Signal-injection-free dual three-phase permanent magnet synchronous motor all-parameter online identification method and device

Abstract

The invention discloses a signal-injection-free dual three-phase permanent magnet synchronous motor full-parameter online identification method, which comprises the following steps: based on a six-phase topological structure of a dual three-phase permanent magnet synchronous motor driving system, deducing an equivalent inductance decoupling parameter model under a dual alpha-beta frame according to a basic mathematical model of the dual three-phase permanent magnet synchronous motor under the dual alpha-beta frame; the model serves as a full-parameter online identification initial model based on a double alpha-beta framework; separating current differential terms in a steady state based on synchronous rotation coordinate transformation, and further reconstructing the all-parameter online identification initial model to obtain an all-parameter online identification reconstruction model; and constructing an all-parameter online identification layered framework, deriving an all-parameter online identification model based on an RLS algorithm by using an all-parameter online identification reconstruction model, realizing equivalent transformation of identification parameters by using a mapping mechanism of each parameter between a double alpha-beta framework and a VSD framework, and finally realizing all-parameter online identification without signal injection. According to the method, extra copper loss, additional noise and torque pulsation cannot be generated in the full-parameter online identification period, and the identification error and the fluctuation amplitude can be effectively reduced.

Description

Technical Field

[0001] This invention relates to a method and device for online identification of all parameters of a three-phase permanent magnet synchronous motor, and particularly to a method and device for online identification of all parameters of a dual three-phase permanent magnet synchronous motor without signal injection. Background Technology

[0002] Currently, dual-three-phase permanent magnet synchronous motors (DT-PMSMs) are widely used in aerospace propulsion systems, marine electric propulsion, and underwater equipment due to their high power density, low torque ripple, and strong fault tolerance. However, their key electrical parameters, such as inductance, stator resistance, and permanent magnet flux linkage, exhibit significant time-varying characteristics during operation. Their values ​​drift due to the combined effects of current and temperature rise, directly leading to reduced control accuracy and dynamic response instability. With the increasing demands for control accuracy and operational reliability in drive systems from industry, online parameter identification technology has become not only an effective method for optimizing control performance but also provides crucial criteria for winding overheat protection and permanent magnet demagnetization fault early warning by monitoring parameter changes, possessing significant theoretical and engineering application value.

[0003] Online parameter identification techniques rely on full-rank equations to achieve real-time parameter calculation and iterative parameter updates through discretized recursive models. Current online parameter identification techniques mainly fall into two categories: signal injection rank enhancement and parameter decoupling dimensionality reduction, to achieve full-rank requirements for the identification equations. Signal injection rank enhancement expands the system's observation space by providing external signal excitation, such as pseudo-random sequence signal injection, sinusoidal voltage signal injection, and DC bias signal injection. These methods essentially rely on injecting different types of signals into the permanent magnet synchronous motor, thereby varying its system state and increasing the number of identification equations to achieve full-rank identification. However, these methods introduce high-frequency electromagnetic noise and additional energy losses. Parameter decoupling dimensionality reduction generally achieves full-rank conditions for the identification equations by reducing the number of parameters to be identified. For example, based on a dual-timescale method, the direct-axis inductance and quadrature-axis inductance with smaller time constants are identified at the same time, while the winding resistance and permanent magnet flux linkage with larger time constants are identified at the same time. The two identifications occur at different times, and the identification frequency of parameters with larger time constants is usually lower. The lookup table-based method fixes some parameters of the permanent magnet synchronous motor through offline calibration, such as offline calibration of its direct-axis inductance and quadrature-axis inductance, and embedding them into the controller as a lookup table. The values ​​of the direct-axis and quadrature-axis inductances are then looked up based on the real-time current and substituted into the identification equations to solve for the remaining unknown parameters. This method does not require the injection of additional signals and has potential applications in underwater equipment, marine electric propulsion systems, and other applications requiring low noise and low energy consumption. In the field of discretized recursive model research, traditional artificial intelligence algorithms such as particle swarm optimization and bacterial foraging algorithms generally require a large amount of preliminary data for model training and have high computational complexity, which is difficult for general industrial controllers to meet in terms of real-time computational overhead. Extended Kalman filter (EKF) eliminates the training stage, but its high computational cost still poses a challenge to the real-time performance of online parameter identification. Recursive least squares (RLS) algorithm is favored by experts and scholars for its fast convergence and superior tracking ability. In addition, the low computational overhead and simple programming of RLS algorithm make it easy to implement in industrial controllers, but its identification performance is also affected by different identification coordinate systems.

[0004] While the synchronous dq coordinate system is advantageous for decoupling control of excitation and torque current, it is not necessarily the best choice for online identification of all parameters. The α-β stationary coordinate system, with its structural advantage of phase difference between input and output vectors, allows for simultaneous identification of two parameters using a single equation, significantly improving identification speed. However, in this method, inherent system disturbances are susceptible to current differential amplification, thus reducing steady-state identification performance—leading to increased identification error and fluctuation amplitude. This phenomenon is more pronounced in dual-three-phase permanent magnet synchronous motor drive systems with complex electromagnetic coupling relationships.

[0005] For a six-phase topology of a dual three-phase permanent magnet synchronous motor, vector space decomposition (VSD) control decomposes the system into a fundamental subspace and harmonic subspaces. The harmonic subspace does not participate in the electromechanical energy conversion process, but its harmonic copper losses reduce the system's output efficiency. Therefore, copper losses are often reduced by constraining the harmonic space reference value to zero. However, this zero-current control method makes its parameters difficult to observe, thus preventing online identification of all parameters. To activate the harmonic subspace, one approach employs a multi-segment DC bias injection strategy. This involves sequentially injecting DC bias signals into both the fundamental and harmonic subspaces. Each injection alters the operating state of the dual three-phase permanent magnet synchronous motor (PMSM). Data on voltage and current are collected after each injection as the PSM recovers to steady-state operation. Solving the equations formed by all collected data allows for the identification of all parameters. Another approach uses a mixed signal injection strategy into both the fundamental and harmonic subspaces. This involves sequentially injecting high-frequency square wave signals into both subspaces. The inductance of both subspaces is calculated using the high-frequency response of the inductor. The winding resistance is then determined using the voltage equation and the calculated harmonic inductance. The permanent magnet flux linkage is obtained by offsetting the injection position. Finally, another approach employs a dual-subspace sinusoidal signal injection strategy. Sinusoidal signals are injected into both subspaces, and a second-order generalized integrator (SOGI) is used to extract the signals, avoiding identification errors caused by higher harmonics. However, all of the above signal injection-based methods inevitably introduce additional copper losses and broadband electromagnetic noise, and cause torque pulsation, which limits their application in noise- and energy-sensitive scenarios such as underwater equipment and ship electric propulsion.

[0006] Although existing online identification methods can achieve full-parameter online identification of dual three-phase permanent magnet synchronous motors, they are all based on the VSD framework and rely on injecting signals into the dual three-phase permanent magnet synchronous motor. Due to the observability of harmonic subspaces, full-parameter online identification cannot be completed without signal injection. Furthermore, the additional copper losses, broadband electromagnetic noise, and torque ripples introduced by the injected signal limit its application in noise- and energy-sensitive scenarios such as underwater equipment and ship electric propulsion. In addition, when existing online identification methods for three-phase permanent magnet synchronous motors without signal injection are extended to dual three-phase permanent magnet synchronous motor drive systems, the inherent disturbances of the system are easily affected by the current differential amplification effect, thus reducing the identification performance under steady-state conditions, i.e., increasing the identification error and fluctuation amplitude. Summary of the Invention

[0007] Although existing online identification methods can achieve full-parameter online identification of dual three-phase permanent magnet synchronous motor drive systems, they are all based on the VSD framework and rely on injecting signals into the dual three-phase permanent magnet synchronous motor. Due to the observability of harmonic subspace, full-parameter online identification cannot be completed without signal injection. Furthermore, the additional copper losses, broadband electromagnetic noise, and torque ripples introduced by the injected signal limit its application in noise- and energy-sensitive scenarios such as underwater equipment and marine electric propulsion. In addition, when existing online identification methods for three-phase permanent magnet synchronous motors without signal injection are extended to dual three-phase permanent magnet synchronous motors, the inherent disturbances of the system are easily affected by the current differential amplification effect, thus reducing the identification performance under steady-state conditions, i.e., increasing the identification error and fluctuation amplitude.

[0008] This invention provides a method and device for online identification of all parameters of a dual three-phase permanent magnet synchronous motor without signal injection, in order to solve the technical problems existing in the prior art.

