A six-phase double-winding motor cooperative driving control system and method based on SiC MOSFET

Abstract

The application relates to the field of motor drive control, and discloses a six-phase double-winding motor cooperative drive control system and method based on SiC MOSFET, the method comprising the following steps: mapping six-phase currents to a fundamental wave subspace axis, a harmonic subspace axis and a zero sequence subspace through generalized Park transformation; constructing a second-order linear extended state observer in the axis current loop, fusing system internal parameter perturbation, electromagnetic coupling between windings and external load mutation into total disturbance and estimating in real time, and performing feedforward compensation; based on a discretized motor prediction model, constructing a global cost function containing a current tracking error term, a harmonic suppression term and a SiC MOSFET switching loss penalty term, and selecting an optimal switching vector to directly drive an inverter by minimizing the cost function. The application can avoid the problem that strong coupling of the six-phase double-winding motor leads to dynamic response lag, fully exert the high-frequency energy efficiency advantage of SiC, and improve reliability through hardware safety bottom-up.

Description

Technical Field

[0001] This invention relates to the field of motor drive control, and specifically to a six-phase dual-winding motor cooperative drive control system and method based on SiC MOSFETs. Background Technology

[0002] Six-phase dual-winding motors, with their advantages of low torque ripple, high power density, and strong fault tolerance, are increasingly widely used in high-performance applications such as electric vehicles, ship propulsion, and robot joint drives. Meanwhile, silicon carbide (SiC) MOSFETs, as wide-bandgap semiconductor devices, possess characteristics such as fast switching speed, low switching losses, and high temperature resistance, providing new possibilities for improving the control bandwidth and power density of motor drive systems. Applying SiC MOSFETs to the drive control of six-phase dual-winding motors has become one of the important development directions of current high-power-density drive technology.

[0003] However, existing six-phase dual-winding motor drive control systems still have the following technical defects when using SiC MOSFETs, which restricts the further improvement of the overall system performance.

[0004] First, the strong electromagnetic coupling of six-phase motors leads to insufficient dynamic performance of traditional control methods. Because the two sets of three-phase windings in a six-phase dual-winding motor have a 30° phase band angle difference in space, there is strong mutual inductive coupling between the magnetic circuits. In a synchronous rotating coordinate system, this coupling manifests as cross-coupled electromotive force between the d and q axes and nonlinear characteristics of inductance parameters varying with the operating point. Traditional proportional-integral (PI) control strategies are typically designed based on linear time-invariant models and cannot compensate for these coupling disturbances and nonlinear parameter changes in real time. When the system faces transient conditions such as rapid acceleration, emergency braking, or sudden load changes, the PI controller is prone to integral saturation, phase lag, and control saturation, resulting in increased torque pulsation, delayed current response, and even system oscillations. This deficiency is particularly prominent in six-phase dual-winding motors because the currents of the two windings not only jointly contribute to the torque output, but their mutual coupling also induces significant circulating current losses in the harmonic subspace, further degrading control quality.

[0005] Secondly, there is a contradiction between the high switching speed of SiC MOSFETs and existing pulse width modulation strategies. Although SiC MOSFETs can theoretically support switching frequencies of hundreds of kilohertz, providing the possibility of reducing current ripple and increasing control bandwidth, the current mainstream space vector pulse width modulation (SVPWM) or sinusoidal pulse width modulation (SPWM) uses a fixed switching frequency, which cannot dynamically adjust the switching action according to the real-time operating status of the motor, such as the rate of change of current and torque command jumps. Simply increasing the switching frequency of SVPWM can improve the current waveform, but it will cause the number of switching operations of the SiC device to increase proportionally in each control cycle, and the switching losses will rise sharply. Since the SiC chip area is relatively small and the heat capacity is limited, excessive switching losses will cause the junction temperature to rise rapidly, triggering overheat protection or even causing device damage. Therefore, existing PWM strategies cannot achieve a balance between fully utilizing the high-frequency advantages of SiC and effectively controlling switching losses, limiting further improvement in system control performance. Summary of the Invention

[0006] The purpose of this invention is to provide a six-phase dual-winding motor cooperative drive control system and method based on SiC MOSFETs, thereby solving at least one of the above-mentioned technical problems.

