Simulation analysis method for switching transformer overvoltage suppression method

Abstract

The invention discloses a simulation analysis method of a switching transformer overvoltage suppression method, and relates to power system simulation analysis. Comprising the following steps: performing dynamic feature extraction on preprocessed data, considering time sequence characteristics and a system topological structure, and fusing multi-source heterogeneous data; based on an improved long-short-term memory network (LSTM) and wavelet transform, a multi-scale overvoltage prediction model is constructed, and accurate prediction of overvoltage of different time scales is realized; designing a self-adaptive overvoltage suppression strategy generation algorithm according to a prediction result and dynamic characteristics of the system, and considering multiple factors such as system stability, equipment life and economy; and implementing an optimal suppression strategy in a simulation environment, and performing strategy evaluation and iterative optimization according to a simulation result. Through simulation verification and iterative optimization, the applicability and reliability of the method are further improved.

Description

Technical Field

[0001] The invention relates to the field of power system simulation analysis, in particular to a simulation analysis method for a switching transformer overvoltage suppression method. Background Art

[0002] With the development and intelligentization of power systems, transformer switching operations play an important role in power systems. However, overvoltage often occurs during transformer switching, which not only threatens the safe operation of system equipment, but also may lead to power quality degradation and system stability problems. Traditional overvoltage suppression methods often rely on static models and empirical rules, which are difficult to adapt to the complex and changeable power system operating environment.

[0003] At present, there are still some problems in the simulation analysis of overvoltage suppression methods for switching transformers. First, it is difficult to fully consider various influencing factors during data acquisition and preprocessing, resulting in low quality of the initial data set. Second, feature extraction methods often ignore the influence of time series characteristics and system topology, and cannot make full use of multi-source heterogeneous data. In addition, the existing overvoltage prediction models are not accurate enough to achieve accurate prediction of overvoltages at different time scales. Finally, the generation and optimization process of the suppression strategy lacks adaptability and comprehensiveness, making it difficult to achieve a good balance between suppression effect, economic cost and system stability.

[0004] Therefore, it is urgent to propose a new simulation analysis method for switching transformer overvoltage suppression method to solve the above problems and improve the overvoltage suppression effect and system operation reliability. Summary of the invention

[0005] In view of the problems existing in the prior art, the present invention is proposed.

[0006] Therefore, the problem to be solved by the present invention is how to improve the overvoltage suppression effect and system operation reliability.

[0007] In order to solve the above technical problems, the present invention provides the following technical solutions:

[0008] In a first aspect, an embodiment of the present invention provides a simulation analysis method for a method for suppressing overvoltage of a switching transformer, which comprises:

[0009] Collecting and preprocessing multi-source data; the multi-source data includes real-time voltage data, historical voltage data, transformer parameters, system topology information, environmental factors and load data;

[0010] Extract and fuse features from preprocessed multi-source data;

[0011] Based on the fused features, an overvoltage prediction model is constructed, and based on the prediction results, an overvoltage suppression generation algorithm is established;

[0012] Based on the suppression strategy generated by the overvoltage suppression generation algorithm, a multi-objective optimization method is used to obtain and implement the optimal suppression strategy, and strategy evaluation and iterative optimization are performed according to the implementation results.

[0013] As a preferred solution of the simulation analysis method of the switching transformer overvoltage suppression method of the present invention, the preprocessing includes removing abnormal values, filling gaps and standardizing.

[0014] The removal of outliers uses the moving median method as shown below:

[0015] x' i =median(X i -k,...,X i ,...,X i +k)

[0016] Among them, x' i is the processed voltage data, X i is the original collected voltage data, k is the window size selected according to the transformer switching frequency;

[0017] The multivariate interpolation method used to fill the gap is shown in the following formula:

[0018]

[0019] in, is the estimated missing voltage value, f is the interpolation function trained based on historical data, X1,X2X n are the known voltage values ​​and related environmental parameters adjacent to the missing data time;

[0020] The standardization uses the Z-score standardization method as shown below:

[0021]

[0022] Among them, z is the standardized voltage data, x is the original or processed voltage data, μ is the mean of the historical voltage data, and σ is the standard deviation of the historical voltage data.