[0009] The technical solution adopted by this invention to solve the technical problems existing in the prior art is as follows: A signal-injection-free online identification method for dual three-phase permanent magnet synchronous motors (PMSMs) is proposed. Based on the six-phase topology of the PSM drive system, and according to the basic mathematical model of the PSM under the dual α-β framework, an equivalent inductance decoupling parameter model under the dual α-β framework is derived as the initial model for online identification of all parameters based on the dual α-β framework. Based on the synchronous rotating coordinate transformation to separate the current differential term in steady state, the initial model for online identification of all parameters is further reconstructed to obtain a reconstructed model for online identification of all parameters. A hierarchical architecture for online identification of all parameters is constructed. An online identification model based on the RLS algorithm is derived from the reconstructed model. The equivalent transformation of identification parameters is achieved by utilizing the mapping mechanism between the dual α-β framework and the VSD framework, ultimately realizing signal-injection-free online identification of all parameters.

[0010] Furthermore, if we consider the windings of the dual three-phase permanent magnet synchronous motor as two independent sets of three-phase windings, then the mathematical model of the dual three-phase permanent magnet synchronous motor based on the dual α-β framework can be expressed as follows: ; ; ; In the formula: This is the α-axis voltage of the first winding; This is the voltage across the α-axis of the second winding. This is the β-axis voltage of the first set of windings; This is the β-axis voltage of the second set of windings; For the α-axis current of the first set of windings; For the α-axis current of the second set of windings; This refers to the β-axis current of the first set of windings; For the β-axis current of the second set of windings; This is the first set of winding α-axis flux linkage; For the second set of windings, the α-axis flux linkage; This is the first set of winding β-axis flux linkage; For the second set of windings, the β-axis flux linkage; This is the inductance matrix under the double α-β framework; For permanent magnet flux linkage; This represents the average value of the winding's self-inductance; The amplitude of the second harmonic of the self-inductance; This represents the average value of the mutual inductance of the windings; This represents the amplitude of the second harmonic of the winding mutual inductance. Stator resistance; It is a differential operator; The rotor electrical angle is given.

[0011] Furthermore, the initial model for full-parameter online identification based on the dual α-β framework is shown below: ; ; ; ; ; ; In the formula: For each winding in a dual α-β frame, it is a direct-axis inductance; For the direct-axis mutual inductance between two sets of windings; For each winding in the dual α-β frame, the cross-axis inductance is used; This refers to the cross-axis mutual inductance between two sets of windings; The rotor's electric angular velocity; This is the inductance coefficient matrix; Let h be the element in the h-th row and v-th column of the inductance coefficient matrix; h = 1, 2, 3, 4, v = 1, 2, 3, 4; This is the differential of the α-axis current of the first winding; This is the differential of the β-axis current of the first winding; This is the differential of the α-axis current of the second winding; This is the differential of the β-axis current of the second winding.

[0012] Furthermore, the method for reconstructing the initial model from the fully parameterized online identification includes the following steps: set up Let be the transformation matrix between the dual dq synchronous coordinate system and the dual α-β coordinate system, then we have: ; ; In the formula: This refers to the d1-axis current in the dual dq synchronous coordinate system. This refers to the d2-axis current in the dual dq synchronous coordinate system. This refers to the q1-axis current in the dual dq synchronous coordinate system. This refers to the q2-axis current in the dual dq synchronous coordinate system. Based on the above equation, the current differential term under the double α-β frame is expressed as follows: ; In the formula: This is the differential of the d1-axis current in the dual synchronous coordinate system; This is the differential of the d2-axis current in the dual synchronous coordinate system; This is the differential of the q1-axis current in the dual synchronous coordinate system; This is the differential of the q2-axis current in the dual synchronous coordinate system; In steady state, the currents along the d and q axes of the dual synchronous rotating coordinate system are DC signals, and their current differential terms... , , , If the value is approximately 0 and can be ignored, then we have: ; Transformation matrix Substituting into the above equation, we further obtain: ; Substituting the simplified results into the initial model of the full-parameter online identification and further deriving, we obtain the following full-parameter online identification and reconstruction model of the dual three-phase permanent magnet synchronous motor: ; ; ; ; In the formula, matrix elements - , - , - , - They are represented as follows: ; ; ; ; In the formula: , This is the inductance coefficient matrix; The set q-axis current differential threshold; This is the actual value of the differential of the q-axis current; This represents the actual value of the differential of the q-axis current under steady-state conditions. This represents the actual differential value of the q-axis current under transient conditions; max() indicates taking the maximum value; min() indicates taking the minimum value; , , , , , , , , , , , , , , , These are the elements in the transformation matrix.

[0013] Furthermore, the method for constructing a fully parameterized online hierarchical identification architecture includes the following steps: When constructing a hierarchical architecture for full-parameter online identification, the criteria for dividing the identification parameters of each layer include physical characteristic criteria and observability criteria. Among them, the physical characteristic criterion is to group parameters with similar time-varying characteristics into the same group; the observability criterion is to avoid grouping parameters with linear correlation into the same group. Let the total number of parameters to be identified for the dual three-phase permanent magnet synchronous motor be 2. N or 2 N +1, dividing the identification parameter set into N +1 layer, where each layer identifies 2 parameters, and each layer represents a subset, specifically as follows: ; In the formula, , , , These correspond to the subsets of identification parameters for the first, second, Nth, and N+1th layers; All parameters to be identified are assigned to the first to Nth layer identification parameter subsets, so that the identification parameter items in the N+1th layer identification parameter subset are the same as the identification parameters in any of the first N layers. Different voltage equations are used for multiple identifications. The average value of the identification parameter items in the N+1th layer identification parameter subset and the same identification parameter items in the first N layers is taken as the final identification parameter value.

[0014] Furthermore, let all the parameters to be identified for the dual three-phase permanent magnet synchronous motor include: , , , , , ; For each winding in a dual α-β frame, it is a direct-axis inductance; For each winding in the dual α-β frame, the cross-axis inductance is used; For the direct-axis mutual inductance between two sets of windings; This refers to the cross-axis mutual inductance between two sets of windings; Stator resistance; For permanent magnet flux linkages; the identification parameter set is divided into the following four subsets of identification parameters: ; In the formula, , , , These are the first, second, third, and fourth layer identification parameter subsets, respectively. Each layer of identification parameter subset includes two parameters to be identified. The four layers of identification parameter subsets are represented as follows: ; In the formula, , To satisfy any two distinct identification parameter terms of the hierarchical core criterion; Make the fourth layer identification parameter the same as any of the first three layers, use different voltage equations for secondary identification, and average the fourth layer identification parameter with the same identification parameter in the first three layers to improve identification accuracy.

[0015] Furthermore, the derivation of the fully parameterized online identification model based on the RLS algorithm from the fully parameterized online identification and reconstruction model includes the following steps: The reconfiguration identification model of the dual three-phase permanent magnet synchronous motor is mapped to the subset of identification parameters according to the following formula: ; In the formula: g To identify the voltage procedure number; Indicates belonging to nThe first dimension of the real space g The voltage equation up to the first i ( g Mapping of a subset of parameters to be identified; This is the g-th identifying voltage equation; for n 3D real space; for g The defined value is 4; The index of the subset of parameters to be identified; The expression for the parameter subset index; For the first A subset of parameters to be identified; The identification parameters are represented as follows: ; In the formula: No. w Layer identification parameter subset; w =1, 2, ..., N , N +1; w To identify the parameter layer number; j for Identify the parameter item number in the middle; It is the first j Each identification parameter item; It is the first i Layer j Solve the equations for the inverse voltage function corresponding to each identified parameter; It is the equation for solving the inverse function of the third-level voltage; It is the equation for solving the inverse function of the fourth-level voltage; i ( j ) is the first i Layer j Each identification parameter item; For the first j The voltage equations corresponding to each identification parameter term; The basic iterative model of the RLS algorithm is as follows: ; In the formula: For time sequence number; For the first k Real-time identification value; For the first k -1 time step identification value; For the first k Time-matrix gain matrix; For the first k Output matrix at time step; For the first kInput matrix at each time step; For the first k The transpose of the input matrix at each time step; For the first k Time-varying covariance matrix; For the first k Covariance matrix at time -1; As a weighting factor; It is the identity matrix; Based on the basic iterative model of the RLS algorithm and the full-parameter online identification and reconstruction model of the dual three-phase permanent magnet synchronous motor, the following identification parameter matrix, input matrix, and output matrix are derived, where: The identification parameter matrix is ​​as follows: ; The input matrix is: ; In the formula, , As an intermediate variable, , The calculation formula is as follows: ; The output matrix is ​​represented as: ; In the formula, , , , The calculation formula is as follows: ; In the formula: To identify the parameter matrix; The input matrix; This is the output matrix; , , , Corresponding to the output matrix Y The elements in the first row and first column, the second row and second column, the third row and third column, and the fourth row and fourth column; The winding resistance is identified using the third voltage equation; The winding resistance is identified using the fourth voltage equation; For permanent magnet flux identified using the third voltage equation; For permanent magnet flux linkages identified using the fourth voltage equation; At the start of the identification process, the nominal parameters of the dual three-phase permanent magnet synchronous motor are used to initialize each identification equation to improve the convergence speed of the online identification. After the identified parameters converge, the identified values ​​are used to replace the initial values ​​in each identification equation, and the identification is performed alternately in each identification equation to update the parameters online.

