New energy station impedance identification method, system and device based on data driving and storage medium

Abstract

The invention discloses a new energy station impedance identification method, system and device based on data driving and a storage medium. The method comprises the steps that two sets of independent disturbance signals are designed; respectively injecting two groups of disturbance signals into each single-machine grid-connected point, respectively measuring voltage and current signals of the single-machine grid-connected point under an abc coordinate system, and converting the voltage and current signals into voltage and current signals of the single-machine grid-connected point under a dq axis; single-machine grid-connected point impedance under the dq axis is calculated, and impedance data of all single-machine grid-connected points are obtained; dividing all impedance data and all voltage and current signals under the dq axis; initializing the ANN model to obtain a basic ANN model; and training and evaluating the basic ANN model, adjusting the basic ANN model according to an evaluation result, and finally outputting an ANN-based impedance identification model. According to the method, internal control parameters of the converter are not needed, the impedance characteristic can be updated online, and the new energy station impedance based on the voltage source converter can be identified under wide operating points.

Description

Technical Field

[0001] The present invention relates to the technology of power electronic system stability analysis, and specifically to a method, system, device and storage medium for impedance identification of new energy power stations based on data driving. Background Art

[0002] The access of large-scale new energy has changed the operation mode of traditional power systems and triggered a series of power system stability problems. Among them, the stability of power electronic power systems is mainly affected by the impedance of power electronic devices. Due to the non-linear and time-varying characteristics of power electronic devices, and the internal control parameters are often unknown, it is difficult to model their impedance. The impedance modeling of new energy power stations first requires obtaining the impedance models of each new energy unit. For new energy units with black box problems, it is necessary to obtain their impedance models through sweep frequency measurement, which usually takes a long time. For new energy units with known control structures and parameters, the order of their impedance models will increase rapidly with the increase of the power station scale, resulting in difficulties in solving theoretical models. Summary of the Invention

[0003] Object of the Invention: The object of the present invention is to provide a method, system, device and storage medium for impedance identification of new energy power stations based on data driving. This method does not require the internal control parameters of the converter, can update the impedance characteristics online, and can identify the impedance of new energy power stations based on voltage source converters at a wide range of operating points.

[0004] Technical Solution: A method for impedance identification of new energy power stations based on data driving according to the present invention includes:

[0005] Design two sets of linearly independent perturbation signals;

[0006] Inject the first set of perturbation signals at each single-unit connection point, measure the voltage and current signals at the single-unit connection point in the abc coordinate system, and convert the voltage and current signals at the single-unit connection point in the abc coordinate system into voltage and current signals at the single-unit connection point in the dq axis;

[0007] Inject the second set of perturbation signals at each single-unit connection point, measure the voltage and current signals at the single-unit connection point in the abc coordinate system, and convert the voltage and current signals at the single-unit connection point in the abc coordinate system into voltage and current signals at the single-unit connection point in the dq axis;

[0008] Calculate the impedance at the single-unit connection point in the dq axis according to the voltage and current signals at the single-unit connection point in the two sets of dq axes, and finally obtain the impedance data of all single-unit connection points;

[0009] Divide the impedance data of all single-unit connection points and the voltage and current signals at all single-unit connection points in the dq axis into a training set and a test set;

[0010] Initialize the ANN model to obtain the basic ANN model; train the basic ANN model using the training set; evaluate the trained basic ANN model using the test set, and adjust the basic ANN model according to the evaluation results, and finally output the impedance identification model based on ANN.

[0011] Further, the calculation formula for the impedance of the single-machine grid connection point under the dq axis is as follows:

[0012]

[0013] In the formula, Z dq represents the converter dq-axis impedance value; Z dd represents the self-impedance of the d-axis; Z dq represents the coupling impedance between the d-axis and the q-axis; Z qd represents the coupling impedance between the q-axis and the d-axis; Z qq represents the self-impedance of the q-axis; V d1 represents the measured value of the d-axis of the grid connection point voltage of the first group of disturbance signals; V q1 represents the measured value of the q-axis of the grid connection point voltage of the first group of disturbance signals; V d2 represents the measured value of the d-axis of the grid connection point voltage of the second group of disturbance signals; V q2 represents the measured value of the q-axis of the grid connection point voltage of the second group of disturbance signals; I d1 represents the measured value of the d-axis of the grid connection point current of the first group of disturbance signals; I q1 represents the measured value of the q-axis of the grid connection point current of the first group of disturbance signals; I d2 represents the measured value of the d-axis of the grid connection point current of the second group of disturbance signals; I q2 represents the measured value of the q-axis of the grid connection point current of the second group of disturbance signals.

