A Zero-Point Calibration Method for Motor Resolvers Based on Multiphase Injection and Phase Normalization Fusion

Abstract

This invention discloses a motor resolver zero-position calibration method based on multiphase injection and phase normalization fusion, relating to the field of motor control technology. The method includes initialization after entering the calibration state, including initializing the phase index, clearing the angle accumulator, and clearing the sampling timer; three-phase polling injection current, adaptive constant current control, and acquiring the original resolver angle values ​​as the original observation values; and performing state machine interlocking based on residual current detection after sampling of each phase and before switching to the next phase; performing theoretical phase compensation and phase normalization calculations on the acquired original observation values ​​to obtain independent observation values ​​for the three phases; and finally obtaining the resolver zero-position offset angle after data fusion. This invention overcomes the systematic bias of the single-phase method, reduces random errors by using multiple average measurements, and solves residual magnetism interference through the "residual current detection interlocking" mechanism, ensuring the independence and accuracy of the measurement sequence.

Description

Technical Field

[0001] This invention relates to the field of motor control technology, specifically, to a motor resolver zero-position calibration method based on the fusion of multiphase injection and phase normalization. Background Technology

[0002] In high-precision motion control fields such as electric drive systems for new energy vehicles, industrial servo drives, and robot joints, permanent magnet synchronous motors (PMSMs) have become the mainstream choice due to their high power density and high efficiency. To achieve its core control algorithm—Field-Oriented Control (FOC)—the control system must acquire the spatial position (angle) of the motor rotor in real time and with high precision. Resolvers, with their robustness, strong anti-interference capabilities, and stable operation under high temperature and pressure, are widely used as position sensors. However, during the assembly of the motor and resolver, it is difficult to ensure that the mechanical zero points of the resolver rotor and the motor rotor are perfectly aligned; this installation deviation is called the "zero-point offset angle." If this angle is not accurately calibrated, the coordinate system used by the FOC algorithm will shift, leading to torque pulsation, reduced efficiency, increased noise, and in severe cases, even starting failure or reverse rotation. Therefore, "resolver zero-point calibration" is an indispensable and critical calibration step in the mass production and maintenance of PMSM drives.

[0003] The most widely used zero-position calibration method for permanent magnet synchronous motor resolvers in existing technologies is the "single-phase DC injection method" (also known as the DC positioning method), such as... Figure 1 As shown, the steps include:

[0004] 1. Selecting the injection phase: The U phase of the motor is usually selected as the calibration phase.

[0005] 2. Applying a fixed magnetic field: The motor controller applies a fixed voltage vector in the three phases U, V, and W through the inverter. This vector causes the current to flow mainly (or entirely) into the U-phase winding (for example, let V_u=V_dc, V_v=0, V_w=0), thereby generating a constant static magnetic field in the direction of the stator U-phase axis.

[0006] 3. Rotor positioning: The permanent magnets on the rotor rotate under the attraction of the stator's static magnetic field until the rotor's magnetic pole axis (d-axis) is aligned with the U-phase stator magnetic field axis and remains in that stable position.

[0007] 4. Angle Reading and Calculation: After the rotor stabilizes, read the angle value θ_measured output by the resolver sensor. Since the electrical angle corresponding to the U-phase axis is theoretically 0 degrees, the calculated zero-offset angle is: θ_offset = θ_measured - 0°.

[0008] 5. End: Store the θ_offset value in the controller's non-volatile memory for coordinate transformation compensation in all subsequent FOC operations.

[0009] The above method has the following drawbacks:

[0010] 1) Limitations of Single-Phase Calibration (Systematic Error): Since this method relies solely on single-phase winding measurements, it implicitly assumes that the three-phase winding parameters of the motor are perfectly symmetrical. However, in actual motors, due to factors such as winding process and core material inhomogeneity, the resistance, inductance, and magnetic saturation characteristics of the three-phase windings inevitably exhibit slight differences. Therefore, the zero-position angle obtained solely from U-phase measurements is essentially the alignment between the "electrical center of U-phase" and the rotor magnetic poles, rather than the overall geometric / magnetic circuit center of the motor. This introduces an inherent systematic error, tied to the characteristics of the U-phase, into the calibration results.