[0016] Furthermore, the method for achieving equivalent transformation of identification parameters by utilizing the mapping mechanism of parameters between the dual α-β framework and the VSD framework includes the following steps: The parameter identification values ​​under the VSD framework are transformed according to the following formula: ; In the formula: A vector composed of the identification values ​​of each parameter of the VSD framework; It is a vector composed of the identification values ​​of each parameter of the double α-β frame; Transformation matrix Represented as: .

[0017] Furthermore, during normal operation and control of the dual three-phase permanent magnet synchronous motor, the actual six-phase current collected is transformed according to the following formula to obtain the actual current in the synchronous rotating coordinate system: ; In the formula: This refers to the d-axis current under the VSD framework. This refers to the q-axis current under the VSD framework. The x-axis current under the VSD framework; This refers to the y-axis current within the VSD framework. This refers to the current in phase A. This refers to the B-phase current. This refers to the C-phase current. This refers to the current in phase D. This refers to the E-phase current. This refers to the F-phase current. This is the transformation matrix from the stationary coordinate system to the VSD frame synchronous coordinate system; The transformation matrix from the six-phase stationary coordinate system to the VSD frame stationary coordinate system; , They are represented as follows: ; ; In the formula: The rotor electrical angle; The differential of the actual q-axis current is calculated using the inverse Euler discretization method according to the following formula: ; In the formula: For the first k The actual value of the differential of the q-axis current at time t; For the first k Actual q-axis current at any given time; For the first k Actual q-axis current at time -1; The system sampling time; The specific calculation method for the inductance coefficient matrix is ​​selected based on the logical relationship between the actual value of the q-axis current derivative and the q-axis current derivative threshold, and then transmitted to the identification layer. The rotor position and speed information required for identification are sampled and imported into the full-parameter online identification equation. The dual α-β frame voltage information is calculated from the reference voltage output by the PI controller according to the following formula: ; In the formula: This is the α-axis voltage of the first winding; This is the voltage across the α-axis of the second winding. This is the β-axis voltage of the first set of windings; This is the β-axis voltage of the second set of windings; For the direct-axis voltage of the VSD control frame; The quadrature-axis voltage of the VSD control frame; The x-axis voltage of the VSD control frame; The y-axis voltage of the VSD control frame; The transformation matrix between the dual dq synchronous coordinate system and the dual α-β coordinate system is expressed as: .

[0018] This invention also discloses an apparatus for an online identification method of full parameters of a dual three-phase permanent magnet synchronous motor without signal injection, comprising a memory and a processor. The memory is used to store a computer program; the processor is used to execute the computer program and, when executing the computer program, implement the steps of the online identification method of full parameters of a dual three-phase permanent magnet synchronous motor without signal injection as described above.

[0019] The advantages and positive effects of this invention are: 1) The method proposed in this invention enables online identification of all parameters of a dual three-phase permanent magnet synchronous motor without signal injection, meaning that no additional copper losses, noise, or torque ripples are generated during the online identification process. To the best of the author's knowledge, existing online identification technologies for dual three-phase permanent magnet synchronous motors are all based on signal injection methods. Therefore, this invention can extend the application boundaries of online identification of all parameters of dual three-phase permanent magnet synchronous motors to scenarios sensitive to noise and energy consumption.

[0020] 2) In traditional methods, inherent system disturbances are easily amplified by the current differential effect, leading to large errors and fluctuations in the online identification of all parameters. This invention separates the current differential term under steady-state conditions based on synchronous rotating coordinate transformation, and then proposes a reconstruction identification algorithm based on a dual α-β framework. In the reconstruction identification algorithm, the q-axis current differential threshold can be adjusted according to different actual operating conditions. Specifically, its value can be flexibly adjusted to adapt to different dual three-phase permanent magnet synchronous motor bodies, different load conditions, and application scenarios. Therefore, the reconstruction identification algorithm based on the dual α-β framework proposed in this invention can effectively reduce the identification error and fluctuation amplitude of the online identification of all parameters, achieving accurate and stable online identification of all parameters, and possessing good adaptability to operating conditions.

[0021] 3) The hierarchical identification architecture constructed in this invention consists of four layers. The identification parameters of the fourth layer can be averaged with those of the same identification parameters in the first three layers to improve identification accuracy. In other words, the identification parameters of the fourth layer can be consistent with those of any one of the first three layers. Since different voltage equations are used for secondary identification, averaging improves identification accuracy. Therefore, the hierarchical identification architecture constructed in this invention possesses the superior characteristics of flexibility and high identification accuracy. Attached Figure Description

[0022] Figure 1 This is a schematic diagram of a dual three-phase permanent magnet synchronous motor drive system.

[0023] Figure 2 This is a schematic diagram of a dual α-β frame for a dual three-phase permanent magnet synchronous motor.

[0024] Figure 3 This is a schematic diagram illustrating the working principle of an online full-parameter identification method for a dual three-phase permanent magnet synchronous motor without signal injection, as per the present invention.

[0025] Figure 4 This is a flowchart illustrating the online identification process of all parameters of a signal-injection-free dual three-phase permanent magnet synchronous motor according to the present invention.

[0026] Figure 5 This is a diagram showing the identification results of resistance and permanent magnet flux linkage using the traditional αβ method under rated steady-state operating conditions.

[0027] Figure 6This is a diagram showing the identification results of the resistance and permanent magnet flux linkage of the method proposed in this invention under rated steady-state operating conditions.

[0028] Figure 7 It is the traditional αβ method under rated steady-state conditions. and Identification results image.

[0029] Figure 8 The method proposed in this invention is under rated steady-state operating conditions. and Identification results image.

[0030] Figure 9 It is the traditional αβ method under rated steady-state conditions. and Identification results image.

[0031] Figure 10 The method proposed in this invention is under rated steady-state operating conditions. and Identification results image.

[0032] Figure 11 This is a diagram showing the identification results of resistance and permanent magnet flux linkage using the traditional αβ method under medium-speed, medium-load steady-state conditions.

[0033] Figure 12 This is a diagram showing the identification results of resistance and permanent magnet flux linkage using the method proposed in this invention under medium-speed, medium-load steady-state conditions.

[0034] Figure 13 It is the traditional αβ method under medium-speed, medium-load steady-state conditions. and Identification results image.

[0035] Figure 14 The method proposed in this invention is for medium-speed, medium-load steady-state operating conditions. and Identification results image.

[0036] Figure 15 It is the traditional αβ method under medium-speed, medium-load steady-state conditions. and Identification results image.

[0037] Figure 16 The method proposed in this invention is for medium-speed, medium-load steady-state operating conditions. and Identification results image.

[0038] Figure 17 This is a diagram showing the identification results of resistance and permanent magnet flux linkage using the traditional αβ method under low-speed, light-load, steady-state conditions.

[0039] Figure 18This is a diagram showing the identification results of resistance and permanent magnet flux linkage using the method proposed in this invention under low-speed, light-load, steady-state conditions.

[0040] Figure 19 It is the traditional αβ method under low-speed, light-load steady-state conditions. and Identification results image.

[0041] Figure 20 The method proposed in this invention is for low-speed, light-load, steady-state operating conditions. and Identification results image.

[0042] Figure 21 It is the traditional αβ method under low-speed, light-load steady-state conditions. and Identification results image.

[0043] Figure 22 The method proposed in this invention is for low-speed, light-load, steady-state operating conditions. and Identification results image.

