PFC starting control method and system of synchronous generator without position sensor based on multi-stage parameter inheritance

CN122993554APending Publication Date: 2026-10-27ZHUZHOU HONGDA MICROELECTRONICS TECH CO LTD
View PDF 0 Cites 0 Cited by

Patent Information

Application Number
CN202611385967.5
Authority / Receiving Office
CN · China
Patent Type
Applications(China)
Current Assignee / Owner
Filing Date
2026-09-08
Publication Date
2026-10-27

AI Technical Summary

Technical Problem

[0005]本发明所要解决的技术问题是现有集成式PFC变换器省去外置电感和电压传感器后,封锁态磁链残留污染采样、滑模观测器冷启动失步及前馈预置相位错位

Benefits of technology

1.在功率开关管封锁态下,提取定子电流基波分量并依据定子电压方程计算磁链耦合电压分量,从桥臂中点电压中扣除后得到纯净反电动势电压信号,消除了体二极管不控整流引起的残余磁链对电压采样的隐性污染,确保预锁相参数提取准确,从源头切断误差传递链条。

✦ Generated by Eureka AI based on patent content.

Smart Images

  • Figure CN122993554A_ABST
    Figure CN122993554A_ABST
Patent Text Reader

Abstract

The application relates to the field of motor control, and particularly discloses a synchronous generator position sensor-free PFC starting control method and system based on multi-stage parameter inheritance, which comprises the following steps: sampling the bridge arm midpoint voltage and the stator current in the power switch tube blocking state, deducting the flux linkage coupling voltage to obtain a pure back electromotive force voltage signal, extracting a steady-state parameter to be packaged into a multi-dimensional state parameter set; the parameter set is written into a sliding mode observer and a matching phase-locked loop, the back electromotive force amplitude convergence degree and the phase convergence degree are monitored, the double convergence outputs a check pass identification and records the corrected residual phase tracking lag angle; the total phase offset compensation amount is calculated, the nominal feedforward voltage value is calculated and preset to an integral register, and a pulse width modulation signal is unlocked; the system comprises functionally matched modules. The application omits external PFC inductance and voltage sensors, eliminates starting current impact and phase out-of-step risk, and realizes seamless and smooth starting of the synchronous generator position sensor-free PFC.
Need to check novelty before this filing date? Find Prior Art

Description

Technical Field

[0001] This invention relates to the field of motor control technology, specifically to a sensorless PFC start-up control method and system for synchronous generators based on multi-stage parameter inheritance, applicable to integrated PFC converters that omit external PFC inductors and directly utilize the stator winding inductance of the synchronous generator as the boost inductor. Background Technology

[0002] In applications with stringent power density requirements, such as aviation power supplies, vehicle-mounted power generation, and portable power supplies, traditional three-phase power factor correction converters (PFCs) employ external boost inductors and input AC voltage sensors, resulting in large size and high cost, which hinders the improvement of system integration. To address this, the industry has proposed replacing external PFC inductors with synchronous generator stator winding inductors and eliminating the input voltage sensor. Instead, a sliding mode observer is used to observe the back electromotive force from the stator current, achieving extreme weight and cost reduction at the hardware level.

[0003] However, when transplanting the mature sliding mode observer from the field of motor drives to the PFC scenario, several deep-seated contradictions exist that existing technologies have not adequately addressed. In the locked-down state, the synchronous generator stator winding forms an uncontrolled rectified current path with the DC bus through the power device's body diode, generating residual flux current. This causes the bridge arm midpoint voltage to be superimposed with a flux-coupled voltage component, making it unsuitable as a pure grid voltage for pre-locking. The back EMF observation state and the phase-locked loop phase-tracking state within the sliding mode observer have different convergence characteristics. Existing unidirectional parameter transfer mechanisms cannot guarantee synchronous convergence in both states, posing a risk of startup before convergence. Without a voltage sensor, the feedforward quantity can only be preset in the locked-down state, while the operating phase is provided by the sliding mode PFC. A residual phase-tracking lag angle exists between the two, causing a spatial misalignment between the feedforward voltage vector and the actual back EMF in the dq coordinate system, triggering a startup current surge.

[0004] Therefore, there is an urgent need for a control method that can achieve shock-free smooth start-up through multi-stage parameter inheritance and closed-loop verification, without completely eliminating the need for external inductors and voltage sensors. Summary of the Invention

[0005] The technical problems to be solved by the present invention are: sampling of residual contamination of the blocked state flux linkage, loss of synchronization of the sliding mode observer during cold start, and misalignment of the feedforward preset phase after the existing integrated PFC converter eliminates the external inductor and voltage sensor.

[0006] The technical solution of the present invention to solve the above-mentioned technical problems is as follows: On the one hand, a sensorless PFC start-up control method for synchronous generators based on multi-stage parameter inheritance is provided, specifically including the following steps: S1. With the power switch in the blocked state, sample the midpoint voltage of the bridge arm and the stator current. Extract the fundamental current component, calculate the flux coupling voltage component according to the stator voltage equation, and subtract the pure back electromotive force voltage signal from the midpoint voltage of the bridge arm.

[0007] The synchronous reference system phase-locked loop is started with this signal, and after convergence, the steady-state AC angular frequency reference value, steady-state electrical angle reference value, and proportional-integral controller integral accumulation value are extracted.

[0008] After Clark transformation, the reference values ​​of steady-state voltage components and current components in the two-phase stationary coordinate system are obtained, and the reference values ​​of steady-state voltage components are stored as the initial value vector of back electromotive force.

[0009] Based on the measured phase relationship between the pure back electromotive force and the stator current, the sampling channel is calibrated for synchronization error, and the inherent delay time parameter is recorded.

[0010] The steady-state AC angular frequency reference value, steady-state electrical angle reference value, reference time identifier, integral accumulation value, current component value, back electromotive force initial value vector and inherent delay time parameter are encapsulated into a multi-dimensional state parameter set, where the back electromotive force initial value vector is the steady-state voltage component reference value of the two-phase stationary coordinate system at the end of the pre-observation window.

[0011] S2. After extrapolating and reconstructing the current component values ​​of the two-phase stationary coordinate system based on the steady-state AC angular frequency reference value to the state writing time, the multi-dimensional state parameters are written into the α-axis and β-axis current observation state variables inside the sliding mode observer; the written electrical angle value and integral accumulation value of the extrapolated and reconstructed steady-state AC angular frequency reference value and steady-state electrical angle reference value to the state writing time are written into the corresponding electrical angular velocity, electrical angle and integral accumulation state variables of the matching phase-locked loop; the initial value vector of the back electromotive force is extrapolated and reconstructed based on the steady-state AC angular frequency reference value to the state writing time and then written into the back electromotive force observation output state register of the sliding mode observer.

[0012] Start the sliding mode observer and its matching phase-locked loop. The observer reconstructs the back electromotive force observation value after compensating for the timing delay of the three-phase AC current signal based on the inherent delay time parameter.

[0013] Independently monitor the amplitude convergence of the back EMF observation value and the initial back EMF vector, as well as the phase convergence of the output phase of the matching phase-locked loop and the extrapolated electric angle reference value.

[0014] When both reach the preset convergence threshold and the difference between the convergence completion times is less than the synchronization tolerance, the dual-state asynchronous convergence is determined to be complete, and a hot migration verification pass flag is output. The steady-state electrical angle reference value is extrapolated to the convergence completion time according to the steady-state AC angular frequency reference value to obtain the extrapolated electrical angle reference value. The difference between the extrapolated electrical angle reference value and the output phase of the matching phase-locked loop is calculated and recorded as the residual phase tracking lag angle after electrical angle extrapolation correction. If the convergence is not completed after the preset maximum convergence waiting time is exceeded, return to S1.

