A maximum four-vector SVPWM-based double three-phase permanent magnet motor harmonic suppression method

Abstract

The application discloses a double three-phase permanent magnet motor harmonic suppression method based on maximum four-vector SVPWM and belongs to the technical field of motor control. The method comprises the following steps: a signal acquisition and coordinate transformation step, which detects the rotating speed and six-phase current and obtains the fundamental wave and harmonic component through vector space decoupling; a nonlinear MTPA instruction generation step, which adopts Newton-Raphson iteration to solve the optimal current angle based on a cross-coupling coefficient table and outputs the d-q axis current given value; a double-plane current regulation and harmonic suppression step, which uses PI control to regulate the fundamental wave and adopts adaptive resonance and fuzzy multi-mode control to realize smooth switching and output voltage instruction; a voltage synthesis and modulation step, which combines the fundamental wave and harmonic voltage into a four-dimensional target vector after inverse transformation of the fundamental wave, realizes the maximum four-vector SVPWM of over-modulation smooth transition through nonlinear partition look-up table and generates inverter pulses; and a closed-loop iteration step. The application realizes the improvement of efficiency, harmonic suppression precision and voltage utilization rate in a wide operating condition range.

Description

Technical Field

[0001] This invention falls within the intersection of the fields of motor control technology and power electronics technology, specifically relating to a harmonic suppression method for dual three-phase permanent magnet motors based on maximum four-vector SVPWM. Background Technology

[0002] Dual three-phase permanent magnet motors are widely used in electric vehicles, ship propulsion, and aerospace due to their advantages of high power density, high efficiency, and low torque ripple. However, dual three-phase motors have complex winding structures and inherent characteristics of strong coupling, high-order harmonics, and multiple harmonics. Their high-performance control relies on accurate mathematical models and advanced control strategies. Existing dual three-phase motor control technologies still have significant shortcomings in parameter adaptability, harmonic suppression capability, and voltage utilization, making it difficult to meet the stringent requirements of high-end applications for motor operating efficiency, output quality, and dynamic response capabilities. Specifically:

[0003] MTPA control neglects cross-saturation effects, resulting in poor parameter adaptability. Existing maximum torque-to-current ratio (MTPA) control is mostly based on ideal linear magnetic circuit models, assuming independent d-axis and q-axis flux linkages and failing to account for cross-saturation effects. In actual operation, the coupling of d-axis and q-axis currents leads to mutual interference of flux linkages. Traditional MTPA, by neglecting the cross-coupling coefficient, suffers from parameter mismatch, causing deviations in the optimal current angle calculation. This prevents the motor from outputting maximum torque per unit current, especially under heavy load and deep field weakening conditions, resulting in significant efficiency losses.

[0004] Harmonic suppression strategies are often simplistic, making it difficult to balance dynamic and steady-state performance. Dual-phase and three-phase motors exhibit abundant harmonic components, and current harmonic suppression methods primarily employ fixed-parameter resonant controllers and simple feedforward compensation. Fixed-parameter resonant controllers cannot adapt to harmonic frequency shifts caused by speed variations, resulting in fluctuating suppression effectiveness depending on operating conditions. Simple feedforward compensation relies on precise mathematical models, leading to poor robustness. More importantly, existing methods lack a coordinated switching mechanism between dynamic and steady-state modes, making it difficult to achieve a balance between high-gain harmonic suppression and rapid dynamic response. This often results in good steady-state suppression while exacerbating dynamic oscillations.

[0005] SVPWM overmodulation handling is crude, highlighting the trade-off between voltage utilization and output smoothness. Existing six-phase SVPWM methods for handling the overmodulation region often employ simple limiting and scaling, failing to systematically consider the continuity of modulation ratio changes across regions. During the transition from the linear to the overmodulation region, traditional methods easily lead to output voltage distortion and increased torque ripple. In the deep overmodulation region, the lack of effective constraints on harmonic plane voltages causes a surge in harmonic components, affecting motor operational stability. Furthermore, existing methods suffer from complex calculations of the action time in the overmodulation region, resulting in poor real-time performance and failing to meet the requirements of control cycle and computational efficiency in high-speed operation scenarios.

[0006] The control architecture lacks systematic integration and has insufficient multi-objective collaborative optimization capabilities. Existing research mostly focuses on improving single components, lacking a systematic solution that organically integrates parameter optimization, harmonic suppression, and overmodulation control. The coupling relationships between various control components have not been effectively addressed, resulting in overall control performance being limited by the weakest link, making it difficult to simultaneously achieve high efficiency, high quality, and high voltage utilization multi-objective collaborative optimization across a wide range of operating conditions.

[0007] In summary, the existing technology has neither disclosed nor taught a comprehensive control method that can systematically integrate nonlinear MTPA considering cross-saturation, adaptive resonance and multi-mode switching composite control, and nonlinear maximum four-vector SVPWM supporting overmodulation smooth transition. There is an urgent need to provide an innovative solution with the above-mentioned synergistic optimization capabilities. Summary of the Invention

[0008] This invention addresses the core problems of existing dual-three-phase permanent magnet motor control technology under complex operating conditions, namely insufficient efficiency optimization, poor output quality, and low voltage utilization due to dispersed control architecture, parameter model mismatch, and fragmented harmonic suppression and overmodulation processing. It proposes an innovative solution. Specifically, existing technologies generally suffer from the following technical challenges: First, regarding the adaptability of the control architecture and parameters, traditional maximum torque-current ratio (MTPA) control is often based on an ideal linear magnetic circuit model, neglecting the cross-saturation effect between the d-axis and q-axis flux linkages. This leads to parameter mismatch and deviations in the optimal current angle calculation, preventing the motor from outputting maximum torque per unit current, especially under heavy load and deep field weakening conditions, resulting in significant efficiency losses. Second, regarding the balance between harmonic suppression and dynamic steady-state performance... Existing harmonic suppression strategies are simplistic, and fixed-parameter resonant controllers cannot adapt to harmonic frequency shifts caused by speed variations. Simple feedforward compensation relies on precise mathematical models, has poor robustness, and lacks a coordinated switching mechanism between dynamic and steady-state modes, making it difficult to achieve a balance between high-gain harmonic suppression and fast dynamic response. Finally, regarding overmodulation processing and multi-objective coordination, traditional SVPWM often uses simple limiting and scaling in the overmodulation region, without systematically considering the continuity of modulation ratio changes across regions. This leads to output voltage distortion and increased torque ripple. Furthermore, it lacks a systematic solution that organically integrates parameter optimization, harmonic suppression, and overmodulation control, making it difficult to achieve high efficiency, high quality, and high voltage utilization in a wide range of operating conditions.

[0009] To address this, the present invention proposes a harmonic suppression method for dual three-phase permanent magnet motors based on maximum four-vector SVPWM. By constructing a control architecture that considers cross-saturation nonlinear MTPA, adaptive resonance and multi-mode switching harmonic suppression, and nonlinear maximum four-vector SVPWM supporting smooth overmodulation transition, the method systematically improves the motor's efficiency optimization capability, harmonic suppression accuracy, and voltage utilization over a wide operating range. The method includes the following steps:

[0010] Signal acquisition and coordinate transformation steps: The motor speed and six-phase current are detected in real time by speed sensor and current sensor. The six-phase current is decomposed into fundamental plane α-β component and harmonic plane xy component by vector space decoupling transformation matrix, and the dq axis current feedback value is obtained by synchronous rotation transformation.

