Doubly-fed fan grid-connected system impedance identification method and system based on data driving

Abstract

The invention relates to the technical field of power electronics, in particular to an impedance identification method and system for a doubly-fed fan grid-connected system based on data driving, and the method specifically comprises the steps: determining the input and output of the system based on a dq theoretical impedance model of the doubly-fed fan grid-connected system; building a time domain simulation model of the single doubly-fed fan grid-connected system, and performing data collection under various operation conditions by using a frequency sweep method; constructing a single doubly-fed fan impedance identification model based on a back propagation neural network optimized by a genetic algorithm, training the single doubly-fed fan impedance identification model, aggregating the station impedance identification model by using the trained single doubly-fed fan impedance identification model, and then performing impedance identification to obtain equivalent impedance after aggregation of the wind power station; and analyzing the stability of the doubly-fed fan grid-connected system by using the impedance identification model based on the generalized Nyquist criterion. According to the method, the impedance identification precision is ensured, and the impedance identification speed is greatly increased at the same time.

Description

Technical Field

[0001] The present invention relates to the field of power electronics technology, and in particular to a data-driven impedance identification method and system for a doubly-fed wind turbine grid-connected system. Background Art

[0002] To address the challenges of climate change, my country has set a "dual carbon" goal. To this end, China is vigorously developing renewable energy, primarily photovoltaic and wind power generation, and the penetration of renewable energy in the electricity supply continues to rise. The integration of renewable energy into the grid inevitably involves the use of extensive power electronics, leading to increasingly prominent subsynchronous and supersynchronous oscillations in the system. Furthermore, due to the inverse spatial distribution of load centers and renewable resources in my country, series capacitors are often used to reduce system reactive power transmission and thus line impedance. This approach also increases the risk of subsynchronous oscillations. Since the research and analysis of subsynchronous oscillations lags far behind the application and iteration of power electronics control methods, there is an urgent need to research methods to address subsynchronous oscillations and ensure sufficient system stability margins. However, due to the large number of wind turbines within a site and the complex operating conditions, traditional state-space equation derivation or swept-frequency measurement methods are difficult and time-consuming to obtain the system's broadband impedance characteristics, limiting their applicability.

[0003] Therefore, data-driven impedance identification, using neural networks as a carrier, replaces the complex small-signal calculations in theoretical impedance models and the multiple time-domain simulation cycles of swept-frequency measurement methods. This accelerates impedance acquisition and improves timeliness, making it of great significance for the analysis of subsynchronous oscillations. Currently, there is no data-driven impedance identification method or system for grid-connected doubly-fed wind turbine systems. Summary of the Invention

[0004] In response to the problems in the prior art, the present invention provides a data-driven impedance identification method and system for a doubly-fed wind turbine grid-connected system. The specific technical solutions are as follows:

[0005] A data-driven impedance identification method for a doubly-fed wind turbine grid-connected system includes the following steps:

[0006] Step S1, determining the input and output of the system based on the dq theoretical impedance model of the doubly fed wind turbine grid-connected system;

[0007] Step S2: Building a time domain simulation model of a single doubly-fed wind turbine grid-connected system and using a frequency sweep method to collect data under various operating conditions;

[0008] Step S3: constructing and training a single doubly-fed wind turbine impedance identification model based on a back propagation neural network optimized by a genetic algorithm, aggregating the station impedance identification model with the trained single doubly-fed wind turbine impedance identification model, and then performing impedance identification to obtain the equivalent impedance of the wind farm after aggregation;

[0009] Step S4: analyzing the stability of the doubly-fed wind turbine grid-connected system using an impedance identification model based on the generalized Nyquist criterion.

[0010] Preferably, determining the input and output of the system includes:

[0011] The impedance characteristics have a nonlinear mapping relationship with the characteristic parameters of the grid connection point port. The theoretical impedance model is as follows:

[0012]

[0013] Among them, Z dq is the theoretical impedance, f is the frequency, F1 represents a mapping relationship, U d , I d Indicates the d-axis voltage and current in the dq coordinate system, U q , I q Represents the q-axis voltage and current in the dq coordinate system; Z dd (f), Z dq (f), Z qd (f), Z qq (f) are the components of the theoretical impedance respectively.

[0014] Preferably, the step S2 specifically includes the following steps:

[0015] Build a time domain simulation model of a single doubly fed wind turbine grid-connected system;

[0016] Inject positive sequence disturbance with frequency (f+f0) and negative sequence disturbance with frequency (f-f0), where f is the impedance characteristic frequency in the dq0 coordinate system and f0 is the grid fundamental frequency;

[0017] Through these two linearly independent perturbations, the dq impedance characteristics at frequency f are obtained, and the calculation results are:

[0018]

[0019] Where Δu d1 , Δi d1 are the voltage and current response components of the d-axis after the first disturbance; Δu d2 , Δi d2 are the voltage and current response components of the d-axis after the second disturbance; Δu q1 , Δi q1 are the voltage and current response components of the q axis after the first disturbance; Δu q2 , Δi q2 are the voltage and current response components of the q-axis after the second disturbance;

[0020] Finally, the impedance characteristics of the doubly fed wind turbine grid-connected system at various frequency points are measured using the sweep frequency method.

