All-working-condition impedance identification method and device for doubly-fed wind turbine generator

Abstract

The invention belongs to the field of electric power, and discloses a double-fed wind turbine generator all-working-condition impedance identification method and device, and the method comprises the steps: determining common features of synchronous control links and electrical control links of different double-fed wind turbine generators, and generating an impedance expression of the double-fed wind turbine generator according to the common features, the impedance expression is used for representing difference and common characteristics of frequency physical characteristics of different doubly-fed wind turbine generators; constructing an impedance identification model according to the impedance expression; generating an impedance characteristic regularization constraint condition according to the frequency physical characteristics of the doubly-fed wind turbine generator, and training the impedance identification model based on the regularization constraint condition; and unknown working condition parameters are input into the trained impedance identification model to obtain the impedance amplitude and phase of the doubly-fed wind turbine generator under the corresponding working condition.

Description

Technical Field

[0001] This invention belongs to the field of power, and in particular relates to a method and device for impedance identification of doubly fed wind turbine generators under all operating conditions. Background Technology

[0002] Doubly-fed induction generator (DFIG) wind turbines have been widely integrated into the power grid due to their high wind energy utilization efficiency and cost-effectiveness. However, as the penetration rate of wind turbines in the power grid continues to increase, instability issues such as subsynchronous / supersynchronous oscillations and harmonic resonances have arisen, seriously affecting the safe and reliable operation of the power grid. Impedance-based stability analysis methods have been widely used to analyze the stability margin of the interconnection system between DFIG wind turbines and the power grid. Therefore, identifying the full operating point impedance of DFIG wind turbines is crucial for analyzing the stability of DFIG-grid interconnection systems considering various operating conditions.

[0003] In related technologies, impedance identification methods mostly rely on extensive parameter tuning experiments to find the optimal neural network structure design, including the number of hidden layers, the number of neurons per layer, the learning rate, and the regularization coefficient. This results in huge model debugging costs and generates millions of training data requirements. Because different new energy generator units have varying impedance characteristics, and the neural network structure parameters are numerous, it is difficult to match neural network parameters that form a strict physical mapping relationship with the impedance characteristics. This reduces model identification performance and increases parameter design costs and training data requirements. Summary of the Invention

[0004] In view of this, the present invention discloses a method and device for impedance identification of doubly fed wind turbines under all operating conditions, which can solve the shortcomings of related technologies.

[0005] To achieve the above objectives, the present invention discloses the following technical solution: According to a first aspect of the present invention, a method for impedance identification of a doubly-fed induction generator (DFIG) under all operating conditions is proposed, the method comprising: The common characteristics of the synchronous control and electrical control links of different doubly-fed wind turbine units are determined, and the impedance expression of the doubly-fed wind turbine units is generated based on the common characteristics. The impedance expression is used to characterize the differences and common characteristics of the frequency physical characteristics of different doubly-fed wind turbine units. Construct an impedance identification model based on the impedance expression; Impedance characteristic regularization constraints are generated based on the frequency physical characteristics of the doubly fed wind turbine, and the impedance identification model is trained based on the regularization constraints. The unknown operating condition parameters are input into the impedance identification model after training to obtain the impedance amplitude and phase of the doubly fed wind turbine under the corresponding operating conditions.

[0006] According to a second aspect of the present invention, a full-condition impedance identification device for a doubly-fed wind turbine is provided, the device comprising: Determining Unit: Determine the common characteristics of the synchronous control and electrical control links of different doubly-fed wind turbine units, and generate the impedance expression of the doubly-fed wind turbine units based on the common characteristics. The impedance expression is used to characterize the differences and common characteristics of the frequency physical characteristics of different doubly-fed wind turbine units. Construction Unit: Construct an impedance identification model based on the impedance expression; Training unit: Generates impedance characteristic regularization constraints based on the frequency physical characteristics of the doubly fed wind turbine, and trains the impedance identification model based on the regularization constraints; Identification Unit: Input the unknown operating condition parameters into the impedance identification model after training to obtain the impedance amplitude and phase of the doubly fed wind turbine under the corresponding operating conditions.

[0007] According to a third aspect of the present invention, an electronic device is provided, comprising: processor; Memory used to store processor-executable instructions; The processor implements the steps of the method as described in the first aspect by running the executable instructions.

[0008] According to a fourth aspect of the invention, a computer-readable storage medium is provided having computer instructions stored thereon that, when executed by a processor, implement the steps of the method as described in the first aspect.