[0011] MTPA command generation steps: The stator current amplitude is given by adjusting the speed PI. Based on the electromagnetic torque equation considering the cross-coupling coefficient, the optimal current angle is solved by Newton-Raphson iteration, and the dq axis current setpoint is calculated and output.

[0012] Dual-plane current regulation and harmonic suppression steps: The fundamental plane uses PI regulation to output the fundamental voltage setpoint; the harmonic plane uses an adaptive resonant controller and fuzzy logic multi-mode switching, smoothly transitioning between PIR mode and feedforward mode according to the operating conditions, and outputting the harmonic plane voltage setpoint.

[0013] Target vector synthesis and SVPWM modulation steps: The fundamental voltage is given and inverse Park transforms to obtain the α-β axis voltage, which is combined with the harmonic voltage to form a four-dimensional target voltage vector; the adjacent maximum vector is selected by sector judgment. The basic vector is the voltage vector with a large projection amplitude in the α-β plane under the given DC bus voltage condition, and the vector combination with a high contribution to the synthesis of the target voltage vector is selected first. The action time is calculated by looking up the coefficient matrix of the partition, and after normalization and zero vector allocation, a six-phase inverter switching pulse sequence is generated.

[0014] Drive and iterative cycle steps: The SVPWM pulse is amplified by the drive circuit and then the inverter outputs six-phase AC power to drive the motor. After the current control cycle is completed, the signal acquisition step is returned to continue the iteration.

[0015] Furthermore, the MTPA instruction generation step further includes:

[0016] The offline calibration sub-step for cross-coupling coefficients is pre-constructed using finite element simulation. , Given a two-dimensional lookup table as input, output the cross-coupling coefficient:

[0017]

[0018] in, , These are the magnetic flux linkages along the d and q axes, respectively. Characterizes the rate of change of d-axis flux linkage with q-axis current; Characterizes the rate of change of q-axis flux linkage with d-axis current;

[0019] The sub-steps for solving the optimal current angle in nonlinear MTPA are based on the electromagnetic torque equation considering cross-coupling:

[0020]

[0021] in, Electromagnetic torque;

[0022] make Substitute into the above formula and... Taking the partial derivative and setting it to zero, we obtain the nonlinear equation:

[0023]

[0024] The solution is obtained using the Newton-Raphson iterative method:

[0025]

[0026] in, For the first The estimated current angle for the next iteration; For function exist The first derivative value at;

[0027] The initial value for iteration is taken from the traditional MTPA solution:

[0028]

[0029] Each iteration is based on the current Table lookup and update To obtain the optimal current angle ;

[0030] Command current distribution sub-step, utilizing Calculate the dq-axis current setting:

[0031]

[0032] When this formula is for surface-mounted motors Then take directly .

[0033] Furthermore, the dual-plane current regulation and harmonic suppression steps further include:

[0034] The fundamental plane PI regulator step will change the command current. Feedback The error is fed into the PI controller:

[0035]

[0036]

[0037] in, The proportional and integral coefficients of the d-axis current loop; , These are the proportional and integral coefficients of the q-axis current loop; Electric angular velocity; For the Laplace operator; the output fundamental voltage is given. ;

[0038] The design sub-steps of the adaptive resonant controller include the design of harmonic currents. The error between the input and its given value (zero) is fed into a proportional-integral-resonant controller, whose transfer function is:

[0039]

[0040] in, This is the proportionality coefficient; The integral coefficient; Let h be the resonant gain of the h-th resonant term; This is the cutoff frequency of the resonant term; The fundamental angular frequency;

[0041] resonant frequency The controller adaptively adjusts with rotational speed to achieve high gain suppression of the 5th, 7th, 11th, and 13th harmonics; the controller output is denoted as... ;

[0042] Fuzzy logic multi-mode switching sub-step, real-time calculation of current change rate and the rate of change of rotational speed The fuzzy classification is divided into five levels (VS, S, M, B, VB), using a triangular membership function.

[0043] in:

[0044] , These are the differences between the x-axis and y-axis harmonic currents in adjacent control cycles;

[0045] The sampling time interval (equal to the control period) ;

[0046] Let be the rate of change of rotational speed; the universes of discourse for input variables Δi and Δω are both normalized to the interval [0,1]. The parameters of the triangular membership function for each fuzzy level are defined as follows:

[0047] VS: Vertex 0, base [0, 0.25];

[0048] S: Vertex 0.25, base [0, 0.5];

[0049] M: Vertex 0.5, base [0.25, 0.75];

[0050] B: Vertex 0.75, base [0.5, 1];

[0051] VB: Vertex 1, bottom edge [0.75, 1];

[0052] The fuzzy rule base is as follows: IF is VS AND is VS THEN mode = steady state (PIR)

[0053] IF is M OR is M THEN mode = Transition (PIR + Feedforward Weighted)

[0054] IF is B OR is B THEN mode = dynamic (feedforward)

[0055] The centroid method is used to defuzzify the image, and the weighting coefficients are calculated as follows:

[0056] in, For the first The activation level of the rule, For the single-point value of the corresponding rule output mode: steady-state mode takes Transition mode Dynamic mode ;get Subsequently, the harmonic voltage command is:

[0057]

[0058] in, The given value for the x-axis harmonic voltage. The given value for the y-axis harmonic voltage. This refers to the x-axis voltage command output by the PIR controller. This refers to the y-axis voltage command output by the PIR controller. This refers to the x-axis voltage command output by the feedforward decoupling module. The y-axis voltage command output by the feedforward decoupling module;

[0059] The feedforward decoupling control sub-step, in dynamic mode, is based on the xy subspace voltage equation:

[0060]

[0061] in, , These are the actual harmonic plane voltages along the x-axis and y-axis, respectively. , These are the harmonic current feedback values ​​along the x-axis and y-axis, respectively. , The harmonic inductances are for the x-axis and y-axis, respectively. , These represent the rates of change of harmonic currents along the x-axis and y-axis, respectively. , This is the product of inductance and current in the coupling term. These are the harmonic back electromotive forces along the x-axis and y-axis, respectively.

[0062] Take the feedforward compensation value:

[0063]

[0064] in, , For x and y axis harmonic inductance; , The back EMFs of the x and y axes harmonics are obtained by looking up tables based on the motor structure and speed using offline calibration data. The offline calibration method is as follows:

[0065] First, based on the finite element model of the motor, several speed sampling points are selected within the rated speed range (e.g., 0 to 1.2 times the rated speed). At each sampling point, the motor is driven to that speed by an external prime mover, with the six-phase windings open-circuited (without excitation), and the open-circuit phase voltages of the six-phase windings are measured. The measured six-phase voltages are then subjected to vector space decoupling transformation to obtain the open-circuit voltage components in the xy subspace. These components represent the harmonic back EMF at the corresponding speed. , .