[0023] As a preferred solution of the simulation analysis method of the switching transformer overvoltage suppression method of the present invention, the feature extraction and fusion includes:

[0024] Perform time-frequency joint analysis on the preprocessed voltage data to extract voltage features;

[0025] Extract system topology features;

[0026] fusing the voltage characteristics and the system topology characteristics;

[0027] The time-frequency joint analysis is shown as follows

[0028] Among them, WPT j,k (z) is the wavelet packet coefficient of voltage data, j is the decomposition scale, k is the node number, h k (n) is the wavelet filter coefficient, ψ(z) is the selected wavelet function, n is the discrete time index, representing the data sampling point, j / 2 is the normalization factor, 2 j z represents the scale transformation of the input signal z;

[0029] The extraction of system topological features using graph convolutional network is shown in the following formula:

[0030] H (l+1) =σ(D -1 / 2 AD -1 / 2 H (l) W (l) )

[0031] Among them, H (l) is the node feature matrix of the lth layer, which represents the state characteristics of each node in the power system at the lth layer, H (l+1) is the node feature matrix of the l+1th layer, A is the system topology adjacency matrix, reflecting the connection relationship between the transformer and other devices, D is the degree matrix, used for normalization, and W (1) is the weight matrix, σ is the activation function, D -1 / 2 AD -1 / 2 is a symmetric normalized Laplace matrix used to smooth features;

[0032] The voltage characteristics and the system topology characteristics are combined as shown in the following formula:

[0033]

[0034] Among them, α i is the attention weight of the i-th feature, e i is the attention score, a is the attention function, W q and W k is the learnable parameter matrix, h i Input features, including wavelet packet coefficients WPT j,k (z) and the features output by the last layer of the graph convolutional network, n is the total number of features; exp is an exponential function used to convert the score into a positive value; is a normalization factor that ensures that the sum of all weights is 1.

[0035] As a preferred solution of the simulation analysis method of the switching transformer overvoltage suppression method of the present invention, the input feature F of the time step t is t Inputting the overvoltage prediction model to obtain a prediction result, and performing multi-scale decomposition and reintegration on the prediction result to obtain a final prediction result;

[0036] The overvoltage prediction model is shown in the following formula:

[0037] i t =σ(W xi x t +W hi h t-1 +b i )

[0038] f t =σ(W xf x t +W hf h t-1 +b f )

[0039] o t =σ(W xo x t +W ho h t-1 +b o )

[0040]

[0041] h t =o t ⊙tanh(c t )+r t

[0042] Among them, i t 、f t , o t They are input gate, forget gate and output gate, which are used to control the flow of information at different time scales during the switching process; is the cell state, storing the long-term voltage change trend; h t is a hidden state, indicating the voltage prediction result at the current moment; c t is the candidate unit state; g t is the input gate; r t is the residual connection; Wxi, W xf W xo W xc W hi W hf W ho W hc is the weight matrix; b i 、b f、b o 、b c is the bias term; σ is the sigmoid function; tanh is the hyperbolic tangent function; ⊙ represents the Hadamard product;

[0043] The multi-scale decomposition is shown in the following formula:

[0044]

[0045] Among them, h(t) is the hidden state sequence output by LSTM, which represents the predicted voltage sequence, D j (t) is the detail coefficient, A J (t) is the approximation coefficient, J is the number of decomposition levels;

[0046] The integration is shown below:

[0047]

[0048] Among them, y is the final predicted voltage value, y i is the prediction result of the i-th scale, w i is the fusion weight.