[0009] As can be seen from the above technical solutions, the full-condition impedance identification method for doubly-fed wind turbines disclosed in this invention is as follows: On the one hand, by deriving the common characteristics of doubly fed wind turbines with different control structures and parameters, the frequency physical characteristics of doubly fed generators were revealed and the establishment of neural network identification models was guided, reducing the parameter design cost of the models. On the other hand, a full-condition impedance identification model of doubly fed wind turbines was established through data-driven methods, and model regularization constraints were designed based on physical characteristics, which improved the accuracy of the models and reduced the amount of data required for model training. Attached Figure Description

[0010] Figure 1 This is a flowchart of an exemplary embodiment of a method for impedance identification of a doubly-fed wind turbine under all operating conditions; Figure 2 This is a schematic diagram of an impedance identification model provided in an exemplary embodiment; Figure 3 This is a schematic diagram illustrating the relationship between mean squared error and the number of training iterations, provided in an exemplary embodiment. Figure 4This is a schematic structural diagram of a device provided in an exemplary embodiment; Figure 5 This is a block diagram of an impedance identification device for a doubly fed wind turbine under all operating conditions, provided in an exemplary embodiment. Detailed Implementation

[0011] Exemplary embodiments will now be described in detail, examples of which are illustrated in the accompanying drawings. When the following description relates to the drawings, unless otherwise indicated, the same numerals in different drawings denote the same or similar elements. The embodiments described in the following exemplary embodiments do not represent all embodiments consistent with one or more embodiments of the present invention. Rather, they are merely examples of apparatuses and methods consistent with some aspects of one or more embodiments of the present invention as detailed in the appended claims.

[0012] It should be noted that the steps of the corresponding methods in other embodiments are not necessarily performed in the order shown and described in this invention. In some other embodiments, the methods may include more or fewer steps than those described in this invention. Furthermore, a single step described in this invention may be broken down into multiple steps in other embodiments; and multiple steps described in this invention may be combined into a single step in other embodiments.

[0013] This invention relates to a full-condition impedance identification method for doubly-fed induction generator (DFIG) wind turbines based on frequency physical characteristic constraints. The control structure of a DFIG wind turbine unit consists of two parts: a synchronous control link and an electrical control link. The synchronous control link provides a reference phase for the system, while the electrical control link ensures stable operation of the system according to given commands. The impedance characteristics of a DFIG wind turbine unit are mainly related to the operating point and system parameters. System parameters are the circuit and control parameters of the DFIG wind turbine unit; the operating point is the voltage and current at the point of common coupling (PCC), which varies according to the operating conditions of the power system. In practice, system parameters are in a black box state, while the operating point can be obtained through measurement.

[0014] Doubly-fed induction generator (DFIG) wind turbines have been widely integrated into the power grid due to their high wind energy utilization efficiency and cost-effectiveness. However, as the penetration rate of wind turbines in the power grid continues to increase, instability issues such as subsynchronous / supersynchronous oscillations and harmonic resonances have arisen, seriously affecting the safe and reliable operation of the power grid. Impedance-based stability analysis methods have been widely used to analyze the stability margin of the interconnection system between DFIG wind turbines and the power grid. Therefore, identifying the full operating point impedance of DFIG wind turbines is crucial for analyzing the stability of DFIG-grid interconnection systems considering various operating conditions.

[0015] In related technologies, impedance identification methods mostly rely on extensive parameter tuning experiments to find the optimal neural network structure design, including the number of hidden layers, the number of neurons per layer, the learning rate, and the regularization coefficient. This results in huge model debugging costs and generates millions of training data requirements. Because different new energy generator units have varying impedance characteristics, and the neural network structure parameters are numerous, it is difficult to match neural network parameters that form a strict physical mapping relationship with the impedance characteristics. This reduces model identification performance and increases parameter design costs and training data requirements.

[0016] To address the shortcomings in related technologies, this invention proposes a full-condition impedance identification method for doubly-fed wind turbine generators.

[0017] Figure 1 This is a flowchart illustrating an exemplary embodiment of a method for impedance identification of a doubly-fed wind turbine under all operating conditions. Figure 1 As shown, the method may include the following steps: Step 101: Determine the common characteristics of the synchronous control and electrical control links of different doubly-fed wind turbine units, and generate the impedance expression of the doubly-fed wind turbine units based on the common characteristics. The impedance expression is used to characterize the differences and common characteristics of the frequency physical characteristics of different doubly-fed wind turbine units.