[0066] Furthermore, the target vector synthesis and SVPWM modulation steps further include:

[0067] The multidimensional target vector synthesis sub-step will Harmonic voltage command Combined, a four-dimensional target voltage vector is formed:

[0068]

[0069] The sector determination and basic vector selection sub-steps are based on... The sign and size relationship of the symbol are used to determine its sector in the α-β plane using direct logical judgment. According to sector number Retrieve the four adjacent largest vectors V1, V2, V3, and V4 from the pre-stored vector table;

[0070] The nonlinear partitioning lookup table calculation sub-step divides the modulation range into multiple sub-regions, and the modulation ratio... Defined as:

[0071]

[0072] in, The amplitude of the fundamental voltage. The maximum amplitude of the fundamental voltage output by the six-phase inverter under square wave conditions; for each sub-region, based on the volt-second balance equation:

[0073]

[0074] in, For the first The projections of the basic vectors onto the α, β, x, and y axes; The duration of action of the corresponding vector; For switching cycles;

[0075] Let the projection matrix

[0076]

[0077] The volt-second equilibrium equation can then be written as: ,in Within the sub-region, the selected basic vectors typically form an effective spanning set of the target voltage vector, and their projection matrix satisfies the full-rank condition in most operating conditions. Therefore, the action time of each vector can be solved using matrix inversion or pseudo-inversion methods. Reversible, solution obtained To facilitate online calculations, a coefficient matrix is ​​defined. , so that:

[0078]

[0079] in: For DC bus voltage; when the projection matrix If a pathological condition occurs, Moore-Penrose pseudoreversal is used. Instead of the ordinary inverse matrix, i.e. To ensure numerical stability;

[0080] The coefficient matrix of all sub-regions is pre-solved and stored in flash memory;

[0081] During online calculation, based on the modulation ratio Locate the current sub-region, retrieve the corresponding coefficient matrix, and quickly obtain the result through multiplication and addition operations. When m crosses the boundary of a sub-region, weighted fusion is used:

[0082]

[0083] in: , , The modulation ratio boundary of adjacent sub-regions; ;

[0084] Normalization and zero vector allocation sub-steps to calculate total effective time = ;when < Then the zero vector action time Select zero vector and Combined to balance the midpoint potential; when ≥ Then scale proportionally. ,make ,at this time ;

[0085] The switching sequence generation sub-step generates a seven-segment sequence based on the principle of minimizing switching losses and ensuring waveform central symmetry.

[0086] The beneficial effects of this invention are as follows:

[0087] (1) This invention is the first to organically integrate three core components: nonlinear MTPA considering cross-saturation, adaptive resonance and multi-mode switching harmonic suppression, and nonlinear maximum four-vector SVPWM supporting smooth overmodulation transition. This forms an integrated closed-loop control architecture from signal acquisition, coordinate transformation, command generation, dual-plane current regulation to multi-dimensional voltage modulation. Compared with the existing technology where the control modules are isolated from each other and parameter optimization and harmonic suppression are disconnected, this invention effectively improves the overall performance of the system over a wide range of operating conditions through information coordination and feedback between steps, and solves the three long-standing technical problems of "parameter mismatch", "contradiction between harmonic suppression and dynamic response", and "coarse overmodulation processing".

[0088] (2) In step S3, this invention innovatively introduces the cross-saturation effect into MTPA modeling. A two-dimensional lookup table is constructed by offline calibration of the cross-coupling coefficients, and the nonlinear optimal current angle equation is solved using Newton-Raphson iteration. This solves the parameter mismatch problem caused by neglecting the mutual coupling of d-axis and q-axis flux linkages in traditional MTPA. This technical solution integrates the nonlinear characteristics of cross-saturation into real-time control for the first time, improving the accuracy of current angle calculation and significantly enhancing the torque output per unit current under heavy load and deep field weakening conditions. Compared with the traditional linear MTPA model, this invention ensures that the motor always operates at the true optimal efficiency point across the entire operating range, effectively avoiding efficiency loss caused by parameter mismatch.

[0089] (3) In step S4 of this invention, the adaptive resonant controller and the fuzzy logic multi-mode switching mechanism are deeply integrated for the first time to construct a composite control strategy of "adaptive resonance + multi-mode switching". By adaptively adjusting the resonant frequency of the PIR controller online with the rotational speed, the problem that fixed resonant parameters cannot adapt to harmonic frequency drift is solved. By using a fuzzy logic switching module based on the rate of change of current and the rate of change of rotational speed (the input variables Δi and Δω are normalized to [0,1], a triangular membership function is used and the parameters are clear, and the centroid method is used for defuzzification), a smooth transition and weighted fusion between the steady-state PIR mode and the dynamic feedforward mode is realized. This combined technical solution enables the system to achieve high gain and zero steady-state error suppression of the 5th, 7th, 11th and 13th harmonics in steady state, and ensures fast response and system stability in dynamic process. It fundamentally solves the contradiction between "high gain suppression and fast dynamic response" in the prior art, and the current harmonic distortion rate can be reduced by more than 60%.

[0090] (4) In step S6 of this invention, the maximum four-vector SVPWM method of "nonlinear partition lookup + weighted fusion smooth transition" is proposed and implemented for the first time. By finely dividing the modulation range into linear region, over-modulation I region, and over-modulation II region and pre-solving the coefficient matrix for each region, the complex nonlinear online calculation is simplified to lookup table and multiplication-addition operation, which greatly improves the real-time performance. At the same time, by using weighted fusion processing of the coefficient matrix output of adjacent sub-regions, a smooth and seamless transition from the linear region to the over-modulation region and even to the square wave condition is realized, which solves the problem of output voltage distortion and torque pulsation caused by simple amplitude limiting and scaling in the over-modulation region of traditional SVPWM. This technical solution maximizes voltage utilization and effectively improves the load-carrying capacity and running stability of the motor in the high-speed weak magnetic region.

[0091] (5) Through the synergistic effect of steps S3, S4, and S6, this invention constructs a complete technical closed loop integrating "parameter optimization, harmonic suppression, and overmodulation control," forming a systematic solution from perception, decision-making, execution to optimization. Compared to the existing technology that only focuses on improving a single link and where each module is independent, this invention achieves multi-objective collaborative optimization through the effective transmission and feedback of information flow between each step: the optimal current command output by the MTPA algorithm provides an ideal fundamental operating point for harmonic suppression; the voltage command output by harmonic suppression directly participates in SVPWM modulation; and the overmodulation strategy ensures smooth output voltage while also considering the constraints of the harmonic plane. This systematic integration enables the motor to effectively reduce current harmonic content, improve motor operating efficiency, and enhance DC bus voltage utilization under typical operating conditions.

[0092] (6) In step S2, this invention employs vector space decoupling transformation to accurately decompose the six-phase current into fundamental and harmonic plane components, laying a theoretical foundation for subsequent independent control. In step S5, a four-dimensional target voltage vector is constructed, unifying the fundamental and harmonic voltage commands within the same mathematical framework, thus achieving precise full-dimensional control of the inverter output voltage. In step S7, a complete control cycle iteration mechanism is formed to ensure the real-time dynamic closed-loop operation of the system. These quantitative models and control architectures provide precise mathematical tools and complete implementation paths for the high-performance control of dual three-phase permanent magnet motors, changing the traditional control method's crude approach of "simple processing" of harmonic components, and significantly enhancing the system's scientific nature, controllability, and reproducibility. Attached Figure Description

[0093] Figure 1 This is a flowchart of a harmonic suppression method for dual three-phase permanent magnet motors based on maximum four-vector SVPWM.

[0094] Figure 2 A flowchart for generating the optimal current command for nonlinear MTPA considering cross-saturation.