[0015] S3. Based on the DC bus voltage, target modulation ratio, power switch on-state voltage drop parameters, dead time parameters, and switching frequency parameters, calculate the equivalent phase offset angle and additional phase offset angle caused by the nonlinear characteristics of the power switch and sum them as the total phase offset compensation amount, which is then added to the extrapolated electrical angle reference value.

[0016] The compensated electrical angle value is used as the first Park transform angle, and a second correction is made based on the residual phase tracking lag angle to obtain the corrected Park transform angle. The steady-state voltage component reference value is extrapolated to the two-phase stationary coordinate system voltage component at the preset time according to the steady-state AC angular frequency reference value. The corrected Park transform angle is used to perform Park transform on the reconstructed two-phase stationary coordinate system voltage component to obtain the corrected d-axis voltage component and q-axis voltage component, and the corrected nominal feedforward voltage value is calculated to align the projection direction of the dq coordinate system with the actual voltage vector direction.

[0017] The corrected nominal feedforward voltage value is preset into the integral accumulator register of the d-axis and q-axis current loop proportional-integral controller as the initial integral value for the first control cycle; the initial value of the current reference value is set to zero and incremented to the target reference value in a preset increment per control cycle.

[0018] In response to the thermal migration verification, the PWM is unlocked by the identifier, and the current closed-loop control based on the sliding mode observer and its matching phase-locked loop is switched on. This enables the current loop to have a decoupling compensation amount that matches the back EMF of the synchronous generator and is orthogonally compensated by the coordinate system in the first control cycle, thus achieving a smooth start-up without impact.

[0019] On the other hand, a sensorless PFC start-up control system for synchronous generators based on multi-stage parameter inheritance is provided to implement a sensorless PFC start-up control method for synchronous generators based on multi-stage parameter inheritance, including: A synchronous generator whose stator winding inductance is configured as the boost inductance of a power factor correction converter; The power conversion circuit includes a three-phase bridge rectifier topology and omits the external power factor correction inductor; The voltage sampling circuit is used to sample the voltage signal between the midpoint of the three-phase bridge arm and the virtual midpoint of the DC bus during the pre-observation window period when all controllable switching devices of the power conversion circuit are in a blocked state. A current sampling circuit is used to sample the three-phase current signal of the stator winding during the pre-observation window period; The control unit is electrically connected to the voltage sampling circuit, the current sampling circuit, and the power conversion circuit. The control unit includes: The flux linkage decoupling pre-observation module is used to receive the bridge arm midpoint voltage signal output by the voltage sampling circuit and the stator current signal output by the current sampling circuit, extract the fundamental component in the stator current signal and calculate the flux linkage coupling voltage component based on the stator winding inductance parameters, subtract the flux linkage coupling voltage component from the bridge arm midpoint voltage signal to obtain the pure back EMF voltage signal, and extract steady-state parameters based on the pure back EMF voltage signal and encapsulate them into a multi-dimensional state parameter set. The full-state hot migration module is used to write the multi-dimensional state parameter set into the corresponding state register of the sliding mode observer and its matching phase-locked loop; The dual-state asynchronous convergence discrimination module is used to independently monitor the convergence degree of back electromotive force amplitude and phase convergence degree after the sliding mode observer and its matching phase-locked loop are started and running. When the convergence degree reaches the threshold and the difference between the convergence completion time is less than the synchronization tolerance, the module outputs a thermal migration verification pass mark and records the residual phase tracking lag angle corrected by electrical angle extrapolation. The phase offset compensation module is used to calculate the total phase offset compensation amount. It extrapolates the steady-state electrical angle reference value to a preset time according to the steady-state AC angular frequency reference value to obtain the extrapolated electrical angle reference value, and superimposes it on the extrapolated electrical angle reference value to obtain the compensated electrical angle value. The coordinate system orthogonal compensation feedforward module is used to extrapolate and reconstruct the initial value vector of the back electromotive force to a preset time based on the compensated electrical angle value and the residual phase tracking lag angle, and then calculate the nominal feedforward voltage value after coordinate system orthogonal compensation. The current soft-start control module is used to preset the nominal feedforward voltage value into the current loop integral register, and in response to the thermal migration verification, unlocks the pulse width modulation signal through the power conversion circuit to achieve a smooth start-up without impact.

[0020] The beneficial effects of this invention are: 1. Under the blocking state of the power switch, the fundamental component of the stator current is extracted and the magnetic flux coupling voltage component is calculated according to the stator voltage equation. After subtracting it from the midpoint voltage of the bridge arm, a pure back electromotive force voltage signal is obtained. This eliminates the hidden pollution of voltage sampling caused by the residual magnetic flux caused by the uncontrolled rectification of the body diode, ensures the accurate extraction of pre-locked loop parameters, and cuts off the error transmission chain from the source.

[0021] 2. After writing the multi-dimensional state parameter set into the sliding mode observer and its matching phase-locked loop to achieve full-state hot migration, the convergence of the back electromotive force amplitude and the phase convergence are independently monitored. The hot migration verification pass flag is output only when both reach the threshold and the difference at the time of convergence is less than the synchronization tolerance, thus eliminating the risk of starting the observer before convergence and achieving disturbance-free switching.

[0022] 3. By calculating the total phase offset compensation caused by the nonlinear characteristics of the power switch and superimposing it onto the extrapolated electrical angle reference value, and by performing a secondary correction on the Park transformation angle based on the residual phase tracking lag angle, the nominal feedforward voltage value preset to the current loop integral register is aligned with the actual voltage vector direction in the dq coordinate system, thereby fundamentally eliminating the starting current surge. Attached Figure Description

[0023] Figure 1 This is a flowchart illustrating the method described in Embodiments 1-4 of the present invention; Figure 2 This is a framework diagram of the system described in Embodiment 5 of the present invention. Detailed Implementation

[0024] The principles and features of the present invention are described below. The examples given are only for explaining the present invention and are not intended to limit the scope of the present invention. Example

[0025] like Figure 1 As shown, this invention provides a sensorless PFC start-up control method for synchronous generators based on multi-stage parameter inheritance, specifically including the following steps: S1. Under the power switch blocking state, sample the midpoint voltage of the bridge arm and the stator current, subtract the magnetic flux coupling voltage to obtain the pure back electromotive force voltage signal, and extract the steady-state parameters to encapsulate them into a multi-dimensional state parameter set. S2. Write the multi-dimensional state parameter set into the sliding mode observer and its matching phase-locked loop, start operation, and independently monitor the convergence degree of back electromotive force amplitude and phase convergence degree; when the convergence degree reaches the threshold and the difference between the convergence completion time is less than the synchronization tolerance, output the thermal migration verification pass mark and record the residual phase tracking lag angle corrected by electrical angle extrapolation. S3. Calculate the total phase offset compensation amount. Extrapolate the steady-state electrical angle reference value to the preset time according to the steady-state AC angular frequency reference value to obtain the extrapolated electrical angle reference value. Add the total phase offset compensation amount to the extrapolated electrical angle reference value to obtain the compensated electrical angle value. Calculate the nominal feedforward voltage value after orthogonal compensation in the coordinate system. Preset the nominal feedforward voltage value to the current loop integral register. Respond to the thermal migration verification pass flag unlock pulse width modulation signal to achieve shockless smooth start. Example

[0026] Preferably, in Embodiment 1, step S1 is specifically implemented through the following steps: S11. Magnetic flux decoupling pre-observation and pure back EMF acquisition, specifically including: S111. After the system completes power-on initialization, the microcontroller sets the drive signals of all power switching transistors to low level, so that the power switching transistors of the upper and lower bridge arms of the three-phase bridge power conversion circuit are in a blocked state of being turned off. At this time, the midpoint of the bridge arm is directly connected to the stator winding of the synchronous generator and there is no pulse width modulation voltage injection.