[0044] In the picture: A represents the A-phase winding of a dual three-phase permanent magnet synchronous motor; B represents the B-phase winding of a dual three-phase permanent magnet synchronous motor; C represents the C-phase winding of a dual three-phase permanent magnet synchronous motor; D represents the D-phase winding of a dual three-phase permanent magnet synchronous motor; E represents the E-phase winding of a dual three-phase permanent magnet synchronous motor; F represents the F-phase winding of a dual three-phase permanent magnet synchronous motor; d1 represents the d1 axis in the dual synchronous rotating coordinate system; q1 represents the q1 axis in the dual synchronous rotating coordinate system; d2 represents the d2 axis in the dual synchronous rotating coordinate system; q2 represents the q2 axis in the dual synchronous rotating coordinate system; α1 represents the dual α-β In the coordinate system, α1 represents the α1 axis; β1 represents the β1 axis in the dual α-β coordinate system; α2 represents the α2 axis in the dual α-β coordinate system; β2 represents the β2 axis in the dual α-β coordinate system; MTPA is maximum torque-current ratio control; PI is proportional-integral controller; PR is resonant controller; SVPWM is space vector pulse width modulation; DT-PMSM is dual three-phase permanent magnet synchronous motor; d is the d-axis in the VSD control framework; q is the q-axis in the VSD control framework; x is the x-axis in the VSD control framework; y is the y-axis in the VSD control framework. ω This is the actual angular velocity of the motor; This is the reference angular velocity for the motor; Electromagnetic torque; This is the reference current for the dq axis; The reference current for the x and y axes; This is the reference voltage for the dq axis; The reference voltage for the x and y axes; The reference voltage for the α1β1 axis; The reference voltage for the α2β2 axis; These are the currents of phases ABC; For the DEF phase current; For the direct-axis inductance identification values ​​of the two sets of windings; For the cross-axis inductance identification values ​​of the two sets of windings; The identification value for the direct-axis mutual inductance between the two sets of windings; The cross-axis mutual inductance identification value between the two sets of windings; and All are direct-axis inductance identification values ​​for the VSD frame; and All are cross-axis inductance identification values ​​for the VSD frame; The x-axis inductance identification value for the VSD frame; The y-axis inductance identification value for the VSD frame; For resistor identification value; For permanent magnet flux linkage identification value; This is the mapping matrix for the parameters from the dual α-β frame to the VSD frame; Ohm is the unit of resistance: ohm; mH is the unit of inductance: millihenry; Wb is the unit of permanent magnet flux linkage: Weber; ms is the millisecond; N1 is the bus voltage; N2 is the neutral point of the first winding; N3 is the neutral point of the second winding; other expressions refer to the definitions in this manual. Detailed Implementation

[0045] The present invention will now be described in detail with reference to the accompanying drawings and embodiments. It should be understood that the preferred embodiments described herein are for illustration and explanation only and are not intended to limit the present invention.

[0046] In the description of this invention, the terms "upper," "lower," "front," "rear," "left," "right," "vertical," "horizontal," "top," and "bottom," etc., indicate the orientation or positional relationship based on the orientation or positional relationship shown in the accompanying drawings. They are used only for the convenience of describing the invention and do not require the invention to be constructed and operated in a specific orientation; therefore, they should not be construed as limitations on the invention. The terms "connected" and "linked" used in this invention should be interpreted broadly. For example, they can refer to a fixed connection or a detachable connection; a direct connection or an indirect connection through intermediate components; or an electrical connection or signal transmission. Those skilled in the art can understand the specific meaning of the above terms according to the specific circumstances.

[0047] The Chinese definitions of the following English words, phrases, and abbreviations are as follows: VSD: Vector Space Decoupling. RLS: Recursive Least Squares. EKF: Kalman Filter. PI Controller: Proportional-Integral Controller. PI-PR Composite Controller: Proportional-Integral-Resonant Composite Controller.

[0048] VSD control, due to its excellent decoupling control capability, is often used as the main control method for dual three-phase permanent magnet synchronous motor drive systems. The mathematical model of a dual three-phase permanent magnet synchronous motor based on the VSD framework is expressed as follows: (1); In the formula: , , , These correspond to the d-axis voltage, q-axis voltage, x-axis voltage, and y-axis voltage, respectively. For stator resistance, , , , These correspond to the d-axis inductance, q-axis inductance, x-axis inductance, and y-axis inductance, respectively. , , , These correspond to the d-axis current, q-axis current, x-axis current, and y-axis current, respectively, where p is the differential operator. It is the rotor's electrical angular velocity. It is a permanent magnet flux linkage.

[0049] For online full-parameter identification of a dual three-phase permanent magnet synchronous motor, six parameters need to be identified. In actual control, to reduce copper losses, a control method with zero harmonic subspace current is generally adopted. , The value is 0. According to equation (1), the harmonic subspace is unobservable without signal injection, meaning that the relevant information and parameters of the harmonic subspace cannot be identified based on the VSD framework.

[0050] To address this issue and expand the application boundaries of online full-parameter identification for dual three-phase permanent magnet synchronous motors, this invention proposes a signal-injection-free online full-parameter identification method for dual three-phase permanent magnet synchronous motors. Please refer to... Figures 1 to 22 This method is based on the six-phase topology of a dual three-phase permanent magnet synchronous motor drive system. Based on the fundamental mathematical model of the dual three-phase permanent magnet synchronous motor under the dual α-β framework, an equivalent inductance decoupling parameter model under the dual α-β framework is derived as the initial model for full-parameter online identification based on the dual α-β framework. By separating the current differential term under steady-state conditions using synchronous rotating coordinate transformation, the initial model for full-parameter online identification is further reconstructed to obtain a reconstructed model for full-parameter online identification. A hierarchical architecture for full-parameter online identification is constructed. The reconstructed model for full-parameter online identification is used to derive a full-parameter online identification model based on the RLS algorithm. The mapping mechanism between the dual α-β framework and the VSD framework is used to achieve equivalent transformation of the identification parameters, ultimately realizing full-parameter online identification without signal injection.

[0051] Preferably, based on the six-phase topology of the dual three-phase permanent magnet synchronous motor drive system, the online identification of all parameters is completed within a dual α-β framework, as shown in the coordinate system of the dual α-β framework. Figure 2 As shown, this framework contains effective information from both the fundamental and harmonic subspaces. The basic mathematical model of the dual three-phase permanent magnet synchronous motor under the dual α-β framework is as follows: (2); (3); (4); In the formula: This is the α-axis voltage of the first winding; This is the voltage across the α-axis of the second winding. This is the β-axis voltage of the first set of windings; This is the β-axis voltage of the second set of windings; For the α-axis current of the first set of windings; For the α-axis current of the second set of windings; This refers to the β-axis current of the first set of windings; For the β-axis current of the second set of windings; This is the first set of winding α-axis flux linkage; For the second set of windings, the α-axis flux linkage; This is the first set of winding β-axis flux linkage; For the second set of windings, the β-axis flux linkage; This is the inductance matrix under the double α-β framework; For permanent magnet flux linkage; This represents the average value of the winding's self-inductance; The amplitude of the second harmonic of the self-inductance; This represents the average value of the mutual inductance of the windings; This represents the amplitude of the second harmonic of the winding mutual inductance. Stator resistance; It is a differential operator; The rotor electrical angle is given.

[0052] Preferably, based on the basic mathematical model of the dual three-phase permanent magnet synchronous motor under the dual α-β framework, the equivalent inductance decoupling parameter model under the dual α-β framework can be further derived as the initial model for online full-parameter identification based on the dual α-β framework: (5); (6); (7); (8); (9); (10); In the formula: For each winding in a dual α-β frame, it is a direct-axis inductance; For the direct-axis mutual inductance between two sets of windings; For each winding in the dual α-β frame, the cross-axis inductance is used; This refers to the cross-axis mutual inductance between two sets of windings; The rotor's electric angular velocity; This is the inductance coefficient matrix; Let h be the element in the h-th row and v-th column of the inductance coefficient matrix; h = 1, 2, 3, 4, v = 1, 2, 3, 4. This is the differential of the α-axis current of the first winding; This is the differential of the β-axis current of the first winding; This is the differential of the α-axis current of the second winding; This is the differential of the β-axis current of the second winding.

[0053] Preferably, the method for reconstructing the initial model of the full-parameter online identification may include the following steps: The voltage equation (5) of the dual three-phase permanent magnet synchronous motor based on the dual α-β frame, derived from the derivation, can serve as the basic model for online identification of all parameters. However, the current differential terms in the system generally need to be discretized according to the reverse Euler method. Taking the α1 axis current as an example, this can be explained as follows: (11); In the actual current sampling calculation of a dual three-phase permanent magnet synchronous motor drive system, noise and errors inevitably exist, which affect the online identification of all parameters in the form of system disturbances.

[0054] We can assume that the sampled current is the sum of the actual current, system noise, and current sampling error, then we have: (12); In the formula: For time sequence number; For the first k Current sampled along axis α1 at time t; For the first k The sampling current of the α1 axis at time -1; For the first k Actual current along axis α1 at time 1; For the first k Actual current of α1 axis at time -1; For the first k The differential of the current along the α1 axis at time t; For the first k System noise at any given moment; For the firstk System noise at time -1; For the first k Current sampling error at any given time; For the first k Current sampling error at time -1; The system sampling time; Assuming the system noise follows a Gaussian distribution, then: (13); In the formula: For the first k The mean at any given time; For the first k The mean at time -1; For the first k Variance over time; For the first k Variance at time -1; N ( ) represents a Gaussian distribution; According to equations (11)–(13), the differential expression of the sampling current of axis α1 is: (14); In the formula: For the first k The differential of the sampling current along the α1 axis at time 1; For the first k Current sampled along axis α1 at time t; For the first k The sampling current of the α1 axis at time -1; For the first k The change in system noise at any given time; For the first k The change in current sampling error at any given time; Noise and error, as system disturbances, are amplified to 1 / 3 of their original value under the action of the differential operator. times ( (Generally very small), which in turn reduces the full-parameter online identification performance under the dual α-β framework, specifically manifested as increased identification error and fluctuation amplitude.