[0095] Figure 3 The flowchart shows the harmonic suppression control process for adaptive resonance and fuzzy multi-mode switching. Figure 4 The flowchart shows the maximum four-vector SVPWM modulation process for nonlinear partition lookup and overmodulation smooth transition.

[0096] Figure 5 The simulation diagram shows the bus current waveform of a dual three-phase permanent magnet motor controlled by this method.

[0097] Figure 6 The simulation diagram shows the Fourier analysis of the current harmonics of a dual three-phase permanent magnet motor controlled by this method.

[0098] Figure 7The waveform of the speed change of the dual three-phase permanent magnet motor controlled by this method is a simulation diagram. Detailed Implementation

[0099] This implementation focuses on high-performance harmonic suppression and efficiency optimization of dual three-phase permanent magnet motors. Relying on a control architecture based on vector space decoupling and maximum four-vector SVPWM, it achieves integrated control from signal acquisition, coordinate transformation, MTPA command generation, dual-plane current regulation, multi-dimensional voltage synthesis to overmodulated SVPWM output. In the three core stages of MTPA command generation, harmonic current suppression, and SVPWM modulation, it deeply integrates innovative algorithms based on cross-saturation nonlinear modeling, adaptive resonance and multi-mode switching, and nonlinear partitioned lookup tables, effectively improving the system's efficiency optimization capability, harmonic suppression accuracy, and voltage utilization over a wide operating range. Figure 1 The following are the specific implementation steps:

[0100] Step S1: The control system is initialized and the motor running status is collected in real time. The motor speed and six-phase current are detected in real time through speed sensors and current sensors. The speed deviation is calculated and the raw six-phase current data is output to provide the basic input for subsequent control.

[0101] Step S2: Based on vector space decoupling, current coordinate transformation and component extraction are performed. Using a preset decoupling transformation matrix and rotation transformation, the six-phase current is decomposed into fundamental plane α-β components, harmonic plane xy components and dq axis current feedback values ​​in synchronous rotating coordinate system, and the decoupled current components are output.

[0102] Step S3: Speed ​​closed-loop regulation and nonlinear MTPA command current generation considering cross saturation. The stator current amplitude is given by speed PI regulation. Based on the offline calibrated cross coupling coefficient lookup table, the optimal current angle is solved by Newton-Raphson iteration. The dq axis current setpoint is calculated and output.

[0103] Step S4: Dual-plane current closed-loop regulation and adaptive resonant harmonic active suppression. The fundamental plane uses PI regulation to output the fundamental voltage setpoint; the harmonic plane uses an adaptive resonant controller and fuzzy logic multi-mode switching, smoothly transitioning between PIR mode and feedforward mode according to the operating conditions, and outputs the harmonic plane voltage setpoint.

[0104] Step S5: Inverse coordinate transformation of voltage reference value and construction of multidimensional target vector. The fundamental voltage is given by inverse Park transformation to obtain α-β axis voltage, which is combined with the harmonic voltage to form a four-dimensional target voltage vector. The multidimensional voltage vector to be synthesized is output.

[0105] Step S6: Support nonlinear maximum four-vector SVPWM with overmodulation transition. Select adjacent maximum vectors by sector judgment. The basic vector is the voltage vector with a large projection amplitude in the α-β plane under a given DC bus voltage condition. Prioritize the selection of vector combinations that have a high contribution to the synthesis of the target voltage vector. Calculate the action time based on the coefficient matrix pre-stored in the partition. After normalization and zero vector allocation, generate a six-phase inverter switching pulse sequence and output drive pulses.

[0106] Step S7: Drive the inverter and control the loop iterate. Amplify the SVPWM pulse through the drive circuit to control the inverter to output six-phase AC power to drive the motor. After completing the current control cycle, return to step S1 to continue the iteration.

[0107] Furthermore, in step S1, the system initialization and signal acquisition module includes a parameter preset unit for storing motor parameters, a speed sensor interface unit for detecting rotational speed, and a current sensor and analog-to-digital converter unit for synchronously sampling six-phase current; the process of performing initialization and real-time acquisition includes:

[0108] 1. Parameter Presetting and Initialization: During system startup, the digital signal processor reads the inherent parameters of the motor from the non-volatile memory, including the number of pole pairs. Stator resistance Direct-axis inductor quadrature axis inductance Permanent magnet flux And load the initial coefficients of the proportional-integral controller for the speed loop and current loop;

[0109] 2. Speed ​​detection and deviation calculation: The mechanical angular velocity ω of the motor is detected in real time using a photoelectric encoder. m , with the externally given rotational speed ω ∗ Compare and calculate the speed deviation Δω=ω ∗ −ω m ;

[0110] 3. Six-phase current synchronous sampling: Six high-precision Hall current sensors and a 12-bit analog-to-digital converter synchronously sample the instantaneous current of the six-phase windings at the trigger moment of each PWM cycle. After digital filtering, the output is sent to the subsequent conversion module.

[0111] Furthermore, in step S2, the coordinate transformation module includes a Clark transformation unit for static coordinate transformation and a Park transformation unit for synchronous rotating coordinate transformation; the process of performing current decoupling and component extraction includes:

[0112] 1. Six-phase stationary coordinate transformation: Transforming the six-phase current... Inputting the Clark transform module, and projecting it using the vector space decoupling transformation matrix T, yields three mutually orthogonal subspace components:

[0113]

[0114] in, For the fundamental energy conversion plane component, The harmonic loss plane component, under neutral point isolation, is replaced by the zero-sequence component. neglect;

[0115] 2. Fundamental current synchronous rotational conversion: [This part is incomplete and requires further context to translate accurately.] Together with rotor position angle (Obtained by integration from the speed sensor) Input to the Park transformation unit, and perform rotational transformation according to the rotor magnetic field orientation:

[0116]

[0117] Output DC excitation current feedback and torque current feedback At the same time, the aforementioned As a harmonic current component, it is directly output to the subsequent control module.

[0118] Furthermore, in step S3, the MTPA command generation module includes a PI regulator unit for speed closed-loop control, a two-dimensional lookup table unit for storing cross-coupling coefficients, and a numerical iterative solver unit for solving the optimal current angle; as shown below. Figure 2 The process of generating the dq-axis current command includes:

[0119] 1. Outer loop PI control of speed: The speed deviation Δω obtained in step S1 is sent to the speed PI controller, and its discretized form is as follows:

[0120]

[0121] in, The given value is the stator current amplitude at the current moment; This represents the current discrete time index; , These are the proportional coefficient and integral coefficient of the speed loop, respectively. The control cycle (PWM switching cycle); To sum the index variables, iterate from 0 to... At each discrete moment; Δ For the first Rotational speed deviation per sampling period;

[0122] 2. Offline calibration of cross-coupling coefficients: Through finite element simulation, a pre-built system is constructed... , Given a two-dimensional lookup table as input, output the cross-coupling coefficient:

[0123]

[0124] in, , These are the magnetic flux linkages along the d and q axes, respectively. Characterizes the rate of change of d-axis flux linkage with q-axis current; Characterizes the rate of change of q-axis flux linkage with d-axis current;

[0125] The table data is stored in a grid format with a step size of 0.1 times the rated current;

[0126] 3. Sub-steps for solving the optimal current angle in nonlinear MTPA, based on the electromagnetic torque equation considering cross-coupling:

[0127]

[0128] in, Electromagnetic torque;

[0129] make Substitute into the above formula and... Taking the partial derivative and setting it to zero, we obtain the nonlinear equation:

[0130]

[0131] The solution is obtained using the Newton-Raphson iterative method:

[0132]

[0133] in, For the first The estimated current angle for the next iteration; For function exist The first derivative value at;

[0134] The initial value for iteration is taken from the traditional MTPA solution:

[0135]

[0136] Each iteration is based on the current Table lookup and update To obtain the optimal current angle To ensure the convergence and computational efficiency of the iterative process, a convergence criterion is set: iteration stops when the absolute value of the change in current angle between two consecutive iterations is less than a preset threshold ε, i.e., the following condition is met:

[0137] <ε

[0138] Where ε is the convergence accuracy threshold, and in this embodiment, ε is taken as 1×10^{-4} rad (approximately 0.0057°). The maximum number of iterations is also set to 10 to prevent control timeouts due to non-convergence during iterations under extreme conditions.

