SG and VSG type electromechanical oscillation mechanism analysis method in deep tuning mode

Abstract

The invention belongs to the technical field of renewable energy source scale utilization technologies and equipment, and particularly relates to an SG and VSG type electromechanical oscillation mechanism analysis method in a deep tuning mode. Comprising the following steps: constructing a VSG-SG system structure diagram considering deep adjustment dynamic characteristics of a synchronous unit; extracting a sequence in the VSG-SG system structure by using a phasor measurement unit (PMU), and establishing a VSG-SG system modal mapping model; calculating an observation function of a VSG-SG system modal mapping model based on dynamic modal decomposition DMD, and obtaining a feature value and a feature vector; and mining a'virtual-physical 'power angle dynamic interaction rule by utilizing the characteristic values and the characteristic vectors, and revealing a VSG-SG type low electromechanical oscillation mechanism in a deep modulation mode. According to the method, on the basis of the high-dimensional modal mapping model of the virtual-physical double-inertia system of the koopman operator, the similar electromechanical oscillation mechanism under the high-dimensional nonlinear system is discussed, the nonlinear characteristics of the system can be adaptively captured, and the modeling process of the system is simplified.

Description

Technical Field

[0001] The present invention belongs to the field of renewable energy scale utilization technology and equipment technology, and in particular relates to a method for analyzing the electromechanical oscillation mechanism of SG and VSG types under deep regulation mode, and more specifically to a method for analyzing the electromechanical oscillation mechanism of synchronous units and virtual synchronous generators under deep regulation mode. Background Art

[0002] In the context of the transformation from traditional power systems to new power systems, virtual inertia based on new energy stations can improve the inertia characteristics of the power grid. However, the uneven distribution of inertia will inevitably affect the power oscillation mode of the new power system. At the same time, when thermal power units are operated in deep peak regulation, the electromechanical and electromagnetic dynamic coupling of the power system is closer, which increases the possibility of system oscillation. The traditional low-frequency oscillation frequency range of 0.1-2.5Hz is difficult to cover the frequency range of "quasi-electromechanical oscillation" caused by the grid connection of new energy, which brings severe challenges to the stability of the new power system and urgently needs in-depth research.

[0003] However, existing technologies have disadvantages in analyzing low-frequency oscillations of nonlinear systems, such as local linearization limitations, difficulty in capturing global dynamics, high computational complexity, model accuracy issues, high data requirements, insufficient robustness, and difficulty in applying to large-scale systems. Summary of the invention

[0004] In view of the shortcomings of the above-mentioned prior art, the present invention provides a method for analyzing the mechanism of electromechanical oscillation of SG and VSG in deep regulation mode. Its purpose is to achieve the invention purpose of analyzing the nonlinear oscillation characteristics of the power system. The technical solution adopted by the present invention to achieve the above-mentioned purpose is:

[0005] A method for analyzing the mechanism of electromechanical oscillation of SG and VSG in deep modulation mode comprises the following steps:

[0006] Construct the VSG-SG system structure diagram considering the dynamic characteristics of deep regulation of synchronous units;

[0007] The sequence in the VSG-SG system structure is extracted using the phasor measurement unit (PMU), and the modal mapping model of the VSG-SG system is established.

[0008] Based on dynamic mode decomposition (DMD), the observation function of the modal mapping model of the VSG-SG system is calculated to obtain the eigenvalues ​​and eigenvectors.

[0009] The eigenvalues ​​and eigenvectors are used to explore the dynamic interaction rules of the "virtual-physical" power angle, revealing the VSG-SG-like low electromechanical oscillation mechanism in the deep modulation mode.

[0010] Furthermore, the phasor measurement unit PMU is used to extract the sequence in the VSG-SG system structure and establish a modal mapping model of the VSG-SG system; wherein the phasor measurement unit PMU includes current and voltage transformers, signal processing circuits, data acquisition units, core control units, wireless transmission units and timing systems;

[0011] First, the voltage and current transformers of the synchronous phasor measurement unit installed at the designated node reduce the high-voltage three-phase electricity in proportion, and then the signal processing circuit filters the preceding voltage and current; the core of the timing system in the synchronous phasor measurement unit is the GPS system, which provides synchronization pulses and time scale information; after the synchronization pulse sent by the GPS, the phasor measurement unit PMU collects the preceding voltage and current signals by controlling the data acquisition unit, and converts them into digital signals; finally, the core control unit adds a time scale to the collected voltage and current data, and transmits the collected data to the control terminal of the system through the wireless transmission unit.

[0012] Furthermore, the sequence extracted by the phasor measurement unit PMU includes:

[0013] (1) Voltage phasor sequence: records the changes in the magnitude and phase of each node voltage over time;

[0014] (2) Current phasor sequence: records the changes in the magnitude and phase of each branch current over time;

[0015] (3) Frequency sequence: monitors the fluctuation of system frequency.