[0049] As a preferred solution of the simulation analysis method of the switching transformer overvoltage suppression method of the present invention, the overvoltage suppression generation algorithm generates a suppression strategy based on the final prediction result as shown in the following formula:

[0050]

[0051] Among them, y t is the predicted voltage value; t is the actual measured voltage value; E t is the current environment information; a t is the suppression action taken at time t; Q([y t ,z t ,E t ],a t ) is the state-action value function; r t+1 To perform an action, a t is the immediate reward obtained; α is the learning rate; γ is the discount factor;

[0052]

[0053] Indicates that in the next state ([y t+1 ,z t+1 ,E t+1 ] and select the action that maximizes the Q value.

[0054] As a preferred solution of the simulation analysis method of the switching transformer overvoltage suppression method of the present invention, the multi-objective optimization method constructs a multi-objective optimization model according to the suppression strategy as shown in the following formula:

[0055] minF(x)=[f1(x),f2(x),…,f m (x)

[0056] stg i (x)≤0,i=1,2,…,p

[0057] h j (x)=0,j=1,2,…,q

[0058] Where x is the decision variable vector, F(x) is the objective function vector; f1(x) represents the voltage deviation minimization target; f2(x) represents the equipment operation number minimization target, f m (x) including other objectives such as maximizing voltage stability margin; g i (x) is the inequality constraint; h j (x) is an equality constraint, such as the power balance equation; p and q are the number of inequality and equality constraints, respectively.

[0059] As a preferred solution of the simulation analysis method of the switching transformer overvoltage suppression method of the present invention, wherein: the overvoltage suppression generation algorithm also includes an optimization control sequence and an adaptive adjustment mechanism;

[0060] The optimized control sequence is shown below:

[0061]

[0062] Among them, u t ,…,u t+N-1 is the control sequence from the current time t to the time t+N-1; l([y t+k ,z t+k ,E t+k ],u t+k ) is the stage cost function; V f ([y t+N ,z t+N ,E t+N ]) is the terminal cost function; N is the prediction time domain length; f([y t+k ,z t+k ,E t+k ],u t+k ) is the system dynamic model, describing how the state changes with the control action; t+k+1 ,z t+k+1 ,E t+k+1 are the predicted voltage, actual voltage and environmental information after k steps respectively; ut+k is the control action after k steps;

[0063] The adaptive regulation mechanism dynamically adjusts the strategy parameters according to the system state and the prediction results of the overvoltage prediction model, as shown in the following formula:

[0064]

[0065] Among them, θ t is the strategy parameter at time t, including the network parameters of the Q function and the weight coefficient of MPC; θ t+1 is the updated strategy parameter; η is the learning rate; is the performance index J with respect to parameter θ t The gradient of J is the measure of the current policy in state y t ,z t ,E t The function of the performance.

[0066] In a second aspect, an embodiment of the present invention provides a simulation analysis system for a method for suppressing overvoltage of a switching transformer, which is characterized by comprising:

[0067] A data acquisition and processing module is used to collect and pre-process multi-source data; the multi-source data includes real-time voltage data, historical voltage data, transformer parameters, system topology information, environmental influencing factors and load data;

[0068] Feature extraction module, used to extract and fuse features from preprocessed multi-source data;

[0069] The model building module is used to build an overvoltage prediction model based on the fused features, and to establish an overvoltage suppression generation algorithm based on the prediction results;

[0070] The optimization and iteration module uses a multi-objective optimization method to obtain and implement the optimal suppression strategy based on the suppression strategy generated by the overvoltage suppression generation algorithm, and performs strategy evaluation and iterative optimization according to the implementation results.

[0071] In a third aspect, an embodiment of the present invention provides a computer device, comprising a memory and a processor, wherein the memory stores a computer program, wherein: when the computer program instructions are executed by the processor, the steps of the simulation and analysis method of the method for suppressing overvoltage of a switching transformer as described in the first aspect of the present invention are implemented.