[0011] 2) Influence of mechanical backlash (random error): During a single DC engagement process, the final position of the rotor is significantly affected by gear backlash and bearing static friction. As the rotor approaches the equilibrium point from different directions, it will stop at different positions within the gear backlash range. Therefore, the results of a single measurement are random, leading to significant fluctuations in calibration accuracy and repeatability due to mechanical disturbances.

[0012] 3) Residual magnetism interference (cumulative error) in multiphase calibration: To improve the above problem, one approach is to continuously inject DC into each of the U, V, and W phases and average the results. However, without effectively dissipating the magnetic field generated by the previous injection, the system directly switches to the next phase. Since the motor windings are large inductors, the current cannot change abruptly. The residual current (inductance energy storage) of the previous phase winding and the residual magnetism in the core will generate a superimposed interference magnetic field at the initial stage of energizing the next phase. This causes a shift in the positioning equilibrium point of the subsequent phases, and the error accumulates with the phase sequence switching, potentially resulting in a worse outcome than single-phase calibration. Summary of the Invention

[0013] To address the aforementioned technical problems, this invention provides a motor resolver zero-position calibration method based on the fusion of multiphase injection and phase normalization. Through three-phase independent polling measurement, theoretical phase normalization, residual current interlocking, and data fusion, it achieves high-precision, high-robustness, and high-reliability automatic calibration of the motor resolver zero position.

[0014] The present invention solves the above problems through the following technical solution:

[0015] A method for zero-position calibration of a motor resolver based on the fusion of multiphase injection and phase normalization includes:

[0016] Step S1: After entering the calibration state, initialization is performed, including initializing the phase index cal_phase, clearing the angle accumulator sum_phi, and clearing the sampling timer.

[0017] Step S2: Three-phase polling injection current, adaptive constant current control, and acquisition of the original resolver angle value as the original observation value. After sampling of each phase and before switching to the next phase, state machine interlocking based on residual current detection is performed.

[0018] Step S3: Perform theoretical phase compensation and phase normalization calculations on the collected raw observation values ​​to obtain independent observation values ​​for U phase, V phase, and W phase;

[0019] Step S4: Perform data fusion on the independent observations of the U phase, V phase, and W phase to obtain the final zero-position offset angle of the resolver.

[0020] The three-phase independent measurement of this invention overcomes the systematic bias of the single-phase method; it reduces random errors by averaging multiple measurements, and solves residual magnetism interference through the "residual current detection interlock" mechanism, thus ensuring the independence and accuracy of the measurement sequence.

[0021] As a further improvement of the present invention, the method of injecting current is as follows: according to the value of phase index cal_phase, the controller applies an initial PWM duty cycle to the corresponding phase winding to start building current.

[0022] As a further improvement of the present invention, the method for adaptive constant current control and acquisition of the original resolver angle value is as follows:

[0023] During current establishment and the entire sampling process, the absolute values ​​of the three-phase currents are monitored in real time, and the maximum value I_max=max(|I_u|,|I_v|,|I_w|) is calculated.

[0024] The proportional-integral (PI) or proportional-adjustment (P) algorithm is used to dynamically adjust the applied PWM duty cycle so that I_max is accurately stabilized near the preset target current I_target.

[0025] Once I_max stabilizes within the allowable error range of the preset target current I_target for a preset time, it is determined that the rotor has been sufficiently and stably engaged.

[0026] While maintaining closed-loop constant current control, the resolver angle value θ_raw is continuously acquired at a preset frequency and accumulated to the angle accumulator sum_phi, while the sampling counter increments.

[0027] When the number of samplings reaches the preset value TOTAL_SAMPLES, the average rotation angle of the current phase is calculated and used as the original rotation angle value of the phase.

