An open-winding surface-mounted permanent magnet synchronous motor (PMSM) online data-driven harmonic injection ripple suppression method

Abstract

The application discloses an open-winding surface-mounted PMSM online data-driven harmonic injection ripple suppression method, belongs to the field of open-winding permanent magnet synchronous motor torque ripple suppression, and is applied to a common DC bus type open-winding surface-mounted permanent magnet synchronous motor driving system. In each sampling period, the method updates three recent sampling values of zero sequence current and zero sequence voltage, constructs a stator current prediction mathematical model containing the zero sequence current, calculates a zero sequence loop motor parameter related polynomial by using an online data-driven technology, obtains a harmonic current injection reference value capable of suppressing the zero sequence torque, and thus the suppression of torque ripple is realized. Meanwhile, the method effectively solves the problems of the decrease of zero sequence current prediction accuracy caused by parameter mismatch and the problem that the system state data-driven identification accuracy is proportional to the amount of online data used without increasing the hardware cost.

Description

Technical Field

[0001] This application belongs to the field of torque pulsation suppression of open-winding permanent magnet synchronous motors, and specifically relates to a method for suppressing pulsation by online data-driven harmonic injection in open-winding surface-mounted PMSMs. Background Technology

[0002] The electric drive system of new energy vehicles is developing towards high efficiency, high power density, high reliability and intelligent control. As the core power unit of the vehicle, the smoothness of operation, energy utilization and wide speed range working capability of the vehicle drive motor directly affect the performance of the vehicle. In existing applications, permanent magnet synchronous motors with rare earth permanent magnet excitation have become an important type of vehicle drive motor due to their high power density, high efficiency and good speed regulation performance.

[0003] With the increasing demands for high speed and high efficiency in automotive electric drive systems, traditional three-phase permanent magnet synchronous motors (PMSMs) are gradually revealing problems such as limited DC bus voltage utilization, increased torque ripple, decreased system efficiency, and demagnetization risks due to increased rotor temperature rise under conditions of field weakening speed extension, high-speed overmodulation, and high load. To address these issues, academia and industry have proposed a dual-inverter powered open-winding PMSM drive topology. This topology opens the stator winding neutral point, connecting each phase winding to two inverter arms, offering advantages such as high DC bus voltage utilization, wide speed extension range, flexible control strategies, and good system reliability. However, in a common DC bus power supply structure, a unique zero-sequence loop is formed between the dual inverters and the motor windings, containing electrical quantities such as zero-sequence voltage, zero-sequence current, and zero-sequence inductance.

[0004] In open-winding permanent magnet synchronous motors with a common DC bus, the combined effect of zero-sequence current and zero-sequence back electromotive force generates zero-sequence torque, resulting in a zero-sequence torque component superimposed on the motor's output torque in addition to the permanent magnet torque and reluctance torque. This zero-sequence torque typically manifests as torque harmonics, exacerbating torque pulsation at the output shaft and affecting torque control accuracy, system energy efficiency, smooth operation, reliability, and NVH performance. Under variable speed and variable torque operating conditions, the zero-sequence torque caused by the combined zero-sequence current and zero-sequence back electromotive force is difficult to suppress effectively, thus impacting the quality of output torque control. Summary of the Invention

[0005] To address the aforementioned issues, this application provides a torque ripple suppression method for open-winding surface-mounted PMSMs based on online data driving and current harmonic injection. This method is applied to a common DC bus type open-winding surface-mounted permanent magnet synchronous motor drive system and is executed cyclically within each sampling cycle of the motor drive control algorithm.

[0006] In the modeling and discretization process, the stator current differential equation of the motor is first established in a synchronous rotating coordinate system fixed to the rotor. The synchronous rotating coordinate system includes the direct axis, the quadrature axis, and the zero-sequence loop. Based on this, a zero-sequence current differential equation is established, which includes a zero-sequence back electromotive force term caused by the harmonics of the third-order and above permanent magnet flux linkages. The stator current differential equation is then discretized into a discrete expression suitable for calculation according to the sampling period. By unifying the modeling and discretization of the zero-sequence loop into the same computational framework, a consistent discrete computational basis is provided for the online calculation of the relevant polynomials of the motor parameters of the subsequent zero-sequence loop and the prediction of the zero-sequence current in the next sampling period.

[0007] The zero-sequence current and zero-sequence voltage calculated within each sampling period are registered online and written to and updated in the online registration database. The online registration database stores at least the zero-sequence current and zero-sequence voltage data for the current sampling period and the previous two sampling periods, used for subsequent online data-driven calculations of the zero-sequence loop motor parameter related polynomials. The online database will continuously register data calculated within the most recent three consecutive sampling periods, providing continuous time-series data without introducing an offline identification process, thus meeting the data retrieval requirements for online algorithm calculations and prediction calculations within each sampling period.

[0008] During the online calculation of the zero-sequence loop motor parameter correlation polynomial, the online data-driven calculation of the zero-sequence loop motor parameter correlation polynomial is performed based on the online registered database to obtain candidate update values ​​for the current sampling period. Simultaneously, it is determined whether to adopt the candidate update values ​​according to the pre-set update constraints to determine the value of the zero-sequence loop motor parameter correlation polynomial for the current sampling period. The update constraints are used to avoid calculation anomalies caused by excessively small or zero denominators during the online calculation process, thereby ensuring the stability of the zero-sequence loop motor parameter correlation polynomial value update and the system control convergence.