[0014] Further, the basic ANN model includes an input layer, a hidden layer, and an output layer. The voltage and current signals of the single-machine grid connection point under the dq axis and the corresponding frequency value f are input to the input layer, and the output layer outputs the impedance amplitude and phase under the dq axis.

[0015] Further, the data of the basic ANN model starts from the input layer, passes through the hidden layer, and reaches the output layer. In the forward propagation, the weighted sum of the inputs is calculated for each node at each level, and then, an output is generated via an activation function, specifically as follows:

[0016] For the j-th neuron in the l-th layer, its output is expressed as:

[0017]

[0018] In the formula, Represents the output of the $i$-th neuron in the upper layer; Represents the weight between the $i$-th neuron and the $j$-th neuron in the $l$-th layer; Represents the bias of the $j$-th neuron in the $l$-th layer; $f$ represents the activation function.

[0019] Furthermore, training the basic ANN model using the training set includes:

[0020] Adopting the multi-period mini-batch stochastic gradient descent algorithm and using the mean squared error loss function as the default metric for evaluating the performance of the regression algorithm, specifically as follows:

[0021]

[0022] In the formula, MSE represents the mean squared error of the loss function; and respectively represent the output of the $d$-th basic ANN model at sample index $n$ and the measured impedance data; $n$ represents the number of mini-batches, $N$ is its maximum value; $d$ represents the norm of the output vector of the basic ANN model, $D$ is its maximum value; $w$ l and $b$ l represent the weight and bias parameters to be learned in the $l$-th layer;

[0023] Initializing the weight and bias parameters using the Gaussian distribution, and the update formula for the learning rate $\eta$ is as follows:

[0024]

[0025] In the formula, and respectively represent the weight and bias parameters of the Gaussian distribution at the current step; and respectively represent the weight and bias parameters of the Gaussian distribution at the previous step; $\eta$ represents the learning rate.

[0026] Furthermore, evaluating the trained basic ANN model using the test set and adjusting the basic ANN model according to the evaluation results, and finally outputting the ANN-based impedance identification model, including:

[0027] Evaluating the generated model using the coefficient of determination ($R$ 2 ) to characterize the training effect

[0028]

[0029] In the formula, $y$ i represents the training data; represents the average value of $y$ i ; $f$ i represents the corresponding data generated by the ANN training model.

[0030] Further, according to the relationship between the coefficient of determination and the number of hidden layer neurons, an appropriate number of hidden layer neurons is determined.

[0031] Based on the same inventive concept, a data-driven impedance identification system for a new energy power station of the present invention includes:

[0032] A disturbance signal design module for designing two sets of linearly independent disturbance signals;

[0033] A measurement module for injecting the first set of disturbance signals at each single-machine connection point, measuring the voltage and current signals of the single-machine connection point in the abc coordinate system, and converting the voltage and current signals of the single-machine connection point in the abc coordinate system into the voltage and current signals of the single-machine connection point in the dq axis;

[0034] The measurement module is used to inject the second set of disturbance signals at each single-machine connection point, measure the voltage and current signals of the single-machine connection point in the abc coordinate system, and convert the voltage and current signals of the single-machine connection point in the abc coordinate system into the voltage and current signals of the single-machine connection point in the dq axis;

[0035] An impedance calculation module for calculating the impedance of the single-machine connection point in the dq axis according to the voltage and current signals of the single-machine connection point in the two sets of dq axes, and finally obtaining the impedance data of all single-machine connection points;

[0036] A data division module for dividing the impedance data of all single-machine connection points and the voltage and current signals of all single-machine connection points in the dq axis into a training set and a test set;

[0037] A model training and evaluation module for initializing the ANN model to obtain a basic ANN model; training the basic ANN model using the training set; evaluating the trained basic ANN model using the test set, and adjusting the basic ANN model according to the evaluation results, and finally outputting an ANN-based impedance identification model.

[0038] Based on the same inventive concept, a data-driven impedance identification device for a new energy power station of the present invention includes a processor and a memory, and computer instructions are stored in the memory. The processor is used to execute the computer instructions stored in the memory. When the computer instructions are executed by the processor, the electronic device implements the steps of the above-mentioned data-driven impedance identification method for a new energy power station.

