A method and device for identifying control mode and parameters of a doubly-fed wind turbine generator converter

Abstract

This invention relates to the field of new energy grid-connected control technology, and particularly to a method and apparatus for identifying control modes and parameters of a doubly-fed induction generator (DFIG) converter. The method involves applying small-amplitude step or pseudo-random voltage disturbances at the turbine terminal or grid connection point, collecting three-phase voltage and three-phase current data at the grid connection point, and obtaining the input vector and measurement output vector in a synchronous rotating coordinate system through Park transformation. An augmented state-space model library containing multiple potential control modes is established, and the PI parameters to be identified are incorporated into the augmented state. Unscented Kalman filters are run in parallel for each model to estimate the state / parameters, calculating the innovation residuals and their covariance. Based on Bayesian inference, the posterior probabilities of each model are recursively updated and normalized. The model with the highest probability is selected to determine the control mode, and the corresponding parameter estimates are output. This method does not require preset modes, improves identification accuracy and parameter physical consistency during mode switching or critical operating conditions, and facilitates stability analysis and operating condition migration. It has wide applicability.

Description

Technical Field

[0001] This invention relates to the field of new energy grid-connected control technology, and in particular to a method and device for identifying control modes and parameters of a doubly fed wind turbine converter. Background Technology

[0002] With the large-scale grid connection of new energy power sources, represented by doubly-fed induction generator (DFIG) wind turbines, the dynamic characteristics and stability mechanisms of the power system have undergone significant changes. Converter control strategies and their parameters directly impact low-frequency oscillations, subsynchronous / wideband oscillations, and transient stability during fault ride-through. In order to conduct stability verification, risk assessment, and control strategy formulation, grid operation and dispatch departments objectively need to master the control modes and key control parameters of wind turbine converters.

[0003] However, in practical engineering, the control of wind turbine converters is usually encapsulated by equipment manufacturers. Due to reasons such as commercial confidentiality, the details of the control topology and PI parameters are often difficult to obtain, which means that the grid side can only conduct "black box / grey box" identification based on external measurable quantities. Existing identification approaches mostly involve obtaining the output characteristics by applying disturbances or utilizing natural disturbances, then constructing an interpretable approximate model and optimizing the parameters to approximate the input-output response of the black box.

[0004] Currently, there are two main approaches to related technologies: one is parameter identification based on a preset single control mode, such as preset the converter to be in P / Q control or P / U control, and using time-domain / frequency-domain methods such as extended Kalman filtering and particle swarm optimization to minimize the error between the model output and the measured result; the other is matching and identification based on waveform features, which involves pre-establishing a multi-mode response waveform library, matching and determining the mode based on morphological features, sensitivity, or empirical rules, and then estimating the parameters.

[0005] However, single-mode presets heavily rely on prior assumptions. Once the preset mode differs from the actual mode, the obtained "parameters" may only be mathematical fitting coefficients, lacking physical meaning and difficult to use for stability analysis after condition transition; at the same time, optimization may fail to converge or reach local optima. Secondly, the effectiveness of waveform library matching depends on the comprehensiveness of the waveform library coverage and the reliability of the matching rules. When parameter perturbations are large, noise is strong, or in the critical region of multiple modes, waveform features are prone to aliasing, leading to mode misjudgment and subsequent parameter identification failure.

[0006] Therefore, there is an urgent need for a DFIG converter control mode and parameter identification method that can take into account both mode recognition and parameter estimation under unknown control mode conditions, and is robust to critical operating conditions and noise. Summary of the Invention

[0007] This invention provides a method and apparatus for identifying the control mode and parameters of a doubly fed wind turbine converter, which can effectively solve the problems in the background art.

[0008] To achieve the above objectives, the technical solution adopted by the present invention is as follows:

[0009] A method for identifying the control mode and parameters of a doubly-fed wind turbine converter includes the following steps:

[0010] S1: Obtain the inherent parameters of the generator, circuit, and converter of the doubly fed wind turbine unit to be identified, as well as the disturbance response sampling period;

[0011] S2: Apply voltage disturbance to the generator terminal or grid connection point of the object to be identified, and collect discrete sampling data of the three-phase voltage at the grid connection point and the three-phase output current of the unit during the disturbance period;

[0012] S3: Perform coordinate transformation on the collected three-phase voltage and three-phase current to obtain the voltage and current quantities in the synchronous rotating coordinate system, and construct the input vector and measurement output vector of the discrete time model accordingly.

[0013] S4: Establish a state-space model library containing multiple potential control modes, where each model corresponds to a control mode. Incorporate the control parameters to be identified into the model to form an augmented state, so that each model outputs the predicted result of the measurement output vector under the same input vector.