[0021] Preferably, in step S3, constructing a single doubly-fed wind turbine impedance identification model based on a back propagation neural network optimized by a genetic algorithm and performing training specifically comprises the following steps:

[0022] The collected data are preprocessed to obtain a data set, which is then divided into a training set, a test set, and a validation set. The data set includes the d-axis voltage U in the dq coordinate system. d , current I d , q-axis voltage U in dq coordinate system q , current I q , frequency f and Z dq (f) The amplitude and phase angle of each element of the matrix;

[0023] A single doubly fed wind turbine impedance identification model based on a back propagation neural network optimized by genetic algorithm is constructed, wherein the back propagation neural network consists of an input layer, several hidden layers, and an output layer; the training set is input into the back propagation neural network for training, and the input is the d-axis voltage U in the dq coordinate system d , current I d , q-axis voltage U in dq coordinate system q , current I q , frequency f, output is Z dq (f) The amplitude and phase angle of each element of the matrix; and the genetic algorithm is used to optimize the weights and bias of the back-propagation neural network. The model is tested with a test set during the training process, and the model is verified with a validation set after the training. When the accuracy of the model reaches the set threshold, the trained single doubly fed wind turbine impedance identification model is output.

[0024] Preferably, in step S3, the station impedance identification model is aggregated with the trained single doubly-fed wind turbine impedance identification model, and then impedance identification is performed, specifically comprising the following steps:

[0025] The dq voltage and current amplitude U of each doubly fed wind turbine AC output port d-i , U q-i , I d-i , I q-i Solve the impedance characteristic matrix Z of all doubly fed wind turbines in the wind farm station dq-i , where i represents the i-th wind turbine, i = 1, 2…n; n is the number of doubly fed wind turbines in the wind farm; U d-i , U q-i are the voltage amplitudes of the d and q axes at the AC output port of the i-th wind turbine respectively; I d-i , I q-iare the d-axis and q-axis current amplitudes of the AC output port of the i-th wind turbine respectively;

[0026] The impedance of all doubly fed wind turbines in the station is aggregated, and a common dq coordinate system is selected as the phase reference point, namely the PCC point, and then the impedance matrix Z of each doubly fed wind turbine is calculated. dq-i Convert to this dq coordinate system;

[0027] The phase angle θ of each doubly fed wind turbine in the local dq coordinate system is measured using a phase-locked loop i , the phase angle θ of the PCC point reference dq coordinate system pcc , then the offset angle Δθ between the local coordinate system and the reference coordinate system is i =θ pcc -θ i , use the following formula to transform the coordinate system:

[0028]

[0029] Where Z dq-i (f) is the impedance matrix of the local dq coordinate system of the i-th wind turbine at frequency f, Z′ dq-i (f) is the impedance matrix of the i-th wind turbine in the reference dq coordinate system after conversion, T trans-i is the conversion matrix of the i-th wind turbine, It's T trans-i The inverse matrix of

[0030] After completing the impedance coordinate system transformation, the impedance characteristics of each doubly fed wind turbine are aggregated to the PCC point of the grid-connected system to obtain the equivalent impedance Z of the wind farm after aggregation. DFIG (s).

[0031] Preferably, the conversion matrix T of the i-th wind turbine is trans-i The specific ones are:

[0032]

[0033] Preferably, the step S4 specifically includes the following steps:

[0034] By using the equivalent impedance of the aggregated wind farms and the grid-side impedance, the generalized Nyquist criterion is used to draw the system characteristic trajectory and determine the stability of the grid-connected system.

[0035] When Z g (s) / Z DFIG (s) satisfies the Nyquist criterion, the system is stable; when Z g (s) / Z DFIG When (s) does not satisfy the Nyquist criterion, the system is unstable, where Z g(s) represents the equivalent impedance of the power grid; Z DFIG (s) represents the equivalent impedance of the wind farm after aggregation.

[0036] A data-driven impedance identification system for a doubly-fed wind turbine grid-connected system, applied to the method, comprises: an input-output determination module for determining the input and output of the system based on a DQ theoretical impedance model of the doubly-fed wind turbine grid-connected system; a data collection module for building a time-domain simulation model of a single doubly-fed wind turbine grid-connected system and using a sweep frequency method to collect data under various operating conditions;

[0037] The impedance identification module is used to construct and train a single doubly fed wind turbine impedance identification model based on a back propagation neural network optimized by a genetic algorithm. The trained single doubly fed wind turbine impedance identification model is then used to aggregate the station impedance identification model. Impedance identification is then performed to obtain the equivalent impedance of the aggregated wind farm station.

[0038] The analysis module is used to analyze the stability of the doubly fed wind turbine grid-connected system based on the generalized Nyquist criterion and the impedance identification model.

[0039] A computer-readable storage medium includes a stored program, wherein when the program is run, the device where the computer-readable storage medium is located is controlled to execute the data-driven double-fed wind turbine grid-connected system impedance identification method.

[0040] A processor is used to run a program, wherein when the program is run, the data-driven doubly-fed wind turbine grid-connected system impedance identification method is executed.