[0018] Step 102: Construct an impedance identification model based on the impedance expression; Step 103: Generate impedance characteristic regularization constraints based on the frequency physical characteristics of the doubly fed wind turbine, and train the impedance identification model based on the regularization constraints. Step 104: Input the unknown operating condition parameters into the impedance identification model after training to obtain the impedance amplitude and phase of the doubly fed wind turbine under the corresponding operating conditions.

[0019] In this embodiment, on the one hand, by deriving the common characteristics of doubly fed wind turbines with different control structures and parameters, the frequency physical characteristics of doubly fed generators are revealed and the establishment of a neural network identification model is guided, reducing the parameter design cost of the model; on the other hand, a full-condition impedance identification model of doubly fed wind turbines is established through data-driven approach, and model regularization constraints are designed through physical characteristics, which improves the accuracy of the model and reduces the amount of data required for model training.

[0020] In one embodiment, in the synchronization control loop and the electrical control loop, the coordinate change small disturbance signal is determined based on multiple linear functions about the operating point, small voltage signals and small current signals of the d-axis and q-axis; in the impedance expression, the impedance elements in the dq coordinate system are determined based on the matrix expression of the operating point polynomial and parameter coefficients.

[0021] The circuit model of a doubly-fed induction generator (DFIG) wind turbine is typically transformed from a three-phase (abc) coordinate system to a dq coordinate system to achieve decoupled control of active and reactive currents. A synchronous control loop generates a reference angle θ for the coordinate transformation from the three-phase to the dq coordinate system. Two different reference components exist in the DFIG wind turbine control system: a control component and an electrical component. The electrical component can be directly sampled from the actual circuit topology, and the control component can be obtained after coordinate transformation. Assuming there are no disturbances in the grid voltage, the control component and the electrical component are identical. However, if a small-signal disturbance exists in the grid voltage, the output angle θ of the synchronous loop in the coordinate transformation will also be affected, resulting in a phase difference Δθ between the control component and the electrical component, i.e., a small disturbance signal in the coordinate transformation.

[0022] Specifically, synchronization mechanisms can be divided into grid-following (GFL) control and grid-forming (GFM) control. Typical control schemes include phase-locked loops (PLLs), droop control, and virtual synchronous generators (VSGs). For all these typical synchronization mechanisms, Δθ can always be expressed in a unified form: ; in, f din , f qin , f did , f qid , f dvn , f dvd , f qvn , f qvd It is a linear function with respect to the operating point.

[0023] The electrical control system is used to control the doubly-fed induction generator (DFIG) wind turbine system to follow given voltage and current commands. It mainly includes the main circuit, current control, voltage control, and PQ power control. In the main circuit, based on the rotor voltage and current equations, we can obtain... ; Where, Δ Us rd, Δ Us rq and Δ Is rd, Δ Isrq represent the small signals of rotor voltage and current in the dq system coordinate system, respectively. P1-4 is the system parameter matrix, which mainly contains information such as circuit parameters, including stator and rotor resistance, stator winding self-inductance, rotor winding self-inductance, stator winding leakage inductance, rotor winding leakage inductance and mutual inductance, etc., which can be simulated through neural network training.

[0024] Considering small-signal interference during system coordinate transformation, the relationship between the control component and the electrical component of the stator voltage and current signals is as follows: ; Similarly, the relationship between the control component and the electrical component of the rotor voltage and current signal is as follows: ; Where Urd, Urq and Ird, Irq are the steady-state values ​​of the rotor voltage and current components in the dq coordinate system.

[0025] The control components of the rotor voltage can be obtained from control elements such as power, voltage, and current, and are expressed as follows: ; H1-3 is a matrix containing the operating point and system fixed parameters. The operating point is the steady-state value of the operating voltage and current, while the system parameter matrix contains PI parameters of control loops such as current loop, voltage loop, and power loop, decoupling coefficients of the current loop, and voltage feedforward coefficients, etc.

[0026] Combining the above equations, we can obtain the common characteristic expression for electrical control loops. ; Among them, f i All ten variables are polynomial functions relating to the operating point and fixed system parameters. Since the system parameters are fixed, the main consideration during doubly fed wind turbine operation is the change in the operating point; therefore, all expressions use the operating point as the variable.

[0027] Furthermore, combining the impedance definition expression, the unified impedance-frequency physical characteristic expression of a doubly-fed induction generator (DFIG) is derived, where the dq admittance is defined as follows: ; The expressions for the synchronization element and the electrical control element can be obtained by combining them. ; Where gi1-4 and gv1-4 are polynomial functions of the operating point.