[0014] S5: Run the filter in parallel for each model in the state-space model library, perform state estimation and parameter estimation for each model, and calculate the residual and its statistics between the output prediction and the measurement output of each model.

[0015] S6: Based on the residuals and their statistics, calculate the matching probability between each model and the measurement data and perform normalization updates. Select the control mode corresponding to the model with the highest matching probability as the identified control mode.

[0016] S7: Output the parameter estimation results of the model corresponding to the control mode as the control parameter identification results.

[0017] Furthermore, in step S2, a voltage disturbance, which is a step signal or a pseudo-random sequence signal, is applied to the end of the object to be identified or the grid connection point, and the amplitude of the voltage disturbance is 0.05 pu.

[0018] Furthermore, in step S3, the coordinate transformation is a Parker transformation, using the grid voltage as a reference to obtain the voltage and current components in the synchronous rotating coordinate system, and the model input vector and measurement output vector are constructed as follows:

[0019] ;

[0020] ;

[0021] Among them, usd (k), u sq (k) represent the d-axis and q-axis stator voltage components at time k, respectively, i wd (k), i wq (k) represents the d-axis and q-axis current components of the unit output current at time k.

[0022] Furthermore, the potential control mode includes at least one or more of active / reactive power control, active power / voltage control, and speed / reactive power control;

[0023] The augmented state includes at least one or more of the following: stator flux linkage, rotor current, grid-side current, DC bus voltage, phase-locked loop angle, and controller integral stage state variables.

[0024] The control parameters to be identified include at least one or more of the following proportional / integral coefficients: the rotor-side converter inner loop proportional / integral coefficient, the rotor-side converter outer loop proportional / integral coefficient, the grid-side converter inner loop proportional / integral coefficient, the grid-side converter outer loop proportional / integral coefficient, and the phase-locked loop controller proportional / integral coefficient.

[0025] Furthermore, the parallel filter is an unscented Kalman filter (UKF), which estimates the state based on the posterior state at time k-1. Covariance Matrix Generate 2L+1 Sigma point vectors that satisfy:

[0026] ;

[0027] in, Let be the vector of the i-th Sigma point at time k-1 (i=1,2,...,2L). This is a scaling factor used to adjust the distribution range of Sigma points; The i-th column represents the square root of the matrix.

[0028] Furthermore, for any unscented Kalman filter (UKF) of model j, state prediction and mean and covariance calculations are performed on the Sigma point, including:

[0029] Calculate the predicted value of the i-th state vector at time k. :

[0030] ;

[0031] Calculate the mean of the 2L predicted state vectors at time k. Covariance Matrix :

[0032] ;

[0033] ;

[0034] in, , Let be the mean and variance weighting coefficients of the i-th Sigma point, respectively. The process noise covariance matrix is... The three-phase voltage measurement data collected at time k-1 are converted into voltage vectors in a synchronous rotating coordinate system with infinite grid voltage as the reference, u sd (k-1), u sq (k-1) represent the d-axis and q-axis stator voltage components at time k-1, respectively;

[0035] Measurement prediction of Sigma points and calculation of innovation residuals and their covariance include:

[0036] Calculate the predicted value of the i-th observation vector at time k. :

[0037] ;

[0038] Calculate the mean of the predicted values ​​of all observed vectors at time k. :

[0039] ;

[0040] Calculate the new residual at time k and its covariance :

[0041] ;

[0042] ;

[0043] in, , These are the nonlinear state transition function (reflecting different control topologies) and output measurement function corresponding to the j-th model, respectively. The three-phase current measurement data of the unit output at time k is converted into a current vector in a synchronous rotating coordinate system with infinite grid voltage as reference by Park transformation, i wd (k), i wq (k) represents the d-axis and q-axis current components of the unit output current at time k.

[0044] Furthermore, Bayesian inference is used to recursively update and normalize the matching probabilities between each model and the measurement data, including:

[0045] According to the new residual and its covariance Calculate the likelihood function of the model:

[0046] ;

[0047] Based on the posterior probability of the previous time step The posterior probability is obtained by recursively updating and normalizing:

[0048] ;

[0049] in, Let A and B represent the posterior probabilities of models A and B being the true models at time k.

[0050] Furthermore, the state-space model library includes at least model A and model B:

[0051] In Model A: The rotor-side converter is in active power / reactive power control mode, and the outer loop takes the active power and reactive power deviation as input to generate the rotor current reference value;

[0052] In Model B: The rotor-side converter is in speed / voltage control mode, and the outer loop uses the rotor speed deviation and terminal voltage deviation as inputs to generate the rotor current reference value;

[0053] In Model B, the state space equations at the rotor-side outer loop integrator state update equations are rewritten based on the speed / voltage control law, replacing the corresponding active / reactive control equations in Model A. The remaining electrical-side dynamic equations, inner loop current control equations, grid-side converter control equations, and output equations remain consistent with Model A.