[0027] Subsequently, a pre-observation window timer is started. This timer, implemented by a software or hardware timer within the microcontroller, measures the duration of the blocking state. The length of the pre-observation window must be preset to ensure that the synchronous reference system phase-locked loop (PLL) completes frequency locking and phase tracking from its initial state and reaches steady-state convergence. Its value is set between 200ms and 500ms, with 300ms being the preferred value. This setting is based on the fact that the synchronous reference system PLL typically requires several to dozens of power frequency cycles to eliminate initial phase errors and frequency deviations. For a synchronous generator with a rated frequency of 50Hz or 60Hz, a 300ms window can cover 15 to 18 power frequency cycles, ensuring sufficient convergence of the PLL's integral state.

[0028] S112. Throughout the entire pre-observation window, a three-phase bridge arm midpoint voltage detection circuit, configured with a voltage divider resistor network between the midpoint of the three-phase bridge arm and the negative terminal or virtual midpoint of the DC bus, continuously samples the three-phase bridge arm midpoint voltage using analog-to-digital conversion at a preset sampling rate. Simultaneously, a three-phase AC current detection circuit, composed of a Hall current sensor or sampling resistor connected in series between the stator winding and the bridge arm midpoint, synchronously acquires the three-phase current signal of the stator winding at the exact same sampling rate as the voltage sampling. The voltage and current sampling rates are both set to a range of 10kHz to 100kHz, with a preferred value of 20kHz. Synchronous sampling is achieved through a multi-channel synchronous sampling analog-to-digital converter built into the microcontroller, ensuring that the three-phase voltage and current data acquired at the same time are strictly aligned in timing, avoiding false phase errors introduced by sampling time deviations. All voltage and current sampling data acquired throughout the pre-observation window are sequentially stored in the microcontroller's random access memory buffer for subsequent fundamental frequency extraction and flux linkage decoupling operations.

[0029] S113. After the pre-observation window period ends, the microcontroller reads the three-phase current sampling data sequence from the random access memory buffer and extracts the amplitude and phase information of its fundamental component using the discrete Fourier transform algorithm. Based on the inductance voltage drop term in the synchronous generator stator voltage equation, the flux linkage coupling voltage component U... mag In the discrete time domain, through umag (n)=L s ×[i(n)-i(n-1)] / T s Calculate, where L s The stator winding inductance parameter typically ranges from 500 μH to 5 mH, with the specific value determined by the electromagnetic design parameters of the stator winding of the synchronous generator used. i(n) represents the fundamental current value at the current sampling moment, and i(n-1) represents the fundamental current value at the previous sampling moment. T s For the sampling period; or through U in the phasor domain mag =jω b L s The calculation is performed using ×I, where I is the fundamental current phasor and ω... b The fundamental angular frequency of the current is obtained by subtracting the time from the phase information of the extracted fundamental component, for example, by dividing the fundamental phase difference between two adjacent time windows by the window length.

[0030] Subsequently, the calculated flux coupling voltage component is subtracted point by point from the sampled voltage values ​​of the bridge arm midpoint recorded at the same time, thus completing the flux decoupling operation. The result is the pure back electromotive force voltage signal after eliminating the influence of flux coupling inside the stator winding. This signal is not contaminated by pulse width modulation high-frequency components in the blocked state and can truly reflect the air gap magnetic field induced electromotive force of the synchronous generator.

[0031] S12. Steady-state parameter extraction and multi-dimensional state parameter set encapsulation, specifically including: S121. The decoupled pure back EMF voltage signal is used as the input to the synchronous reference system phase-locked loop (PLL), and the PLL is started. The synchronous reference system PLL adopts a constraint strategy based on the zero q-axis voltage component. The input three-phase stationary coordinate system voltage is transformed by Park to obtain the d-axis voltage component and q-axis voltage component in the synchronous rotating coordinate system. The q-axis voltage component is used as the error adjustment input to the internal proportional-integral controller.

[0032] The proportional-integral controller performs closed-loop regulation of the three-phase AC angular velocity based on this error, forcing the q-axis voltage component to converge to zero by continuously adjusting the rotational angular velocity of the synchronous rotating coordinate system. The absolute value of the q-axis voltage component output by the synchronous reference system phase-locked loop is continuously monitored and its magnitude is statistically analyzed within a sliding time window. The preset convergence threshold ranges from 0.5V to 2V, with a preferred value of 1V. The preset time window ranges from 20ms to 100ms, with a preferred value of 50ms. When the absolute value of the q-axis voltage component remains below the preset convergence threshold within the preset time window, the synchronous reference system phase-locked loop is deemed to have achieved stable convergence.

[0033] S122. At the end of the pre-observation window, the synchronous reference system phase-locked loop (PLL) has achieved stable convergence. The microcontroller freezes and records the three-phase AC angular velocity output values ​​among the internal state variables of the PLL at this moment, saving them as the steady-state AC angular frequency reference values. Simultaneously, the electrical angle values ​​output by the PLL at this moment are frozen and recorded, saving them as the steady-state electrical angle reference values. Furthermore, the value stored in the integral accumulator register of the proportional-integral controller within the PLL at this moment is read. This value represents the integral history accumulated by the PLL during convergence to eliminate steady-state errors, and is recorded as the integral accumulation value.

[0034] The above three parameters together constitute the core frequency, phase, and integral state reference required for the subsequent hot migration of the sliding mode observer and its associated phase-locked loop (PLL). This ensures that the PLL associated with the sliding mode observer can inherit the established steady-state regulation history of the synchronous reference system PLL at startup, avoiding convergence from zero. If the synchronous reference system PLL has not yet achieved stable convergence by the end of the pre-observation window (i.e., the absolute value of the q-axis voltage component fails to remain below the preset convergence threshold within the preset time window), the microcontroller extends the pre-observation window or restarts the pre-observation window timer and continues to drive the synchronous reference system PLL with a pure back EMF voltage signal until it achieves stable convergence before performing the parameter freezing and recording steps in this section.

[0035] S123. Take the instantaneous value of the pure back electromotive force voltage signal at the end of the pre-observation window period, and transform it from the three-phase stationary coordinate system to the two-phase stationary coordinate system through Clark transformation to obtain the α-axis voltage component and the β-axis voltage component. The α-axis voltage component and the β-axis voltage component together constitute the reference value of the steady-state voltage component in the two-phase stationary coordinate system. At the same time, take the instantaneous value of the stator winding three-phase current sampling at the end of the pre-observation window period, and transform it to the two-phase stationary coordinate system through Clark transformation to obtain the α-axis current component and the β-axis current component. The α-axis current component and the β-axis current component together constitute the current component value in the two-phase stationary coordinate system. Record the end of the pre-observation window period as the reference time marker.

[0036] Furthermore, the reference value of the steady-state voltage component in the above two-phase stationary coordinate system is directly used as the initial value vector of the back electromotive force of the sliding mode observer and stored. This initial value vector of the back electromotive force provides the sliding mode observer with an initial state estimate close to the real back electromotive force, so that the sliding mode observer does not need to reconstruct the back electromotive force from zero after startup, which significantly shortens the convergence time and suppresses the phase oscillation in the early stage of startup.

[0037] S124. Based on the phase relationship between the pure back-EMF voltage signal and the stator winding current sampling value, the sampling channels of the three-phase bridge arm midpoint voltage detection circuit and the three-phase AC current detection circuit are calibrated for synchronization error. Specifically, the zero-crossing time t is obtained by detecting the zero-crossing time of the pure back-EMF voltage signal and the stator winding current sampling value within the same power frequency cycle. v With the current at zero crossing time t i Calculate the difference Δt between detection times. zero =t v -t i Combined with the steady-state AC angular frequency reference value ω ref From τ=Δt zero +Δθ / ω ref Calculate the inherent delay time parameter τ of the voltage sampling channel relative to the current sampling channel. The value of this inherent delay time parameter τ ranges from −10μs to +10μs, and its absolute value is preferably 1μs. In engineering implementation, its absolute value does not exceed one sampling period. A negative value indicates that the voltage sampling channel leads the current sampling channel. Here, Δθ is the theoretical phase angle of the pure back electromotive force voltage signal leading the stator winding current sampling value, in radians. The theoretical phase difference is preferably obtained by offline pre-calibration experiment measurement, or it can be estimated by AC circuit phasor analysis of the stator winding resistance and synchronous inductance parameters of the synchronous generator.

