Search

CN-121150538-B - Space vector control method for flywheel energy storage system of twelve-phase permanent magnet synchronous motor

CN121150538BCN 121150538 BCN121150538 BCN 121150538BCN-121150538-B

Abstract

The invention relates to the technical field of motor control in smart grid industry, and relates to power electronic components such as various semiconductor field effect transistors and the like in a control circuit of a flywheel energy storage system, which are used for high-precision control of a motor, and the invention discloses a space vector control method of a flywheel energy storage system of a twelve-phase permanent magnet synchronous motor, comprising the following steps: S1, twelve-phase space vector decoupling fault-tolerant control, S2, higher-order harmonic energy recovery closed loop, S3, quantum annealing parameter optimization layer, S4, flux linkage self-adaptive observer network, and S5, constructing an 18-dimensional switch vector table. According to the scheme, through multi-reference coordinate system transformation and redundant phase reconstruction algorithm, quick fault-tolerant response of the twelve-phase motor in single-phase or two-phase faults is realized, a current path is dynamically recombined in the fault, torque pulsation is effectively restrained by combining a negative sequence current compensation loop, harmonic distortion rate is detected in real time through a sliding window FFT under high-speed charge and discharge working conditions, and the problem of insufficient dynamic response of a traditional fault-tolerant strategy is solved.

Inventors

  • JIANG XINJIAN
  • LING ZHIJIAN
  • LI FUWANG
  • LI ZHIRU
  • SU JIAXIN

Assignees

  • 山西省能源互联网研究院

Dates

Publication Date
20260512
Application Date
20250910

Claims (7)