[0039] Based on the same inventive concept, a computer-readable storage medium of the present invention stores a computer program, and when the program is executed by a processor, the steps of the above-mentioned data-driven impedance identification method for a new energy power station are implemented.

[0040] Advantageous effects: Compared with the prior art, the remarkable technical effects of the present invention are as follows:

[0041] The present invention proposes a method for impedance identification based on deep learning, which is used to update impedance characteristics online and can accurately identify the impedance of new energy power stations under different working conditions. The present invention can solve the problem that it is difficult to model the impedance due to the nonlinear and time-varying characteristics of power electronic devices and the internal control logic and parameters are difficult to obtain because the manufacturers keep them confidential. Compared with traditional impedance modeling methods, the technical solution of the present invention does not require detailed parameters of the system and its internal control details, and only needs to measure relevant data at the terminal, which is suitable for solving the black box problem of system impedance modeling. Brief description of the drawings

[0042] Figure 1 is a schematic flow chart of a data-driven method for impedance identification of a new energy power station disclosed in an embodiment of the present invention;

[0043] Figure 2 is a schematic structural diagram of a basic ANN model disclosed in an embodiment of the present invention;

[0044] Figure 3 is a schematic diagram of a method for measuring impedance data of a new energy power station disclosed in an embodiment of the present invention;

[0045] Figure 4 is an error analysis diagram of an impedance identification model based on ANN and verification data disclosed in an embodiment of the present invention;

[0046] Figure 5 is a schematic structural diagram of a data-driven system for impedance identification of a new energy power station disclosed in an embodiment of the present invention;

[0047] Figure 6 is a schematic structural diagram of a data-driven device for impedance identification of a new energy power station disclosed in an embodiment of the present invention. Detailed implementation manners

[0048] The present invention will be described in detail below with reference to the drawings and specific embodiments. Those skilled in the art will understand that the purposes and advantages that can be achieved by the present invention are not limited to the above-mentioned beneficial effects specifically described, and the above and other purposes that the present invention can achieve will be more clearly understood according to the following detailed description.

[0049] Those of ordinary skill in the art should understand that the various exemplary components, systems, and methods described in connection with the embodiments disclosed in the present invention can be implemented in hardware, software, or a combination of both. Specifically, whether to implement in hardware or software depends on the specific application and design and tree conditions of the technical solution. Those skilled in the art can use different methods to implement the described functions for each specific application, but such implementation should not be considered to exceed the scope of the present invention.

[0050] The mention of "embodiment" in the present invention means that the specific features, structures, or characteristics described in connection with the embodiment may be included in at least one embodiment of the present invention. The appearance of this phrase at various positions in the specification does not necessarily refer to the same embodiment, nor is it an independent or alternative embodiment mutually exclusive with other embodiments. Those skilled in the art explicitly and implicitly understand that the embodiments described herein can be combined with other embodiments.

[0051] Embodiment 1

[0052] Please refer to Figure 1 , Figure 1 which is a schematic flow diagram of a data-driven impedance identification method for a new energy power station disclosed in an embodiment of the present invention. Among them, Figure 1 the described impedance identification method for a new energy power station is applied in a power system, such as for the stability analysis of a grid-connected system, etc., and the embodiments of the present invention do not make limitations. As Figure 1 shown, the data-driven impedance identification method for a new energy power station may include the following operations:

[0053] S1. Design two sets of linearly independent perturbation signals.

[0054] S2. Inject the first set of perturbation signals at each single-machine grid connection point, measure the voltage and current signals at the single-machine grid connection point in the abc coordinate system, and convert the voltage and current signals at the single-machine grid connection point in the abc coordinate system into voltage and current signals at the single-machine grid connection point in the dq axis.

[0055] In this embodiment, after converting the voltage and current signals at the single-machine grid connection point in the abc coordinate system into voltage and current signals at the single-machine grid connection point in the dq axis, the Fast Fourier Transform (FFT) is used to extract V d1 , V q1 , I d1 , I q1 .

[0056] S3. Inject a second set of disturbance signals at each single - machine connection point, measure the voltage and current signals at the single - machine connection point in the abc coordinate system, and convert the voltage and current signals at the single - machine connection point in the abc coordinate system into the voltage and current signals at the single - machine connection point in the dq axis.