[0038] After calibration, the steady-state AC angular frequency reference value, steady-state electrical angle reference value, reference time identifier, integral accumulation value, current component value in the two-phase stationary coordinate system, back electromotive force initial value vector (i.e., the steady-state voltage component reference value in the two-phase stationary coordinate system at the end of the pre-observation window), and the inherent delay time parameter of the voltage relative to the current sampling channel are encapsulated into a structured multi-dimensional state parameter set, which is stored in the high-speed random access memory of the microcontroller for subsequent hot migration by the sliding mode observer and its matching phase-locked loop. Example

[0039] Preferably, in Embodiment 1, step S2 is specifically implemented through the following steps: S21. Sliding mode observer full-state hot migration write, specifically including: S211. The microcontroller executes the state writing process for the sliding mode observer and its associated phase-locked loop. This process atomically completes the writing operation of all state registers within a single control cycle, preventing state variables from being tampered with by interrupt service routines during the writing process. Simultaneously, the microcontroller resets the convergence timer and starts timing. The convergence timer uses the microcontroller's control cycle as its timing reference, with the control cycle ranging from 50μs to 500μs, preferably 100μs. The timer increments from zero in steps of the control cycle, providing precise time measurement for subsequent two-state asynchronous convergence discrimination. If the timer value exceeds the preset maximum convergence waiting time, convergence is considered to have failed.

[0040] S212. Read the α-axis current component values ​​and β-axis current component values ​​in the two-phase stationary coordinate system from the multi-dimensional state parameter set, and write them into the α-axis current observation state variables and β-axis current observation state variables in the two-phase stationary coordinate system inside the sliding mode observer via direct memory access or data bus transmission of the microcontroller. Before writing, the microcontroller calculates the state variables according to the fundamental frequency rotation extrapolation formula i based on the time interval Δt between the reference time marker and the state writing time. α (t)=i α (t0)cos(ω ref ·Δt)−i β (t0)sin(ω ref ·Δt), i β (t)=i α (t0)sin(ω ref ·Δt)+i β (t0)cos(ω ref ·Δt) reconstructs the two-phase current components to the state writing time, where t0 is the reference time marker, ω ref This is the steady-state AC angular frequency reference value. This write operation ensures that the initial current observation state of the sliding mode observer at startup is completely consistent with the real-time current value obtained by the physical system after sampling and coordinate transformation. This eliminates the process of the sliding mode observer's current state converging from the zero initial value to the actual value, and avoids transient oscillations or amplitude jumps in the back EMF reconstruction caused by deviations in the initial current observation value.

[0041] S213. Read the steady-state AC angular frequency reference value, steady-state electrical angle reference value, reference time marker, and integral accumulation value of the proportional-integral controller from the multi-dimensional state parameter set, calculate the time interval Δt between the state writing time and the reference time marker, and calculate the integral accumulation value of the proportional-integral controller according to θ. write =θ ref +ω ref×Δt extrapolates the steady-state electrical angle reference value to the state writing time to obtain the written electrical angle value. The steady-state AC angular frequency reference value, the written electrical angle value, and the integral accumulation value of the proportional-integral controller are then directly assigned to the corresponding electrical angular velocity state variables, electrical angle state variables, and proportional-integral controller integral accumulation state variables of the sliding mode observer's PLL. This operation ensures that the sliding mode observer's PLL fully inherits the steady-state regulation history established by the synchronous reference system PLL within the pre-observation window at startup, including frequency locking information, phase tracking information, and the integrator's steady-state output. This ensures that the sliding mode PLL does not need to undergo a frequency pulling and phase capturing process starting from zero, achieving seamless switching in the sense of thermal migration.

[0042] S214. The initial back EMF vector is read from the multi-dimensional state parameter set. This initial back EMF vector is composed of the reference values ​​of the steady-state voltage components in the two-phase stationary coordinate system. It is assigned to the back EMF observation output state register of the sliding mode observer as the initial back EMF observation value for the first control cycle after the sliding mode observer starts. Before assigning it, the microcontroller reconstructs the initial back EMF vector to the state writing time using the same fundamental frequency rotation extrapolation formula as in step S212. This initial value is numerically equal to the real back EMF obtained after decoupling the synchronous generator under the locked state, so that the sliding mode observer has an initial estimate that is highly consistent with the physical back EMF in the first control cycle. This compresses the convergence process required for sliding mode equivalent control to the shortest possible time and significantly suppresses the phase tracking lag and current surge risk caused by the initial back EMF value starting from zero.

[0043] S22. Independent monitoring and calculation of the two-state convergence, specifically including: S221. After completing the hot-migration write of all status registers, the microcontroller sends a run enable command to the sliding mode observer and its associated phase-locked loop, enabling both to synchronously enter the closed-loop operation state. The sliding mode observer first reads the inherent delay time parameter of the sampling channel from the multi-dimensional state parameter set. This parameter characterizes the relative inherent delay of the voltage sampling channel relative to the current sampling channel. Its value is equal to the delay time that needs to be added to the current sampling signal to align the timing of the two sampling channels. A negative value indicates that the current sampling signal needs to be advanced by a corresponding time. The microcontroller calculates the number of delayed sampling points based on the ratio of this delay time parameter to the system sampling period, denoted as N for the integer part and α for the fractional part (0 ≤ α < 1).

[0044] When the delay time parameter is greater than or equal to zero, backward linear interpolation compensation is used: i comp (n)=(1−α)×i(n−N)+α×i(n−N−1), where i(n) is the nth current sample value, i comp (n) represents the compensated current value; this formula degenerates into i when N=0. comp(n)=(1−α)×i(n)+α×i(n−1). When the delay time parameter is a negative value, let the fractional part of the ratio of its absolute value to the sampling period be α' (0≤α'<1), and first-order forward extrapolation compensation is adopted: i comp (n)=i(n)+α'×[i(n)−i(n−1)]. This operation only uses current and historical sampling data, with no requirement for future sampling. The above compensation can achieve timing delay alignment with microsecond-level precision.

[0045] Based on the compensated current signal, the sliding mode observer reconstructs the observed back electromotive force value using the sliding mode equivalent control principle, and its current observation equation in the two-phase stationary coordinate system is: dî α / dt=(u α -ê α ) / L s +k×sign(i α -î α ) and dî β / dt=(u β -ê β ) / L s +k×sign(i β -î β ), where î α , î β are observed currents, i α , i β are measured currents, u α , u β are input voltages, ê α , ê β are observed back electromotive force values, k is the sliding mode switching gain, whose value ranges from 10 to 1000, with a preferred value of 100; sign is the sign function. The switching control law drives the current observation error to approach zero, and then the switching control outputs ê α , ê β are subjected to first-order infinite impulse response low-pass filtering with a cutoff frequency of 100Hz to 2kHz, whose discrete implementation is y(n)=a×x(n)+(1-a)×y(n-1), where the filtering coefficient a=2πf c / (f s +2πf c ), f c is the cutoff frequency, f s is the sampling frequency, the preferred cutoff frequency is 500Hz, and continuous observed back electromotive force values are extracted.

[0046] The typical structure of the phase-locked loop supporting the sliding mode observer is: perform a two-parameter arctangent operation atan2 on the filtered observed back electromotive force values ê α , ê β (ê β ,êα The observed phase is obtained, and the difference between the observed phase and the electrical angle state variable inside the phase-locked loop is calculated and normalized to obtain the phase error. The phase error is input to the proportional-integral controller, and the output of the proportional-integral controller is accumulated to the electrical angular velocity state variable. The electrical angular velocity state variable is integrated to obtain the electrical angle state variable, thereby forming a closed-loop phase tracking. The electrical angular velocity state variable, the electrical angle state variable, and the state variable accumulated by the proportional-integral controller are the objects to be written in step S213.