[0028] As a further improvement of the present invention, the state machine interlocking based on residual current detection after each phase sampling is completed and before switching to the next phase specifically includes:

[0029] After each phase sampling is completed, it enters passive discharge mode;

[0030] Perform real-time residual flow monitoring;

[0031] The maximum value I_max at this time is compared with the preset residual current safety threshold I_safe. Only when I_max < I_safe, the controller's state machine is unlocked, allowing the switch to the next phase.

[0032] As a further improvement of the present invention, the state machine interlocking based on residual current detection after each phase sampling is completed and before switching to the next phase specifically includes:

[0033] After each phase sampling is completed, the system enters passive discharge mode and starts timing.

[0034] Perform real-time residual flow monitoring;

[0035] The maximum value I_max at this time is compared with the preset residual current safety threshold I_safe. When I_max < I_safe and the discharge time is greater than the minimum set time, the controller's state machine is unlocked, allowing switching to the next phase.

[0036] In addition to detecting the phase current amplitude, a time threshold can also be used as a criterion. Switching is only allowed after both conditions are met: "current is below the threshold" and "discharge time is greater than the minimum set time", so as to provide a more reliable guarantee.

[0037] As a further improvement of the present invention, step S3 specifically includes:

[0038] The original observation values ​​collected for phases V and W are superimposed with theoretical phase compensation values ​​of 240 degrees and 120 degrees respectively. They are then mapped to the same electrical coordinate system as the U-phase measurement reference and normalized to obtain independent observation values ​​for phases U, V, and W respectively.

[0039] As a further improvement of the present invention, the data fusion method in step S4 is any one of arithmetic mean, weighted average, or median filtering.

[0040] As a further alternative to the present invention, step S2 is replaced by:

[0041] The voltage vector of a specific high-frequency component is injected in a three-phase polling manner. The initial position information of the rotor is estimated as the original observation value by demodulating the high-frequency current response of each phase. After sampling of each phase and before switching to the next phase, a state machine interlock based on residual current detection is performed.

[0042] The core idea of ​​this invention lies in multi-phase independent excitation, phase normalization fusion, and state interlocking, which has universality. The excitation signal is not limited to DC; high-frequency rotating voltage injection or high-frequency pulse injection methods can also be used. The initial positions of each phase can be compensated according to the 'theoretical phase normalization' principle of this invention, and the independence between phases is ensured by 'residual current detection interlocking' (at which time high-frequency residual oscillation needs to be managed), and finally, data fusion is performed.

[0043] Compared with the prior art, the present invention has the following advantages and beneficial effects:

[0044] (1) This invention uses three-phase independent measurement, which is equivalent to sampling the zero position of the motor from three spatial directions. The asymmetric errors of the three-phase windings tend to cancel each other out statistically. After phase normalization and averaging fusion, the final result is closer to the true geometric / magnetic circuit center of the motor, which fundamentally overcomes the systematic deviation of the single-phase method.

[0045] (2) The present invention takes the average rotation angle as the original measurement value, smooths the random jitter of a single mechanical positioning, and reduces the standard deviation of random error to about 1 / √3 (about 58%) of a single measurement, which greatly improves the repeatability and stability of calibration results and reduces the stringent requirements on mechanical assembly accuracy.

[0046] (3) The present invention adopts a “residual current detection interlock” mechanism, which provides a pure initial condition free from historical magnetic field interference for each measurement. That is, the motor is “soft reset” before each measurement, which completely solves the problem of error transmission and accumulation caused by residual magnetism (continuous phase switching can easily lead to magnetic field superposition) interference during multiphase continuous calibration, and ensures the independence and accuracy of the measurement sequence.

[0047] (4) This invention is implemented entirely by motor controller software, without any additional hardware costs. The calibration process is completed automatically when the motor is stationary, without the need for expensive calibration tables, encoders or manual intervention on the production line, which greatly simplifies the production calibration process, improves efficiency, and is suitable for large-scale automated production.