[0016] Furthermore, the VSG-SG system modal mapping model includes:

[0017] Step (1) Based on the original data set of the VSG-SG system, the objective function ELBO of the VSG-SG system modal mapping model is established as a function of maximizing the likelihood conditional probability distribution, which is used to infer the distribution of the observed values ​​of the VSG-SG system modal mapping model; wherein: the observed values ​​of the VSG-SG system modal mapping model include: variables obtained by observing and measuring the system state;

[0018] A nonlinear discrete dynamic system, namely the VSG-SG system, is written as:

[0019] x t+1 =F(x t )

[0020] In the formula, F is a collection of all functions in the infinite-dimensional Hilbert space, i.e., observable functions; x t is the system state variable with a time step of t;

[0021] The observable values ​​of the VSG-SG system are:

[0022] g(x t ,u t )=h(x t )+u t

[0023] In the formula, h(x t ) is the state observation value, u t is the system control input variable with a time step of t, g(x t ,u t ) is the vector observation value;

[0024] The update of the dynamic system is written as:

[0025] h(x t+1 )=Kg(x t ,u t )

[0026] In the formula, h(x t+1 ) is the state observation value with time step t+1, K is the koopman operator;

[0027] The maximum likelihood conditional probability distribution function L is expressed as:

[0028] L = p(x 1:T ∣u 1:T-1 )=∫p(h 1:T ,x 1:T ∣u 1:T-1 )dh.

[0029] In the formula, u 1:T-1 is the system control input variable from time 1 to T-1. Similarly, x 1:T and h 1:T are the system state variables and state observations from time 1 to T, respectively, and p is the likelihood function;

[0030] Step (2) maps the system historical state variable sequence information to a low-dimensional space through the time series encoder LSTM;

[0031] Using Koopman inference network observations, we can build observation models, perform state estimation, feature extraction, modal analysis, predict system behavior, evaluate stability, and design control strategies, thereby achieving an in-depth understanding and effective control of the dynamic behavior of nonlinear systems.

[0032]

[0033] In the formula, and Indicates the current state after being processed by the gating mechanism. is the observed value of the forward time of the VSG-SG system, is the observed value of the reverse time of the VSG-SG system, I t is the sequence of input state variables at time t, which determines what information will be input into the LSTM unit, is the predicted input state variable at time t, and are the corresponding weights and biases, is the weight of the forward gate, The weight of the backward gate is used to determine how to and the previous state or Transmit information between and The bias vectors correspond to the forward gate and the backward gate respectively. The bias vector is used to adjust the transmission of information so that the LSTM unit emphasizes or ignores certain information as needed. t =[x t :u t ], σ is the activation function;

[0034] Step (3) constructs the conditional distribution of the state variable based on the predicted Gaussian distribution using the observation generation block;

[0035] q φ (h t ∣h 1:t-1 ,x 1:t ,u 1:t-1 )

[0036] In the formula, q φ is a probabilistic model, φ is the trainable parameter of the initial recognition network, h t is the state variable at time step t, which is the new state that we want to predict or generate, h 1:t-1 is the sequence of state variables with time steps from 1 to t-1, that is, the historical state sequence;

[0037] Step (4) using the conditional prior network to calculate the KL divergence between the conditional distribution of the state variable constructed based on the predicted Gaussian distribution and the distribution of the actual state variable;

[0038] Step (5) The VSG-SG system modal mapping model constructs a linear model of the nonlinear transient dynamics of the VSG-SG interconnected system by minimizing the error between the output of the linear model of the interconnected system and the KL divergence;

[0039] The function that minimizes the error between the output of the linear model of the interconnected system and the KL divergence is:

[0040]

[0041] In the formula, E is the expected value, that is, the average value of the random variable; D KL [q(h1|x 1:T ,u 1:T-1 )||p(h1)] is the KL divergence, which is used to measure the two probability distributions q(h1|x 1:T ,u 1:T-1 ) and p(h1|x 1:T ,u 1:T-1 ), T is the total number of time steps in the time series, t is a specific time step in the time series, and u t-1 is the control signal with time step t-1, and h1 is the state variable with time step 1.

[0042] Furthermore, the system historical state variable sequence information includes: rotor angle, frequency and voltage amplitude.

[0043] Furthermore, the dynamic mode decomposition (DMD) based calculation of the VSG-SG system modal mapping model observation function to obtain the eigenvalues ​​and eigenvectors includes:

[0044] Step (31) assembles the observation function of the Koopman mapping model into an m-dimensional vector sequence matrix about the time series:

[0045]

[0046] Where Ψ0 is the N observation function ψ i (i=0,1,...,N-1), ψ N-1 is the N-1th conditional vector;

[0047] Step (32) constructs the DMD matrix of the VSG-SG interconnected system time series data set based on singular value decomposition, i.e., SVD decomposition, where the singular value decomposition is:

[0048] Ψ0=UΣV *

[0049] In the above formula, the dimensions of U, V and Σ matrices are: m*r, n*r, r*r, Ψ0 is an initial m-dimensional vector sequence matrix, V * is a matrix in the singular value decomposition and is the conjugate transpose of the matrix V;

[0050] The DMD matrix of the VSG-SG interconnected system time series data set is:

[0051] F DMD =U * Ψ1VΣ -1

[0052] In the above formula, F DMD is the DMD matrix, representing the dynamic modes of the VSG-SG interconnected system time series data set, U * is the conjugate transpose of the left singular matrix U in the singular value decomposition, Σ -1 is the singular value matrix The inverse matrix of , Ψ1 is the observation data after the time difference step;