[0007] The objective of this invention can be achieved through the following technical solutions: A collaborative drive control method for a six-phase dual-winding motor based on SiC MOSFETs, the method comprising the following steps: Step 1: Synchronously acquire the six stator phase currents of the six-phase double Y-winding motor, and map the six stator phase currents to mutually orthogonal fundamental subspaces using the generalized Park transform. Axial and Harmonic Subspace Axis and zero-order subspace axis; Step Two, in The shaft current loop constructs a second-order linear extended state observer (LESO) to integrate internal parameter perturbations, inter-winding electromagnetic coupling, and external load abrupt changes into a total disturbance. The total disturbance is estimated in real time using the LESO. The observed values ​​are then fed forward and compensated to output the results. Axis target reference voltage vector; Step 3: Receive the target reference voltage vector, and based on the discretized motor prediction model, traverse all switching state combinations of the inverter to construct a global cost function that includes current tracking error terms, harmonic suppression terms, and SiC MOSFET switching action penalty terms. By minimizing the cost function The optimal switching vector is selected to directly generate a high-voltage gate drive signal.

[0008] In a preferred embodiment, the second-order linear extended state observer LESO is... shaft and The shaft structure is the same. The discretized observer of the axis updates the current estimate and total disturbance estimate for the next period based on the difference between the current sample value and the current estimate for the current period: The current estimate for the next cycle is based on the current cycle value, plus the control cycle multiplied by (the total disturbance estimate for the current cycle, the proportional term of the estimation error, and the gain term of the control voltage for the current cycle). The total disturbance estimate for the next cycle is the current cycle value plus another proportional term, which is the control cycle multiplied by the estimation error. The estimation error is positively correlated with the current sample value and negatively correlated with the current estimate value.

[0009] In the preferred embodiment, the global cost function It consists of a weighted sum of three terms: The first item reflects the accuracy of torque current tracking, and this value varies with... shaft and The target current of the shaft increases as the sum of the squares of the differences between the target current and the predicted current increases; The second term reflects harmonic loss, and this value varies with... shaft and The predicted current increases as the sum of the squares of the currents on the axis increases; The third item reflects the switching losses of the SiC MOSFET, and this value increases with the total number of switching state changes of the six bridge arms; The smaller the value, the better; the system selects... The smallest switching vector is used as the output.

[0010] In a preferred embodiment, the method further includes step four: When an open-circuit fault is detected in any phase winding, the basis of the generalized Park transformation matrix is ​​adjusted online to reconstruct a virtual voltage vector without order harmonic components, and the optimization calculation of the global cost function J is re-executed within the reduced set of switch vectors to maintain the stator synthesized magnetomotive force as circular.

[0011] In a preferred embodiment, the discretized motor prediction model calculates the current prediction vector for the next cycle based on the current vector, voltage command vector, and back EMF vector of the current cycle. The predicted current decreases linearly with the current current, and the attenuation coefficient is determined by the stator resistance, inductance, and control cycle. Simultaneously, the predicted current and voltage commands minus the back electromotive force show a positive correlation gain relationship; among which... axis, The voltage command for the axis and zero-sequence subspace is set to zero, and the back electromotive force is at... And zero in the zero-order subspace.

[0012] In a preferred embodiment, in step two, the linear state error feedback control law calculates the final target reference voltage vector based on the difference between the target current and the current estimate output by the observer, as well as the total disturbance estimate. The voltage vector is proportional to the current error. After subtracting the total disturbance estimate, the value is divided by the known gain. That is, the larger the current error, the larger the voltage regulation, and the larger the total disturbance estimate, the stronger the feedforward compensation.

[0013] A six-phase dual-winding motor cooperative drive control system based on SiC MOSFETs includes: A control center is used to execute the linear extended state observer and the multi-objective finite control set model predictive control algorithm in parallel. A six-phase full-bridge inverter topology, which consists of 12 SiC MOSFET power switches, is used to receive the switching vector and drive the motor; The hardware current limiting and blocking circuit monitors the phase current analog signal in real time through a hardware comparator. When the current exceeds a preset threshold, the drive pulse sent to the SiC MOSFET is blocked through a logic gate circuit.