[0009] In the zero-sequence current prediction process, based on the relevant polynomial of the zero-sequence loop motor parameters determined in the current sampling period, the zero-sequence current in the next sampling period is predicted and calculated. By completing the prediction of the zero-sequence current in the next sampling period within the current sampling period, the subsequent harmonic injection quantity calculation can directly use the predicted quantity of the next sampling period, reducing the lag of the injected harmonic current on torque ripple suppression.

[0010] During torque ripple suppression, the quadrature-axis stator current harmonic injection amount for suppressing zero-sequence torque is calculated based on the zero-sequence current prediction value of the next sampling period. The quadrature-axis stator current harmonic injection amount is then superimposed on the quadrature-axis stator current reference value of the next sampling period. By combining the zero-sequence current prediction result with the quadrature-axis stator current reference value update process, the quadrature-axis stator current harmonic injection amount can be matched with the zero-sequence torque change of the next sampling period, thereby achieving real-time suppression of torque ripple in open-winding surface-mounted permanent magnet synchronous motors.

[0011] This invention further describes the model discretization, zero-sequence loop back EMF processing, online registered data organization, online calculation of zero-sequence loop motor parameter related polynomials, update constraint determination, zero-sequence current prediction, cross-axis stator current harmonic injection calculation, and control execution process, so that the online calculation process and the current control process within the sampling period are executed in a coordinated manner under the same calculation framework.

[0012] In the model discretization process, a first-order forward Euler discretization method is adopted to convert the stator current differential equation in the synchronous rotating coordinate system, which includes the direct axis, quadrature axis and zero-sequence loop, into a discrete expression executed by the digital controller according to the sampling period. After adopting the discrete expression consistent with the calculation process of the sampling period, the subsequent online calculation of the related polynomials of the zero-sequence loop motor parameters, zero-sequence current prediction and stator current prediction control can call the same model expression form, thereby reducing the inconsistency in calculation caused by model form switching.

[0013] In the zero-sequence loop back EMF processing, the zero-sequence loop back EMF term is set as the zero-sequence loop back EMF caused by the third harmonic of the rotor permanent magnet flux linkage and higher odd harmonics. In the online data-driven calculation of the zero-sequence loop motor parameter related polynomial, the zero-sequence loop back EMF term containing the contribution of each harmonic order is used as a complete whole in the calculation, instead of establishing independent online calculation paths according to harmonic order. The method of using the whole to participate in the calculation is beneficial to maintain a single calculation path within the sampling period and avoid the problems of increased calculation volume of single-step control algorithm and complicated parameter coupling processing caused by order-based processing.

[0014] During the online data storage process, the motor drive control algorithm stores sampled data, including stator voltage and stator current in the synchronous rotating coordinate system, in each sampling period; wherein, the online storage database used for the online data-driven calculation of the zero-sequence loop motor parameter correlation polynomial stores zero-sequence current data and zero-sequence voltage data, and includes at least the data of the current sampling period and the previous two sampling periods.

[0015] By employing a rolling register method with data from three consecutive sampling periods, continuous time-series data required for online computation can be provided without introducing an offline identification process.

[0016] In the online calculation of the zero-sequence loop motor parameter correlation polynomial, the zero-sequence current amplitude increment is calculated based on the discrete zero-sequence current expression of adjacent sampling periods, and the zero-sequence voltage amplitude change is calculated based on the zero-sequence voltage data of adjacent sampling periods. Then, candidate update values ​​of the zero-sequence loop motor parameter correlation polynomial are calculated based on the zero-sequence current amplitude increment and the zero-sequence voltage amplitude change. By using adjacent sampling period data to construct candidate update values, the value of the zero-sequence loop motor parameter correlation polynomial can be adjusted online according to the current operating state.

[0017] In the process of determining the value of the zero-sequence loop motor parameter related polynomial, based on the historical zero-sequence stator voltage data in the online register database, the pre-set update constraints are judged. The update constraints are used to avoid the denominator being too small or zero during the online data-driven calculation. When the update constraints are met, the candidate update value is used to update the zero-sequence loop motor parameter related polynomial; when the update constraints are not met, the historical register value of the zero-sequence loop motor parameter related polynomial is maintained, and the historical register value is called in the calculation process of the current sampling period. The update constraints are established based on the switching state combination of the dual inverter drive topology of the common DC bus open winding permanent magnet synchronous motor and its corresponding stator voltage vector amplitude. Through the above value determination process, the impact of online calculation anomalies on subsequent zero-sequence current prediction and current control can be reduced.

[0018] In the zero-sequence current prediction process, based on the discretized zero-sequence current expression and the related polynomial of the zero-sequence loop motor parameters determined in the current sampling period, the zero-sequence current of the next sampling period is predicted and calculated. Under the condition of small changes in the time interval between adjacent sampling periods, the changes in rotor electrical angle, rotor electrical angular velocity, and zero-sequence current amplitude are treated as approximately constant. After the zero-sequence current prediction of the next sampling period is completed in the current sampling period, it can provide direct input for the subsequent calculation of harmonic injection of quadrature-axis stator current, reducing the impact of control calculation lag on the suppression effect.