[0139] 4. Command current distribution sub-step, utilizing... Calculate the dq-axis current setting:

[0140]

[0141] When it is a surface-mount motor Then take directly .

[0142] Furthermore, the dual-plane current regulation and harmonic suppression steps further include:

[0143] 1. Fundamental plane PI regulation: This adjusts the command current... Feedback The error is fed into the PI controller:

[0144]

[0145]

[0146] in, The proportional and integral coefficients of the d-axis current loop; , These are the proportional and integral coefficients of the q-axis current loop; Electric angular velocity; For the Laplace operator; the output fundamental voltage is given. ;

[0147] 2. Adaptive Resonant Controller Design: This design incorporates harmonic currents. The error between the input and its given value (zero) is fed into a proportional-integral-resonant controller, whose transfer function is:

[0148]

[0149] in, This is the proportionality coefficient; The integral coefficient; Let h be the resonant gain of the h-th resonant term; This is the cutoff frequency of the resonant term; The fundamental angular frequency;

[0150] resonant frequency The controller adaptively adjusts with rotational speed to achieve high gain suppression of the 5th, 7th, 11th, and 13th harmonics; the controller output is denoted as... ;

[0151] 3. Fuzzy logic multi-mode switching: Real-time calculation of current change rate and the rate of change of rotational speed The fuzzy classification is divided into five levels (VS, S, M, B, VB), using a triangular membership function.

[0152] in:

[0153] , These are the differences between the x-axis and y-axis harmonic currents in adjacent control cycles;

[0154] The sampling time interval (equal to the control period) ;

[0155] Let be the rate of change of rotational speed; the universes of discourse for input variables Δi and Δω are both normalized to the interval [0,1]. The parameters of the triangular membership function for each fuzzy level are defined as follows:

[0156] VS: Vertex 0, base [0, 0.25];

[0157] S: Vertex 0.25, base [0, 0.5];

[0158] M: Vertex 0.5, base [0.25, 0.75];

[0159] B: Vertex 0.75, base [0.5, 1];

[0160] VB: Vertex 1, bottom edge [0.75, 1];

[0161] The fuzzy rule base is as follows:

[0162] IF is VS AND is VS THEN mode = steady state (PIR)

[0163] Meaning: The rate of change of harmonic current and the rate of change of rotation speed are both very small. A pure PIR controller is used to perform high-precision, high-gain steady-state suppression of harmonics, aiming to achieve the lowest current harmonic distortion rate.

[0164] IF is M OR is M THEN mode = Transition (PIR + Feedforward Weighted)

[0165] The harmonic current change rate is moderate or the speed change rate is moderate (at least one of the two conditions must be met). PIR control and feedforward weighted fusion control are adopted (each contributing half). While maintaining a certain harmonic suppression capability, feedforward compensation is introduced to accelerate the response and prevent harmonic divergence.

[0166] IF is B OR is B THEN mode = dynamic (feedforward)

[0167] The harmonic current change rate is large or the speed change rate is large (at least one of the two conditions must be met). The system is then fully switched to feedforward control, directly compensating for harmonic coupling terms and back EMF based on the voltage equation, suppressing instantaneous harmonic current surges as quickly as possible to ensure system stability.

[0168] The centroid method is used to defuzzify the image, and the weighting coefficients are calculated as follows:

[0169] in, For the first The activation level of the rule, For the single-point value of the corresponding rule output mode: steady-state mode takes Transition mode Dynamic mode ;get Subsequently, the harmonic voltage command is:

[0170]

[0171] in, The given value for the x-axis harmonic voltage. The given value for the y-axis harmonic voltage. This refers to the x-axis voltage command output by the PIR controller. This refers to the y-axis voltage command output by the PIR controller. This refers to the x-axis voltage command output by the feedforward decoupling module. The y-axis voltage command output by the feedforward decoupling module;

[0172] 4. Feedforward decoupling control: In dynamic mode, based on the xy subspace voltage equation:

[0173]

[0174] in, , These are the actual harmonic plane voltages along the x-axis and y-axis, respectively. , These are the harmonic current feedback values ​​along the x-axis and y-axis, respectively. , The harmonic inductances are for the x-axis and y-axis, respectively. , These represent the rates of change of harmonic currents along the x-axis and y-axis, respectively. , This is the product of inductance and current in the coupling term. These are the harmonic back electromotive forces along the x-axis and y-axis, respectively.

[0175] Take the feedforward compensation value:

[0176]

[0177] in, , For x and y axis harmonic inductance; , The back EMFs of the x and y axes harmonics are obtained by looking up tables from offline calibration data based on the motor structure and speed.

[0178] Furthermore, in step S5, the voltage synthesis module includes an inverse Park transform unit for the fundamental voltage and a multi-dimensional vector combination unit; the process of constructing the target voltage vector includes:

[0179] 1. Inverse Park Transform of Fundamental Voltage: ... Rotor position angle Input to the inverse Park transform module:

[0180]

[0181] Output stationary α-β coordinate system voltage components ;

[0182] 2. Multidimensional target vector synthesis: ... Harmonic voltage command Combined, a four-dimensional target voltage vector is formed:

[0183]

[0184] This vector fully describes the fundamental and harmonic voltage requirements that the inverter needs to synthesize for this control cycle.

[0185] Further, in step S6, the SVPWM modulation module includes a sector determination unit, a basic vector lookup table unit, a partition coefficient matrix storage unit, a multiplication and addition operation unit, a normalization processing unit, and a switch sequence generation unit; as shown below. Figure 4 The entire process of performing PWM pulse modulation includes:

[0186] 1. Sector Determination: The sector determination and basic vector selection sub-steps are based on... The sign and size relationship of the symbol are used to determine its sector in the α-β plane using direct logical judgment. ;

[0187] 2. Basic Vector Selection: Based on sector number Retrieve the four adjacent largest vectors V1, V2, V3, and V4 from the pre-stored vector table; (i.e., the amplitude is...) vector, (DC bus voltage) to ensure that the harmonic amplitude generated in the xy plane is minimized;

[0188] 3. Nonlinear partitioning table lookup calculation for processing time:

[0189] Offline stage: The modulation range is divided into linear regions ( ), overmodulation I region (0.906< Overmodulation II region Each district is further subdivided into sub-districts, totaling Each region; modulation ratio Defined as:

[0190]

[0191] in, The amplitude of the fundamental voltage. The maximum amplitude of the fundamental voltage output by the six-phase inverter under square wave conditions; for each sub-region, based on the volt-second balance equation:

[0192]

[0193] in, For the first The projections of the basic vectors onto the α, β, x, and y axes; The duration of action of the corresponding vector; For switching cycles;

[0194] Let the projection matrix

[0195]

[0196] The volt-second equilibrium equation can then be written as: ,in Within the sub-region, the selected basic vectors typically form an effective spanning set of the target voltage vector, and their projection matrix satisfies the full-rank condition in most operating conditions. Therefore, the action time of each vector can be solved using matrix inversion or pseudo-inversion methods. Reversible, solution obtained To facilitate online calculations, a coefficient matrix is ​​defined. , so that:

[0197]

[0198] in: For DC bus voltage; when the projection matrix If a pathological condition occurs, Moore-Penrose pseudoreversal is used. Instead of the ordinary inverse matrix, i.e. To ensure numerical stability;

[0199] The coefficient matrix of all sub-regions is pre-solved and stored in flash memory;

[0200] During online calculation, based on the modulation ratio Locate the current sub-region, retrieve the corresponding coefficient matrix, and quickly obtain the result through multiplication and addition operations. When m crosses the boundary of a sub-region, weighted fusion is used:

[0201]

[0202] in: , , This represents the modulation ratio boundary between adjacent sub-regions.

[0203] This process ensures that the output voltage remains continuous without jumps.

[0204] 4. Normalization and Zero Vector Assignment: Calculating the Total Effective Time = ;when < Then the zero vector action time Select zero vector and Combined to balance the midpoint potential; when ≥ Then scale proportionally. ,make ,at this time ;

[0205] 5. Switching sequence generation: A seven-segment sequence is generated according to the principle of minimizing switching losses and ensuring waveform central symmetry.

[0206] Furthermore, in step S7, the drive execution module includes a drive amplification unit, a six-phase inverter unit, and a control cycle management unit; the process of executing motor drive and iterative cycles includes:

[0207] 1. Pulse amplification and driving: The PWM pulse signal generated in step S6 is amplified by optocoupler isolation and driver chip to generate sufficient drive current to control the IGBT and MOSFET switching transistors of each bridge arm of the six-phase inverter respectively.

[0208] 2. Inverter and Motor Drive: The inverter converts the DC bus voltage U... dcThe voltage is modulated into a six-phase AC voltage and applied to the stator windings of a dual three-phase permanent magnet motor, generating a rotating magnetic field to drive the motor.

[0209] 3. Control cycle iteration: After the current PWM cycle ends, the system automatically returns to step S1 and uses the newly acquired speed and current data to start the next round of control calculation, realizing real-time dynamic closed-loop control of the motor.

[0210] The present invention will now be described in detail with reference to a practical application scenario to better understand its technical principles and practical application. This embodiment is only used to illustrate the technical solution of the present invention and does not constitute a limitation on the scope of protection of the present invention.

[0211] This embodiment is based on a dual three-phase permanent magnet motor used in a certain type of industrial robot. The motor has a rated power of 50kW, a rated speed of 2000r / min, and a number of pole pairs. =3, stator resistance =0.1Ω, direct-axis inductance =0.5mH, quadrature axis inductance L q =0.8mH, permanent magnet flux linkage =0.25Wb, DC bus voltage U dc =400V. The robot is equipped with the control system described in this invention. In a typical rapid start-stop operation, the technical implementation process of each core step of this method is fully demonstrated, intuitively reflecting the system's advanced nature, accuracy, and practical value.

[0212] In this embodiment, during the acceleration phase of the robot joint, the controller in each PWM cycle (T s The control algorithm is executed once every 100 μs. The following selects a typical moment (motor speed n = 1500 r / min, given speed n). ∗ (1550 r / min, load torque approximately 80 N·m) will be described in detail.

[0213] Based on step S1, the controller first reads the motor parameters from the non-volatile memory and loads the speed loop PI parameters (K). pω =0.8, K iω =15) and current loop PI parameters (K pd =1.2, K id =600, K pq =1.2, K iq =600). The rotary transformer measures the current mechanical angular velocity ω. m =1500r / min=157.08rad / s, given rotational speed ω ∗=1550 r / min = 162.31 rad / s, speed deviation Δω = 50 r / min = 5.24 rad / s. The six-phase Hall current sensor synchronously samples the instantaneous current value (already converted to digital): i A =100A, i B =−50A,i C =−50A,i U =80A, i V =−40A,i W =−40A, forming the current vector i 6s .

[0214] Based on step S2, i 6s Substituting the preset transformation matrix T, the current components of each subspace are calculated:

[0215] Fundamental plane: i α =80A, i β =60A;

[0216] Harmonic plane: i x =4A, i y =−2A;

[0217] Zero-order component i o1、 i o2 Because the neutral point isolation is approximately zero.

[0218] Subsequently, using the rotor position angle θ = 45° (electrical angle) to adjust i α、 i β Perform Park transformation:

[0219]

[0220] Right now A, A (a negative value indicates a weak magnetic field). Harmonic components. A, A is directly output to the harmonic suppression module.

[0221] Based on step S3, the speed PI regulator calculates the stator current amplitude setpoint according to Δω. Since the motor is in an accelerating state, the stator current amplitude setpoint is taken as... =150A.

[0222] The system calls the cross-coupling coefficient lookup table, based on the current i d i q (99 A, A) Find k dq =0.015mH, k qd =0.012mH. Substitute into the nonlinear MTPA equation to solve for the optimal current angle β. Take the initial value of the traditional MTPA:

[0223]

[0224] Calculate the denominator:

[0225] Calculate the numerator: The square root of 0.0625 + 0.0162 = 0.0787 gives 0.2805. The numerator is -0.25 + 0.2805 = 0.0305. Therefore, Using the Newton-Raphson iteration, it converged to [the desired value] after 4 iterations. Then the command current is:

[0226] = cosβ opt =150×cos105°=150×(−0.2588)=−38.82A

[0227] = sinβ opt =150×sin105°=150×0.9659=144.9A

[0228] Based on step S4, fundamental plane PI regulation: current error e d = -i d =−38.82−99.0=−137.8A,e q = -i q =144.9−(−14.14)=159.0A. PI controller output (including decoupling term):

[0229]

[0230] electric angular velocity

[0231] ωL q i q =471.24×0.0008×(−14.14)≈−5.33V, ωL d i d =471.24×0.0005×99.0≈23.33V,

[0232] ω =471.24×0.25≈117.81V

[0233] PI controller output (proportional section):

[0234] Kpd e d =1.2×(−137.8)=−165.36V,K pq e q =1.2 × 159.0 = 190.8 V

[0235] The integral part makes the error zero in steady state, and the final fundamental voltage is given as:

[0236] ≈−165.36−(−5.33)=−160.03V, ≈190.8+(23.33+117.81)=190.8+141.14=331.94V

[0237] (These are theoretical calculations; actual control will be limited by the DC bus voltage.)

[0238] Harmonic plane: i x =4A, i y =−2A. The PIR controller resonant frequency is set to ω0 = 471.24 rad / s, 6th harmonic 2827.4 rad / s, and 12th harmonic 5654.9 rad / s. The output after PIR adjustment... =−8V, =6V. Simultaneously calculate the rate of change of current Δi. x =0.15A / T s ,Δi y =−0.1A / T s Thus, Δi≈1803A / s; the rotational speed change rate is small, and the fuzzy logic output weight λ=0.15 (mainly steady-state mode).