[0014] The system also includes: An auxiliary power supply module is used to supply power to the isolated drive circuit of the SiC MOSFET.

[0015] The beneficial effects of this invention are: (1) The present invention is in A second-order linear extended state observer is constructed for each shaft current loop. The system internal parameter perturbation, electromagnetic coupling between windings, and external load abrupt change are integrated into a total disturbance. The observer estimates and feeds forward to compensate in real time. It does not rely on the precise mathematical model of the motor and does not require a complex decoupling algorithm. It can actively cancel the influence of nonlinear coupling and parameter changes on the current loop. On this basis, a linear state error feedback control law is used to generate the target reference voltage, so that the controlled object is approximately a first-order pure integral linear system in the entire operating domain.

[0016] (2) This invention abandons the traditional fixed switching frequency PWM modulation method and adopts finite control set model predictive control (FCS-MPC) to directly generate inverter switching signals. In addition to the current tracking error term and harmonic suppression term, a penalty term for the number of switching state changes is specially introduced in the cost function to quantify the switching action loss of SiC devices in real time. By adjusting the penalty weight, the system can automatically select the feasible voltage vector with the fewest switching times according to the current operating conditions, thereby effectively suppressing switching losses while ensuring control bandwidth.

[0017] (3) The present invention integrates a pure hardware overcurrent blocking circuit in the drive system. The circuit uses a hardware comparator to monitor the phase current analog signal in real time. Once the preset threshold is exceeded, the drive pulse of SiC MOSFET is directly blocked in nanosecond time through logic gate circuit. Compared with conventional software interrupt protection, the response time is shortened from microsecond to nanosecond, which is fully compatible with the short circuit withstand time of tens of nanoseconds to several microseconds of SiC MOSFET. At the same time, the protection circuit is decoupled from the control software and is not affected by program runaway, electromagnetic interference, etc., providing an independent physical safety net for the power module. Even under extreme faults such as phase-to-phase short circuit, the device can be safely shut down, which greatly improves the reliability of mobile equipment.

[0018] (4) In the fault detection stage, when an open circuit fault occurs in a certain phase winding, the present invention can adjust the basis of the Park transformation matrix in real time, reconstruct the virtual voltage vector using the remaining healthy phase, and automatically shrink the optimization space of the cost function. Under the derating state, it maintains the circular rotating magnetomotive force. This fault-tolerant strategy does not require additional hardware redundancy or switching of control mode, ensuring that the motor can still output torque smoothly after the fault, avoiding system sudden stop. It is suitable for robots and special equipment with strict requirements for continuous operation. Attached Figure Description

[0019] The invention will now be further described with reference to the accompanying drawings.

[0020] Figure 1 This is a diagram showing the overall architecture of the six-phase dual-winding motor cooperative drive system of the present invention; Figure 2 This is a schematic diagram of the auxiliary power supply and +15V isolated drive power supply circuit for the wide voltage input system of the present invention; Figure 3 This is a schematic diagram of a hardware comparator-based high-speed overcurrent protection circuit independent of software logic, as described in this invention. Figure 4 This is a schematic diagram of the physical blocking circuit of the present invention. Detailed Implementation

[0021] The technical solutions of the embodiments of the present invention will be clearly and completely described below with reference to the accompanying drawings. Obviously, the described embodiments are only some embodiments of the present invention, and not all embodiments. Based on the embodiments of the present invention, all other embodiments obtained by those skilled in the art without creative effort are within the scope of protection of the present invention.

[0022] Please see Figures 1-4 As shown; like Figure 1As shown, the overall architecture of the six-phase dual-winding motor cooperative drive system of this invention includes: an FPGA control center, a six-phase full-bridge inverter (composed of 12 SiC MOSFETs), a current and voltage sampling circuit, a hardware overcurrent blocking circuit, an auxiliary power supply module, and a six-phase dual Y-winding motor.