[0019] In the calculation of the quadrature-axis stator current harmonic injection, the zero-sequence torque is calculated based on the predicted value of the zero-sequence current and the zero-sequence loop back electromotive force of the next sampling period. The quadrature-axis stator current harmonic injection is then solved according to the zero-sequence torque suppression condition. In the single-step calculation of the quadrature-axis stator current harmonic injection, the zero-sequence loop back electromotive force of the next sampling period is adopted by the zero-sequence loop back electromotive force component generated by the third-order permanent magnet flux linkage harmonic. The zero-sequence current prediction result is directly used to solve the quadrature-axis stator current harmonic injection, so that the quadrature-axis stator current harmonic injection corresponds to the change of the zero-sequence torque in the next sampling period.

[0020] During the control execution process, the reference value of the quadrature-axis stator current in the next sampling period after superimposing the quadrature-axis stator current harmonic injection amount is input into the stator current prediction control loop to execute the current control of the common DC bus type open winding surface-mounted permanent magnet synchronous motor. In this way, the online data-driven calculation, zero-sequence current prediction and quadrature-axis stator current harmonic injection amount update process can be connected into a continuous control path to achieve torque ripple suppression.

[0021] Compared with the prior art, this application has the following advantages: Compared with existing technologies, this method first establishes a stator current differential equation that includes the back electromotive force term of the zero-sequence loop, and combines an online data-driven method with a rolling registration mechanism of the online registered database. This enables the zero-sequence current prediction equation to be updated in real time during each sampling period, thereby avoiding the problem that the actual values ​​of the zero-sequence inductance and permanent magnet flux harmonics deviate from their rated values, causing distortion of the zero-sequence current state transition equation and leading to a decrease in the accuracy of zero-sequence current prediction.

[0022] Compared with the prior art, this application further improves upon the following: by treating the zero-sequence loop back EMF, which includes contributions from various relevant orders, as a complete whole, it participates in the online data-driven calculation of the zero-sequence loop motor parameter related polynomials. Furthermore, it establishes constraints based on the difference in stator voltage vector amplitude changes within two recent adjacent sampling periods. This allows candidate update values ​​to be updated when the constraints are met and to be recalled when the constraints are not met. In this way, the adverse effects of unmodeled high-order permanent magnet harmonics on the online identification / observation accuracy of zero-sequence back EMF are reduced under variable speed and variable torque conditions. This also optimizes the current harmonic injection method's ability to suppress zero-sequence torque.

[0023] Compared with existing technologies, this invention constructs a data-driven mathematical model based on mathematical analysis and proposes an online data-driven method that can identify the polynomials related to zero-sequence circuit motor parameters using only historical sampling data of four system states—stator current, stator voltage, rotor angular velocity, and rotor angle—from the most recent three sampling periods of the motor drive system. This method reduces the hardware performance requirements of the data-driven algorithm, such as high-performance main control chips and large-capacity random access memory, and expands the applicability of online data-driven methods to high-switching-frequency motor drive engineering scenarios.

[0024] Other features and advantages of this application will be set forth in the description which follows, and will be apparent in part from the description, or may be learned by practicing the application. The objectives and other advantages of this application may be realized and obtained by means of the structures pointed out in the description, claims and drawings. Attached Figure Description

[0025] To more clearly illustrate the technical solutions in the embodiments of this application or the prior art, the drawings used in the description of the embodiments or the prior art will be briefly introduced below. Obviously, the drawings described below are some embodiments of this application. For those skilled in the art, other drawings can be obtained based on these drawings without creative effort.

[0026] Figure 1 This is a flowchart of the open-winding surface-mount PMSM online data-driven harmonic injection suppression method of the present invention; Figure 2 This is a topology diagram of the common DC bus type open winding three-phase permanent magnet synchronous motor drive system involved in this invention; Figure 3 The equivalent circuit diagram of the zero-sequence circuit of the common bus type open winding permanent magnet synchronous motor mentioned in this invention; Figure 4 This is a diagram illustrating the online data-driven calculation process for the polynomial related to the parameters of the zero-sequence loop motor in the method provided by this invention. Figure 5 Simulation results of rotor output torque for a common bus open winding surface-mounted permanent magnet synchronous motor under two conditions: without zero-sequence torque suppression and with the zero-sequence torque suppression method proposed in this invention. Figure 6 The simulation results of the stator phase current of the surface-mounted permanent magnet synchronous motor with open windings on the common bus are shown in the figure, under two conditions: without using the zero-sequence torque suppression method and with using the zero-sequence torque suppression method proposed in this invention. Detailed Implementation

[0027] To make the objectives, technical solutions, and advantages of the embodiments of this application clearer, the technical solutions of the embodiments of this application will be clearly and completely described below with reference to the accompanying drawings. Obviously, the described embodiments are only some embodiments of this application, not all embodiments. Based on the embodiments of this application, all other embodiments obtained by those skilled in the art without creative effort are within the scope of protection of this application.

[0028] like Figures 1 to 3As shown, this embodiment of the invention provides an online data-driven harmonic injection suppression method for open-winding surface-mounted PMSMs. In this embodiment, the torque ripple suppression method is applied to a common DC bus type open-winding surface-mounted permanent magnet synchronous motor drive system and is executed cyclically within each sampling cycle of the motor drive control algorithm. Within the current sampling cycle, the method completes model discretization, online register data update, determination of the polynomial values ​​related to the zero-sequence loop motor parameters, and prediction of the zero-sequence current for the next sampling cycle. Based on the prediction results, the harmonic injection amount of the quadrature-axis stator current for the next sampling cycle is generated to update the reference value of the quadrature-axis stator current for the next sampling cycle.