[0072] In a fourth aspect, an embodiment of the present invention provides a computer-readable storage medium having a computer program stored thereon, wherein: when the computer program instructions are executed by a processor, the steps of the simulation and analysis method of the method for suppressing overvoltage of a switching transformer as described in the first aspect of the present invention are implemented.

[0073] The beneficial effects of the present invention are as follows: the present invention improves the quality and reliability of the initial data set by collecting multi-source data and performing high-quality preprocessing. By extracting and fusion of dynamic features, full use is made of the time series characteristics and system topology information. The multi-scale overvoltage prediction model constructed based on the improved LSTM and wavelet transform significantly improves the prediction accuracy. The adaptive overvoltage suppression strategy generation algorithm takes into account many factors and improves the suppression effect. The multi-objective optimization and decision-making method achieves a balance between the suppression effect, economic cost and system stability. Through simulation verification and iterative optimization, the applicability and reliability of the method are further improved. BRIEF DESCRIPTION OF THE DRAWINGS

[0074] In order to more clearly illustrate the technical solutions of the embodiments of the present invention, the following briefly introduces the drawings required for use in the description of the embodiments. Obviously, the drawings described below are only some embodiments of the present invention. For ordinary technicians in this field, other drawings can be obtained based on these drawings without creative work.

[0075] Figure 1 A flow chart of a simulation analysis method for overvoltage suppression of switching transformers;

[0076] Figure 2 This is the system module diagram of the simulation analysis method for overvoltage suppression of switching transformers. DETAILED DESCRIPTION

[0077] In order to make the above-mentioned objects, features and advantages of the present invention more obvious and easy to understand, the specific implementation methods of the present invention are described in detail below in conjunction with the accompanying drawings.

[0078] In the following description, many specific details are set forth to facilitate a full understanding of the present invention, but the present invention may also be implemented in other ways different from those described herein, and those skilled in the art may make similar generalizations without violating the connotation of the present invention. Therefore, the present invention is not limited to the specific embodiments disclosed below.

[0079] Secondly, the term "one embodiment" or "embodiment" as used herein refers to a specific feature, structure, or characteristic that may be included in at least one implementation of the present invention. The term "in one embodiment" that appears in different places in this specification does not necessarily refer to the same embodiment, nor does it refer to a separate or selective embodiment that is mutually exclusive with other embodiments.

[0080] Example 1

[0081] Reference Figure 1-2 , which is the first embodiment of the present invention, and which provides a simulation analysis method for a switching transformer overvoltage suppression method, comprising:

[0082] S1: Data collection and preprocessing: Collect real-time voltage data, historical voltage data, transformer parameters, system topology information, environmental factors and load data as the initial data set, and preprocess the initial data set, including removing outliers, filling missing values ​​and data standardization;

[0083] Furthermore, data preprocessing includes:

[0084] Use the moving median method to remove outliers, expressed as:

[0085] x' i =median(X i -k,...,X i ,...,X i +k)

[0086] Among them, x' i is the processed voltage data, X i is the original collected voltage data, k is the window size selected according to the transformer switching frequency;

[0087] The multivariate interpolation method used to fill the gap is shown in the following formula:

[0088]

[0089] in, is the estimated missing voltage value, f is the interpolation function trained based on historical data, X1,X2X n are the known voltage values ​​and related environmental parameters adjacent to the missing data time;

[0090] The Z-score standardization method is used to standardize the data, expressed as:

[0091]

[0092] Among them, z is the standardized voltage data, x is the original or processed voltage data, μ is the mean of the historical voltage data, and σ is the standard deviation of the historical voltage data, which is used to eliminate the influence of different voltage levels.

[0093] It should be noted that the moving median method can effectively remove outliers, the multivariate interpolation method can use the relevant variable information to improve the accuracy of missing value filling, and the Z-score standardization method can make data of different dimensions comparable.