[0053] The derivation process is as follows:

[0054] AΨ0=Ψ1

[0055] AUΣV * =Ψ1

[0056]

[0057] In the above formula, A is the time difference operator, which maps Ψ0 to Ψ1;

[0058] Step (33) By performing eigenvalue decomposition on the DMD matrix to obtain the eigenvalues ​​and eigenvectors, the eigenvalues ​​and eigenfunctions of the Koopman operator of the observation function of the Koopman mapping model can be obtained:

[0059]

[0060] Among them, y1...y r F DMD The observation matrix, μ1...μ r is the eigenvalue, Z1 * …Z r * is the conjugate transposed matrix of the characteristic function;

[0061] Step (34) superimposes each mode to retain important information to the greatest extent, so as to better analyze the mode of the VSG-SG transient power system;

[0062] A mode corresponds to a set of eigenvalues ​​and eigenvectors, and the eigenvalues ​​and eigenvectors are obtained from the characteristic equation; the solution containing the conjugate complex roots in the characteristic equation is the oscillation mode of the system, among which there is a pair of closely related electromechanical oscillation circuits that become the quasi-electromechanical oscillation modes of the system. Solving the characteristic equation as 2 The solution of +bs+c=0 is the characteristic root, and -b / 2a is the damping ratio ξ s , is the oscillation frequency ω s , that is, s =ξ s +jω s , where ξ sis the real part of the electromechanical-like mode and determines the damping of the low-frequency power oscillation of the system.

[0063] A device for analyzing the mechanism of electromechanical oscillation of SG and VSG in deep modulation mode, comprising:

[0064] Building module, used to construct the VSG-SG system structure diagram considering the dynamic characteristics of deep regulation of synchronous units;

[0065] Establish a module for extracting the sequence in the VSG-SG system using the phasor measurement unit PMU and establishing a modal mapping model of the VSG-SG system;

[0066] A calculation module is used to calculate the observation function of the modal mapping model of the VSG-SG system based on dynamic mode decomposition (DMD) and obtain the eigenvalues ​​and eigenvectors;

[0067] The mining module is used to use eigenvalues ​​and eigenvectors to mine the dynamic interaction rules of the "virtual-physical" power angle and reveal the VSG-SG type low electromechanical oscillation mechanism in the deep modulation mode.

[0068] Furthermore, the device for analyzing the electromechanical oscillation mechanism of SG and VSG types in a deep tuning mode is used to implement any step of the method for analyzing the electromechanical oscillation mechanism of SG and VSG types in a deep tuning mode.

[0069] A computer device comprises a storage medium, a processor and a computer program stored on the storage medium and executable on the processor, wherein when the processor executes the computer program, the steps of any one of the methods for analyzing the mechanism of SG and VSG type electromechanical oscillations in a deep modulation mode are implemented.

[0070] A computer storage medium having a computer program stored thereon, wherein when the computer program is executed by a processor, the steps of any one of the methods for analyzing the mechanism of SG and VSG type electromechanical oscillations in a deep modulation mode are implemented.

[0071] The present invention has the following beneficial effects and advantages:

[0072] The present invention proposes a method for analyzing the electromechanical oscillation mechanism of SG and VSG in deep tuning mode. The present invention explores the electromechanical oscillation mechanism of high-dimensional nonlinear system based on the virtual-physical dual-inertia system high-dimensional modal mapping model of the Koopman operator. The Koopman operator method can adaptively capture the nonlinear characteristics of the system and simplify the process of system modeling.

[0073] The present invention constructs a deep neural network model with an embedded Koopman operator in the VSG-SG system. The Koopman operator method can use a high-dimensional linear problem to approximate the original nonlinear problem by selecting a suitable function space, while keeping the relevant low-frequency spatiotemporal characteristics unchanged, thereby realizing the analysis of the nonlinear oscillation characteristics of the power system. BRIEF DESCRIPTION OF THE DRAWINGS

[0074] The above and / or additional aspects and advantages of the present invention will become apparent and easily understood from the description of the embodiments in conjunction with the following drawings, in which:

[0075] Figure 1 It is a schematic diagram of the structure of the VSG-SG system for deep adjustment of dynamic characteristics of synchronous units of the present invention. DETAILED DESCRIPTION

[0076] In order to more clearly understand the above-mentioned objectives, features and advantages of the present invention, the present invention will be further described in detail below in conjunction with the accompanying drawings and specific embodiments. It should be noted that the embodiments of the present invention and the features in the embodiments can be combined with each other without conflict.

[0077] In the following description, many specific details are set forth to facilitate a full understanding of the present invention. However, the present invention may also be implemented in other ways different from those described herein. Therefore, the protection scope of the present invention is not limited to the specific embodiments disclosed below.

[0078] Refer to the following Figure 1 The technical solutions of some embodiments of the present invention are described.