[0029] Specifically, firstly, a differential equation for the stator current of the motor is established in a synchronous rotating coordinate system fixed to the rotor. The synchronous rotating coordinate system includes the direct axis, the quadrature axis, and the zero-sequence loop. Based on the differential equation for the stator current, a differential equation for the zero-sequence current, which includes a zero-sequence back electromotive force term caused by the harmonics of the third-order or higher permanent magnet flux linkage, is established. Discretization is then used to obtain a discrete expression suitable for execution by the digital controller according to the sampling period. By unifying the subsequent online calculation and prediction calculation on the discrete expression, it can be ensured that the same model form is called in each calculation link within the sampling period.

[0030] Within each sampling period, the control algorithm calculates the zero-sequence current and zero-sequence voltage, and writes and updates the zero-sequence current and zero-sequence voltage into the online register database. The online register database stores the zero-sequence current data and zero-sequence voltage data of the current sampling period and the previous two sampling periods, so as to be called by the online data-driven calculation within the current sampling period. Through the rolling register of data from three consecutive sampling periods, continuous time-sequence data can be provided for the online calculation of the polynomial related to the zero-sequence loop motor parameters.

[0031] like Figure 4 As shown, online data-driven calculations are performed on the zero-sequence loop motor parameter correlation polynomial based on the online registered database to obtain candidate update values ​​for the zero-sequence loop motor parameter correlation polynomial. The control algorithm determines the candidate update values ​​according to pre-set update constraints. These constraints are used to avoid the denominator being too small or zero during the online data-driven calculation. When the update constraints are met, the candidate update value is used as the value of the zero-sequence loop motor parameter correlation polynomial in the current sampling period. When the update constraints are not met, the historical registered value of the zero-sequence loop motor parameter correlation polynomial is maintained, and the historical registered value is called in subsequent calculations of the current sampling period. Through the above determination process, the impact of abnormal calculation results on subsequent prediction calculations can be reduced.

[0032] After determining the value of the relevant polynomial of the zero-sequence circuit motor parameters in the current sampling period, the zero-sequence current in the next sampling period is predicted and calculated based on the value to obtain the predicted value of the zero-sequence current in the next sampling period. The predicted value of the zero-sequence current is used as the input to the calculation of the harmonic injection amount of the quadrature axis stator current in the current sampling period.

[0033] Finally, the cross-axis stator current harmonic injection amount for suppressing zero-sequence torque is calculated based on the zero-sequence current prediction value of the next sampling period, and the cross-axis stator current harmonic injection amount is superimposed on the cross-axis stator current reference value of the next sampling period. The updated cross-axis stator current reference value of the next sampling period is input into the stator current prediction control loop to execute motor current control, thereby realizing online suppression of torque ripple.

[0034] In one embodiment, the present invention is applied to a common DC bus type open-winding surface-mount permanent magnet synchronous motor drive system, and is executed cyclically within each sampling cycle of the motor drive control algorithm. For ease of explanation, the sampling cycle in which the control calculation is currently being performed is denoted as the current sampling cycle, and the previous sampling cycle and the two previous sampling cycles are denoted as the previous sampling cycle and the two previous sampling cycles, respectively. When calculating adjacent sampling cycles in the following text, the above definitions shall be used to avoid inconsistencies in the understanding of time index.

[0035] In this embodiment, a stator current differential equation model is first established in a synchronous rotating coordinate system fixed to the rotor. The synchronous rotating coordinate system includes a direct axis, a quadrature axis, and a zero-sequence loop. This model includes the dynamic relationship between the direct-axis current, the quadrature-axis current, and the zero-sequence loop current. In the zero-sequence loop current equation, a zero-sequence loop back electromotive force term caused by the third harmonic of the rotor permanent magnet flux linkage and higher odd harmonics is introduced. The purpose of this setting is to ensure that the model on which the zero-sequence loop current prediction is based not only covers the influence of the third-order permanent magnet flux linkage harmonics, but also incorporates the contribution of the zero-sequence loop back electromotive force caused by higher-order harmonics at the model level, thereby reducing the impact of unmodeled higher-order harmonics on the accuracy of subsequent online data-driven calculations.

[0036] The components of the motor stator inductance in the dq rotating coordinate system are equal, that is... ; Based on this, we can establish the following form of differential equation: (1) In the formula, and These represent the stator voltage and stator current along the d-axis, q-axis, and 0-axis in the dq0 coordinate system, respectively; i d i q i0 and i0 represent the stator currents along the d-axis, q-axis, and 0-axis in the dq0 synchronous rotating coordinate system, respectively; u a u b uc These represent the stator phase voltages of phases A, B, and C in the three-phase stationary coordinate system (abc); i a i b i c Let A, B, and C represent the stator phase currents in the three-phase stationary coordinate system (a, b, c), respectively. , ; For rotor flux linkage; The rotor flux linkage is a third harmonic; Stator resistance; For stator inductance; The zero-sequence inductance is the zero-sequence circuit. The electric angular velocity of the rotor; The electrical angle of the rotor; It is a time variable; The zero-sequence loop back electromotive force term caused by the third-order and above harmonics of the rotor permanent magnet flux linkage is shown in Equation (2). in, and Let represent the harmonic flux linkage of the (2n+1)th permanent magnet and the resulting rotating back electromotive force term, respectively.