[0048] (5) This invention is generally applicable to all permanent magnet synchronous motors that require position feedback using resolvers, especially in new energy vehicles, industrial servos, home appliances and compressors, and any other occasions that require fast, automated and high-precision zero-position calibration at the end of the production line. Attached Figure Description

[0049] Figure 1 This is a flowchart of the existing single-phase DC injection method;

[0050] Figure 2 This is a flowchart of the present invention;

[0051] Figure 3 This is a schematic diagram of the adaptive constant current control in this invention. Detailed Implementation

[0052] Before introducing the embodiments of the present invention, the technical terms involved in the present invention will be explained as follows:

[0053] PMSM: Permanent Magnet Synchronous Motor. Its rotor uses permanent magnets to establish a magnetic field, and the stator generates a rotating magnetic field through three-phase alternating current.

[0054] Rotary transformer: an analog position sensor that detects the absolute mechanical angle of a rotor through electromagnetic induction. It has advantages such as high reliability and resistance to harsh environments.

[0055] FOC: Field-Oriented Control. A high-performance control strategy similar to that of a DC motor, which decouples the motor current into an excitation component that generates a magnetic field and a torque component that generates torque through coordinate transformation. Its core relies on accurate rotor position information.

[0056] Zero-point offset angle: The electrical angular deviation between the mechanical zero point of the rotary transformer installation and the magnetic field axis of the permanent magnet of the motor rotor. This deviation is a necessary compensation parameter for FOC control.

[0057] Backlash: In mechanical transmission pairs such as gears and worm gears, due to the presence of meshing clearance, when the driving direction changes, the driven part will only follow the driving part after it has moved a certain distance without moving, resulting in nonlinear position control.

[0058] PWM: Pulse Width Modulation. It controls the effective value of the output voltage or current by adjusting the proportion of the on-time of the switching device within one cycle (duty cycle), and is the core control signal of the motor drive inverter.

[0059] State machine: A mathematical model describing the transitions and triggering conditions between different states of a system. In this invention, it is used to manage the orderly flow of states such as "injection, sampling, discharging, waiting, and switching" during the calibration process.

[0060] The present invention will be further described in detail below with reference to embodiments, but the implementation of the present invention is not limited thereto.

[0061] Combined with appendix Figure 2 As shown, this embodiment of the invention provides a motor resolver zero-position calibration method based on multiphase injection and phase normalization fusion, executed by the motor controller. Its core is a calibration state machine, comprising five core steps: initialization, three-phase polling of injected current and sampling, state machine interlocking based on residual current detection, phase normalization calculation, and data fusion. The state machine manages the switching between two basic modes: "injection sampling" and "discharge waiting." Specifically, it includes:

[0062] 1. Initialization

[0063] The system enters calibration mode (e.g., CTRL_STATE_RESOLVER_BIAS_CAL). The phase index cal_phase (0:U, 1:V, 2:W) is initialized, the angle accumulator sum_phi is cleared, and the sampling counter is set to zero. The target injection current I_target for this phase calibration is set.

[0064] 2. Three-phase polling injection current and sampling

[0065] This invention involves the independent, alternating injection of current into three phases, and adaptive constant current control during current establishment and the entire sampling process. Specifically, it includes:

[0066] (1) Start injection:

[0067] Based on the value of the phase index cal_phase, the controller applies an initial PWM duty cycle to the corresponding phase winding to begin building up current. For example:

[0068] cal_phase=0(U phase): Apply voltage vector (V_u,V_v,V_w)=(injection_pwm_duty,0,0).

[0069] cal_phase=1(V phase): Apply voltage vector (0, injection_pwm_duty, 0).

[0070] cal_phase=2(W phase): Apply voltage vector (0,0,injection_pwm_duty).

[0071] (2) Real-time monitoring and adjustment:

[0072] During current establishment and the entire sampling process, the system monitors the absolute values ​​of the three-phase currents in real time and calculates their maximum value I_max = max(|I_u|,|I_v|,|I_w|). A proportional-integral (PI) or proportional (P) control algorithm is used to dynamically adjust the applied PWM duty cycle injection_pwm_duty to ensure that I_max is precisely stabilized near I_target.