[0054] Furthermore, the state-space model library includes at least model A and model B, and pattern discrimination is performed based on posterior probabilities:

[0055] when When the control mode corresponding to model A is determined, the state vector is output. The part of the parameters to be identified The result is used for identification; otherwise, it is determined to be the control mode corresponding to Model B and a state vector is output. The part of the parameters to be identified As a result of identification;

[0056] Model B, compared to Model A, only rewrites the integral state update related to the outer loop control objective into a speed / voltage control form, and satisfies:

[0057] ;

[0058] ;

[0059] in, and Let T represent the state variables of the integral element of the rotor-side converter d-axis outer loop controller at time k and time k-1, respectively. s Indicates the sampling period. A reference value indicating the rotor speed. This represents the actual rotor speed measured at time k-1. and Let K and K-1 represent the state variables of the integral element of the rotor-side converter q-axis outer loop controller, respectively. This indicates the reference value of the AC voltage at the machine terminals. This represents the actual terminal voltage amplitude measured at time k-1. and All of these represent process noise.

[0060] A control mode and parameter identification method for a doubly-fed wind turbine converter, used to implement the above method, includes:

[0061] The parameter configuration module is used to configure inherent parameters and sampling period.

[0062] The disturbance application module is used to apply voltage disturbances at the generator terminal or grid connection point;

[0063] The data acquisition module is used to acquire three-phase voltage and three-phase current.

[0064] The data preprocessing module is used to perform coordinate transformation and generate model input vectors and measurement output vectors;

[0065] The model library module is used to build an augmented state-space model library containing multiple potential control modes;

[0066] The parallel filtering module is used to estimate the residuals and their statistics in parallel for each model;

[0067] The probability update and decision module is used to calculate the matching probability based on the residual statistics and update the posterior probability, and output the control mode and corresponding parameter estimation results.

[0068] The technical solution of this invention can achieve the following technical effects:

[0069] This invention avoids the problem of parameters losing physical meaning due to mode mismatch by establishing a parallel competition library of state-space models containing multiple potential control modes, thereby improving the interpretability and transferability of the identification results. Parallel filters synchronously estimate the state or parameters of each model, and use the innovation residuals and their statistics for Bayesian probability updates, reducing reliance on human experience.

[0070] By constructing a likelihood function using statistical properties such as residual covariance, high mode discrimination and parameter estimation stability can be maintained even in the presence of noise, large parameter perturbations, or waveform aliasing in the critical region of multiple modes. Small-amplitude step jumps or pseudo-random voltage disturbances can be used to excite the control dynamics, reducing the impact on power grid operation and making it suitable for field tests or offline waveform recording data reproduction. Attached Figure Description

[0071] To more clearly illustrate the technical solutions in the embodiments of the present invention or the prior art, the drawings used in the description of the embodiments or the prior art will be briefly introduced below. Obviously, the drawings described below are only some embodiments recorded in the present invention. For those skilled in the art, other drawings can be obtained based on these drawings without creative effort.

[0072] Figure 1 This is a flowchart illustrating the method disclosed in this invention;

[0073] Figure 2 This is a topological structure diagram of an embodiment of the present invention. Detailed Implementation

[0074] The technical solutions of the present invention will be clearly and completely described below with reference to the accompanying drawings of the embodiments of the present invention. Obviously, the described embodiments are only some embodiments of the present invention, and not all embodiments.

[0075] Unless otherwise defined, all technical and scientific terms used herein have the same meaning as commonly understood by one of ordinary skill in the art to which this invention pertains. The terminology used in this specification is for the purpose of describing particular embodiments only and is not intended to be limiting of the invention. The term "and / or" as used herein includes any and all combinations of one or more of the associated listed items.

[0076] This invention discloses a method for identifying the control mode and parameters of a doubly-fed wind turbine converter, comprising the following steps:

[0077] S1: Obtain the inherent parameters of the generator, circuit, and converter of the doubly-fed induction generator (DFIG) to be identified, as well as the disturbance response sampling period; inherent parameters may include, but are not limited to: asynchronous generator stator resistance R. s Rotor resistance R r Stator inductance L s Rotor inductance L r Magnetizing inductance L m Grid-side converter filter resistor R f Filter inductor L f DC bus capacitor C; synchronous speed ω s Rotor speed ω rStator active power reference value Stator reactive power reference value DC bus voltage reference value Speed ​​reference value AC voltage reference value Sampling period T s Based on the capabilities of the measurement and control system, the dynamic resolution requirements of the disturbance response should be met. Configuring known inherent electrical parameters can significantly reduce the search space to be identified, making subsequent filtering estimation easier to converge.