[0023] Figure 2 This diagram illustrates the auxiliary power supply and +15V isolated drive power supply circuit for a wide-voltage input system provided in this embodiment of the invention. The circuit employs a synchronous buck topology: the high-voltage DC bus (48V~140V) is connected to the buck switching transistor via an input filter capacitor, and a stable +15V voltage is output through inductor energy storage and a freewheeling diode (or synchronous rectifier). The feedback loop uses a resistor divider network to sample the output voltage and send it to the controller for voltage regulation. The +15V output is then processed by an isolation transformer or an isolation DC-DC converter to generate multiple isolated +15V power supplies, which are used to drive the upper SiC MOSFETs in the six-phase bridge arm (e.g., to drive optocouplers or capacitive couplers), while simultaneously generating a +5V low-voltage power supply for the FPGA and other logic circuits. This scheme ensures the stability of the drive power supply over a wide input voltage range and meets the stringent requirements of SiC MOSFETs for drive voltage.

[0024] Figure 3 This is a schematic diagram of a hardware comparator-based high-speed overcurrent protection circuit independent of software logic, provided for embodiments of the present invention. The current in each phase is converted into a voltage signal by a current sensor (such as a Hall element or shunt resistor) and sent to the non-inverting input of a high-speed comparator. The inverting input of the comparator is connected to an overcurrent threshold voltage set by a precision resistor divider network (e.g., set to 1.5 times the peak rated current). When the current in any phase exceeds the threshold, the corresponding comparator output immediately flips, generating a comprehensive fault signal through an OR / AND gate logic combination. This fault signal is directly connected to the enable pin of all SiC MOSFET gate driver chips or uses an AND gate to block the PWM pulse channel, pulling the drive pulse low within nanoseconds to completely shut down all switching transistors. Simultaneously, the fault signal is also sent to the FPGA for software recording and fault handling. This pure hardware protection path does not rely on any software program, and its response time is much shorter than the short-circuit withstand time of SiC MOSFETs (typically 2-3 microseconds), effectively preventing power device burnout.

[0025] Figure 4The schematic diagram of the physical blocking circuit provided in this embodiment of the invention (further detailed logic gate implementation) shows that six PWM pulses are sent to the corresponding driver chips after passing through six AND gates. One input of each AND gate receives the PWM pulse, and the other input receives the inverted signal from the corresponding comparator output (high level during normal operation, low level during overcurrent). When any comparator is triggered, the output of the corresponding AND gate immediately goes low, blocking the driving pulse for that phase. To prevent false triggering caused by transient noise interference, an RC filter network can be set at the comparator output, but the time constant is controlled within 100 nanoseconds to balance anti-interference capability and fast response. In addition, a global fault latch circuit is provided. Once a fault occurs, the latch output remains blocked until an external reset signal clears it, ensuring system safety.

[0026] This invention relates to a coordinated drive control method for a six-phase dual-winding motor based on SiC MOSFETs, comprising the following steps: Step 1: Synchronously acquire the six stator phase currents of the six-phase double Y-winding motor, and map the six stator phase currents to mutually orthogonal fundamental subspaces using the generalized Park transform. Axial and Harmonic Subspace Axis and zero-order subspace axis; Step Two, in The shaft current loop constructs a second-order linear extended state observer (LESO) to integrate internal parameter perturbations, inter-winding electromagnetic coupling, and external load abrupt changes into a total disturbance. Because the mathematical model of a six-phase dual-winding motor is complex and precise decoupling is difficult, the method of lumped disturbance and real-time estimation by observer can eliminate the need to know the specific source of the disturbance, and only need to observe its comprehensive effect and compensate for it. The second-order LESO is used instead of the first-order because the first-order can only estimate the state (current) and cannot estimate the disturbance at the same time. The second-order LESO can output the current estimate and the disturbance estimate at the same time, which is convenient for feedforward compensation. The total disturbance is estimated in real time using the LESO. The observed values ​​are then fed forward and compensated to output the results. Axis target reference voltage vector; The second-order linearly extended state observer LESO in shaft and With the same axis structure, the discretized observer equation for the q-axis is: ; in, Sampled for the current control cycle Actual shaft current, For the observer pair The estimated value of the shaft current and the estimated value of the current in the next cycle are jointly determined by the estimated value of the current cycle, the estimated value of the total disturbance of the current cycle, the estimated error term of the current cycle, and the control input term of the current cycle. For the observer pair Total shaft disturbance Real-time estimated value; The current estimation error represents the deviation between the actual sampled current and the estimated current, reflecting the accuracy of the observer's estimation of the system state. For the output of the current cycle Shaft control voltage; The control cycle is determined by the heat dissipation capability of the SiC MOSFET, current ripple requirements, and control bandwidth, and typically ranges from 20μs to 100μs. A smaller value is chosen if the temperature rise is low and a higher control bandwidth is required; a larger value is chosen if the system has a heavy thermal load. In this embodiment, a value of [value missing] is used. =50μs, corresponding to an equivalent switching frequency of 20kHz; Given the control gain, , for Shaft inductance; , The observer gain coefficient is calculated based on the observer bandwidth. Observer bandwidth The typical value range is 500–2000 rad / s. , The input of the observer is , The output is the estimated current value for the next cycle. Total disturbance estimate Observer bandwidth This determines the tracking speed of the extended state observer for the total disturbance. The larger the value, the faster the disturbance estimation, but the more sensitive it is to current sampling noise. In engineering, it is usually selected as 5 to 10 times the expected current loop bandwidth of the system. In this embodiment, we take... =1000 rad / s; =2000, =1×10 6 During actual debugging, the size can be gradually increased within the above range. Continue until the system starts to oscillate, then adjust the value back to 70% of the oscillation value.