[0055] To address this issue, this invention further proposes a method for separating the current differential term in the steady-state identification model. Based on synchronous rotating coordinate transformation, the current differential term can be separated in steady state, thereby reconstructing the full-parameter online identification model of the dual three-phase permanent magnet synchronous motor, achieving accurate and stable full-parameter online identification.

[0056] Can be set Let be the transformation matrix between the dual dq synchronous coordinate system and the dual α-β coordinate system, then we have: (15); (16); In the formula: This refers to the d1-axis current in the dual dq synchronous coordinate system. This refers to the d2-axis current in the dual dq synchronous coordinate system. This refers to the q1-axis current in the dual dq synchronous coordinate system. This refers to the q2-axis current in the dual dq synchronous coordinate system. Based on the above equation, the current differential term under the double α-β frame is expressed as follows: (17); In the formula: This is the differential of the d1-axis current in the dual synchronous coordinate system; This is the differential of the d2-axis current in the dual synchronous coordinate system; This is the differential of the q1-axis current in the dual synchronous coordinate system; This is the differential of the q2-axis current in the dual synchronous coordinate system; In steady state, the currents along the d and q axes of the dual synchronous rotating coordinate system are DC signals, and their current differential terms... , , , If the value is approximately 0 and can be ignored, then we have: (18); Transformation matrix Substituting into the above equation, we further obtain: (19); Substituting the simplified results into the initial model of the full-parameter online identification and further deriving, we obtain the following full-parameter online identification and reconstruction model of the dual three-phase permanent magnet synchronous motor: (20); In the formula, The range of the q-axis current differential threshold is expressed as follows: (twenty one); , It is the inductance coefficient matrix, represented as: (twenty two); In the formula, matrix elements - , - , - , - They are represented as follows: (twenty three); (twenty four); (25); (26); In the formula: , This is the inductance coefficient matrix; The set q-axis current differential threshold; This is the actual value of the differential of the q-axis current; This represents the actual value of the differential of the q-axis current under steady-state conditions. This represents the actual differential value of the q-axis current under transient conditions; max() indicates taking the maximum value; min() indicates taking the minimum value; , , , , , , , , , , , , , , , These are the elements in the transformation matrix.

[0057] One embodiment of the online full-parameter identification method proposed in this invention is as follows: Figure 3 As shown, the dual three-phase permanent magnet synchronous motor adopts vector control. A PI controller is used in the fundamental subspace to achieve precise tracking of speed and torque, while a PI-PR composite controller is used in the harmonic subspace to effectively suppress 6... k ±1( k = 1st, 3rd, 5th…) harmonic currents. It should be noted that… Figure 3 The control scheme for the dual three-phase permanent magnet synchronous motor is not unique. The proposed implementation scheme is a typical vector control scheme for the dual three-phase permanent magnet synchronous motor drive system. The online identification method for all parameters of the dual three-phase permanent magnet synchronous motor proposed in this invention is based on the normal control operation of the dual three-phase permanent magnet synchronous motor drive system. Therefore, the online identification method for all parameters proposed in this invention is not limited to the control method in this implementation scheme.

[0058] The complete parameter set consisting of the parameters to be identified in a dual three-phase permanent magnet synchronous motor is defined as follows: (27); This invention is based on the online identification model of full parameters of dual three-phase permanent magnet synchronous motor reconfiguration—Equation (20), and constructs a hierarchical identification strategy.

[0059] Preferably, when constructing a hierarchical architecture for full-parameter online identification, the criteria for dividing the identification parameters of each layer include physical characteristic criteria and observability criteria. The physical characteristic criterion is to group parameters with similar time-varying characteristics into the same group; the observability criterion is to avoid grouping parameters with linear correlation into the same group. For example: (1) Physical characteristic criteria: The time constant of each inductance term as a current-related parameter group is relatively small, while the time constant of the resistance and permanent magnet flux linkage as a temperature-related parameter group is relatively large. (2) Observability criterion: Strictly avoid , Divide them into the same group to avoid , They are grouped into the same group. This grouping method can lead to identification failure because the observed vectors are linearly correlated, resulting in an underranked identification matrix.

[0060] Furthermore, the method for constructing a fully parameterized online hierarchical identification architecture includes the following steps: Suppose that there are 2N or 2N+1 parameters to be identified for a dual three-phase permanent magnet synchronous motor. The parameter set is divided into N+1 layers. Based on the above two criteria, the parameter set can be represented as the following N+1 layer subset of identified parameters: ; In the formula, , , , This corresponds to the subsets of identification parameters for the first, second, N, and N+1 layers.

[0061] All parameters to be identified are assigned to the first to Nth layer identification parameter subsets, so that the identification parameter items in the N+1th layer identification parameter subset are the same as the identification parameters in any of the first N layers. Different voltage equations are used for multiple identifications. The average value of the identification parameter items in the N+1th layer identification parameter subset and the same identification parameter items in the first N layers is taken as the final identification parameter value.

[0062] Preferably, the total number of parameters to be identified for the dual three-phase permanent magnet synchronous motor includes: , , , , , The selection of the fourth layer identification parameters in the full-parameter online identification method proposed in this invention is flexible. It can select parameters from the first, second, or third layer and calculate the average value to improve the identification accuracy of that parameter set. Figure 3 The implementation scheme shown presents an example of secondary identification of the third layer, namely, secondary identification of the resistor and permanent magnet flux linkage and calculation of the average value.

[0063] The identification parameter set is divided into the following four levels of identification parameter subsets: (28); In the formula , , , These are the first, second, third, and fourth layer identification parameter subsets, respectively. There are a total of 6 parameters to be identified for the dual three-phase permanent magnet synchronous motor. In the online full-parameter identification method proposed in this invention, each layer of identification parameter group can identify 2 parameters. In principle, any grouping method that meets the above two core criteria can become the full-parameter identification method in this invention. The following is a typical grouping implementation scheme, in which each layer of identification parameter group is set with two parameters. The identification parameter set is represented as follows: (29); In the formula, , To satisfy any two distinct identification parameter terms of the hierarchical core criterion; The fourth-layer identification parameter is made identical to the identification parameters of any of the first three layers. A second identification is performed using different voltage equations, and the average value of the fourth-layer identification parameter and the identical identification parameters from the first three layers is calculated. Because different voltage equations are used for the second identification, the identification accuracy is improved after averaging.

[0064] After the hierarchical identification strategy is constructed, an identification algorithm is needed to implement it. The RLS algorithm, with its fast convergence and superior tracking ability, is widely used in system identification, adaptive control and other fields.

[0065] Preferably, the method for deriving the fully parameterized online identification model based on the RLS algorithm includes the following steps: The mathematical model of the dual three-phase permanent magnet synchronous motor is mapped to the subset of identification parameters according to the following formula: (30); In the formula: g To identify the voltage procedure number; Indicates belonging to n The first dimension of the real space g The voltage equation up to the firsti ( g Mapping of a subset of parameters to be identified; This is the g-th identifying voltage equation; for n 3D real space; for g The defined value is 4; The index of the subset of parameters to be identified; The expression for the parameter subset index; For the first A subset of parameters to be identified; The identification parameters are represented as follows: (31); In the formula: No. w Layer identification parameter subset; w =1, 2, ..., N , N +1; w To identify the parameter layer number; j for Identify the parameter item number in the middle; It is the first j Each identification parameter item; It is the first i Layer j Solve the equations for the inverse voltage function corresponding to each identified parameter; It is the equation for solving the inverse function of the third-level voltage; It is the equation for solving the inverse function of the fourth-level voltage; i ( j ) is the first i Layer j Each identification parameter item; For the first j The voltage equations corresponding to the identified parameter terms.

[0066] The basic iterative model of the RLS algorithm is as follows: (32); In the formula: For time sequence number; For the first k Real-time identification value; For the first k -1 time step identification value; For the first k Time-matrix gain matrix; For the first k Output matrix at time step; For the first k Input matrix at each time step; For the firstk The transpose of the input matrix at each time step; For the first k Time-varying covariance matrix; For the first k Covariance matrix at time -1; As a weighting factor; It is the identity matrix; Based on the basic iterative model of the RLS algorithm and the full-parameter online identification and reconstruction model of the dual three-phase permanent magnet synchronous motor, the following identification parameter matrix, input matrix, and output matrix are derived, where: The identification parameter matrix is ​​as follows: (33); The input matrix is: (34); In the formula, , As an intermediate variable, , The calculation formula is as follows: (35); The output matrix is ​​represented as: (36); In the formula, , , , The calculation formula is as follows: (37); In the formula: To identify the parameter matrix; The input matrix; This is the output matrix; , , , Corresponding to the output matrix Y The elements in the first row and first column, the second row and second column, the third row and third column, and the fourth row and fourth column; The winding resistance is identified using the third voltage equation; The winding resistance is identified using the fourth voltage equation; For permanent magnet flux identified using the third voltage equation; For permanent magnet flux linkages identified using the fourth voltage equation; So if Figure 3 In one embodiment of the present invention, the stator resistance and permanent magnet flux linkage are further calculated as follows: (38); At the start of the identification process, the nominal parameters of the dual three-phase permanent magnet synchronous motor are used to initialize each identification equation to improve the convergence speed of the online identification. After the identified parameters converge, the identified values ​​are used to replace the initial values ​​in each identification equation, and the identification is performed alternately in each identification equation to update the parameters online.