[0057] In this embodiment, after converting the voltage and current signals at the single - machine connection point in the abc coordinate system into the voltage and current signals at the single - machine connection point in the dq axis, use FFT to extract V d2 、V q2 、I d2 、I q2 。

[0058] S4. According to the voltage and current signals at the single - machine connection point in the two sets of dq axes, calculate the impedance at the single - machine connection point in the dq axis, and finally obtain the impedance data of all single - machine connection points.

[0059] In this embodiment, the calculation formula for the impedance at the single - machine connection point in the dq axis is as follows:

[0060]

[0061] In the formula, Z dq represents the converter dq - axis impedance value; Z dd represents the self - impedance of the d - axis; Z dq represents the coupling impedance between the d - axis and the q - axis; Z qd represents the coupling impedance between the q - axis and the d - axis; Z qq represents the self - impedance of the q - axis; V d1 represents the measured value of the d - axis of the connection - point voltage of the first set of disturbance signals; V q1 represents the measured value of the q - axis of the connection - point voltage of the first set of disturbance signals; V d2 represents the measured value of the d - axis of the connection - point voltage of the second set of disturbance signals; V q2 represents the measured value of the q - axis of the connection - point voltage of the second set of disturbance signals; I d1 represents the measured value of the d - axis of the connection - point current of the first set of disturbance signals; I q1 represents the measured value of the q - axis of the connection - point current of the first set of disturbance signals; I d2 represents the measured value of the d - axis of the connection - point current of the second set of disturbance signals; I q2 represents the measured value of the q - axis of the connection - point current of the second set of disturbance signals.

[0062] S5. Divide the impedance data of all single - machine connection points and the voltage and current signals at all single - machine connection points in the dq axis into a training set and a test set;

[0063] S6. Initialize the ANN model to obtain the basic ANN model; train the basic ANN model using the training set; evaluate the trained basic ANN model using the test set, and adjust the basic ANN model according to the evaluation results, and finally output the impedance identification model based on ANN.

[0064] The specific process of step S6 is as follows:

[0065] S6.1. Initialize the ANN model to obtain the basic ANN model.

[0066] As Figure 2 shown, the basic ANN model includes an input layer, a hidden layer, and an output layer. The voltage and current signals and the corresponding frequency value f at the single-machine grid connection point under the dq axis are input into the input layer, and the impedance amplitude and phase under the dq axis are output by the output layer.

[0067] As Figure 2 shown, the measured dq-axis voltages U d , U q and currents I d , I q at the grid connection point and the corresponding frequency value f are used as the five inputs of the basic ANN model in step S6, and the amplitudes and phases of the four elements Z dd , Z dq , Z qd , Z qq in the two-dimensional impedance matrix in Equation (1) are used as the eight outputs of the basic ANN model in step S6.

[0068] The basic ANN model can learn the complex and non-linear relationships in the input data through the hidden layer. In addition, the basic ANN model can adapt to different data types by changing the network structure (such as the number of layers and the number of neurons in each layer) and parameters.

[0069] The overall framework for the basic ANN model to identify the impedance distribution of power electronic devices at different operating points consists of an input layer, a hidden layer, and an output layer. The first step is to obtain the data set and measure the data that can reflect the relevant characteristics of the impedance operating point. Before the measurement starts, the simulation parameter range and interval need to be set, and the impedance data of the dq axis is obtained through impedance measurement. Considering that the impedance model of the converter will change with the change of the operating point, it is necessary to establish an impedance model with multiple operating points.

[0070] As Figure 1 shown, the non-linear modeling method based on neural network establishes a complex non-linear model through input data, and finally obtains the non-linear relationship between the impedance model with multiple operating points and the actual measurement data.

[0071] The data of the basic ANN model starts from the input layer, passes through the hidden layer, and reaches the output layer. In forward propagation, each node at each level calculates the weighted sum of the inputs, and then, through an activation function, generates the output. The principle of each specific step is as follows:

[0072] For the j-th neuron in the l-th layer, its output is expressed as:

[0073]

[0074] In the formula, represents the output of the i-th neuron in the previous layer (or the i-th input of the input layer); represents the weight between the i-th neuron and the j-th neuron in the l-th layer; represents the bias of the j-th neuron in the l-th layer; f represents the activation function.

[0075] S6.2. Train the basic ANN model using the training set, including:

[0076] The loss function is used to represent the difference between the output prediction of the trained neural network model and the target value of the original data set. The loss function adopted in this solution is the mean squared error (MSE), as follows

[0077]

[0078] In the formula, y i represents the true value; represents the predicted value; N represents the number of samples.