[0079] Example 1

[0080] The present invention provides an embodiment, which is a method for analyzing the mechanism of SG and VSG type electromechanical oscillations in a deep regulation mode, and is applicable to an additional damping control method for suppressing electromechanical oscillations caused by wind power grid connection, comprising the following steps:

[0081] Step 1. Construct the VSG-SG system structure diagram considering the dynamic characteristics of the synchronous unit deep regulation;

[0082] like Figure 1 As shown, Figure 1 Schematic diagram of the VSG-SG system structure of the synchronous unit deep adjustment dynamic characteristics of the present invention. H 、V I 、V J 、V K are the bus voltages of each node, X HJ , X JK , respectively, the line impedance between each bus node, X Jis the sum of the filter impedance of the VSG and its line impedance, I HJ ,I JK ,I J are the currents on each line, θ, ψ is the voltage phase of each node.

[0083] Step 2. Use the phasor measurement unit PMU to extract the sequence in the VSG-SG system structure and establish the VSG-SG system modal mapping model;

[0084] The phasor measurement unit PMU includes current and voltage transformers, signal processing circuits, data acquisition units, core control units, wireless transmission units, and timing systems. First, the voltage and current transformers of the synchronous phasor measurement unit installed at the designated node will proportionally reduce the high-voltage three-phase electricity, and then the signal processing circuit will filter the front-stage voltage and current. The core of the timing system in the synchronous phasor measurement unit is the GPS system, which provides synchronization pulses and time scale information. After the synchronization pulse sent by the GPS, the phasor measurement unit PMU collects the front-stage voltage and current signals by controlling the data acquisition unit and converts them into digital signals. Then the core control unit adds a time scale to the collected voltage and current data, and transmits the collected data to the control terminal of the system through the wireless transmission unit. The phasor measurement unit PMU can extract the following sequence:

[0085] (1) Voltage phasor sequence: records the changes in the magnitude and phase of the voltage at each node over time.

[0086] (2) Current phasor sequence: records the changes in the magnitude and phase of each branch current over time.

[0087] (3) Frequency sequence: monitors the fluctuation of system frequency.

[0088] Step 3. Calculate the modal mapping model of the VSG-SG system based on dynamic mode decomposition (DMD), and obtain the eigenvalues ​​and eigenvectors;

[0089] Step 4. Use eigenvalues ​​and eigenvectors to explore the dynamic interaction rules of the "virtual-physical" power angle and reveal the VSG-SG type low electromechanical oscillation mechanism in the deep modulation mode.

[0090] The real part of the eigenvalue represents the stability of the system. A negative real part indicates that the system is stable, and a positive real part indicates that the system is unstable. By analyzing the eigenvalue, the stability of the system and whether there is a potential oscillation mode can be determined. The eigenvector describes the state change trend of the system under a specific oscillation mode. By analyzing the eigenvector, the changes in various variables of the system during the oscillation process, such as voltage, current, frequency, etc., can be understood, thereby revealing the cause and development process of the oscillation. The system eigenvalue and eigenvector are important tools for analyzing the dynamic behavior of the system and revealing the system mechanism. By analyzing the eigenvalue and eigenvector, the stability, oscillation mode, control strategy and fault diagnosis of the system can be understood, thereby providing important theoretical and technical support for the safe and stable operation of the power system.

[0091] Furthermore, the VSG-SG system modal mapping model in step 2 includes the following steps:

[0092] Step (1) Based on the original data set of the VSG-SG system, the objective function ELBO (variational inference) of the modal mapping model of the VSG-SG system is established as a function of maximizing the likelihood conditional probability distribution, which is used to infer the distribution of the observed values ​​of the modal mapping model of the VSG-SG system; wherein: the observed values ​​of the modal mapping model of the VSG-SG system include: variables obtained by observing and measuring the system state;

[0093] A nonlinear discrete dynamic system, namely the VSG-SG system, is written as:

[0094] x t+1 =F(x t )

[0095] In the formula, F is a collection of all functions in the infinite-dimensional Hilbert space, i.e., observable functions; x t is the system state variable with a time step of t.

[0096] The observable values ​​of the VSG-SG system are:

[0097] g(x t ,u t )=h(x t )+u t

[0098] In the formula, h(x t ) is the state observation value, u t is the system control input variable with a time step of t, g(x t ,u t ) is the vector observation value.

[0099] The update of the dynamic system is written as:

[0100] h(xt+1 )=Kg(x t ,u t )

[0101] In the formula, h(x t+1 ) is the state observation value with time step t+1, and K is the koopman operator.

[0102] The maximum likelihood conditional probability distribution function L can be expressed as:

[0103] L = p(x 1:T ∣u 1:T-1 )=∫p(h 1:T ,x 1:T ∣u 1:T-1 )dh.

[0104] In the formula, u 1:T-1 is the system control input variable from time 1 to T-1. Similarly, x 1:T and h 1:T are the system state variables and state observations from time 1 to T respectively, and p is the likelihood function.

[0105] Step (2) maps the system historical state variable sequence information to a low-dimensional space through a time series encoder LSTM, wherein the system historical state variable sequence information includes: rotor angle, frequency and voltage amplitude;

[0106] Using the Koopman inference network observations obtained in step (1), an observation model is established to perform state estimation, feature extraction, modal analysis, predict system behavior, evaluate stability, and design control strategies, thereby achieving an in-depth understanding and effective control of the dynamic behavior of the nonlinear system.