[0067] The method for achieving equivalent transformation of identification parameters by utilizing the mapping mechanism between parameters in the dual α-β framework and the VSD framework includes the following steps: The parameter identification values ​​under the VSD framework are transformed according to the following formula: (39); In the formula: A vector composed of the identification values ​​of each parameter of the VSD framework; It is a vector composed of the identification values ​​of each parameter of the double α-β frame; In the formula, It is by Convert to The parameter transformation matrix is ​​expressed as: (40); To further elaborate on the specific process of the online full-parameter identification method proposed in this invention, a diagram is drawn. Figure 4 The flowchart shown is shown. Now, regarding... Figure 4 The flowchart shown is used for illustration. Preferably, during normal operation control of the dual three-phase permanent magnet synchronous motor, the actual six-phase current collected is transformed according to the following formula to obtain the actual current in the synchronous rotating coordinate system: (41); In the formula: This refers to the d-axis current under the VSD framework. This refers to the q-axis current under the VSD framework. The x-axis current under the VSD framework; This refers to the y-axis current within the VSD framework. This refers to the current in phase A. This refers to the B-phase current. This refers to the C-phase current. This refers to the current in phase D. This refers to the E-phase current. This refers to the F-phase current. This is the transformation matrix from the stationary coordinate system to the VSD frame synchronous coordinate system; The transformation matrix from the six-phase stationary coordinate system to the VSD frame stationary coordinate system; , They are represented as follows: (42); (43); The differential of the actual q-axis current is calculated using the inverse Euler discretization method according to the following formula: (44); In the formula: For the first k The actual value of the differential of the q-axis current at time t; For the first k Actual q-axis current at any given time; For the first k Actual q-axis current at time -1; The specific calculation method for the inductance coefficient matrix is ​​selected based on the logical relationship between the actual value of the q-axis current derivative and the q-axis current derivative threshold, and then transmitted to the identification layer. The rotor position and speed information required for identification are sampled and imported into the full-parameter online identification equation. The dual α-β frame voltage information is calculated from the reference voltage output by the PI controller according to the following formula: (45); In the formula: For the direct-axis voltage of the VSD control frame; The quadrature-axis voltage of the VSD control frame; The x-axis voltage of the VSD control frame; This represents the y-axis voltage of the VSD control frame.

[0068] It should be noted that the dual α-β frame reference voltage used in the flowchart of this invention is compensated for by the inverter nonlinear voltage. This compensation method is based on the lookup table method, that is, the inverter nonlinear voltage is calibrated offline and embedded into the online identification program in the form of a lookup table. Therefore, the reference voltage is equal to the actual voltage acting on the dual three-phase permanent magnet synchronous motor. However, this compensation is not the core of this invention, so it is only explained in the description of the identification process.

[0069] At the start of the identification process, each identification equation is initialized with the nominal parameters of a dual three-phase permanent magnet synchronous motor to improve the convergence speed of online identification. After the identified parameters converge, the identified values ​​are used to replace the initial values ​​in each identification equation, and this process is repeated alternately in each equation to update the parameters online. Finally, the identified parameters are transformed into identification parameters for the VSD framework using a transformation matrix.

[0070] To verify the effectiveness and superiority of the proposed method under wide operating conditions, various operating condition experiments were conducted on a dual three-phase permanent magnet synchronous motor drive system experimental platform. The basic parameters of the dual three-phase permanent magnet synchronous motor drive system experimental platform are shown in Table 1. The two sets of windings of the dual three-phase permanent magnet synchronous motor are driven by a common bus inverter. The TMS320F28335DSP is used as the core controller to implement the algorithm function. The load motor is controlled by a host computer and is coaxially connected to the dual three-phase permanent magnet synchronous motor to form a dual-drive experimental platform.

[0071] Table 1. Parameters of Dual Three-Phase Permanent Magnet Synchronous Motors and Load Motors

[0072] Rated steady-state test To verify the effectiveness and superiority of the proposed online full-parameter identification algorithm for dual three-phase permanent magnet synchronous motors based on a dual α-β frame without signal injection, a rated steady-state test was conducted. For ease of performance comparison, the identification errors for each parameter are defined as follows: (46) In the formula: δX It is the identification error of each identification parameter item. These are the identification values ​​of each identification parameter. These are the nominal values ​​of each parameter. This definition is specifically used in the performance comparison table of identification algorithms.

[0073] Figure 5 , Figure 7 , Figure 9 The waveforms identified by the traditional αβ algorithm are shown respectively. Figure 6 , Figure 8 , Figure 10 The identification waveforms of the algorithm proposed in this invention are shown in Table 2. Table 2 compares the identification performance of the traditional αβ algorithm and the algorithm proposed in this invention, specifically the identification error and fluctuation amplitude. It should be noted that the identification values ​​for each parameter are the average values ​​over the specified time period, calculated directly from the oscilloscope. Δ represents the fluctuation amplitude of each identification parameter, with units consistent with those of each parameter. For all experimental conditions, the resistance identification waveform is 1 Ω / division; the permanent magnet flux identification waveform is 0.083 Wb / division; and all inductance parameters are 5 mH / division.

[0074] from Figure 5 and Figure 6 The comparison Figure 7 and Figure 8 The comparison Figure 9 and Figure 10The comparison in Table 2 shows that, under rated operating conditions, the proposed algorithm outperforms the traditional αβ algorithm in identifying all parameters. Specifically, the identification errors for resistance, direct-axis inductance, quadrature-axis inductance, x-axis inductance, and y-axis inductance are significantly smaller than those of the traditional αβ algorithm, and the fluctuation amplitude is also smaller. This is because, under steady-state conditions, the proposed algorithm effectively suppresses system disturbances, while in the traditional αβ algorithm, system disturbances are directly amplified by the current differential amplification effect, reducing identification performance and increasing identification error and fluctuation amplitude. Under this experimental condition, the superiority of the proposed algorithm in identifying permanent magnet flux linkage is slightly lower than that of the other five parameters. This is because the permanent magnet flux linkage term is mainly affected by rotational speed. Under rated operating conditions, the rotational speed is higher, and its influence from the current differential amplification effect is relatively smaller. However, the permanent magnet flux linkage identification error of the proposed algorithm is still smaller than that of the traditional αβ algorithm.

[0075] Table 2. Comparison of identification performance between the traditional αβ method and the proposed method under rated operating conditions.

[0076] Medium-speed, medium-load steady-state test To further verify the effectiveness and superiority of the method proposed in this invention, a medium-speed, medium-load steady-state test was conducted. The reference motor speed was set to 500 r / min, and the torque was set to 5 N·m. Figure 11 , Figure 13 , Figure 15 The waveforms identified by the traditional αβ algorithm are shown respectively. Figure 12 , Figure 14 , Figure 16 The identification waveforms of the proposed algorithm are shown in Table 3. Table 3 shows the comparison of the identification performance of the traditional αβ algorithm and the proposed algorithm, namely the comparison of identification error and fluctuation amplitude.

[0077] from Figure 11 and Figure 12 The comparison Figure 13 and Figure 14 The comparison Figure 15 and Figure 16The comparison in Table 3 shows that, under medium-speed and medium-load conditions, the proposed algorithm outperforms the traditional αβ algorithm in identifying all parameters. The identification errors and fluctuation amplitudes of all identified parameters are significantly smaller than those of the traditional αβ algorithm. This is because, under steady-state conditions, the proposed algorithm effectively suppresses system disturbances, while in the traditional αβ algorithm, system disturbances are directly amplified by the current differential amplification effect, reducing identification performance and increasing identification error and fluctuation amplitude. Compared to the rated experimental conditions, the superiority of the proposed algorithm is even more pronounced under medium-speed and medium-load conditions. This is because, under medium-speed and medium-load conditions, the signal-to-noise ratio is lower than under rated conditions, and the impact of the current differential amplification effect on identification performance is more significant, especially in the identification of resistance and x-axis inductance. Resistance identification mainly relies on the current signal; under medium-load conditions, the proportion of resistance terms is lower than under rated conditions, and it is more affected by the current differential amplification effect than under rated conditions. The identification of the x-axis inductance relies on the direct-axis current, whose signal-to-noise ratio is lower than that of the quadrature-axis signal. As its proportion decreases, the influence of the current differential amplification effect intensifies. The permanent magnet flux linkage is mainly affected by the rotational speed. Under medium-speed conditions, the rotational speed is lower than under rated conditions, thus its related proportion decreases. The identification errors of the proposed method and the traditional αβ method are both greater than the identification error under rated conditions. However, because the proposed method can effectively suppress the influence of system disturbances, the permanent magnet flux linkage identification performance of the proposed method under this condition is still significantly superior to the traditional αβ method. Table 3 shows that under medium-speed, medium-load conditions, the algorithm proposed in this invention still maintains excellent identification performance. Furthermore, from... Figure 12 , Figure 14 , Figure 16 It can be seen that the method proposed in this invention exhibits good stability of the identified values ​​and small fluctuation amplitude throughout the entire online identification process with all parameters. Figure 11 , Figure 13 , Figure 15 In the traditional αβ algorithm, the identified values ​​are not stable and fluctuate significantly throughout the identification process.