[0079] To improve the learning convergence of the algorithm, the multi-period mini-batch stochastic gradient descent algorithm is adopted, and the mean squared error loss function is used as the default metric for evaluating the performance of the regression algorithm, as follows:

[0080]

[0081] In the formula, MSE represents the mean squared error of the loss function; and respectively represent the output of the d-th basic ANN model at the sample index n and the measured impedance data; n represents the number of mini-batches, N is its maximum value; d represents the modulus of the output vector of the basic ANN model, D is its maximum value; w l and b l represent the weight and bias parameters to be learned in the l-th layer;

[0082] When initializing the weight and bias parameters, considering that parameter initialization based on the Gaussian distribution is applicable to unknown non-linear relationships, the Gaussian distribution is used to initialize the weight and bias parameters. The update formula for the learning rate η is as follows:

[0083]

[0084] In the formula, and respectively represent the weight and bias parameters of the Gaussian distribution at the current step; and respectively represent the weight and bias parameters of the Gaussian distribution at the previous step; η represents the learning rate.

[0085] S6.3. Evaluate the trained basic ANN model using the test set, and adjust the basic ANN model according to the evaluation results, and finally output the impedance identification model based on ANN, including:

[0086] After training is completed, use the coefficient of determination (R 2 ) to evaluate the generated model to characterize the training effect:

[0087]

[0088] In the formula, y i represents the training data; represents the average value of y i ; f i represents the corresponding data generated by the ANN training model. The coefficient of determination R 2 (also known as the coefficient of determination) is an important indicator used to evaluate the goodness of fit of the model in statistics and machine learning. It represents the degree of fit between the predicted value and the actual value of the model.

[0089] Based on the relationship between the coefficient of determination and the number of neurons in the hidden layer, accurately determine the appropriate number of neurons in the hidden layer.

[0090] Adjust the basic ANN model according to the evaluation results, so as to update the impedance of the new energy power station online, and the impedance characteristics of the power station under different working conditions can be identified.

[0091] The present invention proposes a method for calculating impedance characteristics based on a deep learning artificial neural network (ANN) for the black-box modeling of the impedance of a new energy power station based on a voltage source converter (VSC), which can identify the VSC impedance characteristics under different working conditions. A general identification framework for VSC is established using a feedforward neural network, and the accuracy and feasibility of the method are verified by comparing the results obtained by the analytical method with the operation results of the neural network model.

[0092] By training based on ANN offline measurement data, this method does not require the internal control parameters of the converter, can update the impedance characteristics online, and can identify the impedance of new energy power stations based on voltage source converters at a wide range of operating points.

[0093] As Figure 3 shown, the impedance identification model based on the feedforward neural network is verified by simulation. First, the voltage and current data at the PCC are collected and the corresponding impedance information is calculated, and these data are used for the division of the training and test sets. Then, the impedance distribution generated by the trained neural network model is compared and analyzed with the actual distribution, and the accuracy of the model is shown through the error analysis diagram. Figure 4 Figure (a) in Figure 4 shows the error of the modulus of impedance Zdd, Figure 4 Figure (b) in Figure 4 shows the error of the modulus of impedance Zdq, Figure 4 Figure (c) in Figure 4 shows the error of the modulus of impedance Zqd, Figure 4 Figure (d) in Figure 4 shows the error of the modulus of impedance Zqq; Figure 4 Figure (e) in

[0094] Example 2

[0095] Please refer to Figure 5 , Figure 5 which is a schematic structural diagram of a data-driven new energy power station impedance identification system disclosed in an embodiment of the present invention. This system can realize the stability analysis of the grid-connected system, and specifically includes:

[0096] A disturbance signal design module for designing two sets of linearly independent disturbance signals;

[0097] A measurement module for injecting the first set of disturbance signals at each single-machine grid connection point, measuring the voltage and current signals at the single-machine grid connection point in the abc coordinate system, and converting the voltage and current signals at the single-machine grid connection point in the abc coordinate system into the voltage and current signals at the single-machine grid connection point in the dq axis.