[0078] S2: Apply a voltage disturbance to the generator terminal or grid connection point of the object to be identified. During the disturbance, collect discrete sampling data of the three-phase voltage at the grid connection point and the three-phase output current of the unit. The disturbance can be applied at the generator terminal or grid connection point, and the disturbance amplitude should be kept small to avoid triggering protection or causing excessive transitions in operating conditions. The collected signals should include at least the three-phase voltage u. a u b u c and three-phase current i a i b i c It can simultaneously collect auxiliary quantities such as rotational speed, DC bus voltage, and terminal voltage amplitude for model or measurement consistency verification.

[0079] Furthermore, a voltage disturbance, either a step signal or a pseudo-random sequence signal, is applied to the terminal of the object to be identified or the grid connection point, with an amplitude of 0.05 pu. The step disturbance can be set to a short-duration step, and the pseudo-random sequence can cover a wider frequency band to enhance the excitation of the control parameters. An amplitude of 0.05 pu is the recommended value, which can be fine-tuned according to the on-site noise level and measurement accuracy without affecting grid connection safety. Using a small-amplitude disturbance can fully excite the dynamics of the control link while reducing the impact on grid operation, thus improving the feasibility and safety of the project.

[0080] S3: Perform coordinate transformation on the collected three-phase voltage and three-phase current to obtain the voltage and current quantities in the synchronous rotating coordinate system, and construct the input vector and measurement output vector of the discrete time model accordingly.

[0081] In step S3, the coordinate transformation is a Parker transformation. Using the grid voltage as a reference, the voltage and current components in the synchronously rotating coordinate system are obtained, and the model input vector and measurement output vector are constructed as follows:

[0082] ;

[0083] ;

[0084] Among them, u sd (k), u sq(k) represent the d-axis and q-axis stator voltage components at time k, respectively, i wd (k), i wq (k) represents the d-axis and q-axis current components of the unit output current at time k.

[0085] In implementation, a combination of Clarke transform and Park transform can be used to achieve the abc→dq transformation; the angle of the synchronous rotating coordinate system can be estimated by a phase-locked loop (PLL) or with the phase of the infinite grid voltage as a reference. To reduce the impact of noise, anti-aliasing filtering and necessary digital filtering can be applied to the original sampled signal, but excessive phase delay should be avoided to prevent it from affecting the identification accuracy.

[0086] In the synchronous dq coordinate system, the coupling relationship between voltage / current dynamics and control loops is clearer, which facilitates the construction of a unified state-space model library and improves the separability of residual statistical discrimination.

[0087] S4: Establish a state-space model library containing multiple potential control modes, where each model corresponds to a control mode. Incorporate the control parameters to be identified into the model to form an augmented state, so that each model outputs the predicted result of the measurement output vector under the same input vector.

[0088] The potential control mode includes at least one or more of active / reactive power control, active power / voltage control, and speed / reactive power control; the augmented state includes at least one or more of the following: stator flux linkage, rotor current, grid-side current, DC bus voltage, phase-locked loop angle, and controller integral link state variables; the control parameter to be identified includes at least one or more of the following proportional / integral coefficients: rotor-side converter inner loop proportional / integral coefficient, rotor-side converter outer loop proportional / integral coefficient, grid-side converter inner loop proportional / integral coefficient, grid-side converter outer loop proportional / integral coefficient, and phase-locked loop controller proportional / integral coefficient.

[0089] Each model in the model library can be in discretized nonlinear state-space form. For the j-th mode (j=1, 2,...,N), a discretized nonlinear state-space model at time k is established:

[0090]

[0091] in, , Let be the augmented state vectors of the j-th model at time k and time (k-1), respectively, containing the system dynamic variables at time k and time (k-1). , (e.g., inductor current, integrator state) and the parameters to be identified at time k and time k-1. , ; The input vector (terminal voltage) at time k-1. The predicted output vector for the j-th model at time k; , These are the nonlinear state transition function (reflecting different control topologies) and output measurement function corresponding to the j-th model, respectively; Let j be the process noise vector at time k-1 of the j-th model. The noise vector measured at time k for the j-th model all follows a Gaussian distribution.

[0092] S5: Run the filter in parallel for each model in the state-space model library, perform state estimation and parameter estimation for each model, and calculate the residual and its statistics between the output prediction and the measurement output of each model.