[0078] Table 3. Comparison of identification performance between traditional αβ method and proposed method under medium-speed and medium-load conditions.

[0079] Low-speed, light-load steady-state test The motor reference speed was set to 300 r / min and the torque to 3 N·m. A low-speed, light-load steady-state test was conducted to systematically evaluate the performance of the proposed identification algorithm under wide-range operating conditions. Under low-speed, light-load conditions, the system signal-to-noise ratio further decreased, and the impact of the current differential amplification effect on the identification performance in steady state was further exacerbated.

[0080] Figure 17 , Figure 19 , Figure 21 The waveforms identified by the traditional αβ algorithm are shown respectively. Figure 18 , Figure 20 , Figure 22 The identification waveforms of the proposed algorithm are shown in Table 4. Table 4 shows the comparison of the identification performance of the traditional αβ algorithm and the proposed algorithm, namely the comparison of identification error and fluctuation amplitude.

[0081] from Figure 17 and Figure 18 The comparison Figure 19 and Figure 20 The comparison Figure 21 and Figure 22 The comparison in Table 4 shows that, under low-speed and light-load conditions, the proposed algorithm outperforms the traditional αβ algorithm in all parameter identification performance. The identification errors and fluctuation amplitudes of all identification parameters are significantly smaller than those of the traditional αβ algorithm, and the superiority is even more pronounced compared to the previous two conditions. This is because the system signal-to-noise ratio further decreases under low-speed and light-load conditions, and the impact of the current differential amplification effect on identification performance is further exacerbated in steady-state conditions. In steady-state conditions, the proposed algorithm effectively suppresses the impact of system disturbances on identification performance, while in the traditional αβ algorithm, system disturbances directly reduce identification performance and increase identification error and fluctuation amplitude through the current differential amplification effect. Figure 17 , Figure 19 , Figure 21 The results show that the identification errors of the traditional αβ algorithm for resistors, direct-axis inductors, quadrature-axis inductors, x-axis inductors, and y-axis inductors are 38.9%, 21.9%, 20.7%, 67.7%, and 21.3%, respectively, with fluctuation amplitudes reaching 0.56 Ω, 5 mH, 1.4 mH, 3.8 mH, and 1.6 mH, respectively. Figure 18 , Figure 20 , Figure 22It can be seen that the identification errors of the direct-axis inductance, quadrature-axis inductance, x-axis inductance, and y-axis inductance proposed in this invention are only 0.63%, 3.1%, 7.4%, 14%, and 4.3%, respectively, with fluctuation amplitudes of only 0.12 Ω, 0.6 mH, 0.8 mH, 0.8 mH, and 0.6 mH, respectively. Under this condition, the identification errors of resistance and x-axis inductance in the traditional αβ algorithm are relatively large. This is because the resistance identification value is only affected by the current, and the current amplitude is small under light load, resulting in a low proportion of the resistance term, making it more susceptible to system disturbances. The x-axis inductance identification value is mainly affected by the direct-axis signal, which has a lower proportion than the quadrature-axis signal. Under low-speed and light-load conditions, the proportion of x-axis inductance-related terms is low, making it more susceptible to system disturbances. In addition, the nominal value of the x-axis inductance itself is small, which makes it easier to cause a large relative error. While the flux linkage term of permanent magnets is mainly affected by rotational speed, the influence of current differential amplification is relatively weak. However, under this operating condition, the permanent magnet flux linkage identification error and fluctuation amplitude of the algorithm proposed in this invention are both smaller than those of the traditional αβ algorithm. Even under low-speed, light-load conditions, the algorithm proposed in this invention maintains good identification performance, superior to the traditional αβ algorithm.

[0082] Table 4. Comparison of identification performance between traditional αβ method and proposed method under low-speed and light-load conditions.

[0083] The embodiments described above are only used to illustrate the technical ideas and features of the present invention. Their purpose is to enable those skilled in the art to understand the content of the present invention and implement it accordingly. The patent scope of the present invention should not be limited by these embodiments. That is, any equivalent changes or modifications made in accordance with the spirit disclosed in the present invention still fall within the patent scope of the present invention.

Claims

1. A method for online identification of all parameters of a dual three-phase permanent magnet synchronous motor without signal injection, characterized in that, Based on the six-phase topology of a dual-three-phase permanent magnet synchronous motor drive system, and according to the basic mathematical model of the dual-α-β frame, the equivalent inductance decoupling parameter model under the dual-α-β frame is derived as the initial model for full-parameter online identification based on the dual-α-β frame. Based on the synchronous rotating coordinate transformation to separate the current differential term in steady state, the initial model for full-parameter online identification is further reconstructed to obtain the reconstructed model for full-parameter online identification. A hierarchical architecture for full-parameter online identification is constructed, and the full-parameter online identification model based on the RLS algorithm is derived from the reconstructed model. The mapping mechanism between the dual-α-β frame and the VSD frame is used to achieve the equivalent transformation of the identification parameters, ultimately realizing full-parameter online identification without signal injection.

2. The online full-parameter identification method for a signal-injection-free dual three-phase permanent magnet synchronous motor according to claim 1, characterized in that, If the windings of a dual three-phase permanent magnet synchronous motor are considered as two independent sets of three-phase windings, then the mathematical model of the dual three-phase permanent magnet synchronous motor based on the dual α-β framework is expressed as follows: ； ； ； In the formula: This is the α-axis voltage of the first winding; This is the voltage across the α-axis of the second winding. This is the β-axis voltage of the first set of windings; This is the β-axis voltage of the second set of windings; For the α-axis current of the first set of windings; For the α-axis current of the second set of windings; This refers to the β-axis current of the first set of windings; For the β-axis current of the second set of windings; This is the first set of winding α-axis flux linkage; For the second set of windings, the α-axis flux linkage; This is the first set of winding β-axis flux linkage; For the second set of windings, the β-axis flux linkage; This is the inductance matrix under the double α-β framework; For permanent magnet flux linkage; This represents the average value of the winding's self-inductance; The amplitude of the second harmonic of the self-inductance; This represents the average value of the mutual inductance of the windings; This represents the amplitude of the second harmonic of the winding mutual inductance. Stator resistance; It is a differential operator; The rotor electrical angle is given.

3. The online full-parameter identification method for a signal-injection-free dual three-phase permanent magnet synchronous motor according to claim 2, characterized in that, The initial model for full-parameter online identification based on the dual α-β framework is shown below: ； ； ； ； ； ； In the formula: For each winding in a dual α-β frame, it is a direct-axis inductance; For the direct-axis mutual inductance between two sets of windings; For each winding in the dual α-β frame, the cross-axis inductance is used; This refers to the cross-axis mutual inductance between two sets of windings; The rotor's electric angular velocity; This is the inductance coefficient matrix; Let h be the element in the h-th row and v-th column of the inductance coefficient matrix; h = 1, 2, 3, 4, v = 1, 2, 3, 4; This is the differential of the α-axis current of the first winding; This is the differential of the β-axis current of the first winding; This is the differential of the α-axis current of the second winding; This is the differential of the β-axis current of the second winding.

4. The online full-parameter identification method for a signal-injection-free dual three-phase permanent magnet synchronous motor according to claim 3, characterized in that, The method for reconstructing the initial model from the fully parameterized online identification includes the following steps: set up Let be the transformation matrix between the dual dq synchronous coordinate system and the dual α-β coordinate system, then we have: ； ； In the formula: This refers to the d1-axis current in the dual dq synchronous coordinate system. This refers to the d2-axis current in the dual dq synchronous coordinate system. This refers to the q1-axis current in the dual dq synchronous coordinate system. This refers to the q2-axis current in the dual dq synchronous coordinate system. Based on the above equation, the current differential term under the double α-β frame is expressed as follows: ； In the formula: This is the differential of the d1-axis current in the dual synchronous coordinate system; This is the differential of the d2-axis current in the dual synchronous coordinate system; This is the differential of the q1-axis current in the dual synchronous coordinate system; This is the differential of the q2-axis current in the dual synchronous coordinate system; In steady state, the currents along the d and q axes of the dual synchronous rotating coordinate system are DC signals, and their current differential terms... , , , If the value is approximately 0 and can be ignored, then we have: ； Transformation matrix Substituting into the above equation, we further obtain: ； Substituting the simplified results into the initial model of the full-parameter online identification and further deriving, we obtain the following full-parameter online identification and reconstruction model of the dual three-phase permanent magnet synchronous motor: ； ； ； ； In the formula, matrix elements - , - , - , - They are represented as follows: ； ； ； ； In the formula: , This is the inductance coefficient matrix; The set q-axis current differential threshold; This is the actual value of the differential of the q-axis current; This represents the actual value of the differential of the q-axis current under steady-state conditions. This represents the actual value of the differential of the q-axis current under transient conditions. max() means to retrieve the maximum value; min() means taking the minimum value; , , , , , , , , , , , , , , , These are the elements in the transformation matrix.