[0073] Combination Figure 3 As shown, the adjustment logic of this invention is explained using proportional adjustment P as an example:

[0074] float error = I_target - I_max;

[0075] injection_pwm_duty += Kp * error; / / Kp is the scaling factor

[0076] / / Limit the injection_pwm_duty, for example, to the range [0.0, 1.0].

[0077] injection_pwm_duty=clamp(injection_pwm_duty,0.0f,1.0f).

[0078] This invention employs closed-loop control to ensure that:

[0079] Current accuracy: It overcomes the influence of factors such as battery voltage fluctuations and resistance changes caused by motor winding temperature rise, ensuring that the injected magnetic field strength is consistent for each time and each phase, thus improving the consistency and comparability of calibration.

[0080] System safety: Effectively prevents current surges caused by control anomalies or parameter mismatches, protecting power devices and motors.

[0081] Optimized positioning: A stable current helps the rotor position itself to the balance point more quickly and smoothly, reducing vibration. More importantly, a constant magnetic field strength ensures that the three measurements are performed under the same electromagnetic force. This provides a physical basis for subsequent 'phase normalization' and 'data fusion', which is an important guarantee for obtaining high-precision fusion results.

[0082] (3) Stability judgment and sampling:

[0083] Once I_max stabilizes within the allowable error range of I_target for a period of time (e.g., 200ms), it is determined that the rotor has been fully and stably engaged. Subsequently, while maintaining closed-loop constant current control, the resolver angle value θ_raw is continuously and frequently acquired and accumulated to sum_phi, while the sampling counter increments.

[0084] (4) Calculation of average value and end of phase:

[0085] When the number of samplings reaches a preset value TOTAL_SAMPLES (e.g., 1000 times), the average resolver angle of the current phase is calculated: resolver_offset[cal_phase] = sum_phi / TOTAL_SAMPLES. This value is the original measurement value of the phase. Subsequently, the closed-loop injection control is exited, and preparation is made to enter the discharge interlock stage, i.e., the state machine interlock based on residual current detection.

[0086] 3. State machine interlock based on residual current detection (corresponding to "discharge wait" mode)

[0087] To eliminate inter-phase interference, this phase is forcibly entered after each phase sampling is completed and before switching to the next phase.

[0088] (1) Enter the passive discharge mode:

[0089] After the sampling and calculation of the current phase is completed, the state machine switches to the discharge mode. The controller sets all three-phase PWM duty cycles to zero (or applies a zero vector), so that all upper bridge arms of the inverter are turned off, and the energy in the motor winding inductor is naturally discharged through the freewheeling diodes.

[0090] (2) Real-time residual current monitoring:

[0091] The controller reads the instantaneous three-phase current values I_u, I_v, I_w in real time through a current sensor (such as a Hall sensor or a sampling resistor). The absolute values of the three-phase currents are calculated, and the maximum value is taken: I_max=max(|I_u|,|I_v|,|I_w|).

[0092] (3) Threshold judgment and interlock:

[0093] The system continuously compares I_max with the preset residual current safety threshold I_safe:

[0094] If I_max>=I_safe: it is determined that the residual magnetism has not been completely discharged, the state machine remains in this stage and waits, and switching to the next phase injection is prohibited. This ensures that the core residual magnetism and the inductive stored energy are fully attenuated.

[0095] If I_max<I_safe: it is determined that the discharge is completed, and the motor is in a "magnetically neutral" state. The state machine is unlocked, cal_phase is incremented by 1 (pointing to the next phase), and the system re-enters the "injection and sampling" mode.

[0096] In another possible embodiment of the present invention, in addition to detecting the amplitude of the phase current, a time threshold can also be combined as a judgment criterion. For example, timing is started after entering the discharge mode, and switching is allowed only after both conditions of "current below the threshold" and "discharge time greater than the minimum set time" are satisfied, so as to provide more reliable guarantee.