[0093] The parallel filter is an unscented Kalman filter (UKF), which estimates the state based on the posterior state at time k-1. Covariance Matrix Generate 2L+1 Sigma point vectors that satisfy:

[0094] ;

[0095] in, Let λ be the vector of the i-th Sigma point at time k-1 (i=1,2,...,2L), and λ be the scaling factor used to adjust the distribution range of the Sigma points; The i-th column represents the square root of the matrix.

[0096] For any unscented Kalman filter (UKF) of model j, perform state prediction and mean and covariance calculations for the Sigma point, including:

[0097] Calculate the predicted value of the i-th state vector at time k. :

[0098] ;

[0099] Calculate the mean of the 2L predicted state vectors at time k. Covariance Matrix :

[0100] ;

[0101] ;

[0102] in, , Let be the mean and variance weighting coefficients of the i-th Sigma point, respectively. The process noise covariance matrix is... The three-phase voltage measurement data collected at time k-1 are converted into voltage vectors in a synchronous rotating coordinate system with infinite grid voltage as the reference, u sd (k-1), u sq (k-1) represent the d-axis and q-axis stator voltage components at time k-1, respectively;

[0103] Measurement prediction of Sigma points and calculation of innovation residuals and their covariance include:

[0104] Calculate the predicted value of the i-th observation vector at time k. :

[0105] ;

[0106] Calculate the mean of the predicted values ​​of all observed vectors at time k. :

[0107] ;

[0108] Calculate the new residual at time k and its covariance :

[0109] ;

[0110] ;

[0111] in, , These are the nonlinear state transition function (reflecting different control topologies) and output measurement function corresponding to the j-th model, respectively. The three-phase current measurement data of the unit output at time k is converted into a current vector in a synchronous rotating coordinate system with infinite grid voltage as reference by Park transformation, i wd (k), i wq (k) represents the d-axis and q-axis current components of the unit output current at time k. Let be the measurement noise covariance matrix of the j-th model at time k, used to characterize the uncertainty statistical properties of noise in the measurement process.

[0112] S6: Based on the residuals and their statistics, calculate the matching probability between each model and the measurement data and perform normalization updates. Select the control mode corresponding to the model with the highest matching probability as the identified control mode.

[0113] Bayesian inference is used to recursively update and normalize the matching probabilities between each model and the measurement data, including:

[0114] According to the new residual and its covariance Calculate the likelihood function of the model:

[0115] ;

[0116] Based on the posterior probability of the previous time step The posterior probability is obtained by recursively updating and normalizing:

[0117] ;

[0118] in, Let A and B represent the posterior probabilities of models A and B being the true models at time k.

[0119] S7: Output the parameter estimation results of the model corresponding to the control mode as the control parameter identification results. The output may include: control mode number / name, such as P / Q, P / U, ω / Q, etc., and the corresponding PI parameter vectors, such as the proportional and integral coefficients of the rotor-side inner and outer loops, the grid-side inner and outer loops, and the PLL, etc. It may also output parameter covariance or confidence intervals for subsequent small-signal stability analysis, operating condition transition verification, and model reliability assessment.

[0120] Furthermore, the state-space model library includes at least model A and model B:

[0121] In Model A: The rotor-side converter is in active power / reactive power control mode, and the outer loop takes the active power and reactive power deviation as input to generate the rotor current reference value;

[0122] In Model B: The rotor-side converter is in speed / voltage control mode, and the outer loop uses the rotor speed deviation and terminal voltage deviation as inputs to generate the rotor current reference value;

[0123] In Model B, the state space equations at the rotor-side outer loop integrator state update equations are rewritten based on the speed / voltage control law, replacing the corresponding active / reactive control equations in Model A. The remaining electrical-side dynamic equations, inner loop current control equations, grid-side converter control equations, and output equations remain consistent with Model A.

[0124] Models A and B share the same electrical side and inner loop control structure as much as possible, and are only distinguished by the integrator update rules related to the outer loop control objective. This ensures that the identifiability of the mode competition mainly comes from the difference in control objectives, and reduces numerical instability caused by excessive differences in model structure.

[0125] The state-space model library includes at least model A and model B, and pattern discrimination is performed based on posterior probabilities.