[0041] Compared with the prior art, the present invention has the following beneficial effects:

[0042] The data-driven impedance identification method for the doubly fed wind turbine grid-connected system proposed in this invention achieves a seamless transition from "single-machine data drive" to "station physical modeling" by performing multi-dimensional coupling calculations on the output of the single-machine GA-BP model and the network impedance matrix. This enables the model to maintain strong generalization capabilities when the number of units changes, greatly accelerating the speed of impedance identification while ensuring the accuracy of impedance identification. BRIEF DESCRIPTION OF THE DRAWINGS

[0043] To more clearly illustrate the specific embodiments of the present invention or the technical solutions in the prior art, the following briefly describes the drawings required for the specific embodiments or the description of the prior art. Similar elements or parts are generally identified by similar reference numerals throughout the drawings. Elements or parts in the drawings are not necessarily drawn to scale.

[0044] Figure 1 Flow chart of the method of the present invention.

[0045] Figure 2 This is a flow chart of the key steps in the small signal derivation of theoretical impedance modeling for doubly fed wind turbines.

[0046] Figure 3 This is a flow chart of measuring system impedance characteristics using the sweep frequency method of the present invention.

[0047] Figure 4 This is a graph showing how the amplitude and phase angle of the dq impedance matrix measured by the frequency sweep method in the present invention vary with frequency.

[0048] Figure 5 This is the theoretical structure diagram of the BP neural network in the present invention.

[0049] Figure 6 This is the equivalent impedance structure diagram of the doubly-fed wind farm grid-connected system of the present invention.

[0050] Figure 7 This is an equivalent schematic diagram of the simplified grid-connected system of the present invention.

[0051] Figure 8 This is a circuit diagram of the wind farm grid-connected system of the present invention.

[0052] Figure 9 This is a schematic diagram of the station characteristic root locus of the impedance identification result of the present invention.

[0053] Figure 10 1 is a comparison diagram of the GA-BP impedance identification results and the sweep frequency measurement results of the present invention (taking working condition 10 as an example).

[0054] Figure 11 This is the FFT decomposition result diagram of the PCC point voltage of the station of the present invention.

[0055] Figure 12 This is a system principle diagram of the present invention. DETAILED DESCRIPTION

[0056] The following will clearly and completely describe the technical solutions in the embodiments of the present invention in conjunction with the accompanying drawings. Obviously, the described embodiments are only part of the embodiments of the present invention, not all of them. All other embodiments obtained by ordinary technicians in this field based on the embodiments of the present invention without making any creative efforts shall fall within the scope of protection of the present invention.

[0057] It will be understood that when used in this specification and the appended claims, the terms “comprises” and “comprising” indicate the presence of described features, integers, steps, operations, elements and / or components, but do not preclude the presence or addition of one or more other features, integers, steps, operations, elements, components and / or groups thereof.

[0058] It should also be understood that the terms used in the present specification are only for the purpose of describing particular embodiments and are not intended to limit the present invention. As used in the present specification and the appended claims, the singular forms "a", "an", and "the" are intended to include the plural forms unless the context clearly indicates otherwise.

[0059] It should be further understood that the term "and / or" used in the present description and the appended claims refers to and includes any and all possible combinations of one or more of the associated listed items.

[0060] Example 1:

[0061] like Figure 1 As shown, this embodiment provides a data-driven method for identifying impedance of a doubly-fed wind turbine grid-connected system, comprising the following steps:

[0062] Step S1, determining the input and output of the system based on the dq theoretical impedance model of the doubly-fed wind turbine grid-connected system.

[0063] The impedance characteristics have a nonlinear mapping relationship with the characteristic parameters of the grid connection point port. The theoretical impedance model is as follows:

[0064]

[0065] Among them, Z dq is the theoretical impedance, f is the frequency, F1 represents a mapping relationship, U d , I d Indicates the d-axis voltage and current in the dq coordinate system, U q , I q Represents the q-axis voltage and current in the dq coordinate system; Z dd (f), Z dq (f), Z qd (f), Z qq (f) are the components of the theoretical impedance respectively.

[0066] Based on the small signal analysis method, the theoretical impedance model of the doubly fed wind turbine is carried out, and its impedance characteristics are shown in the following formula:

[0067]

[0068] in, is the stator d-axis voltage and current small signal, is the stator q-axis voltage and current small signal, Z DIFG-dq is the theoretical impedance of the doubly fed wind turbine; Z dd , Z dq , Z qd , Z qqThey are the components of the theoretical impedance of the doubly fed wind turbine. The key steps and ideas for the small signal derivation of the theoretical impedance modeling of the doubly fed wind turbine are shown in the attached figure. Figure 2 shown.

[0069] Step S2: Build a time domain simulation model of a single doubly fed wind turbine grid-connected system and use the sweep frequency method to collect data under various operating conditions. Figure 3 As shown, the following steps are included:

[0070] A time domain simulation model of a single doubly fed wind turbine grid-connected system is built; specifically, a time domain simulation of a single 1.5MW doubly fed wind turbine connected to a simple power grid is built. The simulation parameters are shown in Table 1.