[0047] S222. During the continuous operation of the sliding mode observer and its associated phase-locked loop, the microcontroller independently calculates and monitors the amplitude convergence and phase convergence in each control cycle. For amplitude convergence, the microcontroller calculates the difference between the magnitude of the back EMF observed value vector and the magnitude of the initial back EMF vector at the current moment, then divides this difference by the magnitude of the initial back EMF vector and takes the absolute value. The result is expressed as a percentage. The preset convergence threshold for this amplitude error percentage ranges from 1% to 10%, with a preferred value of 5%.

[0048] For phase convergence, the microcontroller first extrapolates the steady-state electrical angle reference value to the current control cycle based on the steady-state AC angular frequency reference value. The extrapolation formula is θ. ext (t)=θ ref +ω ref ×(t−t0), where θ ref The steady-state electrical angle reference value recorded at t0, the end of the pre-observation window, is ω. ref The steady-state AC angular frequency reference value is given, where t is the current time. Subsequently, the electrical angle output by the phase-locked loop (PLL) of the sliding mode observer and the extrapolated electrical angle reference value θ at the current time are calculated. ext The absolute phase difference between (t) is calculated by normalizing the angle to the range of −180° to +180° and then taking the absolute value, resulting in an absolute phase difference within the range of 0° to 180°. The preset convergence threshold ranges from 1° to 10°, with a preferred value of 5°. The above two convergence calculation processes are independent of each other and do not affect each other, ensuring separate quantitative evaluation of the back EMF reconstruction accuracy and the phase tracking accuracy.

[0049] S23. Two-state asynchronous convergence determination and timeout backoff, specifically including: S231. In each control cycle, the microcontroller compares the real-time calculated amplitude convergence with a preset amplitude convergence threshold, and simultaneously compares the real-time calculated phase convergence with a preset phase convergence threshold, and performs continuous compliance determination for each: when the amplitude convergence is less than the preset amplitude convergence threshold for N consecutive control cycles (N is a preset positive integer ranging from 2 to 10, with a preferred value of 5), the moment when this condition is first met (represented by the convergence timer count) is recorded as the amplitude convergence completion moment; when the phase convergence is less than the preset phase convergence threshold for N consecutive control cycles, the moment when this condition is first met is recorded as the phase convergence completion moment. Subsequently, the microcontroller calculates the difference between the amplitude convergence completion moment and the phase convergence completion moment, and takes the absolute value of this difference as the convergence completion moment difference.

[0050] S232. The microcontroller compares the calculated convergence completion time difference with a preset synchronization tolerance. The synchronization tolerance ranges from 5ms to 50ms, with a preferred value of 10ms. When the convergence completion time difference is less than the preset synchronization tolerance, the microcontroller determines that the dual-state asynchronous convergence is complete and immediately sets the hot migration verification pass flag. Simultaneously, the microcontroller reads the electrical angle value output by the phase-locked loop (PLL) of the sliding mode observer at the convergence completion time, extrapolates the steady-state electrical angle reference value to the convergence completion time based on the steady-state AC angular frequency reference value, calculates the difference between the extrapolated electrical angle reference value and the output electrical angle value, and stores this difference as the residual phase tracking lag angle in the high-speed random access memory.

[0051] When the convergence completion time difference is greater than or equal to the preset synchronization tolerance, the microcontroller determines that the two states have not yet converged synchronously, continues to run the sliding mode observer and its associated phase-locked loop, and repeatedly performs convergence monitoring and comparison. If the cumulative time of the convergence timer exceeds the preset maximum convergence waiting time since the start of the sliding mode observer and its associated phase-locked loop, convergence is determined to have failed. The microcontroller clears the integral accumulation register of the d-axis and q-axis current loop proportional-integral controller, resets the d-axis and q-axis current reference values ​​to zero, resets the hot migration verification pass flag, and re-enters the pre-observation window. The maximum convergence waiting time ranges from 200ms to 1000ms, with a preferred value of 500ms. Example

[0052] Preferably, in Embodiment 1, step S3 is implemented through the following steps: S31. Calculation of total phase offset compensation, specifically including: The S311 microcontroller reads the current DC bus voltage value through a DC bus voltage detection circuit. This detection circuit typically consists of a voltage divider resistor network and an isolated operational amplifier, with a sampling accuracy of ±0.2V to ±0.5V. Simultaneously, it reads the preset target modulation ratio from the current loop control parameter table. This target modulation ratio characterizes the ratio of the current loop output modulation voltage to the DC bus voltage, and its value ranges from 0.5 to 0.85. Furthermore, it reads the on-state voltage drop parameter at rated current from the power switch datasheet or pre-calibration parameter table. For silicon-based MOSFETs, this parameter ranges from 0.5V to 1.5V.

[0053] Based on the above parameters, the microcontroller calculates the equivalent phase offset angle and the additional phase offset angle according to the nonlinear modulation model of the power switch. The equivalent phase offset angle φ... on The duty cycle distortion caused by the on-state voltage drop is calculated to satisfy φ. on ≈arcsin[V drop / (V dc ×M)],V drop For the on-state voltage drop, V dc Where V is the DC bus voltage, and M is the target modulation ratio; the equivalent phase offset angle ranges from 0.05° to 2°, with a preferred value of 0.5°; for example, V... drop =1.5V, V dc When V = 200V and M = 0.8, φ on ≈arcsin[1.5 / (200×0.8)]≈0.54°. The additional phase offset angle is determined by the dead time T. dead Switching frequency f sw The dead time is determined by the target modulation ratio M, where the dead time ranges from 0.1μs to 0.3μs, with a preferred value of 0.15μs, and the switching frequency ranges from 10kHz to 20kHz, with a preferred value of 16kHz. The additional phase offset angle is obtained by equivalently reducing the fundamental voltage loss caused by the dead time, and under the small-angle engineering approximation, it satisfies φ. dead ≈(2×T dead ×f sw / M)×180° / π, its value ranges from 0.1° to 1.5°, with a preferred value of 0.3°; for example, take T dead =0.15μs, f sw When φ = 16kHz and M = 0.85, dead ≈(2×0.15×10 -6 ×16×10 3 (0.85)×180° / π≈0.32°.

[0054] S312. The microcontroller algebraically superimposes the equivalent phase offset angle calculated in step S311 with the additional phase offset angle, and the sum of the two is used as the total phase offset compensation. The value of the total phase offset compensation ranges from 0.15° to 3.5°, with a preferred value of 0.8°. Subsequently, the microcontroller extrapolates the steady-state electrical angle reference value to a preset time based on the steady-state AC angular frequency reference value. The extrapolation formula is θ. ext (t3)=θ ref +ω ref ×(t3−t0), where t0 is the end time of the pre-observation window and t3 is the time when the preset nominal feedforward voltage value is set; then, this total phase offset compensation is added to the extrapolated electric angle reference value θ by angle addition. ext (t3) means that the compensated electrical angle value is equal to the extrapolated electrical angle reference value plus the total phase offset compensation. This compensation operation aims to pre-compensate for the equivalent phase offset of the bridge arm midpoint voltage caused by the modulation nonlinearity when the power switch switches from the locked state to the pulse width modulation unlocked state, ensuring that the electrical angle value used for the Park transformation at the moment of unlocking is consistent with the phase of the actual voltage vector under the upcoming pulse width modulation condition, thereby ensuring the accuracy of the coordinate transformation and the precision of the feedforward voltage calculation.