[0126] when When the control mode corresponding to model A is determined, the state vector is output. The part of the parameters to be identified The result is used for identification; otherwise, it is determined to be the control mode corresponding to Model B and a state vector is output. The part of the parameters to be identified As a result of identification;

[0127] Model B, compared to Model A, only rewrites the integral state update related to the outer loop control objective into a speed / voltage control form, and satisfies:

[0128] ;

[0129] ;

[0130] in, and Let T represent the state variables of the integral element of the rotor-side converter d-axis outer loop controller at time k and time k-1, respectively. s Indicates the sampling period. A reference value indicating the rotor speed. This represents the actual rotor speed measured at time k-1. and Let K and K-1 represent the state variables of the integral element of the rotor-side converter q-axis outer loop controller, respectively. This indicates the reference value of the AC voltage at the machine terminals. This represents the actual terminal voltage amplitude measured at time k-1. and All of these represent process noise.

[0131] In the specific implementation process, the inherent parameters of the doubly-fed wind turbine generator, circuit, converter, etc., are obtained in advance, including the stator resistance R of the asynchronous generator. s Rotor resistance R r Stator inductance L s Rotor inductance L r Magnetizing inductance L m Grid-side converter filter resistor R f Filter inductor L f DC bus capacitor C; synchronous speed ω s Rotor speed ω r Stator active power reference value Stator reactive power reference value DC bus voltage reference value Speed ​​reference value AC voltage reference value ; Disturbance response sampling period Ts.

[0132] Apply a small-amplitude (e.g., 0.05 pu) step signal or pseudo-random sequence voltage disturbance to the terminal of the model to be identified, and collect the three-phase voltage at the grid connection point during the disturbance period. and three-phase current Data. The collected data at time k is transformed into a voltage vector in a synchronous rotating coordinate system with an infinite grid voltage as the reference using the Park transformation. and current vector In the formula u sd (k), u sq (k) represent the d-axis and q-axis stator voltage components at time k, respectively, i wd (k), i wq (k) represents the d-axis and q-axis current components of the unit output current at time k.

[0133] Construct a competitive model library, pre-configuring two parallel models, whose state equations are described as follows:

[0134] ;

[0135] Model A:

[0136] ;

[0137] , The subscripts k and k-1 represent time k and time k-1, respectively;

[0138] , , These are the dq components of the stator flux linkage in the global dq coordinate system. , These represent the dq components of the rotor current in the global dq coordinate system. , Do not define the dq component of the grid-side converter output current in the global dq coordinate system. This is the DC bus voltage. The phase-locked loop output angle. , These are the state variables of the integral element of the inner loop controller for the dq-axis of the rotor-side converter. , These are the state variables of the integral element of the outer loop controller for the dq-axis of the rotor-side converter. For the state variables of the integral element of the phase-locked loop controller, , These are the state variables of the integral element of the inner loop controller of the dq axis of the grid-side converter. For the state variables of the integral element of the d-axis outer loop controller of the grid-side converter;

[0139] ,in , These are the proportional and integral coefficients of the inner loop controller of the rotor-side converter to be identified. , These are the proportional and integral coefficients of the outer loop controller of the rotor-side converter to be identified. , These are the proportional and integral coefficients of the inner loop controller of the grid-side converter to be identified. , These are the proportional and integral coefficients of the outer loop controller of the grid-side converter to be identified; , These are the proportional and integral coefficients of the phase-locked loop controller to be identified.

[0140] Equations of state It consists of the following equations:

[0141] ;

[0142] ;

[0143] ;

[0144] ;

[0145] ;

[0146] ;

[0147] ;

[0148] ;

[0149] ;

[0150] ;

[0151] ;

[0152] ;

[0153] ;

[0154] ;

[0155] ;

[0156] ;

[0157] In the formula, The leakage flux coefficient; The stator winding time constant; The rotor winding time constant; Slip rate; ~ These are the first 16 elements of the process noise vector at time k-1;

[0158] , These are the dq-axis components of the rotor voltage in the global dq coordinate system at time k-1, specifically:

[0159] ;

[0160] ;

[0161] , These are the reference values ​​for the dq-axis current of the rotor-side converter at time k-1, specifically:

[0162] ;

[0163] ;

[0164] , , These represent the stator output active power, reactive power, and grid-side converter output active power at time k-1, respectively:

[0165] ;

[0166] ;

[0167] ;

[0168] in, , Let be the stator current dq-axis components in the global dq coordinate system at time k-1, which can be represented by state variables as follows:

[0169] ;

[0170] ;

[0171] , Let dq be the dq-axis components of the grid-side converter output voltage in the global dq coordinate system at time k-1, which can be expressed as:

[0172] ;

[0173] ;

[0174] , These are the dq components of the grid-side converter inner loop current reference value at time k-1, specifically:

[0175] ;

[0176] ;

[0177] Output equation It can be represented by the following equation:

[0178] ;

[0179] ;

[0180] In the formula, , Measure the noise vector at time k The first and second elements;

[0181] Model B:

[0182] ;

[0183] State variables Output variables The output equation is the same as that of Model A. Model B is rewritten according to the speed / voltage control mode as follows:

[0184] ;

[0185] .

