Parameter identification method of doubly-fed induction motor based on adaptive extended Kalman filter

Abstract

The present invention discloses a method for parameter identification of a doubly-fed induction motor based on an adaptive extended Kalman filter, which addresses the problems of low accuracy and integrity, noise sensitivity, and other issues in the prior art of doubly-fed induction motor parameter identification. Based on the dq-axis flux equation and voltage equation of the doubly-fed induction motor, the present invention constructs a state space model with the stator and rotor dq-axis currents as state variables, expands the dimension of the state space model, utilizes the adaptive extended Kalman filter theory, and combines the rank criterion to determine the motor resistance R s 、R r and inductor L s , L r , L m The full observability of all parameters enables full parameter identification of the motor inductance and resistance, and enhances system robustness through adaptive adjustment of noise covariance. This invention effectively eliminates the effects of parameter coupling and improves identification accuracy. The adaptive noise suppression mechanism enhances anti-interference capabilities, meeting the requirements for dynamic online identification of doubly-fed motor parameters.

Description

Technical Field

[0001] The present invention relates to the field of online parameter identification of motors, and in particular to a parameter identification method of a doubly-fed induction motor based on an adaptive extended Kalman filter. Background Art

[0002] Doubly-fed induction motors (DFIGs) have become a mainstream wind turbine generator due to their wide speed range, efficient energy conversion, and flexible power control. However, their operation is affected by complex operating conditions such as wind speed fluctuations and grid disturbances, necessitating real-time dynamic modeling for optimal control. Motor parameters drift with factors such as temperature, magnetic saturation, and aging. Static models identified offline cannot accurately reflect actual operating conditions, necessitating online parameter updates. The stator and rotor windings of a DFIG are coupled via a converter, and the magnetic field exhibits nonlinearities, making it difficult for traditional linear models to accurately describe its dynamic behavior. In actual operation, measurement noise is significant, and traditional Kalman filters lack robustness.

[0003] Compared with the prior art, the differences are as follows:

[0004] Chinese published application CN117978016A proposes a parameter identification and control method for permanent magnet synchronous motors. This method uses a gradient descent method to first obtain the inductance value, then employs an extended Kalman filter to parameterize the resistance and flux linkage. This method requires step-by-step identification of the motor's electrical parameters, resulting in high identification latency and significant error accumulation.

[0005] Chinese published application CN118631112A proposes a parameter identification method using the mean square error of torque as an evaluation function. This method employs a Kalman filter algorithm to identify the inductance, flux linkage, and resistance of a permanent magnet synchronous motor. This method, combined with the evaluation function model, predicts the current and torque at the next moment. However, this application suffers from poor anti-interference capabilities and large torque measurement errors.

[0006] Chinese published application CN111669093A proposes a brushless DC motor parameter identification and control method, establishing a mathematical model based on the αβ axes. However, this application suffers from the problem that the current in the mathematical modeling of the αβ axes is AC, resulting in strong cross-coupling effects, large system disturbances, and poor self-interference rejection.

[0007] Chinese published application CN117978016A proposes a parameter identification and control method for a permanent magnet synchronous motor. The identification object of this application is a permanent magnet synchronous motor. The state equation structure is simple, and the control method is not suitable for a highly complex doubly fed induction motor.

[0008] China's published application CN111293693A proposes a method for identifying control parameters of a doubly-fed wind turbine converter based on an extended Kalman filter. This application combines a mathematical model of a doubly-fed induction motor to identify the control parameters of the doubly-fed induction motor. However, the application has the problem of being unable to identify the electrical parameters of the doubly-fed induction motor. Summary of the Invention

[0009] In response to the shortcomings of the existing technology, the present invention proposes a doubly fed induction motor parameter identification method based on adaptive extended Kalman filtering. Based on the state space equation of the doubly fed induction motor with the stator and rotor dq-axis currents as state variables, the adaptive extended Kalman filtering algorithm theory is utilized to achieve high-precision and low-complexity parameter identification. The core of the present invention lies in the interconnected coupling of parameter identification and adaptive filtering optimization.