[0028] The dq admittance of a doubly-fed wind turbine is defined as follows: ; Among them, Ydd and the other three variables represent the dq admittance of the doubly-fed wind turbine. Combining the above equations, the impedance expression is as follows: ; Among them, Y dd Y dq Y qd Y qq For different impedance elements in the dq coordinate system, the function g i1-4 and g u1-4 This is a matrix expression for the operating point polynomial and parameter coefficients. ; in, ; Where x represents a polynomial consisting of working point variables, ak and bk represent vectors consisting of polynomial coefficients, and a k1 ~a k4 and b k1 ~b k4 All are polynomial coefficients, k=1, 2, 3, 4, T denotes transpose, U d and U q These are the steady-state voltage values ​​along the d-axis and q-axis, respectively. d and I q ...

[0029] It can be seen that x is a matrix that is only related to the operating point, while ak and bk are matrices that are only related to the system parameters. For different doubly-fed wind turbine units, their impedance characteristics can be uniformly represented, which reveals the common characteristics of doubly-fed wind turbine units. However, the elements of the x, ak, and bk vectors in the formula are different, that is, the number of polynomials and the corresponding coefficients are different, which reveals the differences in the impedance characteristics of different doubly-fed wind turbine units.

[0030] In practical stability analysis, sequence impedance has a clearer physical meaning and is easier to measure. dq-domain impedance can be equivalently converted to sequence impedance.

[0031] ; Stability analysis based on admittance is typically performed using Bode plots; therefore, the admittance model in the above equation is presented in the form of logarithmic magnitude and phase, i.e., .

[0032] In one embodiment, the impedance identification model includes an input layer, a hidden layer, and an output layer connected in sequence; wherein the input of the input layer is the operating point parameter and frequency, the output of the output layer is the amplitude and phase of the impedance, and the hidden layer includes a first hidden layer, a second hidden layer, and a third hidden layer; The input to the first hidden layer is the operating point and frequency variable, which is used to simulate the polynomial function contained in the impedance characteristics based on the random deactivation method; The input to the second hidden layer is the output of the first hidden layer, used to simulate the matrix expression; The input to the third hidden layer is the output of the second hidden layer, used to calculate the magnitude and phase of each impedance element, and the output is... , , , , , , , ,in, and For Y dd The amplitude and phase, and For Y dq The amplitude and phase, and For Y qd The amplitude and phase, and For Y qq The amplitude and phase.

[0033] Impedance expressions are used to guide the design of neural network impedance identification models. The input to the impedance identification model is the steady-state value at the operating point. Since the rotor voltage and current can be linearly calculated from the stator voltage and current, and in practice, due to voltage orientation, U... sq Since the value is 0, the actual input variable is set to U. sd I sd I sq And the frequency f, the output is the admittance, amplitude, phase, and eight other variables. , , , , , , , The established neural network model is as follows: Figure 2 As shown, the architecture and number of neurons in each layer are designed to strictly correspond to the theoretical formula.

[0034] The neural network consists of three hidden layers. The first hidden layer corresponds to multinomial factorization. In the first hidden layer, the input is the operating point and frequency, i.e., U. sd I sd I sqAnd f, this layer is designed to simulate the polynomial function contained in the impedance characteristics. Therefore, the number of neurons in this layer should be consistent with the number of operating point polynomials present in the impedance characteristics of a doubly-fed wind turbine in reality. However, due to the black box state, the exact number of polynomials is unknown, but the above polynomial decomposition can still guide the selection of the number of neurons. Specifically, the above polynomial x contains a constant term and a first-order term U. d I d I q quadratic term U d 2 , I d 2 , I q 2 , U d I d , I q I d , U d I q And cubic terms, etc. Due to normalization, Id≤1, Iq≤1, Vd≤1, lower-order terms contribute more to impedance characteristics, so polynomials of cubic degree and above are ignored. At this point, x contains 10 polynomials, and to correspond to x, ak and bk also need to be set to 10. Therefore, the first hidden layer requires a total of 30 neurons. Subsequently, to reflect the differences in doubly-fed wind turbine units, i.e., the different numbers of polynomials, the number of neurons in this layer should be adaptively adjusted according to different doubly-fed wind turbine units to more effectively learn latent features. Therefore, a dropout layer is added after the first hidden layer. By adjusting the dropout rate during training, the model can adaptively deactivate some redundant neurons during training, flexibly adjusting the number of neurons to more accurately reflect the actual impedance characteristics of different doubly-fed wind turbine units. The differences in polynomial coefficients contained in the impedance characteristics of different doubly-fed wind turbine units are reflected by adjusting the network weights through data training.

