

[0033] Further optimization involves the transfer function molecule of the resonant unit including a first-order lead element with lead phase compensation.

[0034] The hierarchical parallel active disturbance rejection control method for permanent magnet synchronous motors to suppress complex multi-frequency disturbances provided by this invention can effectively improve the system's harmonic suppression capability, parameter robustness, and dynamic performance. Attached Figure Description

[0035] Figure 1 This is a diagram illustrating the overall architecture of the hierarchical parallel active disturbance rejection control method for suppressing complex multi-frequency disturbances in a permanent magnet synchronous motor, as described in this invention.

[0036] Figure 2 This is a structural diagram of CESO-RHO in an embodiment of the present invention;

[0037] Figure 3The figure shows a comparison of the current test results of three methods, namely traditional ESO, CESO, and CESO-RHO, when the load torque is 0.15 N·m. Detailed Implementation

[0038] To make the objectives, technical solutions, and advantages of this invention clearer, the invention will be further explained below with reference to specific embodiments, but this explanation does not limit the invention.

[0039] like Figure 1 As shown, this embodiment provides a hierarchical parallel active disturbance rejection control method for suppressing complex multi-frequency disturbances in permanent magnet synchronous motors, including the following steps:

[0040] S1: Signal Acquisition and Coordinate Transformation: Acquiring the three-phase sub-current of the permanent magnet synchronous motor via a current sensor. The actual d-axis current in the synchronously rotating dq coordinate system is obtained through Clark and Park transformations. and q-axis true current The actual rotational speed of the permanent magnet synchronous motor is collected by a motion observer. Using a speed loop PI controller based on the desired speed Compared with actual speed The difference is used to obtain the expected current along the q-axis. In this scheme, the desired current along the d-axis is set. Setting it to 0 simplifies the control logic;

[0041] S2: Construct parallel decoupled dual-branch observers CESO-RHO for the d-axis current loop and q-axis current loop respectively. For each current loop, estimate the two disturbance components and superimpose them to obtain the total disturbance estimate of the corresponding current loop. The parallel decoupled dual-branch observers CESO-RHO for the d-axis current loop and q-axis current loop have the same structure. The dual-branch observer CESO-RHO includes a cascaded extended state observer (CESO) branch and a resonant harmonic observer (RHO) branch with parallel architecture. The transfer functions of the two branches are completely decoupled and the parameters are adjusted independently.

[0042] The hardware and algorithm architectures of the d-axis current loop and the q-axis current loop are exactly the same. Only the complete execution flow of the q-axis is described. The d-axis reuses all observation, compensation and control logic.

[0043] The CESO branch is used to estimate low-frequency and aperiodic disturbances. The CESO branch adopts a two-stage series structure. The first-stage observer is used to make a rough estimate of the disturbance of the current loop system, and the second-stage observer is used to re-estimate and compensate for the estimation error of the first-stage observer, thereby eliminating the steady-state error of the ramp-type aperiodic disturbance.

[0044] The mathematical model for the first-level observer (Standard Loop) is as follows:

[0045] ;

[0046] In the formula, This is the estimated q-axis current output from the first-stage observer. For the estimated perturbation of the first-level observer, This is the amount of compensation for the error in the first-stage current estimation. It is the compensation amount for the estimation error of the first-level disturbance. This represents the actual current along the q-axis. The output voltage of the q-axis current loop is fed back into the observer as a control input feedforward, ensuring that the observer fully reproduces the current loop dynamics. , This is the proportional gain of the first-stage observer; The gain of the controlled object in the current loop is typically set to a value of [value missing]. , The q-axis inductance of the motor;

[0047] The second-level observer (Error Compensation Loop) takes the estimation error of the first-level observer as input and performs secondary estimation compensation for the residual observation error of the first-level observer. The mathematical model is as follows:

[0048] ;

[0049] In the formula, This is the amount of compensation for the error in the first-stage current estimation. It is the compensation amount for the estimation error of the first-level disturbance. , This is the second-level observation gain;

[0050] The total disturbance estimate for CESO is: ;

[0051] This embodiment uses the unified bandwidth method to tune the observation gain:

[0052] ;

[0053] This refers to the observation bandwidth of CESO.

[0054] Traditional single-stage ESOs cannot accurately observe slope disturbances without increasing bandwidth. The two-stage cascaded CESOs used in this invention improve low-frequency gain through residual secondary compensation and derive the final value theorem for slope load disturbances, bringing the steady-state observation error to zero. The entire process does not require increasing the observation bandwidth and does not amplify high-frequency noise from current sampling, thus balancing observation accuracy and signal-to-noise ratio.