[0107]

[0108] In the formula, and Indicates the current state after being processed by the gating mechanism. is the observed value of the VSG-SG system in the forward time (from the previous moment to this moment), is the observed value of the reverse time of the VSG-SG system (from the next moment to this moment), I t is the sequence of input state variables at time t, which determines what information will be input into the LSTM unit, is the predicted input state variable at time t, and are the corresponding weights and biases, is the weight of the forward gate, are the weights of the backward gate. These weight matrices are used to determine how to and the previous state or Transfer information between them. and are the bias vectors corresponding to the forward gate and the backward gate, respectively. These bias vectors are used to adjust the transmission of information so that the LSTM unit can emphasize or ignore certain information as needed. t =[x t :u t ], σ is the activation function.

[0109] Step (3) uses the observation generation block to construct the conditional distribution of the state variable based on the predicted Gaussian distribution:

[0110] q φ (h t ∣h 1:t-1 ,x 1:t ,u 1:t-1 )

[0111] In the formula, q φ is a probabilistic model, φ is the trainable parameter of the initial recognition network, h t is the state variable at time step t, which is the new state that we want to predict or generate, h 1:t-1 is the sequence of state variables with time steps from 1 to t-1, that is, the historical state sequence.

[0112] In step (2), the system historical state variable sequence information is mapped to a low-dimensional space through the time series encoder LSTM, which is a process of feature extraction and dimensionality reduction of the system state. The function is to convert the complex state information of the system into a feature representation that is easier to process and analyze. The features extracted in step (2) are part of the input of the prediction model in step (3). The features extracted by LSTM enable the prediction model in step (3) to better understand the dynamic behavior of the system, thereby more accurately predicting or generating the state variable h t .

[0113] Step (4) using the conditional prior network to calculate the KL divergence between the conditional distribution of the state variable constructed based on the predicted Gaussian distribution and the distribution of the actual state variable;

[0114] The VSG-SG system modal mapping model developed in step (5) constructs a linear model of the nonlinear transient dynamics of the VSG-SG interconnected system by minimizing the error between the output of the linear model of the interconnected system and the KL divergence.

[0115] The role of the KL divergence is to minimize the error indicator. The goal of developing the VSG-SG system modal mapping model is to adjust the parameters of the model by minimizing the KL divergence to make it closer to the nonlinear dynamic behavior of the real system.

[0116] The function that minimizes the error between the output of the linear model of the interconnected system and the KL divergence is:

[0117]

[0118] Where E is the expected value, that is, the average value of the random variable. KL [q(h1|x 1:T ,u 1:T-1 )||p(h1)] is the KL divergence, which is used to measure the two probability distributions q(h1|x 1:T ,u 1:T-1 ) and p(h1|x 1:T ,u 1:T-1 ), T is the total number of time steps in the time series, t is a specific time step in the time series, and u t-1 is the control signal with time step t-1, and h1 is the state variable with time step 1.

[0119] The calculation of the modal mapping model of the VSG-SG system based on dynamic modal decomposition (DMD) described in step 3 includes the following four steps:

[0120] Step (31) first assembles the observation function of the VSG-SG system modal mapping model into an m-dimensional vector sequence matrix about the time series:

[0121]

[0122] Where Ψ0 is the N observation function ψ i (i=0,1,...,N-1). These observation functions are the core components of Koopman operator theory, which are used to describe the dynamic behavior of nonlinear systems. N-1 is the N-1th conditional vector.

[0123] Step (32) constructs the DMD matrix of the VSG-SG interconnected system time series data set based on singular value decomposition (SVD decomposition), where the singular value decomposition is:

[0124] Ψ0=UΣV *

[0125] In the above formula, the dimensions of U, V and Σ matrices are: m*r, n*r, r*r respectively. Ψ0 is an initial m-dimensional vector sequence matrix, V * It is a matrix in the singular value decomposition and is the conjugate transpose of the matrix V.

[0126] The DMD matrix of the VSG-SG interconnected system time series data set is:

[0127] F DMD =U * Ψ1VΣ -1

[0128] In the above formula, F DMD is the DMD matrix, which represents the dynamic mode of the VSG-SG interconnected system time series dataset, U * is the conjugate transpose of the left singular matrix U in the singular value decomposition, Σ -1 is the singular value matrix is the inverse matrix of , and Ψ1 is the observation data after the time difference step.

[0129] The derivation process is as follows:

[0130] AΨ0=Ψ1

[0131] AUΣV * =Ψ1

[0132]

[0133] In the above formula, A is the time difference operator, which maps Ψ0 to Ψ1.

[0134] Step (33) By performing eigenvalue decomposition on the DMD matrix to obtain the eigenvalues ​​and eigenvectors, the eigenvalues ​​and eigenfunctions of the Koopman operator of the observation function of the Koopman mapping model can be obtained:

[0135]

[0136] Among them, y1...y r F DMD The observation matrix, μ1...μ r is the eigenvalue, Z1 * …Z r * is the conjugate transposed matrix of the characteristic function.