[0037] Since the third-order permanent magnet flux linkage is the main order harmonic of the rotor rotation back electromotive force term in the zero-sequence loop, the complete expression of the zero-sequence loop back electromotive force term can be rewritten as shown in equation (3). in, This represents the time-varying rotating back EMF term caused by the mismatch of harmonic parameters of the third-order rotor permanent magnet flux linkage; This represents the zero-sequence loop voltage disturbance term caused by interference factors such as high-order magnetic flux harmonics of the rotor permanent magnet.

[0038] Based on the above modeling, the stator current differential equation is discretized. Specifically, a first-order forward Euler discretization method is used to convert the continuous-time form of the stator current differential equation into a discrete expression executed by the digital controller according to the sampling period. This discrete expression serves as the unified computational basis for subsequent online registered database calls, online data-driven calculations of zero-sequence loop motor parameter correlation polynomials, and zero-sequence current prediction in the next sampling period, ensuring that each calculation stage within the sampling period adopts a consistent model expression.

[0039] Because the amplitude of the rated value of the magnetic flux harmonics of permanent magnets of order three and above is relatively small, the amplitude of the parameter mismatch caused by temperature change is also small. Therefore, in the expression for the zero-sequence loop rotating back electromotive force, this invention does not describe the time-varying rotating back electromotive force term caused by the mismatch of harmonic parameters of the flux linkage of permanent magnets of order three or higher individually, but instead incorporates them all into the total disturbance term. middle; Similarly, in the stator current harmonic injection calculation process involved in the subsequent parts of this invention, only the flux linkage amplitude of the third-order permanent magnet is used to predict the q-axis injection current required to suppress zero-sequence torque in the next sampling period. However, it should be emphasized that the online data-driven calculation method for identifying the relevant polynomials of zero-sequence loop motor parameters proposed in this invention is for online calculation of the back electromotive force term of the zero-sequence loop caused by all orders of permanent magnet flux linkage; Therefore, in the data-driven algorithm proposed in this invention, the order of the zero-sequence back electromotive force term is not specifically distinguished, but is considered as a whole as a complete zero-sequence back electromotive force.

[0040] (2) (3) Furthermore, by discretizing equation (1) using the first-order forward Euler formula, we obtain: (4) In the formula, and These are the stator current sample value and stator voltage sample value in the (k)th sampling period under the dq0 coordinate system, respectively, and so on; The sampling period of the system; Let be the zero-sequence back electromotive force in the zero-sequence loop during the (k)th sampling period.

[0041] Within each sampling period, the drive controller obtains the stator voltage and stator current in the synchronous rotating coordinate system based on the motor sampling signal and coordinate transformation results, and completes the register update. To support subsequent online data-driven calculations, an online register database is set up in the controller. The online register database at least stores the zero-sequence loop current data and zero-sequence loop voltage data for the current sampling period, the previous sampling period, and the two previous sampling periods. When complete state recording is required, the stator voltage and stator current data for both the direct axis and the quadrature axis can also be stored simultaneously to ensure consistent data retrieval within the controller. The online register database can be implemented using a fixed-address register area in the controller's random access memory and is updated in a rolling manner according to the sampling period. Specifically, after each sampling period calculation is completed, the original current sampling period data is transferred to the previous sampling period data, the original previous sampling period data is transferred to the two previous sampling period data, and the new current sampling period data is written. Through the above settings, continuous time-series data can be provided for online data-driven calculations without introducing an offline identification process, while balancing register space and single-cycle calculation burden.

[0042] After obtaining the aforementioned online registered data, online data-driven calculations are performed on the polynomials related to the zero-sequence loop motor parameters. The zero-sequence loop motor parameter related polynomials referred to here are not independent online identifications of the harmonic parameters of the zero-sequence inductance and permanent magnet flux linkage, but rather polynomial quantities formed by combining the zero-sequence inductance related terms and the zero-sequence loop back electromotive force related terms based on the derivation relationship of the zero-sequence loop discretization equation. These polynomial quantities are used for subsequent zero-sequence current prediction and are a comprehensive representation of the zero-sequence loop motor parameters and the zero-sequence loop back electromotive force. The zero-sequence inductance related term refers to the equivalent term jointly characterized by the zero-sequence inductance L0, the sampling period, and the relationship between the zero-sequence current changes in adjacent sampling periods in the zero-sequence loop discretization equation. It reflects the inductive constraint effect of the zero-sequence loop on the zero-sequence current change. In this embodiment, the zero-sequence inductance related term is not identified online as an independent physical parameter, but rather as a component of the zero-sequence loop motor parameter related polynomial, and is calculated online along with the zero-sequence loop back electromotive force related term.

[0043] The zero-sequence current amplitude increments in the (k)th and (k+1)th sampling periods are calculated as follows: (5) Subtracting the two equations shown in equation (5), we get: (6) Considering that the change in zero-sequence current amplitude TermB between two adjacent sampling periods is numerically much smaller than the change in zero-sequence voltage amplitude TermA between two adjacent sampling periods, and furthermore, since the rotor electrical angle, electrical angular velocity, and third-order permanent magnet flux linkage parameters are approximately equal in numerical value between two adjacent sampling periods, i.e. , , This holds true within adjacent sampling periods, therefore the change in zero-sequence back electromotive force TermC is numerically approximately equal to zero. Thus, equation (6) can be approximately simplified to: (7) By polynomial decomposition and merging of equation (7), we can obtain: (8) Since the calculation process shown in equation (8) may result in divergent calculations or even be impossible to calculate due to an excessively small denominator or a denominator that is approximately equal to zero, it is necessary to set corresponding constraints here to ensure that the calculation results shown in equation (8) are accurate. The computational update process is valid. This is represented as a zero-sequence inductance related term.