[0055] The CESO perturbation estimation error transfer function is as follows:

[0056] ;

[0057] Applying the final value theorem to slope disturbances:

[0058] ;

[0059] Therefore, it can completely eliminate the steady-state error of ramp-type non-periodic disturbances, and obtain extremely high gain in the low-frequency band without increasing the observation bandwidth;

[0060] The Resonant Harmonic Observer (RHO) branch is used to achieve accurate observation of periodic disturbances. The RHO branch includes a pre-filter low-frequency disturbance, a 6th harmonic observer, and a 12th harmonic observer. The pre-filter low-frequency disturbance pre-filters out low-frequency disturbances and aperiodic disturbances, retaining only the periodic disturbance components to be sent to the resonant unit, thus avoiding saturation distortion of the resonant unit observation caused by large low-frequency disturbances. The 6th and 12th harmonic observers are connected in series after the pre-filter low-frequency disturbance and are used to extract the 6th and 12th harmonic disturbance components in the q-axis current.

[0061] The pre-amplifier low-frequency disturbance filtering module uses a classic ESO to build the filtering unit, as shown in the following formula:

[0062] ;

[0063] In the formula, This is an estimated value for the q-axis current. The q-axis current observation error is defined as follows: As a feedback signal that drives the convergence of the entire observer, The output voltage of the q-axis current loop serves as the input feedforward for the observer. These variables also apply to the subsequent 6th and 12th harmonic observation branches. The total disturbance estimate for this branch system is composed of the linear superposition of components from three different frequency bands; this frequency domain pre-separation mechanism significantly improves the efficiency and robustness of the resonator. For fundamental / aperiodic disturbance components, The 6th harmonic disturbance component corresponds to the RHO subsystem targeting... The output of the resonant unit with frequency design; This represents the 12th harmonic disturbance component, corresponding to the RHO subsystem for... The output of the resonant unit designed for frequency, It is the electrical angular velocity, and its relationship with the mechanical angular velocity is: p represents the number of pole pairs of the motor; in terms of system function, Its primary function is to establish a clean operating environment for the resonator; without this step, the original error signal... The low-frequency large disturbance components mixed in will directly enter the narrowband resonator, forcing the latter to process signals outside its design frequency band, resulting in a decrease in harmonic estimation accuracy or even stability problems.

[0064] The 6th and 12th resonant observers are used to observe harmonic components, and their models are as follows:

[0065] ;

[0066] In the formula, These are the auxiliary state variables for the 6th and 12th resonance observers, respectively. These two variables are introduced to construct the second-order generalized integrator (SOGI) structure. These are the resonant angular frequencies of the 6th and 12th harmonics, respectively. , , Electric angular velocity; This is the resonant damping coefficient, which determines the bandwidth of the resonant peak. The smaller the value, the sharper the frequency selectivity. It is typically set to a value of [value missing]. ; The observer gain coefficient; This determines the bandwidth and convergence speed of the fundamental wave observer (CESO); , This determines the sensitivity and response speed of the harmonic resonator (RHO) to errors at a specific frequency;

[0067] The parameters are designed as follows:

[0068] ;

[0069] in, Representing the expected convergence bandwidth of the resonant channel, the physical logic behind this set of design parameters is as follows: In the RHO resonant observer, the parameters... and Together, they constitute the numerator of the transfer function, and its tuning logic revolves closely around the observation bandwidth and phase compensation; firstly, Its main function is to perform gain mapping and convergence rate control, serving as the numerator term. coefficient, It determines the imaginary part strength of the system at the resonant frequency, since it is directly related to the observer bandwidth. The gain coefficient is proportional to the square of the gain, which ensures that the convergence speed of the observer for a specific harmonic component can be dynamically adjusted with the bandwidth, thus maintaining a consistent proportional scaling dynamic characteristic at different frequency points. Simultaneously, through the gain coefficient... as well as The introduction of this technology allows the system to achieve a physical-logical balance between the enhancement of the resonance peak and overall robustness, and it is designed to be compatible with... This proportionality ensures that the dynamic characteristics of the observer have a consistent scaling relationship at different frequency points, which can be derived from the subsystem equations of RHO (taking a 6th resonator as an example; the same applies to a 12th resonator):

[0070] ;