[0071] Table 1 Main parameter values ​​of single-machine grid-connected system

[0072] Parameter Symbol Parameter name Value <![CDATA[P n ]]> Doubly fed wind turbine rated power 1.5MW <![CDATA[r s r s ]]> Doubly fed wind turbine stator inductance 0.18pu <![CDATA[l s l s ]]> Doubly fed wind turbine stator resistance 0.023pu <![CDATA[r r r r ]]> Doubly fed wind turbine rotor inductance 0.16pu <![CDATA[l r l r ]]> Doubly fed wind turbine rotor resistance 0.016pu <![CDATA[k p1 ,k I1 ]]> RSC power outer loop PI parameters 0.05 / 20 <![CDATA[k p2 ,k I2 ]]> RSC current inner loop PI parameters 0.6 / 8 <![CDATA[k p3 ,k I3 ]]> GSC voltage outer loop PI parameters 8 / 400 <![CDATA[k p4 ,k I4 ]]> GSC current inner loop PI parameters 0.83 / 5 <![CDATA[C d ]]> DC bus capacitor 450μF <![CDATA[U dc ]]> DC bus voltage 1200V <![CDATA[r g ]]> Transmission line resistance per unit length 0.115Ω / km <![CDATA[l g ]]> Transmission line inductance per unit length 1.05Mh / km

[0073] A positive-sequence disturbance with a frequency of (f+f0) and a negative-sequence disturbance with a frequency of (f-f0) are injected, where f is the impedance characteristic frequency in the dq0 coordinate system and f0 is the grid fundamental frequency. Specifically, based on the frequency sweep method, two sets of linearly independent disturbances are input and impedance data are collected. The relationship between disturbance and impedance is shown in the following formula:

[0074]

[0075] Among them, the superscript represents the corresponding element of the matrix; Tables 1 and 2 below represent the number of disturbance responses; f p Represents the positive sequence disturbance frequency; ω0 represents the grid angular frequency; U represents the voltage disturbance response; I represents the current disturbance response. The superscripts p and n represent the positive sequence and negative sequence disturbance responses respectively. The superscripts pp and pn represent the positive and negative sequence voltages, and the superscripts represent the positive and negative sequence currents. Specifically, Represents the impedance calculated from the measured positive sequence voltage and positive sequence current; represents the impedance calculated from the measured positive sequence voltage and negative sequence current; represents the impedance calculated from the measured negative sequence voltage and positive sequence current; Represents the impedance calculated from the measured negative-sequence voltage and negative-sequence current.

[0076] represents the load positive sequence voltage response measured under the first disturbance injection; represents the load positive sequence voltage response measured under the second disturbance injection; represents the load negative sequence voltage response measured under the first disturbance injection; It represents the load negative sequence voltage response measured under the second disturbance injection. represents the load positive sequence current response measured under the first disturbance injection; represents the load positive sequence current response measured under the second disturbance injection; represents the load negative sequence current response measured under the first disturbance injection; It represents the load negative sequence current response measured under the second disturbance injection.

[0077] When measuring dq impedance, the injected disturbance is still the three-phase voltage or current disturbance. Taking voltage disturbance as an example, the two injected disturbances can be designed according to the following formula:

[0078]

[0079] Where, f is the impedance characteristic frequency in the dq coordinate system; f0 is the grid fundamental frequency; U is the voltage amplitude; u 1a 、u 1b 、u 1c are the voltage amplitudes of phases a, b, and c, respectively; t is time. Tables 1 and 2 below show the first and second disturbance responses, respectively.

[0080] Therefore, to measure the dq impedance characteristics at a frequency of f, a positive sequence disturbance with a frequency of (f+f0) can be injected for the first time, and a negative sequence disturbance with a frequency of (f-f0) can be injected for the second time.

[0081] Through these two linearly independent perturbations, the dq impedance characteristics at frequency f are obtained, and the calculation results are:

[0082]

[0083] Where Δu d1 , Δi d1 are the voltage and current response components of the d-axis after the first disturbance; Δu d2 , Δi d2 are the voltage and current response components of the d-axis after the second disturbance; Δu q1 , Δi q1 are the voltage and current response components of the q axis after the first disturbance; Δu q2 , Δi q2 They are the voltage and current response components of the q-axis after the second disturbance.

[0084] Finally, the impedance characteristics of the doubly fed wind turbine grid-connected system at various frequency points are measured using the sweep frequency method.

[0085] Step S3: construct and train a single doubly fed wind turbine impedance identification model based on a back propagation neural network optimized by a genetic algorithm, aggregate the station impedance identification model with the trained single doubly fed wind turbine impedance identification model, and then perform impedance identification to obtain the equivalent impedance of the wind farm after aggregation.

[0086] The construction and training of a single doubly fed wind turbine impedance identification model based on a back propagation neural network optimized by a genetic algorithm specifically includes the following steps:

[0087] The collected data are preprocessed to obtain a data set, which is then divided into a training set, a test set, and a validation set. The data set includes the d-axis voltage U in the dq coordinate system. d Current I d , q-axis voltage U in dq coordinate system q , current I q , frequency f and Z dq (f) The amplitude and phase angle of each element of the matrix.