[0027] Based on the current estimate and total disturbance estimate output by the linear extended state observer, the final value is calculated using a linear state error feedback control law. The target reference voltage vector of the axis; the control law in shaft and Axisymmetric The expression for the axis is: The final voltage command consists of a proportional control term and a disturbance compensation term. The voltage is adjusted based on the deviation between the target current and the estimated current; the larger the deviation, the larger the voltage adjustment. Disturbance compensation term. The total disturbance estimated by the observer is fed forward to the control output in the form of negative feedback to actively counteract external disturbances and internal changes. The controller output is converted into a physical voltage command. Although the integral term of the PI controller can eliminate steady-state error, it is prone to integral saturation and overshoot under fast transient conditions. This invention replaces the integral term with total disturbance estimation and feedforward compensation to achieve overshoot-free and faster response.

[0028] in, for The target reference current of the shaft, For the final output Axis target reference voltage vector components, for The proportional bandwidth of the axis controller typically ranges from 0.05 to 0.2 and is dimensionless. The value affects the current closed-loop response speed and stability. Specific tuning method: First, set... Set it to a small value (e.g., 0.05), and then slowly increase it under load until the current response shows a slight overshoot (overshoot amount < 5%). This value is the optimal value. In this embodiment, we take 0.1.

[0029] Step 3: Receive the target reference voltage vector, and based on the discretized motor prediction model, traverse all switching state combinations of the inverter to construct a global cost function that includes current tracking error terms, harmonic suppression terms, and SiC MOSFET switching action penalty terms. By minimizing the cost function The optimal switching vector is selected to directly generate a high-voltage gate drive signal.

[0030] The expression for the discretized motor prediction model is: ; in, This is the six-dimensional current vector obtained after sampling and decoupling in step one during the current control cycle. Represents the space vector of the six-phase winding; For the six-dimensional current prediction vector of the next cycle, This is the six-dimensional voltage command vector for the current cycle. All are set to zero; To control the cycle, For stator equivalent inductance, For stator resistance, It is a six-dimensional back electromotive force vector, the components of which are given by the standard voltage equation of the six-phase dual-winding motor. And zero in the zero-order subspace. Motor parameters can be obtained by measuring offline motor parameter identification methods: for example, DC voltammetry to measure resistance, AC injection to measure inductance, and no-load back electromotive force to measure flux linkage.

[0031] The global cost function The expression is: ; The squared term represents the error between the predicted current and the target current; , The current in the subspace does not generate torque, but it does generate copper losses and leakage flux, and the square term penalizes these losses; The sum of the number of state changes of each bridge arm switch directly reflects the number of switch actions; the more actions, the greater the switch losses.

[0032] Tracking only current causes frequent switching states of the MPC, resulting in significant switching losses in SiC MOSFETs; adding harmonic suppression terms avoids this. Circulating current; adding a switching frequency penalty term to suppress unnecessary switching actions, achieving energy saving and thermal management optimization while meeting control performance requirements; the square error has a stronger penalty for large deviations, which is conducive to rapid convergence to the target value; and the square function is differentiable everywhere, which is more convenient in theoretical analysis and hardware implementation.