  1. 1. The space vector control method of the twelve-phase permanent magnet synchronous motor flywheel energy storage system is characterized by comprising the following steps of: S1, twelve-phase space vector decoupling fault-tolerant control, namely adopting multi-reference coordinate system transformation, separating fundamental waves from 5 and 7 harmonics, designing a redundant phase reconstruction algorithm, automatically recombining current paths in any two-phase fault, introducing a negative sequence current compensation loop, and inhibiting torque pulsation in asymmetric operation; s2, a higher-order harmonic energy recovery closed loop is established, a harmonic equivalent circuit model is established, harmonic impedance is calculated in real time, a harmonic selector switch is designed, specific sub-harmonics are led into an auxiliary energy storage capacitor, and a GaN device is used for constructing a high-frequency bidirectional DC-DC loop; S3, a quantum annealing parameter optimization layer is used for constructing QUBO model coding control parameters, designing a cost function, and updating PID parameters through a D-Wave quantum processor every 10 ms; S4, a flux linkage self-adaptive observer network is deployed, an LSTM neural network is deployed to identify the demagnetization degree of the permanent magnet in real time, and the d-q axis current ratio is dynamically adjusted to compensate irreversible demagnetization; S5, constructing an 18-dimensional switch vector table, comprising zero vectors and harmonic injection vectors, rolling an optimization target, and adopting an FPGA to realize a 200 ns-level prediction period; S6, a thermoelectric coupling balancing strategy is adopted, an optical fiber temperature sensor array is arranged to monitor the temperature rise of the 12-phase winding in real time, and the phase current load is dynamically distributed based on a thermal resistance network model; S7, quantum noise suppression communication protocol, namely, BB84 protocol is adopted to encrypt the gigabit optical fiber communication between the motor and the controller, and forward error correction coding is designed to resist electromagnetic pulse interference; S8, a digital twin verification platform constructs a DIGITAL TWIN model containing a bearing friction nonlinear term; Wherein S1 comprises: S1.1, on the basis of the traditional d1q 1/fundamental wave, d3q3/5 harmonic wave and d5q5/7 harmonic wave conversion, adding a sliding window FFT to detect harmonic distortion rate in real time, dynamically adjusting the bandwidth of a harmonic sub-controller, introducing a feedforward compensation term between coordinate systems, and eliminating the coupling effect between d-q axes during high-speed operation; S1.2, when a single-phase fault occurs, balance is maintained through redistribution of adjacent phase currents, when two-phase faults occur, an optimal current path is generated based on minimum torque pulsation optimization of the remaining healthy phase, a virtual neutral point is synthesized through PWM modulation, and magnetomotive force of missing fault phases is compensated; S1.3, extracting a negative sequence component by adopting a second-order generalized integrator, injecting reverse compensation current, establishing a negative sequence voltage equation containing 5 and 7 times of harmonic waves, uniformly calculating compensation quantity by predictive control, adjusting compensation ring parameters on line according to a torque pulsation frequency spectrum, and preferentially inhibiting dominant frequency components; The single-phase fault tolerance control in the S1.2 is based on the A phase fault: By current redistribution principle, including: magnetomotive force balance, fault phase current return to zero The magnetomotive force is required to be reconstructed and synthesized through the residual two phases B, C to be equivalent to the original three-phase symmetrical magnetomotive force; Current constraint, namely keeping total power unchanged, and adjusting B, C-phase current amplitude to be original-phase current The phase difference was adjusted to 90 degrees: Wherein: Is the amplitude of the current of the primary phase, Is an electrical angle; and reconstructing B, C-phase PWM waveforms through an inverter, synthesizing virtual neutral point voltage, and correcting B, C-phase current in real time by adopting current feedback regulation to offset the influence of A-phase loss.
  2. 2. The space vector control method of the flywheel energy storage system of the twelve-phase permanent magnet synchronous motor according to claim 1, wherein the fault-tolerant control of the two-phase fault in S1.2 is based on A, B two-phase fault: optimized by minimum torque ripple, comprising: objective function by optimizing the residual phase current when only the C phase is healthy And possibly zero sequence current Torque ripple is minimized: Wherein: as a function of the torque of the C-phase, For the zero sequence component torque contribution, ; Generating by an optimal current path, comprising: And the harmonic injection method is that specific harmonic is overlapped on the C-phase fundamental wave current to compensate the missing torque component of the fault phase: Coefficients of 、 Determining through off-line optimization or an on-line self-adaptive algorithm; zero sequence current utilization, if the motor allows neutral point access, injecting zero sequence current To balance magnetomotive force: dynamically adjusting voltage vectors through redundant states or capacitance midpoints of inverter bridge arms to synthesize virtual neutral point potential, thereby ensuring And Is a precise tracking of (a).
  3. 3. The space vector control method of the flywheel energy storage system of the twelve-phase permanent magnet synchronous motor according to claim 2, wherein S2 comprises: s2.1, introducing an online parameter identification algorithm into an equivalent circuit model, updating impedance parameters of 5, 7 and 11 harmonics in real time, dynamically adjusting the switching time sequence of a harmonic selector, and realizing the transient response of an impedance matching network through the rapid switching characteristic of a GaN device; s2.2, adopting an adjustable LC resonance network and digital control, generating PWM signals through an FPGA to drive a GaN switch, selectively conducting 5, 7 and 11 times of harmonic paths, adding magnetic coupling resonance compensation, and inhibiting non-target harmonic waves; S2.3, LLC resonance topology is applied to a GaN bidirectional DC-DC loop, zero-voltage switching and zero-current switching are achieved, digital hysteresis control is introduced, and duty ratio is dynamically adjusted according to voltage fluctuation of an energy storage capacitor.