[0098] A measurement module, which is used to inject a second set of disturbance signals at each single-unit grid connection point, measure the voltage and current signals of the single-unit grid connection point in the abc coordinate system, and convert the voltage and current signals of the single-unit grid connection point in the abc coordinate system into the voltage and current signals of the single-unit grid connection point in the dq axis;

[0099] An impedance calculation module, which is used to calculate the impedance of the single-unit grid connection point in the dq axis according to the voltage and current signals of the single-unit grid connection point in two groups of dq axes, and finally obtain the impedance data of all single-unit grid connection points;

[0100] A data partitioning module, which is used to partition the impedance data of all single-unit grid connection points and the voltage and current signals of all single-unit grid connection points in the dq axis into a training set and a test set;

[0101] A model training and evaluation module, which is used to initialize the ANN model to obtain a basic ANN model; use the training set to train the basic ANN model; use the test set to evaluate the trained basic ANN model, and adjust the basic ANN model according to the evaluation results, and finally output an ANN-based impedance identification model.

[0102] In an optional embodiment, the data-driven new energy power station impedance identification method includes: a) designing two sets of linearly independent disturbance signals; b) injecting the first set of disturbance signals at each single-unit grid connection point, measuring the voltage and current signals of the single-unit grid connection point in the abc coordinate system, and converting them into the voltage and current signals of the single-unit grid connection point in the dq axis; c) injecting the second set of disturbance signals at each single-unit grid connection point, measuring the voltage and current signals of the single-unit grid connection point in the abc coordinate system, and converting them into the voltage and current signals of the single-unit grid connection point in the dq axis; d) calculating the impedance of the single-unit grid connection point in the dq axis according to the voltage and current signals of the single-unit grid connection point in two groups of dq axes, and finally obtaining the impedance data of all single-unit grid connection points; e) partitioning the impedance data of all single-unit grid connection points and the voltage and current signals of all single-unit grid connection points in the dq axis into a training set and a test set; f) initializing the ANN model to obtain a basic ANN model; training and evaluating the basic ANN model, and adjusting the basic ANN model according to the evaluation results, and finally outputting an ANN-based impedance identification model.

[0103] Embodiment 3

[0104] Please refer to Figure 6 , Figure 6 which is a schematic structural diagram of a data-driven new energy power station impedance identification device disclosed in an embodiment of the present invention. Among them, Figure 6 the described device can be applied to the power system, such as for the stability analysis of the grid-connected system, etc., and the embodiments of the present invention do not make limitations.

[0105] Such asFigure 6 As shown, the device may include a processor and a memory. Computer instructions are stored in the memory, and the processor is configured to execute the computer instructions stored in the memory. When the computer instructions are executed by the processor, the electronic device implements the steps of the method described in the above embodiments and can achieve the same technical effects as the above method.

[0106] The memory may include a computer system readable medium in the form of volatile memory, such as random access memory (RAM) and / or cache memory. The device may further include other removable / non-removable, volatile / non-volatile computer system storage media. By way of example only, the memory may be used to read and write a non-removable, non-volatile magnetic medium (commonly referred to as a "hard disk drive"). Programs / utilities with a set of (at least one) program modules may be stored, for example, in the memory. Such program modules include, but are not limited to, an operating system, one or more application programs, other program modules, and program data. Implementations of a network environment may be included in each or some combination of these examples. The program modules generally execute the functions and / or methods in the embodiments described in the present invention.

[0107] The processor executes various functional applications and data processing by running the programs stored in the memory, for example, implementing the method provided in Embodiment 1 of the present invention.

[0108] Embodiment 4

[0109] Embodiment 4 of the present invention further provides a computer-readable storage medium, on which a computer program is stored. When the program is executed by a processor, it implements the steps of the method described in the above embodiments and can achieve the same technical effects as the above method.

[0110] The computer storage medium of the embodiments of the present invention may adopt any combination of one or more computer-readable media. The computer-readable medium may be a computer-readable signal medium or a computer-readable storage medium. The computer-readable storage medium may be, for example, but not limited to, an electrical, magnetic, optical, electromagnetic, infrared, or semiconductor system, apparatus, or device, or any combination of the above. More specific examples (non-exhaustive list) of the computer-readable storage medium include: an electrical connection having one or more wires, a portable computer disk, a hard disk, a random access memory (RAM), a read-only memory (ROM), an erasable programmable read-only memory (EPROM or flash memory), an optical fiber, a portable compact disk read-only memory (CD-ROM), an optical storage device, a magnetic storage device, or any suitable combination of the above. In this document, the computer-readable storage medium may be any tangible medium that contains or stores a program that can be used by or in conjunction with an instruction execution system, apparatus, or device.