[0094] S2: Dynamic feature extraction and fusion: Dynamic feature extraction is performed on the preprocessed data, taking into account the time series characteristics and system topology, and fusing multi-source heterogeneous data;

[0095] Furthermore, dynamic feature extraction and fusion include:

[0096] The standardized voltage data z is transformed by wavelet packet and time-frequency joint analysis is performed, which is expressed as:

[0097]

[0098] Among them, WPT j,k (z) is the wavelet packet coefficient of the voltage data z, which is used to capture the transient and steady-state voltage characteristics during the transformer switching process; j is the decomposition scale, usually 3-5 layers, the lower layers capture fast transient characteristics (such as the spike at the switching moment), and the higher layers capture slow changing characteristics (such as the voltage recovery process); k is the node number, indicating a specific frequency band, which is used to distinguish different types of voltage fluctuations; n is the index of the discrete time series, corresponding to the sampling point; h k (n) is the wavelet filter coefficient. Daubechies wavelet is selected because its characteristics are suitable for capturing mutations and oscillations in power systems; ψ(z) is the mother wavelet function; j / 2 is the normalization factor; 2 j z Realizing multi-scale analysis of voltage signals helps to distinguish different stages of the switching process (such as before switching, switching moment, and recovery process).

[0099] The graph convolutional network is used to extract the system topology features, which can be expressed as:

[0100] H (l+1) =σ(D -1 / 2 AD -1 / 2 H (l) W (l) )

[0101] Among them, H (l) is the node feature matrix of the lth layer, which represents the state characteristics of each node (such as bus, transformer, load) in the power system at the lth layer; H (l+1) is the feature of the next layer; A is the system topology adjacency matrix, which reflects the connection relationship between the transformer and surrounding equipment (such as lines, loads, and compensation devices); D is the degree matrix, W (l) is the learnable weight matrix of the lth layer; σ is the activation function, D -1 / 2 AD -1 / 2 is a symmetric and normalized Laplace matrix used to smooth features, and σ is a ReLU activation function that introduces nonlinearity to capture complex system response characteristics.

[0102] The attention mechanism is used to fuse the features extracted by wavelet packet transform and graph convolutional network, which can be expressed as:

[0103]

[0104] Among them, α iis the attention weight of the i-th feature, which is used to dynamically adjust the importance of different features (such as voltage wave features and topological features) in different switching stages; e i is the attention score; a is the scaled dot product attention function; W q and W k is the learnable parameter matrix; h i Input features, including wavelet packet coefficients WPT j,k (z) (reflecting the voltage dynamic characteristics) and graph convolution features (reflecting the impact of system topology).

[0105] It should be noted that wavelet packet transform can effectively capture time-frequency characteristics, graph convolutional network can extract system topological structure information, and attention mechanism can adaptively fuse multi-source features.

[0106] S3: Construction of multi-scale overvoltage prediction model: Based on the improved long short-term memory network (LSTM) and wavelet transform, a multi-scale overvoltage prediction model is constructed to achieve accurate prediction of overvoltage at different time scales;

[0107] Furthermore, the construction of the multi-scale overvoltage prediction model includes:

[0108] The fused features

[0109] F t =[α1h1,α2h2,...,α n h n ]

[0110] Enter the improved LSTM unit, introduce residual connection and gating mechanism, expressed as:

[0111] i t =σ(W xi x t +W hi h t-1 +b i )

[0112] f t =σ(W xf x t +W hf h t-1 +b f )

[0113] o t =σ(W xo x t +V ho h t-1 +b o )

[0114]

[0115] h t =o t ⊙tanh(c t )+r t

[0116] Among them, i t 、f t , o t They are input gate, forget gate and output gate, which are used to control the flow of information at different time scales during the switching process; is the cell state, storing the long-term voltage change trend; h t is a hidden state, indicating the voltage prediction result at the current moment; c t is the candidate unit state; g t For input gating, control the importance of new input features, especially at the switching moment; r t It is a residual connection, which helps to capture the rapid changes of voltage; f t is the input feature of time step t, including voltage fluctuation and system topology information; Wxi, W xf W xo W xc W hi W hf W ho W hc is the weight matrix; b i 、b f 、b o 、b c is the bias term; \σ is the sigmoid function; tanh is the hyperbolic tangent function; ⊙ represents the Hadamard product.