[0088] A single doubly fed wind turbine impedance identification model based on a back propagation neural network (GA-BP) optimized by a genetic algorithm is constructed. The back propagation neural network consists of an input layer, several hidden layers, and an output layer. In essence, it is a neural network that can identify the characteristic relationship between an input data sequence of 5 elements and an output data sequence of 8 elements. The training set is input into the back propagation neural network for training. The input is the d-axis voltage U in the dq coordinate system. d , current I d , q-axis voltage U in dq coordinate system q , current I q , frequency f, output is Z dq (f) The amplitude and phase angle of each element of the matrix; and the genetic algorithm is used to optimize the weights and bias of the back-propagation neural network. The model is tested with a test set during the training process, and the model is verified with a validation set after the training. When the accuracy of the model reaches the set threshold, the trained single doubly fed wind turbine impedance identification model is output.

[0089] The output active power per unit value of this embodiment varies from 0.10 pu to 1.0 pu, with an equal interval and a step size of 0.05 pu. From low to high active power output, they are named working condition 1 (0.10 pu), working condition 2 (0.15 pu) ... working condition 19 (1.00 pu). The dq impedance frequency measured at each working point is between 10-2000 Hz, using a segmented equal interval step size. The low frequency band is 10-49 Hz, with a sweep step size of 1 Hz; the mid-frequency band is 60-500 Hz, with a sweep step size of 10 Hz; the high frequency band is 550-2000 Hz, with a sweep step size of 50 Hz. There are 115 frequency points under each working condition, and a total of 2185 points in 19 working conditions.

[0090] Here, the training set, test set, and validation set are divided into 70%, 15%, and 15% ratios. The impedance sweep measurement data of 13 working conditions are randomly selected as the training set, the data of 3 working conditions are used as the validation set, and the data of 3 working conditions are used as the test set. Taking working condition 19 (1.00pu) and working condition 10 (0.55pu) as examples, the amplitude and phase angle of each element of the dq impedance matrix measured by the sweep method change with frequency as shown in the figure below. Figure 4 shown.

[0091] Normalize the collected data:

[0092]

[0093] Among them, x nor is the normalized data, x is the data before normalization, μ is the mean of the data, and σ is the standard deviation of the data.

[0094] The BP neural network structure is as follows Figure 5 The performance of BP neural networks with different structures in this case is shown in Table 2.

[0095] Table 2 Performance of BP neural networks with different structures

[0096]

[0097]

[0098] The table above shows that when the number of neurons is 30, there is a tendency for overfitting and insufficient generalization ability. Therefore, the number of hidden layer neurons is set to 20. At this time, the weight matrix size from the input layer to the hidden layer is , the weight matrix size from the hidden layer to the output layer is , the hidden layer threshold is 20, and the output layer threshold is 8.

[0099] In this structure, the chromosome length of the GA model individuals is 288, the number of individuals per generation is set to 10, the number of iterations is set to 30, and the mean square error (MSE) of the validation set is used as the fitness function for the individual to the environment. After the iteration is completed, the individual with the best fitness is selected and output, and the weight matrix and threshold of this individual are pre-imported into the neural network. This is used as the basis for further training to derive the final BP neural network.

[0100] The impedance identification model of a single DFIG turbine is aggregated with the station impedance identification model after training, and then the impedance identification is carried out, which specifically includes the following steps:

[0101] The dq voltage and current amplitude U of each doubly fed wind turbine AC output port d-i , U q-i , I d-i , I q-i Solve the impedance characteristic matrix Z of all doubly fed wind turbines in the wind farm stationdq-i , where i represents the i-th wind turbine, i = 1, 2…n; n is the number of doubly fed wind turbines in the wind farm; U d-i , U q-i are the voltage amplitudes of the d and q axes at the AC output port of the i-th wind turbine respectively; I d-i , I q-i are the d-axis and q-axis current amplitudes of the AC output port of the i-th wind turbine respectively; at this time, each doubly fed wind turbine can be equivalent to a controlled current source in parallel with the admittance. Combining the impedance models of the doubly fed wind turbine, transmission line, transformer and other parts, taking the chain structure as an example, the equivalent impedance structure diagram of the doubly fed wind turbine station grid-connected system is as follows: Figure 6 As shown, there are n feeders in total, and multiple wind turbines are connected in parallel on each feeder. The current source and impedance in parallel represent a wind turbine, and the remaining impedances represent the feeder line impedance.

[0102] The impedance matrix Z of each doubly fed wind turbine dq-i It is established in its own local dq coordinate system. The impedance of the doubly fed wind turbines in the whole station is aggregated, and a common dq coordinate system is selected as the phase reference point, namely the PCC point, and then the impedance matrix Z of each doubly fed wind turbine is converted into dq-i Convert to this dq coordinate system.

[0103] The phase angle θ of each doubly fed wind turbine in the local dq coordinate system is measured using a phase-locked loop i , the phase angle θ of the PCC point reference dq coordinate system pcc , then the offset angle Δθ between the local coordinate system and the reference coordinate system is i =θ pcc -θ i , use the following formula to transform the coordinate system:

[0104]

[0105] Where Z dq-i (f) is the impedance matrix of the local dq coordinate system of the i-th wind turbine at frequency f, Z′ dq-i (f) is the impedance matrix of the i-th wind turbine in the reference dq coordinate system after conversion, T trans-i is the conversion matrix of the i-th wind turbine, It's T trans-i The inverse matrix of .