[0010] The purpose of the present invention can be achieved through the following technical solutions:

[0011] The method for parameter identification of a doubly-fed induction motor based on an adaptive extended Kalman filter comprises the following steps:

[0012] 1) Build a state-space model of a doubly-fed induction motor with the stator and rotor dq-axis currents as state variables;

[0013] 2) Based on the state-space equation of the doubly-fed induction motor, where the stator and rotor dq-axis currents are state variables, the parameters to be identified are incorporated into the state variables, the dimension of the motor state-space model is expanded, and the observability of the parameters is determined;

[0014] 3) Using the adaptive extended Kalman filter theory and the parameter iterative identification method, the stator resistance R of the doubly fed induction motor is realized. s , rotor resistance R r , stator inductance L s , rotor inductance L r and the stator-rotor mutual inductance L m Dynamic online identification.

[0015] As a further improvement of the present invention, the step 1) is specifically as follows:

[0016] S11. According to the voltage equation and flux equation of the doubly fed induction motor, convert the flux differential term into the current differential term and ignore the stator resistance R s , rotor resistance R r , stator inductance L s , rotor inductance L r and the stator-rotor mutual inductance L m The differential term of , where t is time, the mathematical relationship between current differential and flux differential is obtained:

[0017] Magnetic flux equation:

[0018] , , , ;

[0019] Voltage equation:

[0020] , , , ;

[0021] In the formula 、 、 、 、 、 、 、 、 、 、 、 They are stator d-axis current, stator q-axis current, rotor d-axis current and rotor q-axis current, stator d-axis voltage, stator q-axis voltage, rotor d-axis voltage and rotor q-axis voltage, stator d-axis flux, stator q-axis flux, rotor d-axis flux and rotor q-axis flux, is the stator electrical angular velocity, is the rotor electrical angular velocity, 、 、 are the stator inductance, rotor inductance and mutual inductance, 、 is the stator and rotor resistance; known:

[0022] ;

[0023] Obtain the relationship between the flux differential term and the current differential term,

[0024] ;

[0025] S12. Extract the magnetic flux differential term in the voltage equation and make the slip ,

[0026] ;

[0027] S13, bring the flux differential term of S12 into S11 to eliminate the flux,

[0028] ;

[0029] S14. Write the state equations with the stator and rotor dq axis currents of the doubly fed induction motor as state variables.

[0030] .

[0031] As a further improvement of the present invention, the step 2) is specifically as follows:

[0032] S21. Construct the state equation F:

[0033] ;

[0034] X=[i ds ,i qs ,i dr ,i qr ,R s ,R r ,L s ,L r ,L m ] T ; ;

[0035] ;

[0036] The X is the state variable, the A is the Jacobian matrix, and the H is the measurement matrix;

[0037] S22. Use rank criterion to judge:

[0038] ;

[0039] The observability of the system state variables meets the standards.

[0040] As a further improvement of the present invention, the step 3) is specifically as follows:

[0041] S31. Initialize the extended Kalman filter algorithm:

[0042] Set the initial state vector and the initial error covariance matrix P0; setting the initial values ​​of the process noise covariance matrix Q0 and the observation noise covariance matrix R0; the are the initial value of stator d-axis current, the initial value of stator q-axis current, the initial value of rotor d-axis current and the initial value of rotor q-axis current, the initial value of stator inductance, the initial value of rotor inductance and the initial value of mutual inductance, the initial value of stator resistance and the initial value of rotor resistance;

[0043] S32, Status Prediction:

[0044] Based on the current state estimate =[i ds ,i qs ,i dr ,i qr ,Rs ,R r ,L s ,L r ,L m ] T and nine-dimensional equation of state , predict the next moment state : ; is the current state estimate, is the estimated value of the state at the next moment at the current moment, and the Δt is the sampling time interval; update the prediction error covariance matrix :

[0045] , is the noise covariance matrix at that moment;

[0046] The is the covariance matrix of the current moment to the next moment,

[0047] ;

[0048] is the Jacobian matrix at the current moment; Find the partial derivative of the state equation F with respect to the state variable X at the current moment;

[0049] S33. Observation update:

[0050] Get the actual observation value at the next moment ; Calculate Kalman gain :

[0051]

[0052] The is the actual observation value at the next moment, is the Kalman gain at the next moment, and H is the observation matrix; the modified state estimate ;

[0053] ;

[0054] The The state estimate for the next moment is updated; the error covariance matrix is ​​updated ;

[0055] ;

[0056] The The covariance matrix of the next moment to the next moment, is the covariance matrix from the current moment to the next moment.