[0055] S32. Coordinate system orthogonal compensation feedforward calculation, specifically including: S321. The microcontroller directly assigns the compensated electrical angle value obtained in step S312 to the first Park transform angle variable. Subsequently, it reads the residual phase tracking lag angle recorded in step S232 from the high-speed random access memory. The residual phase tracking lag angle ranges from −1° to +1°, and it is the fixed tracking deviation between the extrapolated electrical angle reference value and the output phase of the phase-locked loop (PLL) of the sliding mode observer. It takes a positive value when the PLL lags, and the deviation is approximately constant during the steady-state operation of the PLL, so it can still be directly used at the preset time. The microcontroller performs a secondary correction calculation, subtracting the residual phase tracking lag angle from the first Park transform angle, and the result is used as the corrected Park transform angle.

[0056] Simultaneously, the microcontroller extrapolates the steady-state voltage component reference value according to the steady-state AC angular frequency reference value to reconstruct the two-phase stationary coordinate system voltage components at the preset time. The reconstruction formula is: u α (t3)=u α (t0)cos( ω ref ·Δt)−u β (t0)sin(ω ref ·Δt), u β (t3)=u α (t0)sin(ω ref ·Δt)+u β(t0)cos(ω ref ·Δt), where Δt=t3−t0. Subsequently, the microcontroller uses the corrected Park transformation angle as the rotation angle to perform Park coordinate transformation on the reconstructed two-phase stationary coordinate system voltage components. Through sine and cosine trigonometric function operations, the α-axis voltage component and β-axis voltage component are projected onto the synchronous rotating coordinate system to obtain the corrected d-axis voltage component and q-axis voltage component.

[0057] S322. The microcontroller performs normalization processing on the corrected d-axis and q-axis voltage components obtained in step S321. Specifically, the corrected d-axis voltage component is divided by the product of the DC bus voltage value and the target modulation ratio to obtain the nominal feedforward voltage value of the d-axis of the current inner loop; the corrected q-axis voltage component is divided by the product of the DC bus voltage value and the target modulation ratio to obtain the nominal feedforward voltage value of the q-axis of the current inner loop. This normalization operation converts the voltage amplitude in the synchronous rotating coordinate system into a per-unit feedforward quantity with the same dimensions as the output of the current loop proportional-integral controller. After the above two correction and normalization processes, the projection direction of the obtained corrected nominal feedforward voltage values ​​of the d-axis and q-axis of the current inner loop in the dq coordinate system is precisely aligned with the direction of the actual back electromotive force voltage vector of the synchronous generator, thereby completely eliminating the coordinate system projection deviation caused by the residual phase tracking lag angle and ensuring the phase accuracy of the feedforward compensation quantity.

[0058] S33. Feedforward preset and shock-free smooth start-up, specifically including: S331. The microcontroller directly writes the corrected nominal feedforward voltage value of the d-axis to the integral accumulation register of the d-axis current loop proportional-integral controller via its internal data bus, and simultaneously writes the corrected nominal feedforward voltage value of the q-axis to the integral accumulation register of the q-axis current loop proportional-integral controller. This write operation is completed atomically within a single control cycle, ensuring that the initial integral values ​​of the d-axis and q-axis current loops take effect at the same time, avoiding cross-coupling disturbances caused by asynchronous write timing between the two axes. The written initial integral value is numerically equal to the back EMF voltage component after coordinate system orthogonal compensation and normalization, with a per-unit range of -0.85 to +0.85. This preset operation ensures that the current loop proportional-integral controller has an integral output matching the back EMF of the synchronous generator in the first control cycle after entering the closed loop, rather than integrating from zero, thereby eliminating sudden changes in current loop output and starting current surges caused by missing initial integral values.

[0059] S332. When the power switch is still in the blocked state, the microcontroller sets the initial values ​​of the d-axis and q-axis current reference values ​​of the current loop to zero at the moment of switching in. The target d-axis current reference value is obtained by limiting the output of the DC bus voltage outer loop regulator, and the target q-axis current reference value is set according to the target power factor, and is set to zero when operating at unity power factor. Incremental values ​​are configured for the d-axis and q-axis current reference values ​​per control cycle, calculated based on a preset total rise time: the total rise time ranges from 10ms to 100ms, with a preferred value of 50ms. Taking a 100μs control cycle as an example, the preferred total rise time corresponds to an increase of 0.2% of the rated current per control cycle, i.e., the proportion of the increment to the rated current per control cycle ranges from 0.1% to 1%. The microcontroller linearly increments the current reference values ​​in each control cycle until the corresponding target reference values ​​are reached. This ramp-up strategy ensures that the inductor current rises smoothly at a controllable rate during the build-up process, avoiding current overshoot caused by step changes in the reference value.

[0060] S333. The microcontroller continuously monitors the status of the thermal migration verification pass flag. When the flag is set, it immediately sends an unlock command to the drive circuit of the three-phase bridge power converter, causing the power switching transistors to start outputting pulse width modulation drive signals, thus officially ending the lockout state. At the same time, the microcontroller switches the feedback signal source of the current closed-loop control to the phase information provided by the sliding mode observer and its matching phase-locked loop, enabling the d-axis current loop and q-axis current loop to enter a closed-loop operation mode based on sliding mode observation.

[0061] In the first control cycle, the d-axis current loop proportional-integral (PI) controller and the q-axis current loop PI controller begin calculations from the preset initial integral values ​​in their respective integral accumulator registers. The sum of their proportional and integral outputs directly constitutes the modulation voltage, which includes back EMF feedforward compensation. Simultaneously, the d-axis and q-axis current reference values ​​smoothly increase from zero according to a preset increment per control cycle, allowing the inductor current to build up slowly after the back EMF has been offset by feedforward. The current loop output modulation voltage is matched in real-time with the synchronous generator's back EMF in both amplitude and phase, achieving a smooth, shock-free start-up. Example

[0062] like Figure 2 As shown, a sensorless PFC start-up control system for synchronous generators based on multi-stage parameter inheritance is used to implement a sensorless PFC start-up control method for synchronous generators based on multi-stage parameter inheritance, including: A synchronous generator whose stator winding inductance is configured as the boost inductance of a power factor correction converter; The power conversion circuit includes a three-phase bridge rectifier topology and omits the external power factor correction inductor; The voltage sampling circuit is used to sample the voltage signal between the midpoint of the three-phase bridge arm and the virtual midpoint of the DC bus during the pre-observation window period when all controllable switching devices of the power conversion circuit are in a blocked state. The current sampling circuit is used to sample the three-phase current signal of the stator winding during the pre-observation window period; The control unit is electrically connected to the voltage sampling circuit, the current sampling circuit, and the power conversion circuit. The control unit includes: The flux linkage decoupling pre-observation module is used to receive the bridge arm midpoint voltage signal output by the voltage sampling circuit and the stator current signal output by the current sampling circuit, extract the fundamental component in the stator current signal and calculate the flux linkage coupling voltage component based on the stator winding inductance parameters, subtract the flux linkage coupling voltage component from the bridge arm midpoint voltage signal to obtain the pure back EMF voltage signal, and extract steady-state parameters based on the pure back EMF voltage signal and encapsulate them into a multi-dimensional state parameter set. The full-state hot migration module is used to write a multi-dimensional set of state parameters into the corresponding state registers of the sliding mode observer and its associated phase-locked loop. The dual-state asynchronous convergence discrimination module is used to independently monitor the convergence degree of back electromotive force amplitude and phase convergence degree after the sliding mode observer and its matching phase-locked loop are started and running. When the convergence degree reaches the threshold and the difference between the convergence completion time is less than the synchronization tolerance, the module outputs a thermal migration verification pass mark and records the residual phase tracking lag angle corrected by electrical angle extrapolation. The phase offset compensation module is used to calculate the total phase offset compensation amount. It extrapolates the steady-state electrical angle reference value to a preset time according to the steady-state AC angular frequency reference value to obtain the extrapolated electrical angle reference value, and superimposes it on the extrapolated electrical angle reference value to obtain the compensated electrical angle value. The coordinate system orthogonal compensation feedforward module is used to extrapolate and reconstruct the initial value of the back electromotive force to a preset time based on the compensated electrical angle value and the residual phase tracking lag angle, and then calculate the nominal feedforward voltage value after coordinate system orthogonal compensation. The current soft-start control module is used to preset the nominal feedforward voltage value into the current loop integral register and, in response to the thermal migration verification, unlocks the pulse width modulation signal through the power conversion circuit to achieve a smooth start-up without impact. The embodiments and / or implementation methods described above are merely preferred embodiments and / or implementation methods for implementing the technology of the present invention, and are not intended to limit the implementation methods of the technology of the present invention in any way. Any person skilled in the art can make some modifications or alterations to other equivalent embodiments without departing from the scope of the technical means disclosed in the content of the present invention, but they should still be regarded as the technology or embodiments that are substantially the same as the present invention.