[0106] T trans-i The calculation method is:

[0107]

[0108] After the impedance coordinate system conversion is completed, the impedance matrix is ​​merged and simplified by series-parallel connection, Y-Δ transformation, etc. The simplification method varies according to different circuits. The present invention uses the specific embodiment drawings as an example to illustrate the simplification. Figure 6 The n wind turbines in each feeder can be simplified into a series line by connecting them in series and parallel, and the two feeders can be simplified into a parallel line by connecting them in parallel. The equivalent schematic diagram of the simplified grid-connected system is shown in the figure below. Figure 7 As shown, where Z DFIG (s) represents the equivalent impedance of the wind farm after aggregation, i DFIG (s) represents the equivalent output current source at the end of the doubly fed wind turbine, Z net (s) represents the grid transmission line impedance, i g (s) represents the current at the network end, u pcc (s) represents the PCC point voltage, u g (s) represents the grid terminal voltage.

[0109] The impedance characteristics of each doubly fed wind turbine are aggregated to the PCC point of the grid-connected system to obtain the equivalent impedance Z of the wind farm after aggregation. DFIG (s).

[0110] Step S4, based on the generalized Nyquist criterion and using the impedance identification model to analyze the stability of the doubly fed wind turbine grid-connected system, specifically includes the following steps:

[0111] Using the GA-BP impedance identification results, i.e. the equivalent impedance of the wind farm after aggregation, combined with the grid-side impedance, the generalized Nyquist criterion is used to draw the system characteristic trajectory and determine the grid-connected system stability;

[0112] When Z g (s) / Z DFIG (s) satisfies the Nyquist criterion, the system is stable; when Z g (s) / Z DFIG When (s) does not satisfy the Nyquist criterion, the system is unstable, where Z g (s) represents the equivalent impedance of the power grid; Z DFIG (s) represents the equivalent impedance of the wind farm after aggregation.

[0113] The subsynchronous oscillation frequency can be calculated using the generalized Nyquist curve, and the crossing frequency corresponding to the intersection of its characteristic trajectory and the unit circle is the subsynchronous oscillation frequency.

[0114] This embodiment builds a wind farm grid-connected system with three feeders, each of which has six 1.5MW double-fed wind turbines connected in parallel. Figure 8 shown.

[0115] On the DFIG side, the fan output of feeder 1 is set to 1.00 pu, the fan output of feeder 2 is set to 0.85 pu, and the fan output of feeder 3 is set to 0.70 pu. Using the GA-BP neural network, the impedance of each DFIG is first identified, and then the coordinate system is transformed by the equation, and finally the impedance matrix is ​​connected in parallel to the PCC point. The generalized characteristic root locus of the system is as follows: Figure 9 As shown, the vertical axis is the imaginary axis, the horizontal axis is the real axis, the line is the generalized root locus curve, and the red dotted line represents the unit circle.

[0116] Comparison of GA-BP impedance identification results and sweep frequency measurement results Figure 10 As shown, the red curve is the impedance curve obtained by frequency sweep measurement, and the blue curve is the impedance curve obtained by the method proposed in the present disclosure.

[0117] Secondly, the FFT decomposition results of the PCC point voltage at the station are as follows: Figure 11 As shown in Figure 2, the FFT decomposition results show that except for the power frequency of 50 Hz, the PCC point voltage does not have obvious harmonics of other frequencies, and the system operates stably.

[0118] The time domain simulation results are compared with the impedance identification results to verify the effectiveness of the model.

[0119] Impedance analysis is an effective tool for studying subsynchronous oscillation phenomena in renewable energy grid-connected systems. However, due to the large number of wind turbines in a station and the complex operating conditions, the traditional state-space equation method is computationally intensive, and the frequency sweep measurement method is labor-intensive. The data-driven impedance identification method for a doubly fed wind turbine grid-connected system proposed in this invention achieves a seamless transition from "single-machine data drive" to "station physical modeling" by performing multi-dimensional coupling calculations on the output of a single-machine GA-BP model and the network impedance matrix. This allows the model to maintain strong generalization capabilities even when the number of units changes, greatly accelerating the speed of impedance identification while ensuring its accuracy.

[0120] Example 2:

[0121] Based on the same inventive concept as in Example 1, Figure 12 As shown, this embodiment provides a data-driven doubly-fed wind turbine grid-connected system impedance identification system, which is applied to the method described above, including:

[0122] The input and output determination module is used to determine the input and output of the system based on the DQ theoretical impedance model of the doubly fed wind turbine grid-connected system. The data collection module is used to build a time-domain simulation model of a single doubly fed wind turbine grid-connected system and use the sweep frequency method to collect data under various operating conditions.

[0123] The impedance identification module is used to construct and train a single doubly fed wind turbine impedance identification model based on a back propagation neural network optimized by a genetic algorithm. The trained single doubly fed wind turbine impedance identification model is then used to aggregate the station impedance identification model. Impedance identification is then performed to obtain the equivalent impedance of the aggregated wind farm station.

[0124] The analysis module is used to analyze the stability of the doubly fed wind turbine grid-connected system based on the generalized Nyquist criterion and the impedance identification model.

[0125] Example 3:

[0126] Based on the same inventive concept as Example 1, this embodiment provides a computer-readable storage medium, which includes a stored program, wherein when the program is running, the device where the computer-readable storage medium is located is controlled to execute the data-driven doubly fed wind turbine grid-connected system impedance identification method.