[0044] In the online data-driven calculation process, the zero-sequence current amplitude increment is first calculated based on the discrete zero-sequence loop current expression of adjacent sampling periods, and the zero-sequence voltage amplitude change is calculated based on the zero-sequence loop voltage data of adjacent sampling periods. Subsequently, based on the zero-sequence current amplitude increment and the zero-sequence voltage amplitude change, candidate update values ​​of the zero-sequence loop motor parameter related polynomials are calculated. To avoid ambiguity, in this embodiment, the candidate update value refers to the temporary calculated value calculated based on recent data in the online registered database within the current sampling period, which has not yet passed the update constraint condition judgment.

[0045] To ensure the numerical stability of online data-driven calculations, this embodiment sets preset update constraints. These preset update constraints are used to determine whether candidate update values ​​can be used as valid values ​​for the current sampling period. Their purpose is to avoid numerical divergence or incalculability caused by excessively small or zero denominators in the calculation formula. The update constraints are established based on the switching state combinations of the dual-inverter drive topology of the common DC bus open-winding permanent magnet synchronous motor and its corresponding stator voltage vector amplitude.

[0046] Within the current sampling period, the controller calls historical zero-sequence circuit stator voltage data from the online registered database to perform constraint determination. When the update constraint condition is met, the candidate update value is used to update the zero-sequence circuit motor parameter related polynomial, and the updated value is used as the valid value for the current sampling period in subsequent calculations. When the update constraint condition is not met, only the update of the zero-sequence circuit motor parameter related polynomial is frozen, its historical registered value is maintained, and this historical registered value is called in subsequent calculations of the current sampling period to ensure the continuity of the control flow without interrupting the control calculation of the current sampling period.

[0047] Considering the 64 switching state combinations of the dual-inverter drive topology of the open-loop permanent magnet synchronous motor drive system and the corresponding stator voltage vector amplitude generated in the dq0 synchronous rotating coordinate system, the following equation (8) is executed. The constraints that must be satisfied during the computation and update process are: (9) That is, the algorithm proposed in this invention will use historical zero-sequence stator voltage data stored in the online database to judge the constraint conditions shown in equation (9) during the calculation process of each sampling period. If the constraint conditions are satisfied, then the algorithm shown in equation (8) will be executed. Update the calculation steps; conversely, if the conditions are not met, then all involved calculations within this sampling period will be updated. The calculation process uses data stored in an online database. Registered value.

[0048] At this point, the zero-sequence inductance related terms... Discretization can then be performed using the online data-driven computation method shown in equation (8).

[0049] After determining the relevant polynomials of the zero-sequence loop motor parameters valid for the current sampling period, the zero-sequence current for the next sampling period is predicted and calculated based on the discretized zero-sequence loop current expression. To balance engineering implementation and single-cycle computation, numerical equivalence processing is applied to state variables with small changes within a prediction step of one sampling period.

[0050] The numerical equivalence processing refers to treating the changes in rotor electrical angle, rotor electrical angular velocity, and zero-sequence current amplitude as approximately unchanged in adjacent sampling periods, under the condition that the sampling period is much smaller than the time scale of the motor mechanical time constant and the electrical state changes significantly, and using the corresponding values ​​of the current sampling period to replace the corresponding values ​​in the prediction calculation of the next sampling period.

[0051] The above processing can reduce computational complexity while ensuring that the prediction accuracy meets control requirements.

[0052] The corresponding derivation relationships and the expression for predicting the zero-sequence current in the next sampling period are as follows: (10) Similarly, considering that the changes in rotor electrical angle, rotor electrical angular velocity, and zero-sequence current amplitude are all small in two adjacent sampling periods, the values ​​of the above state variables in the (k)th and (k+1)th adjacent sampling periods can be equivalent, that is: (11) Thus, the motor parameter correlation polynomial in the zero-sequence loop within the (k)th adjacent sampling period can be discretized in the DSP chip using the online data-driven calculation method shown in Equation (11).

[0053] It is particularly important to emphasize that the online data-driven calculation method used here to identify the polynomials related to the zero-sequence loop motor parameters is used to perform online calculations of the back electromotive force term of the zero-sequence loop caused by all orders of the permanent magnet flux linkage.

[0054] That is, the order of the zero-sequence back electromotive force term is not specifically distinguished based on the order of the permanent magnet flux linkage, but is considered and calculated as a complete whole including all orders of zero-sequence back electromotive force.

[0055] Therefore, the predicted zero-sequence current in the zero-sequence loop during the (k+1)th adjacent sampling period can be obtained by the following formula: (12) It should be noted that in the online data-driven calculation stage of the zero-sequence loop motor parameter related polynomial, the zero-sequence loop back EMF is calculated as a complete whole including the contributions of each relevant order of the permanent magnet flux linkage, without being identified separately according to the harmonic order; while in the single-step solution stage of the quadrature axis stator current harmonic injection, considering the computational constraints within a single sampling period, the zero-sequence loop back EMF component generated by the third-order permanent magnet flux linkage harmonic can be used to participate in the solution of the quadrature axis stator current harmonic injection in the next sampling period.