[0117] The voltage fluctuation during the transformer switching process contains information of multiple time scales, from millisecond-level transient fluctuations to minute-level recovery processes. Through wavelet decomposition, we can separate these different scales of information and conduct targeted modeling and prediction. Combined with wavelet transform, multi-scale decomposition can be achieved, which can be expressed as:

[0118]

[0119] Where h(t) is the hidden state sequence output by LSTM, which represents the predicted voltage sequence; D j (t) is the detail coefficient of the jth layer, reflecting the rapid voltage fluctuation during the switching process; A J (t) is the approximate coefficient of the Jth layer, reflecting the voltage recovery trend after switching; J is the maximum number of layers of wavelet decomposition, usually 3-5 layers to cover voltage changes from milliseconds to minutes.

[0120] Design a multi-scale fusion layer to integrate the prediction results of different scales, expressed as:

[0121]

[0122] Among them, y is the final predicted voltage value; y i is the prediction result of the i-th scale, including the peak at the switching moment, short-term oscillation and long-term recovery trend; w i is the fusion weight of the i-th scale, which is adaptively adjusted through training to adapt to different types of switching situations.

[0123] In a feasible embodiment, the workflow of the overvoltage prediction model is as follows:

[0124] Data preparation: Collect multi-dimensional time series data including voltage, current, power, etc.

[0125] LSTM prediction: Input the preprocessed data into the LSTM network.

[0126] The LSTM network outputs a preliminary voltage prediction sequence h(t).

[0127] Wavelet decomposition: Perform wavelet decomposition on h(t) to obtain coefficients of different scales: h(t) = D1(t) + D2(t) + ... + D j (t)+A J (t)

[0128] Among them, D j (t) is the detail coefficient, A J (t) is the approximate coefficient.

[0129] Scale-specific processing: The coefficients of each scale are processed separately. For example, D1(t) may correspond to millisecond-level spikes, using a peak detection algorithm.

[0130] D2(t) and D3(t) may correspond to short-term oscillations, using Fourier analysis.

[0131] A J (t) Corresponding to the long-term trend, a regression model can be used.

[0132] Multi-scale fusion: weighted fusion of the processing results of each scale: y = w1 y1 + w2 y2 + ... + w n y n

[0133] The weights w_i are adaptively adjusted based on historical data through a machine learning algorithm.

[0134] Output results: The final output y is the overvoltage prediction result, which contains comprehensive information from fast transients to slow changes.

[0135] Although the LSTM model can capture long-term dependencies, it may not be sensitive enough to short-term rapid changes. After wavelet decomposition, we can model different frequency components separately, such as using LSTM or other suitable models with different parameters. Multi-scale fusion allows us to comprehensively consider the prediction results of different time scales and obtain a more comprehensive prediction through weighted averaging and other methods. This method can better adapt to different types of switching situations. For example, some switching mainly affects high-frequency components, while others mainly affect low-frequency components. By adjusting the weights of different scales, the model can adaptively focus on the most relevant time scale.

[0136] It should be noted that the improved LSTM unit can better capture long-term dependencies, the wavelet transform realizes the multi-scale decomposition of the signal, and the multi-scale fusion layer integrates the prediction results of different scales to improve the prediction accuracy.