[0071] The denominator determines the resonant frequency. At a specific resonant frequency, a large gain (resonance peak) is generated. The molecular part is a typical first-order lead element, which introduces a left-half-plane zero into the system. The existence of this zero directly provides phase thrust. Below, the phase frequency characteristics of the molecules determine the strength of the compensation; let The numerator term is transformed into a complex number: The phase angle of this complex number in the complex plane for:

[0072] ;

[0073] When change When the value is , The changes will change simultaneously (Imaginary part of numerator) and The ratio of (real parts of the molecule) in actual practice, due to ,therefore, The value is always negative, which means that the molecular vector falls in the second quadrant of the complex plane, and its phase angle is... Greater than 90, because the denominator contributes at the resonance point. The phase shift, therefore the total phase of the system is equal to If the system requires more phase lead (to cope with greater delay), reduce (i.e., increase) The absolute value of the resultant vector is used to increase the proportion of the real part (negative value), thereby shifting the angle of the synthesized vector towards... The "pull" direction design provides the system with more phase margin.

[0074] Total disturbance estimate of the q-axis current loop Provide compensation;

[0075] S3: The total disturbance estimate is compensated into the control law to obtain the control voltage. The control law consists of three parts: a proportional feedback term, a disturbance compensation term, and a reference differential feedforward term, to realize the system's active disturbance rejection control.

[0076] q-axis control voltage For example, the calculation method is as follows:

[0077] Consider the q-axis current equation of a permanent magnet synchronous motor , For system gain, The total disturbance, including model uncertainties, harmonic disturbances, and coupling terms, is controlled to make the current... Follow reference instructions Define tracking error And design a control law to make the error move dynamically as desired. Convergence, of which, For proportional gain; by dynamically expanding the error, we can obtain Substituting the expected dynamics, we get Due to actual disturbances Unknown, total disturbance estimate provided by the CESO-RHO observer. Compensation is performed, thereby obtaining the final control law:

[0078] ;

[0079] The control law consists of three parts: a proportional feedback term. The disturbance compensation term is used to eliminate tracking errors and determine the system's dynamic response speed. Used to cancel various disturbances in real time to enhance robustness, referencing the differential feedforward term. This is used to provide the acceleration signal required for tracking a time-varying reference, avoiding the lag caused by relying solely on feedback; substituting the control law into the system equations yields the dynamic closed-loop error. When the observer accurately estimates the disturbance The tracking error converges exponentially, achieving zero steady-state error tracking;

[0080] Ultimately The output is then sent to the next stage of the PMSM control system.

[0081] S4: Input the d-axis and q-axis control voltages output from the ADRC controller and the real-time rotor electrical angle into the inverse Park transform module, and convert them into... Coordinate system voltage The signal is fed into the SVPWM for pulse width modulation, outputting six control signals. These six control signals are then input to the inverter to output three-phase control current. Drives the permanent magnet synchronous motor to rotate.

[0082] To verify the harmonic suppression capability, dynamic response performance, and robustness of the CESO-RHO proposed in this invention within the current loop of a permanent magnet synchronous motor, a dual-motor counter-drive experimental platform was constructed. The experimental platform consists of a drive-side system and a load-side system. The two motors are rigidly connected coaxially via couplings, achieving mechanical coupling and adjustable load application. The motor used in the experiment is a 42JSF840AS-1000 model, a 24V, 80W three-phase permanent magnet synchronous motor with a 1000-line encoder. Both systems are powered by a 24V DC power supply, and each system provides a 3.3V operating power supply to the LaunchPad through its respective LMR16006 step-down regulator.

[0083] The experimental setup consists of a power supply module, a control and drive module, and a tandem permanent magnet synchronous motor. The drive-side system comprises the LAUNCHXL-F28027F development board and the BOOSTXL-DRV8305EVM drive board. The LAUNCHXL-F28027F serves as the main control core, running the motor control algorithm and controlling the three-phase inverter bridge on the drive board via PWM signals to enable the 80W motor to operate at a given speed. The BOOSTXL-DRV8305EVM board features an onboard DRV8305 gate driver, which drives a three-phase full-bridge inverter circuit composed of six CSD18540Q5B power MOSFETs, converting the 24V DC bus voltage into the three-phase AC voltage required to drive the motor. An onboard voltage sampling resistor divider network and low-side current sampling resistors, in conjunction with the DRV8305's internal differential amplifier, enable real-time acquisition of bus voltage, phase voltage, and three-phase current, which are then fed back to the main control chip via an ADC to form a closed-loop control. The load-side system has the same hardware composition as the drive-side system, but its functional positioning differs. The LAUNCHXL-F28027F on the load side operates in torque control mode. By driving another 80W motor to generate adjustable resistance torque, load disturbances are applied to the drive-side motor, thereby testing the current control performance of the controlled motor under different load conditions. The two systems are electrically independent and can both read fault diagnosis information from the internal registers of the DRV8305 via the SPI bus. Rapid fault response and interlocking protection are achieved through GPIO signals such as nFAULT, ensuring the safe operation of the experimental platform. Simulink's Embedded Coder function is used for code compilation and signal status monitoring and protection.