[0097] 4. Theoretical phase compensation and normalization

[0098] Since the U, V, and W three phases are spatially offset from each other by 120 electrical degrees, the previously collected resolver_offset[0], [1], and [2] correspond to the resolver readings when the rotor is at three different theoretical positions of 0°, 240°, and 120°, respectively. To fuse them, they need to be compensated to the same reference frame (with the U phase as the reference).

[0099] (1) U-phase reference: The U-phase measurement value itself already corresponds to the 0° theoretical position, so: θ_U_norm=resolver_offset[0].

[0100] (2) V-phase compensation: The theoretical position of V-phase is 240° (i.e. 4π / 3 radians), and its normalized value is: θ_V_norm=resolver_offset[1]+4π / 3.

[0101] (3) W phase compensation: The theoretical position of W phase is 120° (i.e. 2π / 3 radians), and its normalized value is: θ_W_norm=resolver_offset[2]+2π / 3.

[0102] Note that angle calculations should be performed within the range of [0, 2π). Furthermore, angle compensation must be matched with the three-phase phase of each motor.

[0103] 5. Multi-source data fusion

[0104] The three normalized independent observations are then combined to obtain the final optimal estimate. The most basic and effective method is the arithmetic mean:

[0105] θ_final_offset=(θ_U_norm+θ_V_norm+θ_W_norm) / 3;

[0106] Finally, the calculated θ_final_offset is stored as the system's resolver zero offset angle resolver_angle_offset, which is used for all subsequent FOC coordinate transformation compensations.

[0107] Compared with the prior art, the present invention has the following features and advantages:

[0108] a. Adaptive constant current control during the injection process: During the DC injection stage, the PWM duty cycle is adjusted through real-time current feedback closed-loop control to achieve precise and stable current control and ensure the consistency of measurement conditions.

[0109] b. Measurement strategy "spatial circular sampling": By measuring three phases independently and normalizing the phase, "single-point sampling" is upgraded to "spatial three-point sampling", which effectively offsets the system error introduced by single-phase asymmetry.

[0110] c. "Residual current detection interlock" during phase switching: Forced discharge and current detection during phase switching ensure that each measurement is performed in a zero residual magnetism environment, eliminating error accumulation.

[0111] These three factors work together to form the basis for the high precision and robustness of this solution.

[0112] In another possible embodiment of the present invention, data fusion can also be a weighted average:

[0113] When it is known that a certain phase winding has better process consistency or a higher signal-to-noise ratio, the normalized values ​​of the three phases can be assigned different weights for weighted averaging, such as:

[0114] θ_final=(k1*θ_U_norm+k2*θ_V_norm+k3*θ_W_norm) / (k1+k2+k3), where k1,k2,k3 are weight coefficients.

[0115] In another possible embodiment of the present invention, data fusion can also be median filtering:

[0116] To improve the ability to resist interference from wild values, the median of the three normalized values ​​can be selected as the final result, i.e., θ_final=median(θ_U_norm,θ_V_norm,θ_W_norm).

[0117] The core ideas of this invention (multi-phase independent excitation, phase normalization fusion, and state interlocking) are universally applicable. In another feasible embodiment of this invention, the excitation signal is not limited to DC; a high-frequency rotating voltage injection method or a high-frequency pulse injection method can also be used. Specifically, voltage vectors containing specific high-frequency components can be applied sequentially to the U, V, and W phase windings of the motor. The initial position information of the rotor is estimated by demodulating the high-frequency current response of each phase. The obtained initial position information of each phase can also be compensated according to the "theoretical phase normalization" principle of this invention, and the independence between phases is ensured by "residual current detection interlocking" (at which point high-frequency residual oscillation needs to be managed), and finally, data fusion is performed. This demonstrates the broad applicability of the method framework of this invention.

[0118] Although the present invention has been described herein with reference to illustrative embodiments, the above embodiments are merely preferred embodiments of the present invention, and the implementation of the present invention is not limited to the above embodiments. It should be understood that those skilled in the art can devise many other modifications and implementations, which will fall within the scope and spirit of the principles disclosed in this application.