[0033] in, , They are respectively shaft and The target reference current for the shaft; , They are respectively shaft and Predicted value of shaft current; , They are respectively shaft and Predicted current values ​​in the axial harmonic subspace; For the first The switching state of each inverter arm in the current control cycle is either 0 or 1; 0 indicates that the lower arm is on and the upper arm is off, and 1 indicates that the upper arm is on and the lower arm is off. For the first The switching state of each inverter arm in the previous control cycle; , , These are the torque tracking weight, harmonic suppression weight, and switching loss penalty weight, respectively, all of which are dimensionless positive real numbers. The weighting coefficients can be determined through offline simulation or online self-tuning in engineering applications. For example... =1 is used as the baseline. =0.5, =0.01. If the harmonic current in the system is too large, causing severe rotor overheating, the value can be appropriately increased. (For example, 0.5~2); if the SiC MOSFET has excessively high temperature rise and large switching losses, the value can be increased. .

[0034] The global cost function Select to make The smallest switching vector is taken as the optimal output.

[0035] In this embodiment, the hardware parallelism of the FPGA is utilized to achieve this in one control cycle. The following operations can be completed simultaneously within 50μs: The system synchronously acquires six-phase currents and performs generalized Park transform and inverse Park transform. The required FPGA clock cycles are approximately 20, and the time consumption is less than 1μs.

[0036] Parallel operation shaft and The axis has two linearly extended state observers (LESO), each of which independently performs multiplication and addition operations, with a computation time of approximately 0.5 μs.

[0037] Based on the estimated value of the LESO output, the linear state error feedback control law is calculated in parallel to obtain... Axis target reference voltage , .

[0038] For Finite Control Set Model Predictive Control (FCS-MPC), the FPGA pre-stores voltage vector tables corresponding to 64 switching states of the six-phase inverter. Within each control cycle, 64 parallel cost function calculation units are used to calculate the predicted current for the next cycle based on the discretized prediction model for each switching state. And then calculate the cost function value in that state. All 64 The values ​​are generated in parallel within the same clock cycle, and then the minimum value and its corresponding switch state number are found within 10 clock cycles through a hardware comparison tree (64-input comparator). Finally, the switch vector corresponding to this number is directly sent to the driver chips of the six bridge arms through the parallel output port; the entire MPC optimization process takes about 2μs, which is much less than the control cycle, leaving sufficient time margin for fault detection and auxiliary functions.

[0039] The method also includes step four: When an open-circuit fault is detected in any phase winding, the basis of the generalized Park transformation matrix is ​​adjusted online to reconstruct a virtual voltage vector without order harmonic components, and the optimization calculation of the global cost function J is re-executed within the reduced set of switch vectors to maintain the stator synthesized magnetomotive force as circular.

[0040] When an open-circuit fault is detected in any phase winding, the system performs fault-tolerant operation according to the following sub-steps: Fault detection: Real-time monitoring of the effective or instantaneous values ​​of the six-phase current, and utilization of the zero-sequence current component after generalized Park transform. , Make a judgment. If the current of a certain phase is consistently below 10% of the normal value for more than one control cycle, and the drive pulse of the corresponding bridge arm is normal, then it is determined that the winding of that phase is open.

[0041] Transformation matrix reconstruction: Based on the location of the fault phase, the basis of the generalized Park transformation matrix is ​​dynamically adjusted, reducing the original 6×6 decoupling matrix to a 5×5 matrix, and redefining the basis vectors of the fundamental subspace and harmonic subspace, so that the remaining five phase currents can still generate a circular rotating magnetomotive force after mapping.

[0042] Virtual voltage vector reconstruction: Discard all switching states involving the faulty phase from the original 64 voltage vectors, retaining only the switching combinations that the remaining healthy phases can generate. Using these effective switching states, a virtual voltage vector with the same amplitude and continuous phase as before the fault is synthesized through offline calculation or real-time table lookup, ensuring the continuity of the magnetomotive force trajectory.

[0043] Current command reallocation: Based on the magnitude of the torque command before the fault, The target reference current is proportionally distributed to the remaining five-phase windings; simultaneously, The current command in the harmonic subspace is adjusted to a non-zero value to balance the current in each phase and prevent overload of a certain phase.