[0056] The two processing methods mentioned above correspond to the comprehensive representation requirements of online data-driven computation and the real-time computation requirements of single-cycle control solution, respectively. The two methods complement each other technically and do not conflict.

[0057] After obtaining the predicted value of the zero-sequence current in the next sampling period, the zero-sequence torque related quantity is determined based on the predicted value of the zero-sequence current in the next sampling period and the back electromotive force of the zero-sequence circuit in the next sampling period. The harmonic injection quantity of the quadrature axis stator current in the next sampling period is then solved according to the zero-sequence torque suppression condition.

[0058] The zero-sequence torque correlation quantity referred to here refers to the part of the rotor output torque expression that is formed by the zero-sequence current and the zero-sequence circuit back electromotive force and causes torque pulsation; the zero-sequence torque suppression condition refers to the compensation relationship used to determine the harmonic injection quantity of the quadrature shaft stator current, so that the torque compensation term generated by the injected current and the zero-sequence torque correlation quantity satisfy the suppression relationship within the target sampling period.

[0059] The rotor output torque equation of a common-bus surface-mounted open-winding permanent magnet synchronous motor that takes into account stator current harmonic injection can be expressed as formula (13): (13) in, , , , These represent the q-axis reference current, the harmonic current to be injected into the q-axis, the zero-sequence back electromotive force, and the predicted value of the zero-sequence current, respectively, during the (k+1)th sampling period. It should be emphasized that, considering the computational complexity of the stator current harmonic injection algorithm, only the zero-sequence back electromotive force generated by the third-order permanent magnet flux linkage is considered here, i.e. Therefore, the zero-sequence torque generated by the zero-sequence back electromotive force and the zero-sequence current in the (k+1)th sampling period is: (14) Considering that the rotor electrical angle and electrical angular velocity are approximately equal in two adjacent sampling periods, that is ,

[0060] The approximate processing relationship between adjacent sampling periods is shown in formula (15). (15) The zero-sequence torque suppression condition is shown in equation (16). According to the principle of the stator current harmonic injection method, the amount of stator current harmonic injection used to suppress torque pulsation caused by zero-sequence torque in the (k+1)th sampling period must meet the following condition: (16) Therefore, the amount of q-axis stator current harmonic injection used to suppress zero-sequence torque in the (k+1)th sampling period is: (17) After obtaining the harmonic injection amount of the quadrature-axis stator current in the next sampling period, it is superimposed on the reference value of the quadrature-axis stator current in the next sampling period to obtain the updated reference value of the quadrature-axis stator current.

[0061] The updated quadrature-axis stator current reference value is input into the stator current prediction control loop, and combined with other axis current reference values ​​and the current state quantity, the current control of the common DC bus type open winding surface-mounted permanent magnet synchronous motor is executed to output the dual inverter switching state control quantity or equivalent modulation control quantity.

[0062] Then, the q-axis stator current reference value in the (k+1)th sampling period is updated as follows: (18) Through the above processing, torque ripple caused by zero-sequence torque can be suppressed online without adding additional hardware suppression components.

[0063] Thus, under stator current predictive control, the above-mentioned online data driving method and stator current harmonic injection method can effectively suppress torque pulsation caused by zero-sequence torque in a common bus open winding permanent magnet synchronous motor.

[0064] To verify the effectiveness of the online data-driven zero-sequence torque suppression method proposed in this invention, a surface-mounted permanent magnet synchronous motor with common bus open windings and motor parameters as shown in Table 1 was used as the controlled object. Simulation verification of the proposed method was performed, and the results are as follows: Figure 5 , Figure 6 As shown.

[0065] Table 1. Numerical values ​​of relevant parameters for open-winding surface-mounted permanent magnet synchronous motors in simulation verification.

[0066] Figure 5The figure shows the rotor output torque simulation results under two control strategies: one without the zero-sequence torque suppression strategy and the other with the zero-sequence torque suppression strategy proposed in this invention. It can be seen that the method proposed in this invention can provide a significant torque ripple suppression effect under operating conditions with a large duty cycle.

[0067] Figure 6 The simulation results of stator phase current under two control strategies are shown. It can be seen that the stator current harmonic injection method used in this invention achieves the control effect of suppressing the above-mentioned torque ripple only under the premise of finitely increasing the phase current amplitude. Although this application has been described in detail with reference to the foregoing embodiments, those skilled in the art should understand that modifications can still be made to the technical solutions described in the foregoing embodiments, or equivalent substitutions can be made to some of the technical features; and these modifications or substitutions do not cause the essence of the corresponding technical solutions to deviate from the spirit and scope of the technical solutions of the embodiments of this application.