[0137] S4: Adaptive overvoltage suppression strategy generation: Based on the prediction results and system dynamic characteristics, an adaptive overvoltage suppression strategy generation algorithm is designed, taking into account multiple factors such as system stability, equipment life and economy;

[0138] Furthermore, the adaptive overvoltage suppression strategy generation includes:

[0139] Based on the prediction result y in claim 4, a strategy generation algorithm based on reinforcement learning is designed, which is expressed as:

[0140]

[0141] Among them, y t is the voltage value predicted in claim 4; t is the actual measured voltage value; E t is the current environment information, including load level, system impedance, etc.; a t is the suppression action taken at time t, such as adjusting the capacity of the reactive power compensation device, changing the transformer tap position, etc.; Q([y t ,z t ,E t ],a t ) is the state-action value function; r t+1 To perform action a t The immediate reward obtained after α is the learning rate; γ is the discount factor;

[0142]

[0143] Indicates that in the next state ([y t+1 ,z t+1 ,E t+1 ] and select the action that maximizes the Q value.

[0144] The model predictive control (MPC) method is introduced to optimize the control sequence, which is expressed as:

[0145]

[0146] Among them, u t ,…,u t+N-1 is the control sequence from the current time t to the time t+N-1; l([y t+k ,z t+k ,E t+k ],u t+k ) is the stage cost function; V f ([y t+N ,z t+N ,E t+N ]) is the terminal cost function; N is the prediction time domain length; f([y t+k ,z t+k ,E t+k ],u t+k ) is the system dynamic model, describing how the state changes with the control action; t+k+1 ,z t+k+1 ,E t+k+1 are the predicted voltage, actual voltage and environmental information after k steps respectively; u t+k is the control action after k steps.

[0147] Design an adaptive adjustment mechanism to dynamically adjust the strategy parameters according to the system status and prediction results, expressed as:

[0148]

[0149] Among them, θ t is the strategy parameter at time t, including the network parameters of the Q function and the weight coefficient of MPC; θ t+1 is the updated strategy parameter; η is the learning rate; is the performance index J with respect to parameter θ t The gradient of J is the measure of the current policy in state y t ,z t ,E t The function of the performance.

[0150] It should be noted that the strategy generation algorithm based on reinforcement learning can adaptively learn the optimal suppression strategy, the model predictive control method takes into account the system behavior in the future, and the adaptive adjustment mechanism can dynamically adjust the strategy parameters according to the system status.

[0151] S5: Multi-objective optimization and decision-making: Use multi-objective optimization methods to comprehensively consider the suppression effect, economic cost and system stability to select the optimal suppression strategy;

[0152] Furthermore, multi-objective optimization and decision-making include:

[0153] Based on the generated suppression strategy, a multi-objective optimization model is constructed, which is expressed as:

[0154] minF(x)=[f1(x),f2(x),…,f m (x)

[0155] stg i (x)≤0,i=1,2,…,p

[0156] h j (x)=0,j=1,2,…,q

[0157] Among them, x is the decision variable vector, including the capacity of the reactive compensation device, the transformer tap position and other control parameters; F(x) is the objective function vector; f1(x) represents the voltage deviation minimization target, reflecting the overvoltage suppression effect; f2(x) represents the equipment operation number minimization target, which is related to the equipment life; f m (x) may include other objectives such as maximizing voltage stability margin; g i (x) is an inequality constraint, such as equipment capacity limitation, voltage allowable range, etc.; h j (x) is an equality constraint, such as the power balance equation; p and q are the number of inequality and equality constraints, respectively.

[0158] The improved NSGA-III algorithm is used to solve multi-objective optimization problems, including non-dominated sorting, congestion calculation and reference point selection;

[0159] Use fuzzy comprehensive evaluation method to make decisions, expressed as:

[0160]

[0161] Among them, B is the comprehensive evaluation result vector, which represents the final score of each suppression scheme; A is the weight vector, which reflects the relative importance of different objectives (such as voltage quality, economy, and equipment life); R is the fuzzy relationship matrix, whose element r_{ij} represents the degree to which the i-th suppression scheme satisfies the j-th objective;

[0162] As the fuzzy synthesis operator, a weighted average operator is used here.