[0111] A computer-readable signal medium may include a data signal propagated in a baseband or as part of a carrier wave, in which computer-readable program code is carried. Such a propagated data signal may take many forms, including but not limited to electromagnetic signals, optical signals, or any suitable combination of the foregoing. A computer-readable signal medium may also be any computer-readable medium other than a computer-readable storage medium, which can send, propagate, or transmit a program for use by or in connection with an instruction execution system, apparatus, or device.

[0112] The program code contained on a computer-readable medium may be transmitted using any appropriate medium, including but not limited to wireless, wireline, optical fiber cable, RF, etc., or any suitable combination of the foregoing.

[0113] Computer program code for performing the operations of the present invention may be written in one or more programming languages or combinations thereof. The programming languages include object-oriented programming languages such as Java, Smalltalk, C++, and also include conventional procedural programming languages such as the "C" language or similar programming languages. The program code may be executed entirely on the user's computer, partially on the user's computer, executed as a stand-alone software package, partially on the user's computer and partially on a remote computer, or entirely on a remote computer or server. In the case of a remote computer, the remote computer may be connected to the user's computer through any type of network, including a local area network (LAN) or a wide area network (WAN), or may be connected to an external computer (e.g., through the Internet using an Internet service provider).

[0114] Of course, a storage medium containing computer-executable instructions provided by an embodiment of the present invention, the computer-executable instructions are not limited to the method operations as described above, and may also execute related operations in the methods provided by any embodiment of the present invention.

[0115] The specific embodiments described above further elaborate on the objectives, technical solutions, and beneficial effects of the present invention. It should be understood that the above description is only specific embodiments of the present invention and is not intended to limit the present invention. Any modifications, equivalent replacements, improvements, etc. made within the spirit and principles of the present invention shall be included within the protection scope of the present invention.

Claims

1. A data-driven new energy station impedance identification method, characterized in that: include: Design two sets of linearly independent disturbance signals; A first set of disturbance signals is injected into each single-machine grid-connected point, the voltage and current signals of the single-machine grid-connected point in the abc coordinate system are measured, and the voltage and current signals of the single-machine grid-connected point in the abc coordinate system are converted into the voltage and current signals of the single-machine grid-connected point in the dq axis; A second set of disturbance signals is injected into each single-machine grid-connected point, the voltage and current signals of the single-machine grid-connected point in the abc coordinate system are measured, and the voltage and current signals of the single-machine grid-connected point in the abc coordinate system are converted into the voltage and current signals of the single-machine grid-connected point in the dq axis; According to the voltage and current signals of the single-machine grid-connected points under the two sets of dq axes, the impedance of the single-machine grid-connected points under the dq axes is calculated, and finally the impedance data of all single-machine grid-connected points are obtained; The impedance data of all single-machine grid-connected points and the voltage and current signals of all single-machine grid-connected points under the dq axis are divided into a training set and a test set; Initialize the ANN model to obtain a basic ANN model; use the training set to train the basic ANN model; use the test set to evaluate the trained basic ANN model, adjust the basic ANN model according to the evaluation result, and finally output an impedance identification model based on ANN.

2. The data-driven impedance identification method for new energy stations according to claim 1 is characterized in that: The calculation formula of the impedance of the single machine grid connection point under the dq axis is as follows: In the formula, Z dq Indicates the impedance value of the converter dq axis; Z dd represents the self-impedance of the d-axis; Z dq It represents the coupling impedance between the d-axis and the q-axis; Z qd It represents the coupling impedance between the q-axis and the d-axis; Z qq represents the self-impedance of the q-axis; V d1 Represents the d-axis measurement value of the grid-connected point voltage of the first group of disturbance signals; V q1 Represents the q-axis measurement value of the grid-connected point voltage of the first group of disturbance signals; V d2 Represents the d-axis measurement value of the grid-connected point voltage of the second group of disturbance signals; V q2 I represents the q-axis measurement value of the grid-connected point voltage of the second group of disturbance signals; d1 I represents the d-axis measurement value of the grid-connected point current of the first group of disturbance signals; q1 I represents the q-axis measurement value of the grid-connected point current of the first group of disturbance signals; d2 I represents the d-axis measurement value of the grid-connected point current of the second group of disturbance signals; q2 The q-axis measurement value of the grid-connected point current representing the second set of disturbance signals.