[0035] The second hidden layer is for simulating g. i1-4 and g v1-4 The first hidden layer contains eight neurons, whose input is the output of the first hidden layer; the third hidden layer contains eight neurons and calculates the final impedance amplitude and phase of the doubly fed wind turbine.

[0036] In one embodiment, the impedance expression is used to establish impedance regularization constraints.

[0037] As can be seen from the above derivation, although doubly-fed wind turbines differ in control parameters and structure, they share common characteristics and can be characterized using a unified expression and neural network. Furthermore, by simultaneously solving the impedance definition and the common characteristic expression, and canceling ΔIs sd, we can obtain ΔIs sq. ; In small-signal analysis, the applied disturbance is small and linear, and does not significantly affect the steady-state operating point of the system. This linearization allows the dynamic characteristics of the d-axis and q-axis to be analyzed separately. On the other hand, in the control system of doubly-fed induction generator (DFIG) wind turbines, voltage-oriented control or vector control is usually used, and the dynamic characteristics of the d-axis and q-axis can be decoupled. Therefore, Δ Us sd, Δ Us sq represents independent small perturbations. To satisfy the above equation, the coefficients α and β should be 0. After simplification, we can obtain... ; The above equation is a common constraint for different doubly fed wind turbine units. Therefore, the constructed neural network should also satisfy this condition as much as possible to further enhance the physical meaning and make the neural network better reflect the actual characteristics of the impedance model of the doubly fed wind turbine unit.

[0038] In practice, the mean squared error (MSE) between the actual and predicted values ​​is used as the loss function to train the impedance identification neural network model, that is, ; In the formula, yi represents the neural network prediction value; yi represents the training sample value; N represents the number of samples. MSE, as the loss function, directly reflects the accuracy of the constructed model.

[0039] The trained model also needs to satisfy the constraints in the above formula, that is, Lst needs to approach 0 during training. Therefore, this method uses regularization to satisfy this condition, and the loss function expression is then rewritten as follows: ; Where λ is the regularization coefficient, such as Figure 3 As shown, during the training and verification process, the neural network gradually reduces the error between the actual sample value and the predicted value in order to better reflect the impedance characteristics, enhance the recognition performance, and reduce the data requirements.

[0040] In one embodiment, training the impedance identification model based on the regularization constraints includes: guiding the impedance identification model to train and converge quickly based on the regularization constraints.

[0041] Furthermore, the step of guiding the impedance identification model to train and converge quickly based on the regularization constraints includes: performing frequency sweep measurements on the doubly-fed wind turbine generator using the perturbation injection method to obtain online measured impedance data at different operating points; dividing the online measured impedance data into a training set and a validation set; setting the training target threshold to 1e-7 to indicate whether the neural network model has learned the potential characteristics of the impedance, and setting the ratio threshold of validation loss to training loss to 1.5 to check whether the model has sufficient generalization performance; adaptively changing the random inactivation rate and the regularization coefficient of the constraints during training until the impedance identification model reaches the set threshold.

[0042] For different doubly-fed induction generator (DFIG) wind turbines, this implementation method constructs a full-condition impedance identification model through data training to improve the model's identification accuracy. The specific process is as follows: To achieve impedance identification of doubly fed wind turbines, a neural network model was first built on the MATLAB platform with a learning rate of 0.01. Since the model has fixed the number of hidden layers and the number of neurons in each layer based on frequency physical information, the cost of parameter adjustment experiments was greatly reduced.

[0043] Obtaining impedance data requires injecting frequency disturbance signals into the doubly-fed induction generator (DFIG) multiple times under stable operating conditions. The signals are then sampled from the common coupling point, and the voltage and current signals at the disturbance frequency are extracted using a fast Fourier transform. The impedance can then be expressed as follows: ; Where: V sp V sp2 They represent frequencies f, respectively. p and f p Voltage disturbance at -2f1, I sp I sp2 Indicates at frequency f p and f p The current response generated by -2f1.

[0044] Repeat the above steps until all specified frequency points have been measured. This is the frequency scanning process. After this process, the impedance data of the doubly-fed induction generator (DFIG) at one operating point will be obtained. By changing the operating point of the DFIG and continuing the frequency scanning process, the impedance dataset required for measurement can be obtained. Specifically, f ranges from 1 Hz to 200 Hz, with intervals of 2 Hz, and I... sd The range is from 0.2 pu to 1 pu, with intervals of 0.2 pu, I sq The range is 0 to 1 pu, with intervals of 0.25 pu, U sdThe range is from 0.9 pu to 1.1 pu, with intervals of 0.1 pu, for a total of 75 operating points and corresponding 7.5 k impedance data.