[0057] S34. Adaptively adjust the noise covariance matrix:

[0058] According to the residual sequence , dynamically update the observation noise covariance matrix :

[0059] ;

[0060] The To dynamically update the observation noise covariance matrix.

[0061] The α is a smoothing factor, and its value range is ; Adjust the process noise covariance matrix based on the parameter estimate change rate :

[0062] ;

[0063] The is the adjustment coefficient, and Q0 is the initial process noise covariance;

[0064] S35, loop iteration: repeat steps S2 to S4, and output identification parameters in real time .

[0065] As a further improvement of the present invention, the residual sequence in the noise covariance matrix is ​​adaptively adjusted in step 3). The calculation of includes a sliding window mechanism, specifically: calculating its mean and covariance to suppress mutation noise interference; the window size is adaptively adjusted according to the dynamic response speed of the system, and the window size is the most recent N residual data.

[0066] As a further improvement of the present invention, the values ​​of the smoothing factor α and the adjustment coefficient β in step 3) are calibrated through offline experiments, or dynamically optimized online according to the parameter convergence speed.

[0067] Compared with the prior art, the present invention has the following beneficial effects:

[0068] First, by eliminating parameter coupling and combining it with adaptive extended Kalman filter algorithm optimization, the identification accuracy of resistance and inductance is significantly improved;

[0069] Secondly, the adaptive noise covariance adjustment mechanism simplifies the model complexity and combines with the sliding window mechanism to optimize the computational load and effectively suppress measurement noise interference.

[0070] Thirdly, it can skip the difficulty of magnetic flux observation, reduce the observation cost and identification difficulty;

[0071] Fourthly, online identification of motor parameters is realized. BRIEF DESCRIPTION OF THE DRAWINGS

[0072] Figure 1 This is a technical flowchart for the state space equation of a doubly-fed induction motor based on the stator and rotor dq-axis currents as state variables and its adaptive extended Kalman parameter identification method;

[0073] Figure 2 Flowchart for modeling the state space equations of a doubly fed induction motor with the stator and rotor dq axis currents as state variables;

[0074] Figure 3 This is the observability judgment flow chart;

[0075] Figure 4 Adaptive extended Kalman algorithm for online identification of the stator resistance R of the doubly fed induction motor s , rotor resistance R s , stator inductance L s , rotor inductance L r and the stator-rotor mutual inductance L m flow chart;

[0076] Figure 5 R s Simulation result diagram;

[0077] Figure 6 R s Simulation result diagram;

[0078] Figure 7 For L s Simulation result diagram;

[0079] Figure 8 For L r Simulation result diagram;

[0080] Figure 9 For L m Simulation result diagram. DETAILED DESCRIPTION

[0081] The present invention is further described in detail below with reference to the accompanying drawings and specific embodiments:

[0082] Combine Figure 1 As shown in the figure, the state space equation of the doubly fed induction motor based on the stator and rotor dq axis current of the doubly fed induction motor and its adaptive extended Kalman parameter identification method are constructed by building the state equation with the stator and rotor dq axis current of the doubly fed induction motor as the state variables, and building the [i ds ,i qs ,i dr ,i qr ,R s ,R r ,L s ,Lr ,L m ] is the state equation F of the state variable, the state variable is expanded to nine dimensions, and the rank criterion is used to judge the observability of the system according to the observation matrix H and Jacobian matrix A. ds ,i qs ,i dr ,i qr ,R s ,R r ,L s ,L r ,L m ] is the state variable of the adaptive extended Kalman filter algorithm to identify the stator resistance R s , rotor resistance R r , stator inductance L s , rotor inductance L r and the stator-rotor mutual inductance L m .

[0083] Combine Figure 2 As shown, the state space equation modeling of the doubly fed induction motor with the stator and rotor dq axis currents as state variables includes the following steps:

[0084] S1. According to the voltage equation and flux equation of the doubly fed induction motor, the flux differential term is converted into the current differential term, and the stator resistance R is ignored. s , rotor resistance R r , stator inductance L s , rotor inductance L r and the stator-rotor mutual inductance L m The differential term of , we can get the mathematical relationship between current differential and flux differential: flux equation:

[0085] , , , .