[0137] Step (34) superimposes each mode to basically retain important information to the greatest extent, thereby better analyzing the modes of the VSG-SG transient power system.

[0138] A mode corresponds to a set of eigenvalues ​​and eigenvectors, and the eigenvalues ​​and eigenvectors are obtained from the characteristic equation. The solution containing the conjugate complex roots of the characteristic equation is the oscillation mode of the system, among which there is a pair of closely related electromechanical oscillation circuits that become the quasi-electromechanical oscillation modes of the system. Solving the characteristic equation as 2The solution of +bs+c=0 is the characteristic root, and -b / 2a is the damping ratio ξ s , is the oscillation frequency ω s That is, s =ξ s +jω s Where ξ s is the real part of the quasi-electromechanical mode, which determines the damping of the low-frequency power oscillation of the system.

[0139] Example 2

[0140] The present invention further provides an embodiment, which is a device for analyzing the mechanism of electromechanical oscillation of SG and VSG in deep modulation mode, comprising:

[0141] Building module, used to construct the VSG-SG system structure diagram considering the dynamic characteristics of deep regulation of synchronous units;

[0142] Establish a module for extracting the sequence in the VSG-SG system using the phasor measurement unit PMU and establishing a modal mapping model of the VSG-SG system;

[0143] A calculation module is used to calculate the observation function of the modal mapping model of the VSG-SG system based on dynamic mode decomposition (DMD) and obtain the eigenvalues ​​and eigenvectors;

[0144] The mining module is used to use eigenvalues ​​and eigenvectors to mine the dynamic interaction rules of the "virtual-physical" power angle and reveal the VSG-SG type low electromechanical oscillation mechanism in the deep modulation mode.

[0145] Another device for analyzing the mechanism of electromechanical oscillations of SG and VSG types in deep tuning mode of the present invention is used to implement the steps of a method for analyzing the mechanism of electromechanical oscillations of SG and VSG types in deep tuning mode described in Example 1.

[0146] Example 4

[0147] Based on the same inventive concept, an embodiment of the present invention further provides a computer device, including a storage medium, a processor, and a computer program stored on the storage medium and executable on the processor. When the processor executes the computer program, the steps of the method for analyzing the mechanism of SG and VSG type electromechanical oscillations in a deep tuning mode described in Example 1 are implemented.

[0148] Example 5

[0149] Based on the same inventive concept, an embodiment of the present invention further provides a computer storage medium, on which a computer program is stored. When the computer program is executed by a processor, the steps of a method for analyzing the mechanism of electromechanical oscillation of SG and VSG types in a deep adjustment mode described in Example 1 are implemented.

[0150] Those skilled in the art will appreciate that embodiments of the present invention may be provided as methods, systems, or computer program products. Therefore, the present invention may take the form of a complete hardware embodiment, a complete software embodiment, or an embodiment combining software and hardware. Moreover, the present invention may take the form of a computer program product implemented on one or more computer-usable storage media (including but not limited to disk storage, CD-ROM, optical storage, etc.) containing computer-usable program code.

[0151] The present invention is described with reference to flowcharts and / or block diagrams of methods, devices (systems), and computer program products according to embodiments of the present invention. It should be understood that each process and / or block in the flowchart and / or block diagram, as well as the combination of processes and / or blocks in the flowchart and / or block diagram, can be implemented by computer program instructions. These computer program instructions can be provided to a processor of a general-purpose computer, a special-purpose computer, an embedded processor, or other programmable data processing device to produce a machine, so that the instructions executed by the processor of the computer or other programmable data processing device generate instructions for implementing the processes in the flowchart and / or block diagram. Figure 1 A process or multiple processes and / or boxes Figure 1 A device that provides the functions specified in a block or multiple blocks.

[0152] These computer program instructions may also be stored in a computer-readable memory capable of directing a computer or other programmable data processing device to operate in a specific manner, so that the instructions stored in the computer-readable memory produce an article of manufacture comprising an instruction device, which implements the process Figure 1 A process or multiple processes and / or boxes Figure 1 A function specified in one or more boxes.

[0153] These computer program instructions can also be loaded onto a computer or other programmable data processing device so that a series of operating steps are executed on the computer or other programmable device to produce a computer-implemented process, thereby providing instructions for implementing the process. Figure 1 A process or multiple processes and / or boxes Figure 1 A step that specifies a function in one or more boxes.

[0154] 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, ordinary technicians in the relevant field should understand that the specific implementation methods of the present invention can still be modified or replaced by equivalents. Any modification or equivalent replacement that does not depart from the spirit and scope of the present invention should be covered within the scope of protection of the claims of the present invention.

Claims

1. A method for analyzing the mechanism of electromechanical oscillation of SG and VSG in deep modulation mode, characterized by: The following steps are involved: Construct the VSG-SG system structure diagram considering the dynamic characteristics of deep regulation of synchronous units; The sequence in the VSG-SG system structure is extracted using the phasor measurement unit (PMU), and the modal mapping model of the VSG-SG system is established. Based on dynamic mode decomposition (DMD), the observation function of the modal mapping model of the VSG-SG system is calculated to obtain the eigenvalues ​​and eigenvectors. The eigenvalues ​​and eigenvectors are used to explore the dynamic interaction rules of the "virtual-physical" power angle, revealing the VSG-SG type low electromechanical oscillation mechanism in the deep modulation mode.