[0045] The data is fed into a neural network model for training. Due to the limited amount of data, the dataset is randomly divided into a training set and a validation set. The training set is used to train the neural network model to learn the latent characteristics of impedance. The validation set is used to evaluate the success of the training process and to improve predictive performance by optimizing the model parameters. The training algorithm is adaptive moment estimation, while the validation algorithm for optimizing the model parameters is Bayesian optimization. After training, an impedance identification model is obtained.

[0046] Figure 4 This is a schematic structural diagram of a device provided in an exemplary embodiment. Please refer to... Figure 4 At the hardware level, the device includes a processor 402, an internal bus 404, a network interface 406, memory 408, and non-volatile memory 410, and may also include other hardware required for its functions. One or more embodiments of the present invention can be implemented in software, for example, the processor 402 reads the corresponding computer program from the non-volatile memory 410 into memory 408 and then runs it. Of course, in addition to software implementation, one or more embodiments of the present invention do not exclude other implementation methods, such as logic devices or a combination of hardware and software, etc. That is to say, the execution subject of the following processing flow is not limited to each logic unit, but can also be hardware or logic devices.

[0047] Please refer to Figure 5 A full-condition impedance identification device for doubly-fed wind turbines can be applied to, for example... Figure 5 The device shown, in order to implement the technical solution of the present invention, includes: The determining unit 501 is used to determine the common characteristics of the synchronous control link and electrical control link of different doubly-fed wind turbine units, and generate the impedance expression of the doubly-fed wind turbine unit based on the common characteristics. The impedance expression is used to characterize the differences and common characteristics of the frequency physical characteristics of different doubly-fed wind turbine units. Construction unit 502 is used to construct an impedance identification model based on the impedance expression; Training unit 503 is used to generate impedance characteristic regularization constraints based on the frequency physical characteristics of the doubly fed wind turbine, and to train the impedance identification model based on the regularization constraints. The identification unit 504 is used to input unknown operating condition parameters into the impedance identification model after training in order to obtain the impedance amplitude and phase of the doubly fed wind turbine under the corresponding operating condition.

[0048] Optionally, in the synchronous control link and the electrical control link, the small disturbance signal of coordinate change is determined based on multiple linear functions about the operating point, small voltage signals and small current signals of the d-axis and q-axis; In the impedance expression, the impedance elements in the dq coordinate system are determined based on the matrix expression of the operating point polynomial and the parameter coefficients.

[0049] Furthermore, the impedance expression is as follows: ; Among them, Y dd Y dq Y qd Y qq For different impedance elements in the dq coordinate system, the function g i1-4 and g u1-4 This is a matrix expression for the operating point polynomial and parameter coefficients. ; in, ; Where x represents a polynomial consisting of working point variables, ak and bk represent vectors consisting of polynomial coefficients, and a k1 ~a k4 and b k1 ~b k4 All are polynomial coefficients, k=1, 2, 3, 4, T denotes transpose, U d and U q These are the steady-state voltage values ​​along the d-axis and q-axis, respectively. d and I q ...

[0050] Furthermore, the impedance identification model includes an input layer, a hidden layer, and an output layer connected in sequence; wherein, the input of the input layer is the operating point parameter and frequency, the output of the output layer is the amplitude and phase of the impedance, and the hidden layer includes a first hidden layer, a second hidden layer, and a third hidden layer. The input to the first hidden layer is the operating point and frequency variable, which is used to simulate the polynomial function contained in the impedance characteristics based on the random deactivation method; The input to the second hidden layer is the output of the first hidden layer, used to simulate the matrix expression; The input to the third hidden layer is the output of the second hidden layer, used to calculate the magnitude and phase of each impedance element, and the output is... , , , , , , , ,in, and For Y dd The amplitude and phase, and For Y dq The amplitude and phase, and For Y qd The amplitude and phase, and For Y qq The amplitude and phase.

[0051] Optionally, the training loss expression for the regularization constraint is: ; Where λ is the regularization coefficient, L st For regularization constraints: .

[0052] Optionally, the training unit 503 is specifically used for: The impedance identification model is trained and converged quickly based on the regularization constraints.