[0086] Voltage equation:

[0087] , , , , where 、 、 、 、 、 、 、 They are stator d-axis voltage, stator q-axis voltage, rotor d-axis voltage and rotor q-axis voltage, stator d-axis flux, stator q-axis flux, rotor d-axis flux and rotor q-axis flux, respectively. is the stator electrical angular velocity, is the rotor electrical angular velocity, 、 、 are the stator inductance, rotor inductance and mutual inductance, 、 is the stator and rotor resistance; known:

[0088] ; Obtain the relationship between the flux differential term and the current differential term,

[0089] ;

[0090] S2. Extract the magnetic flux differential term in the voltage equation and let ,

[0091] ;

[0092] S3, bring the magnetic flux differential term of S2 into S1 to eliminate the magnetic flux,

[0093] ;

[0094] S4. Write the state equation with the stator and rotor dq axis currents of the doubly fed induction motor as state variables.

[0095] ;

[0096] Combine Figure 3 As shown, the state space equation of the doubly fed induction motor based on the stator and rotor dq axis currents of the doubly fed induction motor as state variables incorporates the parameters to be identified into the state variables, expands the dimension of the motor state space model, and determines the observability of the parameters, including the following steps:

[0097] S1. Build the state equation F:

[0098] ;

[0099] X=[i ds ,i qs ,i dr ,i qr ,R s ,R r ,L s ,L r ,L m ] T ;

[0100] ;

[0101] ;

[0102] S2. Use rank criterion to judge:

[0103] ;

[0104] The observability of system state variables meets the standards;

[0105] Combine Figure 4 As shown, the adaptive extended Kalman filter theory is used to adopt the parameter iterative identification method to realize the stator resistance R of the doubly fed induction motor. s , rotor resistance R r , stator inductance L s , rotor inductance L r and the stator-rotor mutual inductance L m The dynamic online identification includes the following steps:

[0106] S1. Initialize the extended Kalman filter algorithm:

[0107] Set the initial state vector and the initial error covariance matrix P0; set the initial values ​​of the process noise covariance matrix Q0 and the observation noise covariance matrix R0;

[0108] S2, state prediction: based on the current state estimation =[i ds ,i qs ,i dr ,i qr ,R s ,R r ,L s ,L r ,L m ] T and nine-dimensional equation of state , predict the next moment state :

[0109] ;

[0110] The Δt is the sampling time interval; update the prediction error covariance matrix :

[0111] ;

[0112] The Jacobian matrix;

[0113] S3. Observation update: Get actual observation value ; Calculate Kalman gain :

[0114] ;

[0115] The H is the observation matrix; the modified state estimate :

[0116] ;

[0117] Update the error covariance matrix :

[0118] ;

[0119] S4. Adaptively adjust the noise covariance matrix:

[0120] According to the residual sequence , dynamically update the observation noise covariance matrix :

[0121] ;

[0122] The α is a smoothing factor, and its value range is ; Adjust the process noise covariance matrix based on the parameter estimate change rate :

[0123] ;

[0124] The is the adjustment coefficient, and Q0 is the initial process noise covariance;

[0125] S5, loop iteration: repeat steps S2 to S4, and output identification parameters in real time .

[0126] Combine Figures 5 to 9 As shown, the adaptive extended Kalman algorithm is used to identify the parameters , change L in 0.3 seconds r The value of L changes in 0.5 seconds s The value of , the identification result is high in accuracy, the response speed is fast, and the tracking parameters are accurate.

[0127] The above description is merely a preferred embodiment of the present invention and does not constitute any other form of limitation to the present invention. Any modification or equivalent variation based on the technical essence of the present invention shall still fall within the scope of protection claimed by the present invention.