[0127] Example 4:

[0128] Based on the same inventive concept as that of Example 1, this embodiment provides a processor, which is used to run a program, wherein when the program is running, the data-driven method for identifying the impedance of a doubly-fed wind turbine grid-connected system is executed.

[0129] Those skilled in the art will appreciate that the modules of the various examples described in conjunction with the embodiments disclosed herein can be implemented in electronic hardware, computer software, or a combination of the two. In order to clearly illustrate the interchangeability of hardware and software, the composition of each example has been generally described in terms of function in the above description. Whether these functions are performed in hardware or software depends on the specific application and design constraints of the technical solution. Professional and technical personnel can use different methods to implement the described functions for each specific application, but such implementation should not be considered to be beyond the scope of the present invention.

[0130] In the embodiments provided by the present invention, it should be understood that the division of modules is merely a logical function division, and there may be other division methods in actual implementation, for example, multiple modules can be combined into one module, one module can be split into multiple modules, or some features can be ignored, etc.

[0131] In addition, the functional modules in various embodiments of the present invention may be integrated into a single processing module, or each module may exist physically separately, or two or more modules may be integrated into a single module. The aforementioned integrated modules may be implemented in the form of hardware or software functional modules.

[0132] If the integrated module is implemented in the form of a software functional module and sold or used as an independent product, it can be stored in a computer-readable storage medium. Based on this understanding, the technical solution of the present invention, or the part that contributes to the prior art, or all or part of the technical solution can be embodied in the form of a software product. The computer software product is stored in a storage medium and includes several instructions for enabling a computer device (which can be a personal computer, server or network device, etc.) to perform all or part of the steps of the method described in each embodiment of the present invention. The aforementioned storage medium includes: U disk, read-only memory (ROM, Read-0nly Memory), random access memory (RAM, Random Access Memory), mobile hard disk, magnetic disk or optical disk, etc., various media that can store program code.

[0133] Finally, it should be noted that the above embodiments are only used to illustrate the technical solutions of the present invention, rather than to limit it. Although the present invention has been described in detail with reference to the above embodiments, those skilled in the art should understand that they can still modify the technical solutions described in the above embodiments, or make equivalent replacements for some or all of the technical features therein. These modifications or replacements do not deviate the essence of the corresponding technical solutions from the scope of the technical solutions of the embodiments of the present invention, and they should all be included in the scope of the claims and description of the present invention.

Claims

1. A data-driven impedance identification method for a doubly-fed wind turbine grid-connected system, characterized in that: The following steps are involved: Step S1, determining the input and output of the system based on the dq theoretical impedance model of the doubly fed wind turbine grid-connected system; Step S2: Building a time domain simulation model of a single doubly-fed wind turbine grid-connected system and using a frequency sweep method to collect data under various operating conditions; Step S3: constructing and training a single doubly-fed wind turbine impedance identification model based on a back propagation neural network optimized by a genetic algorithm, aggregating the station impedance identification model with the trained single doubly-fed wind turbine impedance identification model, and then performing impedance identification to obtain the equivalent impedance of the wind farm after aggregation; Step S4: analyzing the stability of the doubly-fed wind turbine grid-connected system using an impedance identification model based on the generalized Nyquist criterion.

2. The data-driven impedance identification method for a doubly-fed wind turbine grid-connected system according to claim 1, characterized in that: Determine the inputs and outputs of the system including: The impedance characteristics have a nonlinear mapping relationship with the characteristic parameters of the grid connection point port. The theoretical impedance model is as follows: Among them, Z dq is the theoretical impedance, f is the frequency, F1 represents a mapping relationship, U d , I d Indicates the d-axis voltage and current in the dq coordinate system, U q , I q Represents the q-axis voltage and current in the dq coordinate system; Z dd (f), Z dq (f), Z qd (f), Z qq (f) are the components of the theoretical impedance respectively.

3. The data-driven impedance identification method for a doubly-fed wind turbine grid-connected system according to claim 1, characterized in that: The step S2 specifically includes the following steps: Build a time domain simulation model of a single doubly fed wind turbine grid-connected system; Inject positive sequence disturbance with frequency (f+f0) and negative sequence disturbance with frequency (f-f0), where f is the impedance characteristic frequency in the dq0 coordinate system and f0 is the grid fundamental frequency; Through these two linearly independent perturbations, the dq impedance characteristics at frequency f are obtained, and the calculation results are: Where Δu d1 , Δi d1 are the voltage and current response components of the d-axis after the first disturbance; Δu d2 , Δi d2 are the voltage and current response components of the d-axis after the second disturbance; Δu q1 , Δi q1 are the voltage and current response components of the q axis after the first disturbance; Δu q2 , Δi q2 are the voltage and current response components of the q-axis after the second disturbance; Finally, the impedance characteristics of the doubly fed wind turbine grid-connected system at various frequency points are measured using the sweep frequency method.