[0053] Furthermore, the training unit 503 is specifically used for: Frequency sweep measurement of doubly fed wind turbines was performed using the perturbation injection method to obtain online measurement impedance data at different operating points. The online impedance measurement data is divided into a training set and a validation set; The training target threshold is set to 1e-7 to indicate whether the neural network model has learned the latent features of impedance, and the ratio threshold of validation loss to training loss is set to 1.5 to check whether the model has sufficient generalization performance. During training, the random inactivation rate and the regularization coefficient of the constraints are adaptively changed until the impedance identification model reaches the set threshold.

[0054] The systems, devices, modules, or units described in the above embodiments can be implemented by computer chips or entities, or by products with certain functions. A typical implementation device is a computer, which can take the form of a personal computer, laptop computer, cellular phone, camera phone, smartphone, personal digital assistant, media player, navigation device, email sending and receiving device, game console, tablet computer, wearable device, or any combination of these devices.

[0055] In a typical configuration, a computer includes one or more processors (CPU), input / output interfaces, network interfaces, and memory.

[0056] Memory may include non-persistent storage in computer-readable media, such as random access memory (RAM) and / or non-volatile memory, such as read-only memory (ROM) or flash RAM. Memory is an example of computer-readable media.

[0057] Computer-readable media, including both permanent and non-permanent, removable and non-removable media, can store information using any method or technology. Information can be computer-readable instructions, data structures, modules of programs, or other data. Examples of computer storage media include, but are not limited to, phase-change memory (PRAM), static random access memory (SRAM), dynamic random access memory (DRAM), other types of random access memory (RAM), read-only memory (ROM), electrically erasable programmable read-only memory (EEPROM), flash memory or other memory technologies, CD-ROM, digital versatile optical disc (DVD) or other optical storage, magnetic tape, disk storage, quantum memory, graphene-based storage media or other magnetic storage devices, or any other non-transferable medium that can be used to store information accessible by a computing device. As defined herein, computer-readable media does not include transient computer-readable media, such as modulated data signals and carrier waves.

[0058] For any other form of computer-readable medium (or computer-readable storage medium) as described above, computer instructions may be stored thereon, which, when executed by a processor, implement one or more of the above embodiments, thereby realizing the technical solution of the present invention.

[0059] The present invention also proposes a computer program that, when executed by a processor, implements one or more of the embodiments described above, thereby realizing the technical solution of the present invention. This computer program may be specifically recorded on the above-described or other computer-readable media, and the present invention does not impose any limitations on this.

[0060] It should also be noted that the terms "comprising," "including," or any other variations thereof are intended to cover non-exclusive inclusion, such that a process, method, article, or apparatus that comprises a list of elements includes not only those elements but also other elements not expressly listed, or elements inherent to such a process, method, article, or apparatus. Without further limitation, an element defined by the phrase "comprising one..." does not exclude the presence of other identical elements in the process, method, article, or apparatus that includes said element.

[0061] The foregoing has described specific embodiments of the invention. Other embodiments are within the scope of the appended claims. In some cases, the actions or steps described in the claims may be performed in a different order than that shown in the embodiments and may still achieve the desired results. Furthermore, the processes depicted in the drawings do not necessarily require the specific or sequential order shown to achieve the desired results. In some embodiments, multitasking and parallel processing are also possible or may be advantageous.

[0062] The terminology used in one or more embodiments of the invention is for the purpose of describing particular embodiments only and is not intended to limit the invention. The singular forms “a,” “the,” and “the” used in one or more embodiments of the invention and in the appended claims are also intended to include the plural forms unless the context clearly indicates otherwise. It should also be understood that the term “and / or” as used herein refers to and includes any or all possible combinations of one or more associated listed items.

[0063] It should be understood that although the terms first, second, third, etc., may be used to describe various information in one or more embodiments of the present invention, such information should not be limited to these terms. These terms are only used to distinguish information of the same type from one another. For example, first information may also be referred to as second information without departing from the scope of one or more embodiments of the present invention, and similarly, second information may also be referred to as first information. Depending on the context, the word "if" as used herein may be interpreted as "when," "when," or "in response to a determination."

[0064] The above description is merely a preferred embodiment of one or more embodiments of the present invention and is not intended to limit the present invention. Any modifications, equivalent substitutions, improvements, etc., made within the spirit and principles of one or more embodiments of the present invention should be included within the protection scope of one or more embodiments of the present invention.