Claims

1. A parameter identification method for a doubly-fed induction motor based on an adaptive extended Kalman filter is characterized in that: The following steps are involved: 1) Build a state-space model of a doubly-fed induction motor with the stator and rotor dq-axis currents as state variables; The step 1) is specifically as follows: S11. According to the voltage equation and flux equation of the doubly fed induction motor, convert the flux differential term into the current differential term and ignore the stator resistance R s , rotor resistance R r , stator inductance L s , rotor inductance L r and the stator-rotor mutual inductance L m The differential term of , where t is time, the mathematical relationship between current differential and flux differential is obtained: Magnetic flux equation: ， ， ， ； Voltage equation: ， ， ， ; In the formula 、 、 、 、 、 、 、 、 、 、 、 They are stator d-axis current, stator q-axis current, rotor d-axis current and rotor q-axis current, stator d-axis voltage, stator q-axis voltage, rotor d-axis voltage and rotor q-axis voltage, stator d-axis flux, stator q-axis flux, rotor d-axis flux and rotor q-axis flux, is the stator electrical angular velocity, is the rotor electrical angular velocity, 、 、 are the stator inductance, rotor inductance and mutual inductance, 、 is the stator and rotor resistance; Known: ； Obtain the relationship between the flux differential term and the current differential term, ； S12. Extract the magnetic flux differential term in the voltage equation and make the slip , ； S13, bring the flux differential term of S12 into S11 to eliminate the flux, ； S14. Write the state equations with the stator and rotor dq axis currents of the doubly fed induction motor as state variables. ； 2) Based on the state-space equation of the doubly-fed induction motor, where the stator and rotor dq-axis currents are state variables, the parameters to be identified are incorporated into the state variables, the dimension of the motor state-space model is expanded, and the observability of the parameters is determined; 3) Using the adaptive extended Kalman filter theory and the parameter iterative identification method, the stator resistance R of the doubly fed induction motor is realized. s , rotor resistance R r , stator inductance L s , rotor inductance L r and the stator-rotor mutual inductance L m Dynamic online identification.

2. The method for parameter identification of a doubly-fed induction motor based on adaptive extended Kalman filtering according to claim 1, characterized in that: The step 2) is specifically as follows: S21. Construct the state equation F: ； X=[i ds ,i qs ,i dr ,i qr ,R s ,R r ,L s ,L r ,L m ] T ; ; ； The X is the state variable, the A is the Jacobian matrix, and the H is the measurement matrix; S22. Use rank criterion to judge: ； The observability of the system state variables meets the standards.

3. The method for parameter identification of a doubly-fed induction motor based on adaptive extended Kalman filtering according to claim 2, characterized in that: The step 3) is specifically as follows: S31. Initialize the extended Kalman filter algorithm: Set the initial state vector and the initial error covariance matrix P0; Set the initial values ​​of the process noise covariance matrix Q0 and the observation noise covariance matrix R0; are the initial value of stator d-axis current, the initial value of stator q-axis current, the initial value of rotor d-axis current and the initial value of rotor q-axis current, the initial value of stator inductance, the initial value of rotor inductance and the initial value of mutual inductance, the initial value of stator resistance and the initial value of rotor resistance; S32, Status Prediction: Based on the current state estimate =[i ds ,i qs ,i dr ,i qr ,R s ,R r ,L s ,L r ,L m ] T and nine-dimensional equation of state , predict the next moment state : ; is the current state estimate, is the estimated value of the state at the next moment at the current moment, and the Δt is the sampling time interval; update the prediction error covariance matrix : , is the noise covariance matrix at that moment; The is the covariance matrix of the current moment to the next moment, ； is the Jacobian matrix at the current moment; Find the partial derivative of the state equation F with respect to the state variable X at the current moment; S33. Observation update: Get the actual observation value at the next moment ; Calculate Kalman gain : ； The is the actual observation value at the next moment, is the Kalman gain at the next moment, and H is the observation matrix; the modified state estimate : ； The The state estimate for the next moment is updated; the error covariance matrix is ​​updated : ； The The covariance matrix of the next moment to the next moment, is the covariance matrix from the current moment to the next moment; S34. Adaptively adjust the noise covariance matrix: According to the residual sequence , dynamically update the observation noise covariance matrix : ； The To dynamically update the observation noise covariance matrix; The α is a smoothing factor, and its value range is ; Adjusting the process noise covariance matrix based on the parameter estimation change rate : ； The is the adjustment coefficient, and Q0 is the initial process noise covariance; S35, loop iteration: repeat steps S2 to S4, and output identification parameters in real time .

4. The method for parameter identification of a doubly-fed induction motor based on adaptive extended Kalman filtering according to claim 3, characterized in that: In step 3), the residual sequence in the noise covariance matrix is ​​adaptively adjusted. The calculation of includes a sliding window mechanism, specifically: calculating its mean and covariance to suppress mutation noise interference; the window size is adaptively adjusted according to the dynamic response speed of the system, and the window size is the most recent N residual data.

5. The method for parameter identification of a doubly-fed induction motor based on adaptive extended Kalman filtering according to claim 3, characterized in that: The values ​​of the smoothing factor α and the adjustment coefficient β in step 3) are calibrated through offline experiments, or dynamically optimized online according to the parameter convergence speed.

Images