[0063] This document uses specific examples to illustrate the principles and implementation methods of this application. The descriptions of the above embodiments are only for the purpose of helping to understand the methods and core ideas of this application. The above descriptions are only preferred embodiments of this application. It should be noted that due to the limitations of written expression, while there are objectively infinite specific structures, those skilled in the art can make several improvements, modifications, or changes without departing from the principles of this application, and can also combine the above technical features in an appropriate manner. These improvements, modifications, changes, or combinations, or the direct application of the inventive concept and technical solution to other situations without modification, should all be considered within the scope of protection of this application.

Claims

1. A sensorless PFC start-up control method for synchronous generators based on multi-stage parameter inheritance, characterized in that, include: When the power switch is in the blocked state, the midpoint voltage of the bridge arm and the stator current are sampled, the magnetic flux coupling voltage is subtracted to obtain the pure back electromotive force voltage signal, and the steady-state parameters are extracted and packaged into a multi-dimensional state parameter set. The multi-dimensional state parameter set is written into the sliding mode observer and its matching phase-locked loop, started and run, and the amplitude convergence and phase convergence of the back electromotive force are monitored independently. When the convergence reaches the threshold and the difference between the convergence completion time is less than the synchronization tolerance, output the hot migration verification pass mark and record the residual phase tracking hysteresis angle corrected by electrical angle extrapolation. Calculate the total phase offset compensation amount, extrapolate the steady-state electrical angle reference value to the preset time to obtain the extrapolated electrical angle reference value, and superimpose it on the total phase offset compensation amount to obtain the compensated electrical angle value. Calculate the nominal feedforward voltage value after orthogonal compensation of the coordinate system. The nominal feedforward voltage value is preset into the current loop integral register, and the pulse width modulation signal is unlocked in response to the thermal migration verification pass flag, so as to achieve a shockless smooth start-up.

2. The sensorless PFC start-up control method for synchronous generators based on multi-stage parameter inheritance according to claim 1, characterized in that, In the power switch's locked-out state, the midpoint voltage of the bridge arm and the stator current are sampled, and the pure back electromotive force voltage signal is obtained by subtracting the flux linkage voltage. Specifically, this includes: After the system is powered on and initialized, all power switches are placed in the locked state, and the pre-observation window timer is started. The duration of the pre-observation window is preset to be sufficient for the synchronous reference system phase-locked loop to complete stable convergence. During the pre-observation window, the voltage signal between the midpoint of the bridge arm and the virtual midpoint of the DC bus is continuously sampled at a preset sampling rate by the three-phase bridge arm midpoint voltage detection circuit, and the three-phase current signal of the stator winding is sampled simultaneously by the three-phase AC current detection circuit. The voltage sampling data and current sampling data during the pre-observation window are recorded. After the pre-observation window ends, the amplitude and phase information of the fundamental component of the three-phase current sampling data are extracted. Based on the inductance voltage drop term in the stator voltage equation, the flux coupling voltage component is calculated based on the fundamental component and the stator winding inductance parameters. The decoupled pure back EMF voltage signal is obtained by subtracting it from the voltage sampling value at the midpoint of the bridge arm.

3. The sensorless PFC start-up control method for synchronous generators based on multi-stage parameter inheritance according to claim 2, characterized in that, The extracted steady-state parameters are encapsulated into a multi-dimensional set of state parameters, specifically including: The synchronous reference system phase-locked loop is started with the pure back electromotive force voltage signal. The three-phase AC angular velocity is adjusted by a proportional-integral controller closed loop with the q-axis voltage component as the error adjustment amount. When the absolute value of the q-axis voltage component is continuously less than the preset convergence threshold within a preset time window, the synchronous reference system phase-locked loop is determined to have completed stable convergence. At the end of the pre-observation window, the three-phase AC angular velocity output by the synchronous reference system phase-locked loop is frozen and recorded as the steady-state AC angular frequency reference value. The electrical angle at that moment is frozen and recorded as the steady-state electrical angle reference value. The value of the integral accumulation register of the internal proportional-integral controller is read and recorded as the integral accumulation value. The pure back EMF voltage signal and stator current sampling data at the end of the pre-observation window period are transformed by Clark to obtain the steady-state voltage component reference value and current component value in the two-phase stationary coordinate system. The steady-state voltage component reference value is stored as the initial value vector of the back EMF of the sliding mode observer, and the end time is used as the reference time identifier. Based on the phase relationship between the pure back EMF voltage signal and the stator current sampling value, the synchronization error of the voltage and current sampling channels is calibrated, and the relative inherent delay time of the voltage sampling channel relative to the current sampling channel is recorded as the inherent delay time parameter. The steady-state AC angular frequency reference value, steady-state electrical angle reference value, reference time identifier, integral accumulation value, current component value, back EMF initial value vector and inherent delay time parameter are encapsulated into a multi-dimensional state parameter set.

4. The sensorless PFC start-up control method for synchronous generators based on multi-stage parameter inheritance according to claim 3, characterized in that, The process of writing the multi-dimensional state parameter set into the sliding mode observer and its associated phase-locked loop specifically includes: Start the state writing process of the sliding mode observer and its associated phase-locked loop, and simultaneously start the convergence timer to zero and begin timing; Read the current component values ​​and reference time identifier in the two-phase stationary coordinate system from the multi-dimensional state parameter set, reconstruct the current component values ​​from the reference time to the state writing time according to the steady-state AC angular frequency reference value, and then write them into the two-phase stationary coordinate system α-axis current observation state variables and β-axis current observation state variables inside the sliding mode observer respectively. The steady-state AC angular frequency reference value, steady-state electrical angle reference value, reference time identifier and integral accumulation value are read from the multi-dimensional state parameter set. The steady-state electrical angle reference value is extrapolated to the state writing time according to the steady-state AC angular frequency reference value to obtain the written electrical angle value. The steady-state AC angular frequency reference value, written electrical angle value and integral accumulation value are written into the electrical angular velocity, electrical angle and integral accumulation state variables of the matching phase-locked loop. The initial value vector of the back electromotive force and the reference time identifier are read from the multi-dimensional state parameter set. The vector is extrapolated from the reference time to the state writing time according to the steady-state AC angular frequency reference value. The back electromotive force observation output state register of the sliding mode observer is assigned as the initial value of the back electromotive force observation in the first control cycle after the sliding mode observer is started.

5. The sensorless PFC start-up control method for synchronous generators based on multi-stage parameter inheritance according to claim 4, characterized in that, The startup and operation, and independent monitoring of the convergence of the back electromotive force amplitude and phase, specifically include: After completing the full-state hot migration write, the sliding mode observer and its matching phase-locked loop are started. The inherent delay time parameters of the sampling channel are read from the multi-dimensional state parameter set. The timing delay compensation is performed on the input three-phase AC current signal. The back electromotive force observation value is reconstructed based on the compensated current signal. During the operation of the sliding mode observer and its matching phase-locked loop, the percentage of amplitude error between the observed back electromotive force and the initial back electromotive force vector is independently calculated as the amplitude convergence. The steady-state electrical angle reference value is extrapolated to the current control cycle according to the steady-state AC angular frequency reference value to obtain the extrapolated electrical angle reference value. The absolute phase difference between the output phase of the matching phase-locked loop and the extrapolated electrical angle reference value is calculated as the phase convergence.