[0186] This invention further discloses a control mode and parameter identification method for a doubly-fed wind turbine converter, used to implement the above method, including:

[0187] The parameter configuration module is used to configure inherent parameters and sampling period.

[0188] The disturbance application module is used to apply voltage disturbances at the generator terminal or grid connection point;

[0189] The data acquisition module is used to acquire three-phase voltage and three-phase current.

[0190] The data preprocessing module is used to perform coordinate transformation and generate model input vectors and measurement output vectors;

[0191] The model library module is used to build an augmented state-space model library containing multiple potential control modes;

[0192] The parallel filtering module is used to estimate the residuals and their statistics in parallel for each model;

[0193] The probability update and decision module is used to calculate the matching probability based on the residual statistics and update the posterior probability, and output the control mode and corresponding parameter estimation results.

[0194] Although this application has been described in conjunction with specific features and embodiments, it is obvious that various modifications and combinations can be made thereto without departing from the spirit and scope of this application. Accordingly, this specification and drawings are merely exemplary illustrations of the application as defined herein, and are to be considered as covering any and all modifications, variations, combinations, or equivalents within the scope of this application. Clearly, those skilled in the art can make various alterations and modifications to this application without departing from its scope. Thus, if such modifications and modifications fall within the scope of this application and its equivalents, this application intends to include such modifications and modifications.

Claims

1. A method for identifying the control mode and parameters of a doubly-fed wind turbine converter, characterized in that, Includes the following steps: S1: Obtain the inherent parameters of the generator, circuit, and converter of the doubly fed wind turbine unit to be identified, as well as the disturbance response sampling period; S2: Apply voltage disturbance to the generator terminal or grid connection point of the object to be identified, and collect discrete sampling data of the three-phase voltage at the grid connection point and the three-phase output current of the unit during the disturbance period; S3: Perform coordinate transformation on the collected three-phase voltage and three-phase current to obtain the voltage and current quantities in the synchronous rotating coordinate system, and construct the input vector and measurement output vector of the discrete time model accordingly. S4: Establish a state-space model library containing multiple potential control modes, where each model corresponds to a control mode. Incorporate the control parameters to be identified into the model to form an augmented state, so that each model outputs the predicted result of the measurement output vector under the same input vector. S5: Run the filter in parallel for each model in the state-space model library, perform state estimation and parameter estimation for each model, and calculate the residual and its statistics between the output prediction and the measurement output of each model. S6: Based on the residuals and their statistics, calculate the matching probability between each model and the measurement data and perform normalization updates. Select the control mode corresponding to the model with the highest matching probability as the identified control mode. S7: Output the parameter estimation results of the model corresponding to the control mode as the control parameter identification results; The state-space model library includes at least model A and model B: In Model A: The rotor-side converter is in active power / reactive power control mode, and the outer loop takes the active power and reactive power deviation as input to generate the rotor current reference value; In Model B: The rotor-side converter is in speed / voltage control mode, and the outer loop uses the rotor speed deviation and terminal voltage deviation as inputs to generate the rotor current reference value; In Model B, the state space equations at the rotor-side outer loop integrator state update equations are rewritten based on the speed / voltage control law, replacing the corresponding active / reactive control equations in Model A. The remaining electrical-side dynamic equations, inner loop current control equations, grid-side converter control equations, and output equations remain consistent with Model A.

2. The method for identifying the control mode and parameters of a doubly-fed wind turbine converter according to claim 1, characterized in that, In step S2, a voltage disturbance, either a step signal or a pseudo-random sequence signal, is applied to the terminal of the object to be identified or the grid connection point, and the amplitude of the voltage disturbance is 0.05 pu.

3. The method for identifying the control mode and parameters of a doubly-fed wind turbine converter according to claim 1, characterized in that, In step S3, the coordinate transformation is a Parker transformation, which obtains the voltage and current components in the synchronous rotating coordinate system with the grid voltage as a reference, and constructs the model input vector and measurement output vector as follows: ； ； Among them, u sd (k), u sq (k) represent the d-axis and q-axis stator voltage components at time k, respectively, i wd (k), i wq (k) represents the d-axis and q-axis current components of the unit output current at time k.