5. The method for online identification of all parameters of a signal-injection-free dual three-phase permanent magnet synchronous motor according to claim 1, characterized in that, The method for constructing a fully parameterized online hierarchical identification architecture includes the following steps: When constructing a hierarchical architecture for full-parameter online identification, the criteria for dividing the identification parameters of each layer include physical characteristic criteria and observability criteria. Among them, the physical characteristic criterion is to group parameters with similar time-varying characteristics into the same group; the observability criterion is to avoid grouping parameters with linear correlation into the same group. Let the total number of parameters to be identified for the dual three-phase permanent magnet synchronous motor be 2. N or 2 N +1, dividing the identification parameter set into N +1 layer, where each layer identifies 2 parameters, and each layer represents a subset, specifically as follows: ； In the formula, , , , These correspond to the subsets of identification parameters for the first, second, Nth, and N+1th layers; All parameters to be identified are assigned to the first to Nth layer identification parameter subsets, so that the identification parameter items in the N+1th layer identification parameter subset are the same as the identification parameters in any of the first N layers. Different voltage equations are used for multiple identifications. The average value of the identification parameter items in the N+1th layer identification parameter subset and the same identification parameter items in the first N layers is taken as the final identification parameter value.

6. The online full-parameter identification method for a signal-injection-free dual three-phase permanent magnet synchronous motor according to claim 5, characterized in that, Suppose that all the parameters to be identified for the dual three-phase permanent magnet synchronous motor include: , , , , , ; For each winding in a dual α-β frame, it is a direct-axis inductance; For each winding in the dual α-β frame, the cross-axis inductance is used; For the direct-axis mutual inductance between two sets of windings; This refers to the cross-axis mutual inductance between two sets of windings; Stator resistance; For permanent magnet flux linkages; the identification parameter set is divided into the following four subsets of identification parameters: ； In the formula, , , , These are the first, second, third, and fourth layer identification parameter subsets, respectively. Each layer of identification parameter subset includes two parameters to be identified. The four layers of identification parameter subsets are represented as follows: ； In the formula, , To satisfy any two distinct identification parameter terms of the hierarchical core criterion; Make the fourth layer identification parameter the same as any of the first three layers, use different voltage equations for secondary identification, and average the fourth layer identification parameter with the same identification parameter in the first three layers to improve identification accuracy.

7. The method for online identification of all parameters of a signal-injection-free dual three-phase permanent magnet synchronous motor according to claim 6, characterized in that, The derivation of the fully parameterized online identification model based on the RLS algorithm from the fully parameterized online identification and reconstruction model includes the following steps: The reconfiguration identification model of the dual three-phase permanent magnet synchronous motor is mapped to the subset of identification parameters according to the following formula: ； In the formula: g To identify the voltage procedure number; Indicates belonging to n The first dimension of the real space g The voltage equation up to the first i ( g Mapping of a subset of parameters to be identified; This is the g-th identifying voltage equation; for n 3D real space; for g The defined value is 4; The index of the subset of parameters to be identified; The expression for the parameter subset index; For the first A subset of parameters to be identified; The identification parameters are represented as follows: ； In the formula: No. w Layer identification parameter subset; w =1, 2, ..., N , N +1; w To identify the parameter layer number; j for Identify the parameter item number in the middle; It is the first j Each identification parameter item; It is the first i Layer j Solve the equations for the inverse voltage function corresponding to each identified parameter; It is the equation for solving the inverse function of the third-level voltage; It is the equation for solving the inverse function of the fourth-level voltage; i ( j ) is the first i Layer j Each identification parameter item; For the first j The voltage equations corresponding to each identification parameter term; The basic iterative model of the RLS algorithm is as follows: ； In the formula: For time sequence number; For the first k Real-time identification value; For the first k -1 time step identification value; For the first k Time-matrix gain matrix; For the first k Output matrix at time step; For the first k Input matrix at each time step; For the first k The transpose of the input matrix at each time step; For the first k Time-varying covariance matrix; For the first k Covariance matrix at time -1; As a weighting factor; It is the identity matrix; Based on the basic iterative model of the RLS algorithm and the full-parameter online identification and reconstruction model of the dual three-phase permanent magnet synchronous motor, the following identification parameter matrix, input matrix, and output matrix are derived, where: The identification parameter matrix is ​​as follows: ； The input matrix is: ； In the formula, , As an intermediate variable, , The calculation formula is as follows: ； The output matrix is ​​represented as: ； In the formula, , , , The calculation formula is as follows: ； In the formula: To identify the parameter matrix; The input matrix; This is the output matrix; , , , Corresponding to the output matrix Y The elements in the first row and first column, the second row and second column, the third row and third column, and the fourth row and fourth column; The winding resistance is identified using the third voltage equation; The winding resistance is identified using the fourth voltage equation; For permanent magnet flux identified using the third voltage equation; For permanent magnet flux linkages identified using the fourth voltage equation; At the start of the identification process, the nominal parameters of the dual three-phase permanent magnet synchronous motor are used to initialize each identification equation to improve the convergence speed of the online identification. After the identified parameters converge, the identified values ​​are used to replace the initial values ​​in each identification equation, and the identification is performed alternately in each identification equation to update the parameters online.

8. The method for online identification of all parameters of a signal-injection-free dual three-phase permanent magnet synchronous motor according to claim 7, characterized in that, The method for achieving equivalent transformation of identification parameters by utilizing the mapping mechanism between parameters in the dual α-β framework and the VSD framework includes the following steps: The parameter identification values ​​under the VSD framework are transformed according to the following formula: ； In the formula: A vector composed of the identification values ​​of each parameter of the VSD framework; It is a vector composed of the identification values ​​of each parameter of the double α-β frame; Transformation matrix Represented as: 。 9. The method for online identification of all parameters of a signal-injection-free dual three-phase permanent magnet synchronous motor according to claim 1, characterized in that, During normal operation and control of a dual three-phase permanent magnet synchronous motor, the actual six-phase current collected is obtained by coordinate transformation according to the following formula to obtain the actual current in the synchronous rotating coordinate system: ； In the formula: This refers to the d-axis current under the VSD framework. This refers to the q-axis current under the VSD framework. The x-axis current under the VSD framework; This refers to the y-axis current within the VSD framework. This refers to the current in phase A. This refers to the B-phase current. This refers to the C-phase current. This refers to the current in phase D. This refers to the E-phase current. This refers to the F-phase current. This is the transformation matrix from the stationary coordinate system to the VSD frame synchronous coordinate system; The transformation matrix from the six-phase stationary coordinate system to the VSD frame stationary coordinate system; , They are represented as follows: ； ； In the formula: The rotor electrical angle; The differential of the actual q-axis current is calculated using the inverse Euler discretization method according to the following formula: ； In the formula: For the first k The actual value of the differential of the q-axis current at time t; For the first k Actual q-axis current at any given time; For the first k Actual q-axis current at time -1; The system sampling time; The specific calculation method for the inductance coefficient matrix is ​​selected based on the logical relationship between the actual value of the q-axis current derivative and the q-axis current derivative threshold, and then transmitted to the identification layer. The rotor position and speed information required for identification are sampled and imported into the full-parameter online identification equation. The dual α-β frame voltage information is calculated from the reference voltage output by the PI controller according to the following formula: ； In the formula: This is the α-axis voltage of the first winding; This is the voltage across the α-axis of the second winding. This is the β-axis voltage of the first set of windings; This is the β-axis voltage of the second set of windings; For the direct-axis voltage of the VSD control frame; The quadrature-axis voltage of the VSD control frame; The x-axis voltage of the VSD control frame; The y-axis voltage of the VSD control frame; The transformation matrix between the dual dq synchronous coordinate system and the dual α-β coordinate system is expressed as: 。 10. A device for online identification of all parameters of a signal-injection-free dual three-phase permanent magnet synchronous motor, comprising a memory and a processor, characterized in that, The memory is used to store a computer program; the processor is used to execute the computer program and, when executing the computer program, implement the steps of the online identification method for full parameters of a dual three-phase permanent magnet synchronous motor without signal injection as described in any one of claims 1 to 9.

Images