Claims

1. A method for suppressing pulsation through online data-driven harmonic injection in an open-winding surface-mount PMSM, characterized in that: The method, applied to a common DC bus type open-winding surface-mounted permanent magnet synchronous motor drive system and executed cyclically within the sampling period of the motor drive control algorithm, includes: In the synchronous rotating coordinate system dq0 which is fixed to the rotor, the stator current differential equation is constructed, in which a zero-sequence current differential equation is established that includes a zero-sequence back electromotive force term caused by the harmonic flux of the third-order and above permanent magnets. The stator current differential equation is discretized to form a discrete expression that is calculated step by step by the digital controller according to the sampling period. Based on the discrete expression, the zero-sequence current and zero-sequence voltage of the current sampling period are calculated in each sampling period, and the zero-sequence current and zero-sequence voltage are updated and registered to construct an online registered database containing data of the current sampling period and the previous two sampling periods. Based on the online registered database, online data-driven calculation is performed on the zero-sequence loop motor parameter related polynomial to obtain candidate update values, and the value of the zero-sequence loop motor parameter related polynomial in the current sampling period is determined according to the preset update constraint conditions. The preset update constraint conditions are used to avoid the denominator being too small or zero in the online data-driven calculation. Based on the zero-sequence loop motor parameter related polynomial determined in the current sampling period, the zero-sequence current in the next sampling period is predicted and calculated. The q-axis stator current harmonic injection amount used to suppress zero-sequence torque is calculated based on the zero-sequence current prediction value of the next sampling period, and the q-axis stator current harmonic injection amount is superimposed on the q-axis stator current reference value of the next sampling period.

2. The method for suppressing pulsation by harmonic injection in an open-winding surface-mount PMSM according to claim 1, characterized in that: The stator current differential equation is discretized by using a first-order forward Euler discretization method, which converts the stator current differential equation in the dq0 synchronous rotating coordinate system into a discrete expression calculated by the digital controller according to the sampling period.

3. The method for suppressing pulsation by harmonic injection in an open-winding surface-mount PMSM according to claim 1, characterized in that: The zero-sequence back EMF term represents the zero-sequence loop back EMF caused by the third harmonic and higher odd harmonics of the rotor permanent magnet flux linkage. In the online data-driven calculation of the zero-sequence loop motor parameter related polynomial, the zero-sequence back EMF term containing contributions of each order is included as a complete whole in the calculation.

4. The method for suppressing pulsation by harmonic injection in an open-winding surface-mount PMSM according to claim 1, characterized in that: The data stored in each sampling period includes the stator voltage and stator current in the dq0 synchronous rotating coordinate system. The data used for online data-driven calculation in the online registered database includes the zero-sequence current data and zero-sequence voltage data of the current sampling period and the previous two sampling periods.

5. The method for suppressing pulsation by harmonic injection in an open-winding surface-mount PMSM according to claim 1, characterized in that: The online data-driven calculation of the polynomial related to the zero-sequence loop motor parameters includes: The corresponding zero-sequence current amplitude increment is calculated based on the discrete zero-sequence current expression for the k-th and (k+1)-th sampling periods. The change in zero-sequence voltage amplitude is calculated based on zero-sequence voltage data from adjacent sampling periods. The candidate update value of the zero-sequence loop motor parameter correlation polynomial is calculated based on the zero-sequence current amplitude increment and the zero-sequence voltage amplitude change.

6. The method for suppressing pulsation by harmonic injection in an open-winding surface-mount PMSM according to claim 1, characterized in that: The step of determining the value of the zero-sequence loop motor parameter correlation polynomial in the current sampling period according to the preset update constraint conditions includes: Based on historical zero-sequence stator voltage data in the online registered database, the update constraints used to avoid the denominator being too small or zero are determined. When the update constraint condition is met, the candidate update value is used to update the zero-sequence loop motor parameter related polynomial; If the update constraint is not met, the historical register value of the zero-sequence loop motor parameter related polynomial is maintained, and the historical register value is called in the calculation of the current sampling period. The update constraint is established based on the switching state combination of the common DC bus open winding permanent magnet synchronous motor dual inverter drive topology and its corresponding stator voltage vector magnitude.

7. The method for suppressing pulsation by harmonic injection in an open-winding surface-mount PMSM according to claim 1, characterized in that: When predicting the zero-sequence current for the next sampling period, the calculation is performed based on the discretized zero-sequence current expression and the related polynomials of the zero-sequence loop motor parameters determined in the current sampling period. Numerical equivalent processing is also performed on the changes in rotor electrical angle, rotor electrical angular velocity, and zero-sequence current amplitude in adjacent sampling periods.

8. The method for suppressing pulsation by harmonic injection in an open-winding surface-mount PMSM according to claim 1, characterized in that: The zero-sequence loop motor parameter correlation polynomial is a combination polynomial characterizing the zero-sequence inductance correlation term and the zero-sequence back electromotive force correlation term in the zero-sequence loop. The zero-sequence back electromotive force correlation term corresponds to the overall zero-sequence back electromotive force quantity, which includes the contribution of each relevant order of the permanent magnet flux linkage.

9. The method for suppressing pulsation by harmonic injection in an open-winding surface-mount PMSM according to claim 1, characterized in that: The calculation of the q-axis stator current harmonic injection based on the zero-sequence current prediction value of the next sampling period includes: The zero-sequence torque related quantities are determined based on the predicted value of the zero-sequence current in the next sampling period and the zero-sequence back electromotive force in the next sampling period. The q-axis stator current harmonic injection amount for the next sampling period is determined based on the zero-sequence torque suppression condition. In the single-step calculation of the q-axis stator current harmonic injection, the zero-sequence back electromotive force of the next sampling period adopts the zero-sequence back electromotive force component generated by the third-order permanent magnet flux linkage harmonic.

10. The method for suppressing pulsation by harmonic injection in an open-winding surface-mount PMSM according to claim 1, characterized in that: The reference value of the q-axis stator current in the next sampling period, after superimposing the q-axis stator current harmonic injection amount, is input into the stator current prediction and control loop to perform current control of the common DC bus type open winding surface-mounted permanent magnet synchronous motor.

Images