Claims

1. A method for impedance identification of a doubly-fed induction generator (DFIG) under all operating conditions, characterized in that, The method includes: The common characteristics of the synchronous control and electrical control links of different doubly-fed wind turbine units are determined, and the impedance expression of the doubly-fed wind turbine units is generated based on the common characteristics. The impedance expression is used to characterize the differences and common characteristics of the frequency physical characteristics of different doubly-fed wind turbine units. Construct an impedance identification model based on the impedance expression; Impedance characteristic regularization constraints are generated based on the frequency physical characteristics of the doubly fed wind turbine, and the impedance identification model is trained based on the regularization constraints. The unknown operating condition parameters are input into the impedance identification model after training to obtain the impedance amplitude and phase of the doubly fed wind turbine under the corresponding operating conditions.

2. The method according to claim 1, characterized in that, In the synchronous control and electrical control links, the small disturbance signals of coordinate changes are determined based on multiple linear functions about the operating point, small voltage signals and small current signals of the d-axis and q-axis; In the impedance expression, the impedance elements in the dq coordinate system are determined based on the matrix expression of the operating point polynomial and the parameter coefficients.

3. The method according to claim 2, characterized in that, The impedance expression is as follows: ； Among them, Y dd Y dq Y qd Y qq For different impedance elements in the dq coordinate system, the function g i1-4 and g u1-4 This is a matrix expression for the operating point polynomial and parameter coefficients. ； in, ； Where x represents a polynomial consisting of working point variables, ak and bk represent vectors consisting of polynomial coefficients, and a k1 ~a k4 and b k1 ~b k4 All are polynomial coefficients, k=1, 2, 3, 4, T denotes transpose, U d and U q These are the steady-state voltage values ​​along the d-axis and q-axis, respectively. d and I q ...

4. The method according to claim 3, characterized in that, The impedance identification model includes an input layer, a hidden layer, and an output layer connected in sequence; wherein, the input of the input layer is the operating point parameter and frequency, the output of the output layer is the amplitude and phase of the impedance, and the hidden layer includes a first hidden layer, a second hidden layer, and a third hidden layer. The input to the first hidden layer is the operating point and frequency variable, which is used to simulate the polynomial function contained in the impedance characteristics based on the random deactivation method; The input to the second hidden layer is the output of the first hidden layer, used to simulate the matrix expression; The input to the third hidden layer is the output of the second hidden layer, used to calculate the magnitude and phase of each impedance element, and the output is... , , , , , , , ,in, and For Y dd The amplitude and phase, and For Y dq The amplitude and phase, and For Y qd The amplitude and phase, and For Y qq The amplitude and phase.

5. The method according to claim 3, characterized in that, The training loss expression for the regularization constraint is: ； Where λ is the regularization coefficient, L st For regularization constraints: 。 6. The method according to claim 1, characterized in that, The training of the impedance identification model based on the regularization constraints includes: The impedance identification model is trained and converged quickly based on the regularization constraints.

7. The method according to claim 6, characterized in that, The method of guiding the impedance identification model to fast training and convergence based on the regularization constraints includes: Frequency sweep measurement of doubly fed wind turbines was performed using the perturbation injection method to obtain online measurement impedance data at different operating points. The online impedance measurement data is divided into a training set and a validation set; The training target threshold is set to 1e-7 to indicate whether the neural network model has learned the latent features of impedance, and the ratio threshold of validation loss to training loss is set to 1.5 to check whether the model has sufficient generalization performance. During training, the random inactivation rate and the regularization coefficient of the constraints are adaptively changed until the impedance identification model reaches the set threshold.

8. A full-condition impedance identification device for a doubly-fed wind turbine generator, characterized in that, The device includes: Determining Unit: Determine the common characteristics of the synchronous control and electrical control links of different doubly-fed wind turbine units, and generate the impedance expression of the doubly-fed wind turbine units based on the common characteristics. The impedance expression is used to characterize the differences and common characteristics of the frequency physical characteristics of different doubly-fed wind turbine units. Construction Unit: Construct an impedance identification model based on the impedance expression; Training unit: Generates impedance characteristic regularization constraints based on the frequency physical characteristics of the doubly fed wind turbine, and trains the impedance identification model based on the regularization constraints; Identification Unit: Input the unknown operating condition parameters into the impedance identification model after training to obtain the impedance amplitude and phase of the doubly fed wind turbine under the corresponding operating conditions.

9. An electronic device, characterized in that, include: processor; Memory used to store processor-executable instructions; The processor implements the steps of the method as described in any one of claims 1-7 by running the executable instructions.

10. A computer-readable storage medium storing computer instructions thereon, characterized in that, When executed by the processor, this instruction implements the steps of the method as described in any one of claims 1-7.

Images