4. The method for identifying the control mode and parameters of a doubly-fed wind turbine converter according to claim 1, characterized in that, The potential control modes include at least one or more of active / reactive power control, active power / voltage control, and speed / reactive power control; The augmented state includes at least one or more of the following: stator flux linkage, rotor current, grid-side current, DC bus voltage, phase-locked loop angle, and controller integral stage state variables. The control parameters to be identified include at least one or more of the following proportional / integral coefficients: the rotor-side converter inner loop proportional / integral coefficient, the rotor-side converter outer loop proportional / integral coefficient, the grid-side converter inner loop proportional / integral coefficient, the grid-side converter outer loop proportional / integral coefficient, and the phase-locked loop controller proportional / integral coefficient.

5. The method for identifying the control mode and parameters of a doubly-fed wind turbine converter according to claim 1, characterized in that, The parallel filter is an unscented Kalman filter (UKF), which estimates the state based on the posterior state at time k-1. Covariance Matrix Generate 2L+1 Sigma point vectors that satisfy: ； in, Let λ be the vector of the i-th Sigma point at time k-1, i=1,2,...,2L, and λ be the scaling factor used to adjust the distribution range of the Sigma points. The i-th column represents the square root of the matrix.

6. The method for identifying the control mode and parameters of a doubly-fed wind turbine converter according to claim 5, characterized in that, For any unscented Kalman filter (UKF) of model j, perform state prediction and mean and covariance calculations for the Sigma point, including: Calculate the predicted value of the i-th state vector at time k. : ； Calculate the mean of the 2L predicted state vectors at time k. Covariance Matrix : ； ； in, , Let be the mean and variance weighting coefficients of the i-th Sigma point, respectively. The process noise covariance matrix is... The three-phase voltage measurement data collected at time k-1 are converted into voltage vectors in a synchronous rotating coordinate system with infinite grid voltage as the reference, u sd (k-1), u sq (k-1) represent the d-axis and q-axis stator voltage components at time k-1, respectively; Measurement prediction of Sigma points and calculation of innovation residuals and their covariance include: Calculate the predicted value of the i-th observation vector at time k. : ； Calculate the mean of the predicted values ​​of all observed vectors at time k. : ； Calculate the new residual at time k and its covariance : ； ； in, , These are the nonlinear state transition function (reflecting different control topologies) and output measurement function corresponding to the j-th model, respectively. The three-phase current measurement data of the unit output at time k is converted into a current vector in a synchronous rotating coordinate system with infinite grid voltage as reference by Park transformation, i wd (k), i wq (k) represents the d-axis and q-axis current components of the unit output current at time k. Let be the measurement noise covariance matrix of the j-th model at time k.

7. The method for identifying the control mode and parameters of a doubly-fed wind turbine converter according to claim 6, characterized in that, Bayesian inference is used to recursively update and normalize the matching probabilities between each model and the measurement data, including: According to the new residual and its covariance Calculate the likelihood function of the model: ； Based on the posterior probability of the previous time step The posterior probability is obtained by recursively updating and normalizing: ； in, Let A and B represent the posterior probabilities of models A and B being the true models at time k.

8. The method for identifying the control mode and parameters of a doubly-fed wind turbine converter according to claim 7, characterized in that, The state-space model library includes at least model A and model B, and pattern discrimination is performed based on posterior probabilities: when When the control mode corresponding to model A is determined, the state vector is output. The part of the parameters to be identified As a result of identification; Otherwise, it is determined to be the control mode corresponding to Model B and the state vector is output. The part of the parameters to be identified As a result of identification; Model B, compared to Model A, only rewrites the integral state update related to the outer loop control objective into a speed / voltage control form, and satisfies: ； ； in, and Let T represent the state variables of the integral element of the rotor-side converter d-axis outer loop controller at time k and time k-1, respectively. s Indicates the sampling period. A reference value indicating the rotor speed. This represents the actual rotor speed measured at time k-1. and Let K and K-1 represent the state variables of the integral element of the rotor-side converter q-axis outer loop controller, respectively. This indicates the reference value of the AC voltage at the machine terminals. This represents the actual terminal voltage amplitude measured at time k-1. and All of these represent process noise.

9. A control mode and parameter identification method for a doubly-fed wind turbine converter, characterized in that, To implement the method according to any one of claims 1 to 8, comprising: The parameter configuration module is used to configure inherent parameters and sampling period. The disturbance application module is used to apply voltage disturbances at the generator terminal or grid connection point; The data acquisition module is used to acquire three-phase voltage and three-phase current. The data preprocessing module is used to perform coordinate transformation and generate model input vectors and measurement output vectors; The model library module is used to build an augmented state-space model library containing multiple potential control modes; The parallel filtering module is used to estimate the residuals and their statistics in parallel for each model; The probability update and decision module is used to calculate the matching probability based on the residual statistics and update the posterior probability, and output the control mode and corresponding parameter estimation results.

Images