6. The sensorless PFC start-up control method for synchronous generators based on multi-stage parameter inheritance according to claim 5, characterized in that, When the convergence rate reaches the threshold and the time difference between convergence completion is less than the synchronization tolerance, a hot migration verification pass flag is output and the residual phase tracking lag angle corrected by electrical angle extrapolation is recorded, specifically including: The amplitude convergence is compared with the preset amplitude convergence threshold, and the phase convergence is compared with the preset phase convergence threshold. It is determined whether the two are less than their respective preset thresholds for multiple consecutive control cycles. The moment when the amplitude convergence and phase convergence first meet the above continuous compliance conditions is recorded as the amplitude convergence completion time and the phase convergence completion time, respectively. The difference between the two convergence completion times is calculated. When the difference between the convergence completion time and the preset synchronization tolerance is less than the preset synchronization tolerance, the dual-state asynchronous convergence is determined to be complete, and a hot migration verification pass flag is output. The steady-state electrical angle reference value is extrapolated to the convergence completion time according to the steady-state AC angular frequency reference value to obtain the extrapolated electrical angle reference value. The difference between the extrapolated electrical angle reference value and the output phase at the convergence completion time of the matching phase-locked loop is calculated and stored as the residual phase tracking lag angle. When the difference is greater than or equal to the preset synchronization tolerance, the monitoring continues. If the convergence is not completed after the preset maximum convergence waiting time, the convergence is determined to be failed and the system is reset and re-enters the pre-observation window period.

7. The sensorless PFC start-up control method for synchronous generators based on multi-stage parameter inheritance according to claim 1, characterized in that, The calculation of the total phase offset compensation involves extrapolating the steady-state electrical angle reference value to a preset time to obtain the extrapolated electrical angle reference value, and then superimposing it on the total phase offset compensation to obtain the compensated electrical angle value. Specifically, this includes: Read the DC bus voltage value sampled by the DC bus voltage detection circuit, the target modulation ratio preset by the current loop, the on-state voltage drop parameter of the power switch under rated current, and the dead time parameter and switching frequency parameter preset by the microcontroller pulse width modulation module; calculate the equivalent phase offset angle and additional phase offset angle caused by the nonlinear characteristics of the power switch. The equivalent phase offset angle is superimposed with the additional phase offset angle to obtain the total phase offset compensation amount; the steady-state electrical angle reference value in the multi-dimensional state parameter set is extrapolated to the preset time according to the steady-state AC angular frequency reference value to obtain the extrapolated electrical angle reference value; the total phase offset compensation amount is superimposed on the extrapolated electrical angle reference value to obtain the compensated electrical angle value.

8. The sensorless PFC start-up control method for synchronous generators based on multi-stage parameter inheritance according to claim 7, characterized in that, The calculation of the nominal feedforward voltage value after coordinate system orthogonal compensation specifically includes: The compensated electrical angle value is used as the first Park transformation angle, and the residual phase tracking lag angle is subtracted from it to obtain the corrected Park transformation angle. The initial value vector of the back electromotive force in the multi-dimensional state parameter set is extrapolated and reconstructed to the preset time according to the steady-state AC angular frequency reference value. The corrected Park transformation angle is used to perform Park transformation on the reconstructed two-phase stationary coordinate system voltage components to obtain the corrected d-axis voltage component and q-axis voltage component. The corrected d-axis voltage components and q-axis voltage components are normalized to obtain the corrected nominal feedforward voltage values ​​of the inner current loop d-axis and q-axis, so that the projection direction of the nominal feedforward voltage values ​​in the dq coordinate system is aligned with the actual voltage vector direction, thus eliminating coordinate system projection deviation.

9. The sensorless PFC start-up control method for synchronous generators based on multi-stage parameter inheritance according to claim 8, characterized in that, The step of presetting the nominal feedforward voltage value into the current loop integration register, in response to the thermal migration verification unlocking pulse width modulation signal through the identifier, to achieve a shockless smooth start-up, specifically includes: Write the corrected nominal feedforward voltage value of the d-axis into the integral accumulator register of the d-axis current loop proportional-integral controller, and write the corrected nominal feedforward voltage value of the q-axis current loop proportional-integral controller into the integral accumulator register of the q-axis current loop as the initial integral value of the current loop in the first control cycle when the closed loop is entered. The initial values ​​of the d-axis and q-axis current reference values ​​are set to zero at the moment of cut-in. The target reference value of the d-axis current is obtained by limiting the output of the DC bus voltage outer loop regulator. The target reference value of the q-axis current is set according to the reactive current command corresponding to the target power factor. The increment of each control cycle is configured according to the preset total rise time, so that the current reference value increases from zero to its respective target reference value by the increment. In response to the thermal migration verification pass flag, the pulse width modulation output is unlocked and the current closed-loop control based on the matching phase-locked loop phase feedback is switched in, so that the current loop starts to adjust from the preset integral initial value in the first control cycle, and the current reference value rises smoothly from zero in preset increments per control cycle, so as to achieve a smooth start-up without impact.

10. A sensorless PFC start-up control system for synchronous generators based on multi-stage parameter inheritance, characterized in that, The method for implementing the sensorless PFC start-up control of a synchronous generator based on multi-stage parameter inheritance as described in any one of claims 1-9 includes: A synchronous generator whose stator winding inductance is configured as the boost inductance of a power factor correction converter; The power conversion circuit includes a three-phase bridge rectifier topology and omits the external power factor correction inductor; The voltage sampling circuit is used to sample the voltage signal between the midpoint of the three-phase bridge arm and the virtual midpoint of the DC bus during the pre-observation window period when all controllable switching devices of the power conversion circuit are in a blocked state. A current sampling circuit is used to sample the three-phase current signal of the stator winding during the pre-observation window period; The control unit is electrically connected to the voltage sampling circuit, the current sampling circuit, and the power conversion circuit. The control unit includes: The flux linkage decoupling pre-observation module is used to extract the fundamental component of the stator current signal, calculate the flux linkage coupling voltage component based on the stator winding inductance parameters and subtract it from the bridge arm midpoint voltage signal to obtain a pure back electromotive force voltage signal, and extract steady-state parameters based on it and encapsulate it into a multi-dimensional state parameter set. The full-state hot migration module is used to write the multi-dimensional state parameter set into the corresponding state register of the sliding mode observer and its matching phase-locked loop; The dual-state asynchronous convergence discrimination module is used to independently monitor the convergence degree of back electromotive force amplitude and phase convergence degree after the sliding mode observer and its matching phase-locked loop are started and running. When the convergence degree reaches the threshold and the difference between the convergence completion time is less than the synchronization tolerance, the module outputs a thermal migration verification pass mark and records the residual phase tracking lag angle corrected by electrical angle extrapolation. The phase offset compensation module is used to calculate the total phase offset compensation amount. It extrapolates the steady-state electrical angle reference value to a preset time according to the steady-state AC angular frequency reference value to obtain the extrapolated electrical angle reference value, and superimposes it on the extrapolated electrical angle reference value to obtain the compensated electrical angle value. The coordinate system orthogonal compensation feedforward module is used to extrapolate and reconstruct the initial value vector of the back electromotive force to a preset time based on the compensated electrical angle value and the residual phase tracking lag angle, and then calculate the nominal feedforward voltage value after coordinate system orthogonal compensation. The current soft-start control module is used to preset the nominal feedforward voltage value into the current loop integral register, and in response to the thermal migration verification, unlocks the pulse width modulation signal through the power conversion circuit to achieve a smooth start-up without impact.