4. The data-driven impedance identification method for a doubly-fed wind turbine grid-connected system according to claim 1, characterized in that: The step S3 of constructing a single doubly fed wind turbine impedance identification model based on a back propagation neural network optimized by a genetic algorithm and training the model specifically includes the following steps: The collected data are preprocessed to obtain a data set, which is then divided into a training set, a test set, and a validation set. The data set includes the d-axis voltage U in the dq coordinate system. d , current I d , q-axis voltage U in dq coordinate system q , current I q , frequency f and Z dq (f) The amplitude and phase angle of each element of the matrix; A single doubly fed wind turbine impedance identification model based on a back propagation neural network optimized by genetic algorithm is constructed, wherein the back propagation neural network consists of an input layer, several hidden layers, and an output layer; the training set is input into the back propagation neural network for training, and the input is the d-axis voltage U in the dq coordinate system d , current I d , q-axis voltage U in dq coordinate system q , current I q , frequency f, output is Z dq (f) The amplitude and phase angle of each element of the matrix; and the genetic algorithm is used to optimize the weights and bias of the back-propagation neural network. The model is tested with a test set during the training process, and the model is verified with a validation set after the training. When the accuracy of the model reaches the set threshold, the trained single doubly fed wind turbine impedance identification model is output.

5. The data-driven impedance identification method for a doubly-fed wind turbine grid-connected system according to claim 1, characterized in that: In step S3, the station impedance identification model is aggregated with the trained single doubly-fed wind turbine impedance identification model, and then impedance identification is performed, which specifically includes the following steps: The dq voltage and current amplitude U of each doubly fed wind turbine AC output port d-i , U q-i , I d-i , I q-i Solve the impedance characteristic matrix Z of all doubly fed wind turbines in the wind farm station dq-i , where i represents the i-th wind turbine, i = 1, 2…n; n is the number of doubly fed wind turbines in the wind farm; U d-i , U q-i are the voltage amplitudes of the d and q axes at the AC output port of the i-th wind turbine respectively; I d-i , I q-i are the d-axis and q-axis current amplitudes of the AC output port of the i-th wind turbine respectively; The impedance of all doubly fed wind turbines in the station is aggregated, and a common dq coordinate system is selected as the phase reference point, namely the PCC point, and then the impedance matrix Z of each doubly fed wind turbine is calculated. dq-i Convert to this dq coordinate system; The phase angle θ of each doubly fed wind turbine in the local dq coordinate system is measured using a phase-locked loop i , the phase angle θ of the PCC point reference dq coordinate system pcc , then the offset angle Δθ between the local coordinate system and the reference coordinate system is i =θ pcc -θ i , use the following formula to transform the coordinate system: Where Z dq-i (f) is the impedance matrix of the local dq coordinate system of the i-th wind turbine at frequency f, Z d ' q-i (f) is the impedance matrix of the i-th wind turbine in the reference dq coordinate system after conversion, T trans-i is the conversion matrix of the i-th wind turbine, It's T trans-i The inverse matrix of After completing the impedance coordinate system transformation, the impedance characteristics of each doubly fed wind turbine are aggregated to the PCC point of the grid-connected system to obtain the equivalent impedance Z of the wind farm after aggregation. DFIG (s).

6. The data-driven impedance identification method for a doubly-fed wind turbine grid-connected system according to claim 1, characterized in that: The conversion matrix T of the i-th wind turbine trans-i The specific ones are:

7. The data-driven impedance identification method for a doubly-fed wind turbine grid-connected system according to claim 1, characterized in that: The step S4 specifically includes the following steps: By using the equivalent impedance of the aggregated wind farms and the grid-side impedance, the generalized Nyquist criterion is used to draw the system characteristic trajectory and determine the stability of the grid-connected system. When Z g (s) / Z DFIG (s) satisfies the Nyquist criterion, the system is stable; when Z g (s) / Z DFIG When (s) does not satisfy the Nyquist criterion, the system is unstable, where Z g (s) represents the equivalent impedance of the power grid; Z DFIG (s) represents the equivalent impedance of the wind farm after aggregation.

8. A data-driven impedance identification system for a doubly-fed wind turbine grid-connected system, characterized in that: The method applied to any one of claims 1 to 7, comprising: The input and output determination module is used to determine the input and output of the system based on the DQ theoretical impedance model of the doubly fed wind turbine grid-connected system. The data collection module is used to build a time-domain simulation model of a single doubly fed wind turbine grid-connected system and use the sweep frequency method to collect data under various operating conditions. The impedance identification module is used to construct and train a single doubly fed wind turbine impedance identification model based on a back propagation neural network optimized by a genetic algorithm. The trained single doubly fed wind turbine impedance identification model is then used to aggregate the station impedance identification model. Impedance identification is then performed to obtain the equivalent impedance of the aggregated wind farm station. The analysis module is used to analyze the stability of the doubly fed wind turbine grid-connected system based on the generalized Nyquist criterion and the impedance identification model.

9. A computer-readable storage medium, characterized in that The computer-readable storage medium includes a stored program, wherein when the program is running, the device where the computer-readable storage medium is located is controlled to execute the data-driven impedance identification method for the doubly-fed wind turbine grid-connected system according to any one of claims 1 to 7.

10. A processor, characterized in that: The processor is used to run a program, wherein when the program is run, the data-driven doubly-fed wind turbine grid-connected system impedance identification method according to any one of claims 1 to 7 is executed.

Images