3. The data-driven impedance identification method for new energy stations according to claim 1 is characterized in that: The basic ANN model includes an input layer, a hidden layer and an output layer. The voltage and current signals of the single-machine grid-connected point under the dq axis and the corresponding frequency value f are input into the input layer, and the output layer outputs the impedance amplitude and phase under the dq axis.

4. The data-driven impedance identification method for new energy stations according to claim 3 is characterized in that: The data of the basic ANN model starts from the input layer, passes through the hidden layer and reaches the output layer. In the forward propagation, each node at each level calculates the weighted sum of the input, and then generates the output through an activation function, as follows: For the jth neuron in the lth layer, its output is expressed as: In the formula, Represents the output of the i-th neuron in the previous layer; represents the weight between the i-th neuron and the j-th neuron in the l-th layer; represents the bias of the jth neuron in the lth layer; f represents the activation function.

5. The data-driven impedance identification method for new energy stations according to claim 1 is characterized in that: Use the training set to train the basic ANN model, including: A multi-period mini-batch stochastic gradient descent algorithm is used, and the mean square error loss function is used as the default indicator for evaluating the performance of the regression algorithm, as follows: In the formula, MSE represents the mean square error of the loss function; and They represent the output and measured impedance data of the dth basic ANN model at sample index n, respectively; n represents the number of small batches, N is its maximum value; d represents the modulus of the output vector of the basic ANN model, D is its maximum value; w l and b l Represents the weights and bias parameters that need to be learned in the lth layer; Use Gaussian distribution to initialize the weights and bias parameters, and the learning rate η update formula is as follows: In the formula, and Respectively represent the weight and bias parameters of the current step Gaussian distribution; and They represent the weight and bias parameters of the Gaussian distribution in the previous step respectively; η represents the learning rate.

6. The data-driven impedance identification method for new energy stations according to claim 1 is characterized in that: The trained basic ANN model is evaluated using the test set, and the basic ANN model is adjusted according to the evaluation results, and finally an ANN-based impedance identification model is output, including: Using the coefficient of determination (R 2 )Evaluate the generated model to characterize the effect of training: In the formula, y i represents training data; Represents y i The average value of i Represents the corresponding data generated by the ANN training model.

7. The data-driven impedance identification method for new energy stations according to claim 6 is characterized in that: The appropriate number of hidden layer neurons is determined based on the relationship between the determination coefficient and the number of hidden layer neurons.

8. A data-driven new energy station impedance identification system, characterized in that: include: The disturbance signal design module is used to design two sets of linearly independent disturbance signals; A measurement module is used to inject a first set of disturbance signals into each single-machine grid-connected point, measure the voltage and current signals of the single-machine grid-connected point in the abc coordinate system, and convert the voltage and current signals of the single-machine grid-connected point in the abc coordinate system into the voltage and current signals of the single-machine grid-connected point in the dq axis; The measuring module is used to inject a second set of disturbance signals into each single-machine grid-connected point, measure the voltage and current signals of the single-machine grid-connected point in the abc coordinate system, and convert the voltage and current signals of the single-machine grid-connected point in the abc coordinate system into the voltage and current signals of the single-machine grid-connected point in the dq axis; The impedance calculation module is used to calculate the impedance of the single-machine grid-connected points under the dq axes according to the voltage and current signals of the single-machine grid-connected points under the two sets of dq axes, and finally obtain the impedance data of all single-machine grid-connected points; A data partitioning module, used to partition the impedance data of all single-machine grid-connected points and the voltage and current signals of all single-machine grid-connected points under the dq axes into a training set and a test set; The model training and evaluation module is used to initialize the ANN model and obtain the basic ANN model; train the basic ANN model using the training set; evaluate the trained basic ANN model using the test set, and adjust the basic ANN model according to the evaluation results, and finally output the impedance identification model based on ANN.

9. A data-driven new energy station impedance identification device, characterized in that: The electronic device comprises a processor and a memory, wherein the memory stores computer instructions, and the processor is used to execute the computer instructions stored in the memory. When the computer instructions are executed by the processor, the electronic device implements the steps of the data-driven new energy station impedance identification method as described in any one of claims 1 to 7.

10. A computer-readable storage medium, characterized in that: The computer-readable storage medium stores a computer program, which, when executed by a processor, implements the steps of the data-driven new energy station impedance identification method as described in any one of claims 1 to 7.

Images