  4. 4. The space vector control method of the flywheel energy storage system of the twelve-phase permanent magnet synchronous motor according to claim 3, wherein the mode of restraining non-target harmonic waves in the S2.2 through the non-target harmonic wave trap is as follows: an LC series resonance trap is added at the total input port: 3 rd harmonic, l=1.2 μh, c=9.4 nF, resonating at 150kHz, fundamental frequency 50kHz; The 9 th harmonic is L=130nH, C=1.2nF, and resonates at 450kHz; Magnetic isolation, namely adopting a NiZn ferrite magnetic ring.
  5. 5. The space vector control method of the flywheel energy storage system of the twelve-phase permanent magnet synchronous motor according to claim 4, wherein the magnetic coupling resonance compensation optimization in S2.2 comprises: The layered winding structure is adopted: the inner layer is special for 5 th harmonic, the Litz line and the stock number is more than or equal to 100; the outer layer is a flat copper strip shared by 7 and 11 times of harmonic waves; controlling a coupling coefficient, namely controlling k=0.3-0.5, and inhibiting non-target frequency band energy transfer through COMSOL simulation optimization; the auxiliary harmonic injection circuit is used for injecting reverse harmonic current through a GaN H bridge to offset residual 3 times and 9 times of interference; The energy storage capacitor dynamic management in S2.3 includes: And adding a bidirectional GaN switch into the capacitor branch, conducting only when the target harmonic is detected, and counteracting the ESR of the capacitor through an operational amplifier feedback network by the negative impedance converter.
  6. 6. The space vector control method of the flywheel energy storage system of the twelve-phase permanent magnet synchronous motor according to claim 5, wherein the dynamic PID parameter layering based on quantum annealing in the step S3 is further optimized, and the method comprises the following steps: S3-A1, dividing a control system of the twelve-phase permanent magnet synchronous motor into three layers, wherein the top layer is a torque performance monitoring layer and is responsible for evaluating a torque pulsation index; S3-A2, designing a specific QUBO model aiming at each level, and decomposing torque pulsation into twelve-phase current unbalance degree and rotor position correlation by a top-level model; the middle layer model models the harmonic loss as a function of the THD of each phase current, and the switching loss is related to the switching frequency and the dead time; S3-A3, adopting a layered annealing strategy, firstly determining a torque optimization direction by a top layer, searching a loss balance point by a middle layer under the constraint, finally solving a specific parameter combination by a bottom layer, forming an integral optimization problem by a QUBO model of three layers through chained coupling, and completing one full parameter update on a D-Wave processor every 10 ms; S3-A4, according to the flywheel energy storage working condition, comprising three items of charging, discharging and maintaining, and automatically adjusting the weight coefficients of three indexes in the cost function, wherein the charging stage is used for focusing on switching loss optimization, the discharging stage is used for prioritizing torque stability, and the maintaining state is used for balancing harmonic suppression; S3-A5, establishing a database of quantum annealing solutions, identifying high-quality solution characteristics under different working conditions through a classical machine learning algorithm, providing initial solution and tunnel effect parameters for a subsequent annealing process, and accelerating convergence to a better region.
  7. 7. The space vector control method of the twelve-phase permanent magnet synchronous motor flywheel energy storage system according to claim 6, wherein the quantum cooperative control optimization scheme of the twelve-phase current loop in S3 is as follows: S3-B1, analyzing the spatial distribution characteristics of twelve-phase windings, explicitly encoding phase-to-phase mutual inductance coupling items in a QUBO model, and converting the electromagnetic interference relation of adjacent phases, interval phases and orthogonal phases into quadratic term constraint; S3-B2, dividing the twelve-phase current loop into three groups of four-phase systems, wherein a strong coupling QUBO model is adopted in each group, the groups are subjected to cooperative optimization through an annealing chain, independent Kp and Ki parameter combinations are respectively allocated in each group, bandwidth constraint is shared, and dynamic response consistency is ensured; S3-B3, embedding a redundant variable in the QUBO model, automatically adjusting the current distribution proportion of adjacent phases when a certain phase fault is detected, rapidly redistributing the current reference value of the healthy phase through quantum annealing, and maintaining the total torque stable; S3-B4, aiming at the specific 5 th, 7 th, 11 th and 13 th harmonic waves of a twelve-phase system, setting harmonic suppression zone bits in a QUBO model, and when the zone bit corresponding to an annealing solution is activated, automatically switching PID parameters to a high harmonic suppression mode, increasing the bandwidth of a current loop and adjusting an integral coefficient; S3-B5, dynamically adjusting annealing parameters according to the real-time unbalance degree of twelve-phase current, increasing annealing iteration times when the unbalance is high, adopting a rapid annealing mode when the unbalance is low, and balancing and optimizing precision and response speed.