[0084] In the experiment, the number of motor pole pairs was set to 4, the bus voltage was 24V, and the sampling frequency was 20kHz. The input speed for the steady-state load experiment was set to 50rpm, and the experiments were conducted under load torque conditions of 0.15N·m and 0.20N·m, respectively. The calculation interval for THD was uniformly selected from 1.5s to 3s. The comparison methods included the traditional ADRC-ESO, the current loop ADRC cascaded extended state observer CESO, and the IESO designed in this invention. The experimental results were mainly evaluated from the aspects of total harmonic distortion (THD) of phase current, q-axis current ripple, d-axis current ripple, main harmonic components, and speed step response.

[0085] Table 1 Main parameters of the experimental motor

[0086]

[0087] The steady-state current performance under different load torques was tested with an input speed of 50 rpm. Figure 3 Experimental waveforms for three control methods are presented when the load torque is 0.15 N·m. The three columns in the figure correspond to ESO, CESO, and CESO-RHO, respectively; the first row is i. a The second row is a magnified image in the 1.5s to 3s interval, i. a The current graphs are as follows: the third row is the iq current graph, the fourth row is the id current graph, and the fifth row is the i... q Spectral analysis in the 1.5s to 3s interval. Figure 3 It can be seen that under a load torque of 0.15 N·m, the i of ADRC-ESO a The current exhibits significant distortion, with a THD of 5.99%. CESO improves its low-frequency disturbance estimation capability through a cascaded observation structure, reducing the THD to 5.18%, but its suppression of periodic harmonics remains insufficient. In contrast, IESO's i a The waveform is closer to a sine wave, and THD has decreased to 1.96%. Meanwhile, under ISO... q with i d The current fluctuations are approximately 0.10A and 0.12A, respectively, both less than ADRC-ESO and CESO, indicating that the method proposed in this invention can simultaneously improve the steady-state quality of torque current and excitation current.

[0088] Those skilled in the art will understand that embodiments of this application can be provided as methods, systems, or computer program products. Therefore, this application can take the form of a completely hardware embodiment, a completely software embodiment, or an embodiment combining software and hardware aspects. Furthermore, this application can take the form of a computer program product implemented on one or more computer-usable storage media (including but not limited to disk storage, CD-ROM, optical storage, etc.) containing computer-usable program code. The solutions in the embodiments of this application can be implemented in various computer languages, such as the object-oriented programming language Python and the interpreted scripting language JavaScript.

[0089] Other embodiments of the invention will readily occur to those skilled in the art upon consideration of the specification and practice of the invention disclosed herein. This application is intended to cover any variations, uses, or adaptations of the invention that follow the general principles of the invention and include common knowledge or customary techniques in the art not disclosed herein. The specification and examples are to be considered exemplary only, and the true scope and spirit of the invention are indicated by the following claims.

[0090] It should be understood that the present invention is not limited to the content already described above, and various modifications and changes can be made without departing from its scope. The scope of the present invention is limited only by the appended claims.

Claims

1. A hierarchical parallel active disturbance rejection control method for suppressing complex multi-frequency disturbances in a permanent magnet synchronous motor, characterized in that, include: S1: Collect the three-phase sub-currents of the permanent magnet synchronous motor, and obtain the actual d-axis current and q-axis current in the synchronous rotating dq coordinate system through Clark transformation and Park transformation; The actual speed of the permanent magnet synchronous motor is collected; the desired q-axis current is obtained by using the speed loop PI controller and the difference between the desired speed and the actual speed, and the desired d-axis current is set to 0. S2: Construct parallel decoupled dual-branch observers for the d-axis current loop and the q-axis current loop respectively, estimate the two disturbance components respectively, and superimpose them to obtain the total disturbance estimate of the corresponding current loop. The parallel decoupled dual-branch observers for the d-axis and q-axis current loops have the same structure. The dual-branch observer includes a cascaded extended state observer branch and a resonant harmonic observer branch with parallel architecture. The transfer functions of the two branches are completely decoupled, and the parameters are independently adjustable. The cascaded extended state observer branch is used to estimate low-frequency and aperiodic disturbances, employing a two-stage series structure. The first-stage observer is used to coarsely estimate the disturbance of the current loop system, and the second-stage observer is used to re-estimate and compensate for the estimation error of the first-stage observer. The resonant harmonic observer branch is used to estimate the 6th and 12th current harmonic disturbances in the current loop. S3: The total disturbance estimate is compensated into the control law to obtain the control voltage, wherein the control law consists of three parts: a proportional feedback term, a disturbance compensation term, and a reference differential feedforward term; S4: Input the control voltage and the real-time rotor electrical angle into the inverse Park transform module, and convert it into... Coordinate system voltage The signal is fed into the SVPWM (Space Vector Pulse Width Modulation) module for pulse width modulation, outputting six control signals. These six control signals are then input into the inverter to output three-phase control current. Drives the permanent magnet synchronous motor to rotate.