2. The method for analyzing the mechanism of SG and VSG type electromechanical oscillations in deep modulation mode according to claim 1 is characterized by: The phase measurement unit PMU is used to extract the sequence in the VSG-SG system structure and establish a modal mapping model of the VSG-SG system; wherein the phase measurement unit PMU includes current and voltage transformers, signal processing circuits, data acquisition units, core control units, wireless transmission units and timing systems; First, the voltage and current transformers of the synchronous phasor measurement unit installed at the designated node reduce the high-voltage three-phase electricity in proportion, and then the signal processing circuit filters the preceding voltage and current; the core of the timing system in the synchronous phasor measurement unit is the GPS system, which provides synchronization pulses and time scale information; after the synchronization pulse sent by the GPS, the phasor measurement unit PMU collects the preceding voltage and current signals by controlling the data acquisition unit, and converts them into digital signals; finally, the core control unit adds a time scale to the collected voltage and current data, and transmits the collected data to the control terminal of the system through the wireless transmission unit.

3. The method for analyzing the mechanism of SG and VSG type electromechanical oscillations in deep modulation mode according to claim 2 is characterized by: The sequence extracted by the phasor measurement unit PMU includes: (1) Voltage phasor sequence: records the changes in the magnitude and phase of each node voltage over time; (2) Current phasor sequence: records the changes in the magnitude and phase of each branch current over time; (3) Frequency sequence: monitors the fluctuation of system frequency.

4. The method for analyzing the mechanism of electromechanical oscillation of SG and VSG in deep modulation mode according to claim 1 is characterized by: The VSG-SG system modal mapping model includes: Step (1) Based on the original data set of the VSG-SG system, the objective function ELBO of the VSG-SG system modal mapping model is established as a function of maximizing the likelihood conditional probability distribution, which is used to infer the distribution of the observed values ​​of the VSG-SG system modal mapping model; wherein: the observed values ​​of the VSG-SG system modal mapping model include: variables obtained by observing and measuring the system state; A nonlinear discrete dynamic system, namely the VSG-SG system, is written as: x t+1 =F(x t ) In the formula, F is a collection of all functions in the infinite-dimensional Hilbert space, i.e., observable functions; x t is the system state variable with a time step of t; The observable values ​​of the VSG-SG system are: g(x t ,u t )=h(x t )+u t In the formula, h(x t ) is the state observation value, u t is the system control input variable with a time step of t, g(x t ,u t ) is the vector observation value; The update of the dynamic system is written as: h(x t+1 )=Kg(x t ,u t ) In the formula, h(x t+1 ) is the state observation value with time step t+1, K is the koopman operator; The maximum likelihood conditional probability distribution function L is expressed as: L=p(x 1:T ∣u 1:T-1 )=∫p(h 1:T ,x 1:T ∣u 1:T-1 )dh. In the formula, u 1:T-1 is the system control input variable from time 1 to T-1. Similarly, x 1:T and h 1:T are the system state variables and state observations from time 1 to T, respectively, and p is the likelihood function; Step (2) maps the system historical state variable sequence information to a low-dimensional space through the time series encoder LSTM; Using Koopman inference network observations, we can build observation models, perform state estimation, feature extraction, modal analysis, predict system behavior, evaluate stability, and design control strategies, thereby achieving an in-depth understanding and effective control of the dynamic behavior of nonlinear systems. In the formula, and Indicates the current state after being processed by the gating mechanism. is the observed value of the forward time of the VSG-SG system, is the observed value of the reverse time of the VSG-SG system, I t is the sequence of input state variables at time t, which determines what information will be input into the LSTM unit, is the predicted input state variable at time t, and are the corresponding weights and biases, is the weight of the forward gate, The weight of the backward gate is used to determine how to and the previous state or Transmit information between and The bias vectors correspond to the forward gate and the backward gate respectively. The bias vector is used to adjust the transmission of information so that the LSTM unit emphasizes or ignores certain information as needed. t =[x t :u t ], σ is the activation function; Step (3) constructs the conditional distribution of the state variable based on the predicted Gaussian distribution using the observation generation block; q φ (h t ∣h 1:t-1 ,x 1:t ,u 1:t-1 ) In the formula, q φ is a probabilistic model, φ is the trainable parameter of the initial recognition network, h t is the state variable at time step t, which is the new state that we want to predict or generate, h 1:t-1 is the sequence of state variables with time steps from 1 to t-1, that is, the historical state sequence; Step (4) using the conditional prior network to calculate the KL divergence between the conditional distribution of the state variable constructed based on the predicted Gaussian distribution and the distribution of the actual state variable; Step (5) The VSG-SG system modal mapping model constructs a linear model of the nonlinear transient dynamics of the VSG-SG interconnected system by minimizing the error between the output of the linear model of the interconnected system and the KL divergence; The function that minimizes the error between the output of the linear model of the interconnected system and the KL divergence is: -D KL [q(h1∣x 1:T ,u 1:T-1 )||p(h1)] In the formula, E is the expected value, that is, the average value of the random variable; D KL [q(h1|x 1:T ,u 1:T-1 )||p(h1)] is the KL divergence, which is used to measure the two probability distributions q(h1|x 1:T ,u 1:T-1 ) and p(h1|x 1:T ,u 1:T-1 ), T is the total number of time steps in the time series, t is a specific time step in the time series, and u t-1 is the control signal with time step t-1, and h1 is the state variable with time step 1.