[0239] Feedforward compensation: Harmonic inductance L x =L y =0.2 mH, from the table the harmonic back electromotive force e x =1.5V, e y =−1.2V, then

[0240] = −471.24×0.0002×(−2)+1.5=0.1885+1.5=1.6885V

[0241]

[0242] Final harmonic voltage command:

[0243]

[0244]

[0245] Based on step S5, the fundamental voltage inverse Park transform ( =45 o ):

[0246] = cosθ− sinθ=(−160.03)×0.7071−331.94×0.7071=−113.16−234.74=−347.9V

[0247] = sinθ+ cosθ=(−160.03)×0.7071+331.94×0.7071=−113.16+234.74=121.58V

[0248] (These are theoretical calculations; actual control will be limited by the DC bus voltage.)

[0249] The multidimensional target vector is:

[0250]

[0251] Based on step S6, the fundamental voltage amplitude is first calculated. V.

[0252] Modulation ratio:

[0253]

[0254] A result greater than 1 indicates that the signal has entered the overmodulation region. After actual limiting, the maximum fundamental voltage amplitude is: Therefore, the voltage component after limiting is:

[0255]

[0256]

[0257] Amplitude V, modulation ratio 1.0, at the boundary of the overmodulation II region. The system adjusts according to the amplitude-limited value. Determine the sector and retrieve the corresponding coefficient matrix. .

[0258] The duration of action of the four largest vectors is calculated using multiplication and addition operations:

[0259]

[0260] Design calculation s, s, s, s, total 100 s The zero vector has a zero-time operation. Then, a seven-segment symmetrical sequence is generated according to the switching order corresponding to the sector, and PWM pulses are output.

[0261] Based on step S7, the PWM pulse is amplified by the drive circuit and controls the IGBT switch. The inverter converts the DC bus voltage of 400V into a six-phase AC voltage to drive the motor. After the current PWM cycle ends, the system automatically returns to step S1 to start the next cycle calculation, realizing real-time closed-loop control.

[0262] Through the application of this invention, the servo motor of this industrial robot has achieved significant results across the entire operating range. At a rated speed of 2000 r / min and 100% load, it performs significantly better than traditional dual three-phase motor vector control (using i... d Compared to 0 control and conventional SVPWM: motor efficiency increased from 92.5% to 94.3%, an improvement of 1.8 percentage points, mainly due to the precise control of MTPA considering cross saturation; stator current total harmonic distortion (THD) decreased from 9.2% to 3.5%, a reduction of 62%, thanks to the synergistic effect of the adaptive resonant controller and the maximum four-vector SVPWM; in the high-speed field weakening region (speed 3000 r / min), the overmodulation strategy improved the output voltage utilization rate by 13% and the maximum torque output capability of the motor by 10%; in terms of dynamic response, speed overshoot decreased from 9% to 3.5%, and the settling time was shortened by 35%.

[0263] As can be seen from the above embodiments, the "Harmonic Suppression Method for Dual Three-Phase Permanent Magnet Motors Based on Maximum Four-Vector SVPWM" provided by this invention possesses outstanding creativity, practicality, and advancement. Its core advantages are reflected in three aspects: First, it achieves optimal efficiency control of the motor over a wide operating range. By using the nonlinear MTPA algorithm considering cross-saturation in step S3, the optimal current angle is accurately solved, overcoming the efficiency reduction problem caused by parameter mismatch in traditional methods. Maximum torque output per unit current can be achieved over a wide torque-speed range. Second, it constructs a high-quality harmonic suppression system. By employing the adaptive resonance and multi-mode switching composite control strategy in step S4, combined with the maximum four-vector SVPWM in step S6, high-gain, zero-steady-state-error suppression of specific harmonics is achieved in steady state, while ensuring rapid system response and stability in dynamic conditions, significantly reducing current harmonic distortion rate and torque ripple. Thirdly, a smooth, seamless transition in the overmodulation region and maximum voltage utilization are achieved. Through the nonlinear partitioning lookup table and weighted fusion processing in step S6, the system can output smoothly from the linear region to the overmodulation region and even into the square wave condition, effectively improving the motor's high-speed field-weakening load-carrying capacity and DC bus voltage utilization. This invention organically integrates parameter optimization, harmonic suppression, and overmodulation control, forming a complete closed loop of "perception-decision-execution-optimization," with a clear technical architecture, providing a complete solution for high-performance dual three-phase permanent magnet motor drive systems.

[0264] Figure 5 The bus current simulation results provided in this embodiment of the invention show that when the control algorithm of this invention is adopted, the motor operating bus current is reduced from 26.6A to 24.1A, a reduction of about 9% in current, which significantly reduces energy consumption.

[0265] Figure 6 The simulation results of the Fourier analysis of the current provided in the embodiments of the present invention show that when the control algorithm of the present invention is adopted, the harmonics of the motor operating current are reduced from 9.96% to 0.75%, which effectively reduces the current harmonics.

[0266] Figure 7 The simulation results of the speed waveform provided in the embodiments of the present invention show that when the control algorithm of the present invention is adopted, the speed of the motor can be stabilized quickly when the load changes, and it has good dynamic performance.

[0267] It should be noted that the aforementioned unit modules are mostly implemented through software programming and algorithm calls. In this document, relational terms such as "first" and "second" are used merely to distinguish one entity or operation from another, and do not necessarily require or imply any such actual relationship or order between these entities or operations. Furthermore, the terms "comprising," "including," or any other variations thereof are intended to cover non-exclusive inclusion, such that a process, method, article, or apparatus that comprises a list of elements includes not only those elements but also other elements not expressly listed, or elements inherent to such a process, method, article, or apparatus. Without further limitations, an element defined by the phrase "comprising one..." does not exclude the presence of other identical elements in the process, method, article, or apparatus that includes said element.

[0268] Although embodiments of the invention have been shown and described, it will be understood by those skilled in the art that various changes, modifications, substitutions and alterations can be made to these embodiments without departing from the principles and spirit of the invention, the scope of which is defined by the appended claims and their equivalents.

Claims

1. A harmonic suppression method for a dual three-phase permanent magnet motor based on maximum four-vector SVPWM, characterized in that, Includes the following steps: Signal acquisition and coordinate transformation steps: The motor speed and six-phase current are detected in real time by speed sensor and current sensor. The six-phase current is decomposed into fundamental plane α-β component and harmonic plane xy component by vector space decoupling transformation matrix, and the dq axis current feedback value is obtained by synchronous rotation transformation. MTPA command generation steps: The stator current amplitude is given by adjusting the speed PI. Based on the electromagnetic torque equation considering the cross-coupling coefficient, the optimal current angle is solved by Newton-Raphson iteration, and the dq axis current setpoint is calculated and output. Dual-plane current regulation and harmonic suppression steps: The fundamental plane uses PI regulation to output the fundamental voltage setpoint; the harmonic plane uses an adaptive resonant controller and fuzzy logic multi-mode switching, smoothly transitioning between PIR mode and feedforward mode according to the operating conditions, and outputting the harmonic plane voltage setpoint. Target vector synthesis and SVPWM modulation steps: The fundamental voltage is given and inverse Park transforms to obtain the α-β axis voltage, which is combined with the harmonic voltage to form a four-dimensional target voltage vector; the adjacent maximum vector is selected by sector judgment. The basic vector is the voltage vector with a large projection amplitude in the α-β plane under the given DC bus voltage condition, and the vector combination with a high contribution to the synthesis of the target voltage vector is selected first. The action time is calculated by looking up the coefficient matrix of the partition, and after normalization and zero vector allocation, a six-phase inverter switching pulse sequence is generated. Drive and iterative cycle steps: The SVPWM pulse is amplified by the drive circuit and then the inverter outputs six-phase AC power to drive the motor. After the current control cycle is completed, the signal acquisition step is returned to continue the iteration.