Claims

1. A method for zero-position calibration of a motor resolver based on the fusion of multiphase injection and phase normalization, characterized in that, include: Step S1: After entering the calibration state, initialization is performed, including initializing the phase index cal_phase, clearing the angle accumulator sum_phi, and clearing the sampling timer. Step S2: Three-phase polling injection current, adaptive constant current control, and acquisition of the original resolver angle value as the original observation value. After sampling of each phase and before switching to the next phase, state machine interlocking based on residual current detection is performed. Step S3: Perform theoretical phase compensation and phase normalization calculations on the collected raw observation values ​​to obtain independent observation values ​​for U phase, V phase, and W phase; Step S4: Perform data fusion on the independent observations of the U phase, V phase, and W phase to obtain the final zero-position offset angle of the resolver.

2. The motor resolver zero-position calibration method based on multiphase injection and phase normalization fusion according to claim 1, characterized in that, The method for injecting current is as follows: based on the value of the phase index cal_phase, the controller applies an initial PWM duty cycle to the corresponding phase winding to start building current.

3. The motor resolver zero-position calibration method based on multiphase injection and phase normalization fusion according to claim 1, characterized in that, The method for adaptive constant current control and acquisition of the original resolver angle value is as follows: During current establishment and the entire sampling process, the absolute values ​​of the three-phase currents are monitored in real time, and the maximum value I_max=max(|I_u|,|I_v|,|I_w|) is calculated. The proportional-integral (PI) or proportional-adjustment (P) algorithm is used to dynamically adjust the applied PWM duty cycle so that I_max is accurately stabilized near the preset target current I_target. Once I_max stabilizes within the allowable error range of the preset target current I_target for a preset time, it is determined that the rotor has been sufficiently and stably engaged. While maintaining closed-loop constant current control, the resolver angle value θ_raw is continuously acquired at a preset frequency and accumulated to the angle accumulator sum_phi, while the sampling counter increments. When the number of samplings reaches the preset value TOTAL_SAMPLES, the average rotation angle of the current phase is calculated and used as the original rotation angle value of the phase.

4. The motor resolver zero-position calibration method based on multiphase injection and phase normalization fusion according to claim 3, characterized in that, The state machine interlock based on residual current detection, performed after sampling of each phase and before switching to the next phase, specifically includes: After each phase sampling is completed, it enters passive discharge mode; Perform real-time residual flow monitoring; The maximum value I_max at this time is compared with the preset residual current safety threshold I_safe. Only when I_max < I_safe, the controller's state machine is unlocked, allowing the switch to the next phase.

5. The motor resolver zero-position calibration method based on multiphase injection and phase normalization fusion according to claim 3, characterized in that, The state machine interlock based on residual current detection, performed after sampling of each phase and before switching to the next phase, specifically includes: After each phase sampling is completed, the system enters passive discharge mode and starts timing. Perform real-time residual flow monitoring; The maximum value I_max at this time is compared with the preset residual current safety threshold I_safe. When I_max < I_safe and the discharge time is greater than the minimum set time, the controller's state machine is unlocked, allowing switching to the next phase.

6. The motor resolver zero-position calibration method based on multiphase injection and phase normalization fusion according to claim 1, characterized in that, Step S3 specifically includes: The original observation values ​​collected for phases V and W are superimposed with theoretical phase compensation values ​​of 240 degrees and 120 degrees respectively. They are then mapped to the same electrical coordinate system as the U-phase measurement reference and normalized to obtain independent observation values ​​for phases U, V, and W respectively.

7. The motor resolver zero-position calibration method based on multiphase injection and phase normalization fusion according to claim 1, characterized in that, The data fusion method in step S4 is any one of arithmetic mean, weighted average, or median filtering.

8. The motor resolver zero-position calibration method based on multiphase injection and phase normalization fusion according to claim 1, characterized in that, Step S2 is replaced by: The voltage vector of a specific high-frequency component is injected in a three-phase polling manner. The initial position information of the rotor is estimated as the original observation value by demodulating the high-frequency current response of each phase. After sampling of each phase and before switching to the next phase, a state machine interlock based on residual current detection is performed.

Images