5. The method for analyzing the mechanism of electromechanical oscillation of SG and VSG in deep modulation mode according to claim 4 is characterized by: The system historical state variable sequence information includes: rotor angle, frequency and voltage amplitude.

6. The method for analyzing the mechanism of SG and VSG type electromechanical oscillations in deep modulation mode according to claim 1 is characterized by: The method of calculating the observation function of the modal mapping model of the VSG-SG system based on dynamic modal decomposition (DMD) to obtain eigenvalues ​​and eigenvectors includes: Step (31) assembles the observation function of the Koopman mapping model into an m-dimensional vector sequence matrix about the time series: Where Ψ0 is the N observation function ψ i (i=0,1,...,N-1), ψ N-1 is the N-1th conditional vector; Step (32) constructs the DMD matrix of the VSG-SG interconnected system time series data set based on singular value decomposition, i.e., SVD decomposition, where the singular value decomposition is: Ψ0=UΣV * In the above formula, the dimensions of U, V and Σ matrices are: m*r, n*r, r*r, Ψ0 is an initial m-dimensional vector sequence matrix, V * is a matrix in the singular value decomposition and is the conjugate transpose of the matrix V; The DMD matrix of the VSG-SG interconnected system time series data set is: F DMD =U * Ψ1VΣ -1 In the above formula, F DMD is the DMD matrix, representing the dynamic modes of the VSG-SG interconnected system time series data set, U * is the conjugate transpose of the left singular matrix U in the singular value decomposition, Σ -1 is the singular value matrix The inverse matrix of , Ψ1 is the observation data after the time difference step; The derivation process is as follows: AΨ0=Ψ1 AUSV * =Ψ1 In the above formula, A is the time difference operator, which maps Ψ0 to Ψ1; Step (33) By performing eigenvalue decomposition on the DMD matrix to obtain the eigenvalues ​​and eigenvectors, the eigenvalues ​​and eigenfunctions of the Koopman operator of the observation function of the Koopman mapping model can be obtained: Among them, y1...y r F DMD The observation matrix, μ1...μ r is the eigenvalue, Z1 * …Z r * is the conjugate transposed matrix of the characteristic function; Step (34) superimposes each mode to retain important information to the greatest extent, so as to better analyze the mode of the VSG-SG transient power system; A mode corresponds to a set of eigenvalues ​​and eigenvectors, and the eigenvalues ​​and eigenvectors are obtained from the characteristic equation; the solution containing the conjugate complex roots in the characteristic equation is the oscillation mode of the system, among which there is a pair of closely related electromechanical oscillation circuits that become the quasi-electromechanical oscillation modes of the system. Solving the characteristic equation as 2 The solution of +bs+c=0 is the characteristic root, and -b / 2a is the damping ratio ξ s , is the oscillation frequency ω s , that is, s =ξ s +jω s , where ξ s is the real part of the electromechanical-like mode and determines the damping of the low-frequency power oscillation of the system.

7. A device for analyzing the mechanism of electromechanical oscillation of SG and VSG in deep modulation mode, characterized by comprising: Building module, used to construct the VSG-SG system structure diagram considering the dynamic characteristics of deep regulation of synchronous units; Establish a module for extracting the sequence in the VSG-SG system using the phasor measurement unit PMU and establishing a modal mapping model of the VSG-SG system; A calculation module is used to calculate the observation function of the modal mapping model of the VSG-SG system based on dynamic mode decomposition (DMD) and obtain the eigenvalues ​​and eigenvectors; The mining module is used to use eigenvalues ​​and eigenvectors to mine the dynamic interaction rules of "virtual-physical" power angles and reveal the VSG-SG type low electromechanical oscillation mechanism in deep modulation mode.

8. The device for analyzing the mechanism of SG and VSG type electromechanical oscillations in deep modulation mode according to claim 7 is characterized in that: The device for analyzing the electromechanical oscillation mechanism of SG and VSG types in a deep tuning mode is used to implement the steps of a method for analyzing the electromechanical oscillation mechanism of SG and VSG types in a deep tuning mode as described in any one of claims 1 to 6.

9. A computer device comprising a storage medium, a processor, and a computer program stored in the storage medium and executable on the processor, characterized in that: When the processor executes the computer program, the steps of a method for analyzing the mechanism of electromechanical oscillation of SG and VSG types in a deep modulation mode as described in any one of claims 1 to 6 are implemented.

10. A computer storage medium, characterized in that: The computer storage medium stores a computer program, and when the computer program is executed by the processor, the steps of the method for analyzing the mechanism of SG and VSG type electromechanical oscillations in a deep modulation mode described in any one of claims 1 to 6 are implemented.

Images