2. The hierarchical parallel active disturbance rejection control method for suppressing complex multi-frequency disturbances in a permanent magnet synchronous motor according to claim 1, characterized in that: The mathematical model of the first-stage observer in the cascaded extended state observer branch of the q-axis current loop is as follows: ； In the formula, This is the estimated q-axis current output from the first-stage observer. For the estimated perturbation of the first-level observer, This represents the actual current along the q-axis. This is the output voltage of the q-axis current loop. , This is the proportional gain of the first-stage observer; For the gain of the controlled object in the current loop; The mathematical model for the second-level observer is as follows: ； In the formula, This is the amount of compensation for the error in the first-stage current estimation. It is the compensation amount for the estimation error of the first-level disturbance. , This is the second-level observation gain; The total perturbation estimate for the cascaded extended state observer is: .

3. The hierarchical parallel active disturbance rejection control method for suppressing complex multi-frequency disturbances in a permanent magnet synchronous motor according to claim 2, characterized in that: , This represents the observation bandwidth of the cascaded extended state observer.

4. The hierarchical parallel active disturbance rejection control method for suppressing complex multi-frequency disturbances in a permanent magnet synchronous motor according to claim 1, characterized in that: The perturbation estimation error transfer function of the cascaded extended state observer is as follows: ； in, This represents the observation bandwidth of the cascaded extended state observer.

5. The hierarchical parallel active disturbance rejection control method for suppressing complex multi-frequency disturbances in a permanent magnet synchronous motor according to claim 1, characterized in that: The resonant harmonic observer branch includes a pre-low-frequency disturbance filtering module, a 6th resonant observer, and a 12th resonant observer. The pre-low-frequency disturbance filtering module filters out low-frequency disturbances and aperiodic disturbances, retaining only the periodic disturbance components. The 6th and 12th resonant observers are connected in series after the pre-low-frequency disturbance filtering module to extract the 6th and 12th harmonic disturbance components in the current.

6. The hierarchical parallel active disturbance rejection control method for suppressing complex multi-frequency disturbances in a permanent magnet synchronous motor according to claim 5, characterized in that: The pre-amplifier low-frequency disturbance filtering module uses a classic ESO to build the filtering unit, as shown in the following formula: ； In the formula, This is an estimated value for the q-axis current. The q-axis current observation error is defined as follows: , This represents the actual current along the q-axis. This is the output voltage of the q-axis current loop. This is the estimated total disturbance of the resonant harmonic observer branch; For fundamental / aperiodic disturbance components, This is the 6th harmonic disturbance component; It is the 12th harmonic disturbance component.

7. The hierarchical parallel active disturbance rejection control method for suppressing complex multi-frequency disturbances in a permanent magnet synchronous motor according to claim 6, characterized in that: A 6th-order and 12th-order resonant observers are constructed using a second-order generalized integrator structure, as shown in the following models: ； In the formula, These are the orthogonal auxiliary state variables of the 6th and 12th resonance observers, respectively; These are the resonant angular frequencies of the 6th and 12th harmonics, respectively. This is the resonant damping coefficient; is the observer gain coefficient.

8. The hierarchical parallel active disturbance rejection control method for suppressing complex multi-frequency disturbances in a permanent magnet synchronous motor according to claim 7, characterized in that: ； in, The expected convergence bandwidth of the resonant channel. , This is the gain coefficient.

9. The hierarchical parallel active disturbance rejection control method for suppressing complex multi-frequency disturbances in a permanent magnet synchronous motor according to claim 7, characterized in that: The transfer function molecule of the resonant unit contains a first-order lead element with lead phase compensation.

Images