[0044] Optimization range narrowing: The optimization range of the cost function of FCS-MPC is limited to the set of effective switching states after reconstruction. The cost function is minimized again and the optimal switching vector is output. The whole process is completed within one control cycle after the fault occurs, realizing sensorless derating fault-tolerant operation and maintaining the stable output torque of the motor.

[0045] A six-phase dual-winding motor cooperative drive control system based on SiC MOSFETs includes: The control center, employing an FPGA or SoC chip, is used to execute the linear extended state observer and the multi-objective finite control set model predictive control algorithm in parallel. The FPGA chip integrates hardware multiply-accumulators, distributed RAM, and sufficient logic units, enabling parallel execution of matrix transformations, LESO iterations, MPC optimization, and other operations. The FPGA's operating frequency is 50MHz–200MHz, and the control cycle is… Corresponding timer interrupts; all algorithms are written in a hardware description language (Verilog / VHDL) as a parallel pipeline structure.

[0046] The six stator phase currents are converted into voltage signals by Hall current sensors or precision shunt resistors. These voltage signals are then filtered by signal conditioning circuits, such as second-order low-pass filters, with the cutoff frequency set to 1 / 5 to 1 / 10 of the switching frequency to remove high-frequency noise. The filtered signals are then sent to the FPGA's built-in analog-to-digital converter (ADC) or an external multi-channel high-speed ADC (12-bit resolution, sampling rate above 1Msps). The FPGA synchronously triggers sampling of all channels in each control cycle.

[0047] The six-phase full-bridge inverter topology consists of 12 SiC MOSFET power switches, with the upper and lower switches of each phase forming a bridge arm. Each SiC MOSFET's gate is driven by an isolated driver chip, such as a magnetically or capacitively isolated driver IC. This driver chip provides a +15V turn-on voltage and a -3V to -5V turn-off negative voltage, and has an independent fault feedback pin. The driver chip's input receives PWM pulses from the FPGA, and its output is connected to the SiC MOSFET gate via a gate resistor with a resistance value of 2Ω to 10Ω. The hardware current limiting and blocking circuit is independent of the software logic of the control center. It monitors the phase current analog signal in real time through a hardware comparator. When the current exceeds a preset threshold, it blocks the drive pulse sent to the SiC MOSFET through a logic gate circuit. Overcurrent threshold voltage setting method: Use a precision resistor divider network, such as 1% precision resistors, to divide the voltage from a +5V reference voltage source. Use the peak rated current as the reference voltage. For example, if the sensitivity of the current sensor is K (V / A), then the threshold voltage... =1.5× ×K. Different threshold values ​​can be achieved by adjusting the ratio of the voltage divider resistors. The comparator output can be connected to a latch (such as an RS flip-flop). Once an overcurrent is triggered, the fault state is maintained until it is manually reset.

[0048] The auxiliary power supply module receives a 48V–140V high-voltage DC bus input. It first steps down the voltage to +15V from the main power supply using a synchronous step-down controller (e.g., LM5116). Then, multiple isolated DC-DC converters generate isolated +15V drive power (one for each upper transistor). Simultaneously, a low-dropout regulator generates +5V to supply the control center and other low-voltage circuits. All power supplies are designed with overvoltage, overcurrent, and thermal shutdown protection.

[0049] The foregoing has provided a detailed description of one embodiment of the present invention, but this description is merely a preferred embodiment and should not be construed as limiting the scope of the invention. All equivalent variations and modifications made within the scope of the claims of this invention should still fall within the patent coverage of this invention.

Claims

1. A collaborative drive control method for a six-phase dual-winding motor based on SiC MOSFETs, characterized in that, The method includes the following steps: Step 1: Synchronously acquire the six stator phase currents of the six-phase double Y-winding motor, and map the six stator phase currents to mutually orthogonal fundamental subspaces using the generalized Park transform. Axial and Harmonic Subspace Axis and zero-order subspace axis; Step Two, in The shaft current loop constructs a second-order linear extended state observer (LESO) to integrate internal parameter perturbations, inter-winding electromagnetic coupling, and external load abrupt changes into a total disturbance. The total disturbance is estimated in real time using the LESO. The observed values ​​are then fed forward and compensated to output the results. Axis target reference voltage vector; Step 3: Receive the target reference voltage vector, and based on the discretized motor prediction model, traverse all switching state combinations of the inverter to construct a global cost function that includes current tracking error terms, harmonic suppression terms, and SiC MOSFET switching action penalty terms. By minimizing the cost function The optimal switching vector is selected to directly generate a high-voltage gate drive signal.