2. The method according to claim 1, characterized in that, The MTPA instruction generation step further includes: The offline calibration sub-step for the cross-coupling coefficient is performed by pre-constructing a model with the dq-axis current through finite element simulation. , Given a two-dimensional lookup table as input, output the cross-coupling coefficient: in, , These are the magnetic flux linkages along the d and q axes, respectively. Characterizes the rate of change of d-axis flux linkage with q-axis current; Characterizes the rate of change of q-axis flux linkage with d-axis current; The sub-steps for solving the optimal current angle in nonlinear MTPA are based on the electromagnetic torque equation considering cross-coupling: in, Electromagnetic torque; For extreme logarithms, It is a permanent magnet flux chain. For a direct-axis inductor, L q It is a quadrature axis inductor. make , Given the stator current amplitude, substitute it into the above formula and... Taking the partial derivative and setting it to zero, we obtain the nonlinear equation: The solution is obtained using the Newton-Raphson iterative method: in, For the first The estimated current angle for the next iteration; For function exist The first derivative value at; The initial value for iteration is taken from the traditional MTPA solution: Each iteration is based on the current Table lookup and update To obtain the optimal current angle ; Command current distribution sub-step, utilizing Calculate the dq-axis current setting: When this formula is for surface-mounted motors Then take directly .

3. The method according to claim 1, characterized in that, The dual-plane current regulation and harmonic suppression steps further include: The fundamental plane PI regulator step will control the command current. Feedback The error is fed into the PI controller: in, The proportional and integral coefficients of the d-axis current loop; , These are the proportional and integral coefficients of the q-axis current loop; ω is the electric angular velocity; m It is the mechanical angular velocity; For the Laplace operator; the output fundamental voltage is given. ; The design sub-steps of the adaptive resonant controller include the design of harmonic currents. The error between the input and its given value (zero) is fed into a proportional-integral-resonant controller, whose transfer function is: in, This is the proportionality coefficient; The integral coefficient; Let h be the resonant gain of the h-th resonant term; This is the cutoff frequency of the resonant term; The fundamental angular frequency; resonant frequency The controller adaptively adjusts with rotational speed to achieve high gain suppression of the 5th, 7th, 11th, and 13th harmonics; the controller output is denoted as... ; Fuzzy logic multi-mode switching sub-step, real-time calculation of current change rate and the rate of change of rotational speed The fuzzy classification is divided into five levels (VS, S, M, B, VB), using a triangular membership function. in: , These are the differences between the x-axis and y-axis harmonic currents in adjacent control cycles; The sampling time interval (equal to the control period) ; Let be the rate of change of rotational speed; the universes of discourse for input variables Δi and Δω are normalized to the interval [0,1], and the parameters of the triangular membership function for each fuzzy level are defined as follows: VS: Vertex 0, base [0, 0.25]; S: Vertex 0.25, base [0, 0.5]; M: Vertex 0.5, base [0.25, 0.75]; B: Vertex 0.75, base [0.5, 1]; VB: Vertex 1, bottom edge [0.75, 1]; The fuzzy rule base is as follows: IF is VS AND is VS THEN mode = steady state (PIR) IF is M OR is M THEN mode = Transition (PIR + Feedforward Weighted) IF is B OR is B THEN mode = dynamic (feedforward) The centroid method is used to defuzzify the image, and the weighting coefficients are calculated as follows: in, For the first The activation level of the rule, For the single-point value of the corresponding rule output mode: steady-state mode takes Transition mode Dynamic mode ;get Subsequently, the harmonic voltage command is: in, The given value for the x-axis harmonic voltage. The given value for the y-axis harmonic voltage. This refers to the x-axis voltage command output by the PIR controller. This refers to the y-axis voltage command output by the PIR controller. This refers to the x-axis voltage command output by the feedforward decoupling module. The y-axis voltage command output by the feedforward decoupling module; The feedforward decoupling control sub-step, in dynamic mode, is based on the xy subspace voltage equation: in, , These are the actual harmonic plane voltages along the x-axis and y-axis, respectively. , These are the harmonic current feedback values ​​along the x-axis and y-axis, respectively. , The harmonic inductances are for the x-axis and y-axis, respectively. , These represent the rates of change of harmonic currents along the x-axis and y-axis, respectively. , This is the product of inductance and current in the coupling term. These are the harmonic back electromotive forces along the x-axis and y-axis, respectively. Take the feedforward compensation value: The values ​​are obtained by looking up offline calibration data based on the motor structure and speed.

4. The method according to claim 1, characterized in that, The target vector synthesis and SVPWM modulation steps further include: The multidimensional target vector synthesis sub-step transforms the given value of the fundamental voltage in the stationary two-phase coordinate system (α-β axis). Harmonic voltage command Combined, a four-dimensional target voltage vector is formed: The sector determination and basic vector selection sub-steps are based on... The sign and size relationship of the symbol are used to determine its sector in the α-β plane using direct logical judgment. According to sector number Retrieve the four adjacent largest vectors V1, V2, V3, and V4 from the pre-stored vector table; The nonlinear partitioning lookup table calculation sub-step divides the modulation range into multiple sub-regions, and the modulation ratio... Defined as: in, The amplitude of the fundamental voltage. The maximum amplitude of the fundamental voltage output by the six-phase inverter under square wave conditions; for each sub-region, based on the volt-second balance equation: in, For the first The projections of the basic vectors onto the α, β, x, and y axes; The duration of action of the corresponding vector; For switching cycles; Let the projection matrix The volt-second equilibrium equation can then be written as: ,in Within the sub-region, the selected basic vectors typically form an effective spanning set of the target voltage vector, and their projection matrix satisfies the full-rank condition in most operating conditions. Therefore, the action time of each vector can be solved using matrix inversion or pseudo-inversion methods. Reversible, solution obtained To facilitate online calculations, a coefficient matrix is ​​defined. , so that: in: For DC bus voltage; when the projection matrix If a pathological condition occurs, Moore-Penrose pseudoreversal is used. Instead of the ordinary inverse matrix, i.e. To ensure numerical stability; The coefficient matrix of all sub-regions is pre-solved and stored in flash memory; During online calculation, based on the modulation ratio Locate the current sub-region, retrieve the corresponding coefficient matrix, and quickly obtain the result through multiplication and addition operations. When m crosses the boundary of a sub-region, weighted fusion is used: in: , , The modulation ratio boundary of adjacent sub-regions; ; Normalization and zero vector allocation sub-steps to calculate total effective time = ;when < Then the zero vector action time Select zero vector and Combined to balance the midpoint potential; when ≥ Then scale proportionally. ,make ,at this time ; The switching sequence generation sub-step generates a seven-segment sequence based on the principle of minimizing switching losses and ensuring waveform central symmetry.

Images