Description

Space vector control method for flywheel energy storage system of twelve-phase permanent magnet synchronous motor Technical Field The invention relates to the technical field of motor control in smart grid industry, in particular to a space vector control method of a twelve-phase permanent magnet synchronous motor flywheel energy storage system. Background The control circuit of the flywheel energy storage system relates to various power electronic components such as semiconductor field effect transistors, is used for high-precision switch control of a motor, is used as an efficient and environment-friendly energy storage mode, has important application value in the fields of grid frequency modulation, renewable energy stabilization, aerospace energy storage and the like, and the twelve-phase permanent magnet synchronous motor of the core component becomes an ideal driving device for flywheel energy storage by virtue of high power density, low torque pulsation and strong fault tolerance. The twelve-phase permanent magnet synchronous motor flywheel energy storage system combines the advantages of the flywheel energy storage system and the permanent magnet synchronous motor, the flywheel energy storage technology stores kinetic energy through a flywheel rotating at a high speed, the permanent magnet synchronous motor is known to have the characteristics of high efficiency, high power density and stable operation, and the motor adopts a twelve-phase winding design, so that more stable electromagnetic torque can be generated, and the efficiency and stability of the system are further improved. In actual operation, the following key problems still exist to be solved: 1. The traditional fault-tolerant control strategy, such as based on fault phase current reconstruction or neutral point adjustment, can ensure the basic operation of the motor in the phase failure or winding short circuit, but is difficult to consider the dynamic response requirement of flywheel energy storage in the high-speed charge and discharge process; 2. If the inherent space harmonics of the twelve-phase motor are not effectively utilized, not only copper loss and iron loss are increased, but also electromagnetic vibration is caused, the existing control strategy mostly adopts a harmonic suppression technology, but the active recovery potential of harmonic energy is ignored, so that the energy conversion efficiency of flywheel energy storage in a high frequency band is reduced, and the overall energy efficiency of the system is reduced by 10% -15% particularly in a transient process; 3. the running condition of flywheel energy storage is highly nonlinear, such as a wide rotating speed range and frequent load mutation, traditional PID or model predictive control relies on an accurate mathematical model, and quantum computation can realize ultra-high-speed optimization solution through quantum parallelism. Disclosure of Invention The invention aims to solve the defects in the prior art, and provides a space vector control method of a twelve-phase permanent magnet synchronous motor flywheel energy storage system. In order to achieve the purpose, the invention adopts the following technical scheme that the space vector control method of the twelve-phase permanent magnet synchronous motor flywheel energy storage system comprises the following steps: S1, twelve-phase space vector decoupling fault-tolerant control, namely adopting multi-reference coordinate system transformation, separating fundamental waves from 5 and 7 harmonics, designing a redundant phase reconstruction algorithm, automatically recombining current paths in any two-phase fault, introducing a negative sequence current compensation loop, and inhibiting torque pulsation in asymmetric operation; s2, a higher-order harmonic energy recovery closed loop is established, a harmonic equivalent circuit model is established, harmonic impedance is calculated in real time, a harmonic selector switch is designed, specific sub-harmonics are led into an auxiliary energy storage capacitor, and a GaN device is used for constructing a high-frequency bidirectional DC-DC loop; S3, a quantum annealing parameter optimization layer is used for constructing QUBO model coding control parameters, designing a cost function, and updating PID parameters through a D-Wave quantum processor every 10 ms; S4, a flux linkage self-adaptive observer network is deployed, an LSTM neural network is deployed to identify the demagnetization degree of the permanent magnet in real time, and the d-q axis current ratio is dynamically adjusted to compensate irreversible demagnetization; S5, constructing an 18-dimensional switch vector table, comprising zero vectors and harmonic injection vectors, rolling an optimization target, and adopting an FPGA to realize a 200 ns-level prediction period; S6, a thermoelectric coupling balancing strategy is adopted, an optical fiber temperature sensor array is arranged to monitor the temper