2. The six-phase dual-winding motor cooperative drive control method based on SiC MOSFET according to claim 1, characterized in that, The second-order linearly extended state observer LESO in shaft and The shaft structure is the same. The discretized observer of the axis updates the current estimate and total disturbance estimate for the next period based on the difference between the current sample value and the current estimate for the current period: The current estimate for the next cycle is based on the current cycle value, plus the control cycle multiplied by (the total disturbance estimate for the current cycle, the proportional term of the estimation error, and the gain term of the control voltage for the current cycle). The total disturbance estimate for the next cycle is the current cycle value plus another proportional term, which is the control cycle multiplied by the estimation error. The estimation error is positively correlated with the current sample value and negatively correlated with the current estimate value.

3. The six-phase dual-winding motor cooperative drive control method based on SiC MOSFET according to claim 2, characterized in that, The global cost function It consists of a weighted sum of three terms: The first item reflects the accuracy of torque current tracking, and this value varies with... shaft and The target current of the shaft increases as the sum of the squares of the differences between the target current and the predicted current increases; The second term reflects harmonic loss, and this value varies with... shaft and The predicted current increases as the sum of the squares of the currents on the axis increases; The third item reflects the switching losses of the SiC MOSFET, and this value increases with the total number of switching state changes of the six bridge arms; The smaller the value, the better; the system selects... The smallest switching vector is used as the output.

4. The six-phase dual-winding motor cooperative drive control method based on SiC MOSFET according to claim 3, characterized in that, The method also includes step four: When an open-circuit fault is detected in any phase winding, the basis of the generalized Park transformation matrix is ​​adjusted online to reconstruct a virtual voltage vector without order harmonic components, and the optimization calculation of the global cost function J is re-executed within the reduced set of switch vectors to maintain the stator synthesized magnetomotive force as circular.

5. The six-phase dual-winding motor cooperative drive control method based on SiC MOSFET according to claim 4, characterized in that, The discrete motor prediction model calculates the current prediction vector for the next cycle based on the current vector, voltage command vector, and back electromotive force vector of the current cycle. The predicted current decreases linearly with the current current, and the attenuation coefficient is determined by the stator resistance, inductance, and control cycle. Simultaneously, the predicted current and voltage commands minus the back electromotive force show a positive correlation gain relationship; among which... axis, The voltage command for the axis and zero-sequence subspace is set to zero, and the back electromotive force is at... And zero in the zero-order subspace.

6. The six-phase dual-winding motor cooperative drive control method based on SiC MOSFET according to claim 2, characterized in that, In step two, the linear state error feedback control law calculates the final target reference voltage vector based on the difference between the target current and the current estimate output by the observer, as well as the total disturbance estimate. The voltage vector is proportional to the current error. After subtracting the total disturbance estimate, the value is divided by the known gain. That is, the larger the current error, the larger the voltage regulation, and the larger the total disturbance estimate, the stronger the feedforward compensation.

7. A system for implementing the cooperative drive control method as described in any one of claims 1-6, characterized in that, include: A control center is used to execute the linear extended state observer and the multi-objective finite control set model predictive control algorithm in parallel. A six-phase full-bridge inverter topology, which consists of 12 SiC MOSFET power switches, is used to receive the switching vector and drive the motor; The hardware current limiting and blocking circuit monitors the phase current analog signal in real time through a hardware comparator. When the current exceeds a preset threshold, the drive pulse sent to the SiC MOSFET is blocked through a logic gate circuit.

8. The six-phase dual-winding motor cooperative drive control system based on SiC MOSFETs according to claim 7, characterized in that, The system also includes: An auxiliary power supply module is used to supply power to the isolated drive circuit of the SiC MOSFET.

Images