[0163] It should be noted that the multi-objective optimization model can consider multiple objectives at the same time, the NSGA-III algorithm can effectively solve the multi-objective optimization problem, and the fuzzy comprehensive evaluation rule can make trade-offs and decisions among multiple objectives.

[0164] S6: Simulation verification and iterative optimization: Implement the optimal suppression strategy in a simulation environment, and perform strategy evaluation and iterative optimization based on the simulation results.

[0165] Furthermore, simulation verification and iterative optimization include:

[0166] Build power system simulation models, including transformers, lines, loads and other components;

[0167] Implement the selected optimal suppression strategy and record system response and performance indicators;

[0168] Design an evaluation index system, including overvoltage amplitude, duration, system stability margin and economic cost;

[0169] Based on the simulation results, the Bayesian optimization method is used to tune the strategy parameters, which is expressed as:

[0170] θ=arg maxE[u(θ)|D 1:t ]

[0171] Among them, θ is the optimal strategy parameter, including the controller gain coefficient, threshold, etc.; θ is the set of strategy parameters to be optimized; u(θ) is the utility function, which comprehensively evaluates the performance of the strategy in terms of overvoltage suppression effect, economy and reliability; D 1:t is the historical simulation data, including the system response and performance indicators under different parameter settings; E represents the expected value, which is the expectation of the performance of the unknown parameters based on historical data; arg max represents finding the parameters that maximize the objective function.

[0172] Iterate the above steps until the termination condition is met.

[0173] It should be noted that simulation verification can evaluate the effect of the suppression strategy without affecting the actual system, and the Bayesian optimization method can efficiently adjust the strategy parameters to improve the suppression effect.

[0174] The above is a schematic scheme of a simulation analysis method for a switching transformer overvoltage suppression method of this embodiment. It should be noted that the technical scheme of the simulation analysis system of the switching transformer overvoltage suppression method and the technical scheme of the simulation analysis method of the switching transformer overvoltage suppression method described above belong to the same concept, and the details of the technical scheme of the simulation analysis system of the switching transformer overvoltage suppression method not described in detail in this embodiment can all be referred to the description of the technical scheme of the simulation analysis method of the switching transformer overvoltage suppression method described above.

[0175] The simulation analysis system of the switching transformer overvoltage suppression method in this embodiment includes:

[0176] The data acquisition and preprocessing module is used to collect real-time voltage data, historical voltage data, transformer parameters, system topology information, environmental factors and load data as the initial data set, and preprocess the initial data set, including removing outliers, filling missing values ​​and data standardization;

[0177] Dynamic feature extraction and fusion module, used to extract dynamic features from preprocessed data, taking into account time series characteristics and system topology, and fusing multi-source heterogeneous data;

[0178] Multi-scale overvoltage prediction model building module, which is used to build a multi-scale overvoltage prediction model based on the improved long short-term memory network (LSTM) and wavelet transform to achieve accurate prediction of overvoltages at different time scales;

[0179] The adaptive overvoltage suppression strategy generation module is used to design an adaptive overvoltage suppression strategy generation algorithm based on the prediction results and system dynamic characteristics, taking into account multiple factors such as system stability, equipment life and economy;

[0180] Multi-objective optimization and decision-making module, which is used to select the optimal suppression strategy by adopting a multi-objective optimization method and comprehensively considering the suppression effect, economic cost and system stability;

[0181] The simulation verification and iterative optimization module is used to implement the optimal suppression strategy in the simulation environment and perform strategy evaluation and iterative optimization based on the simulation results.

[0182] This embodiment further provides a computing device, which is applicable to the simulation analysis of the overvoltage suppression method of switching transformers, and includes:

[0183] A memory and a processor; the memory is used to store computer executable instructions, and the processor is used to execute computer executable instructions to implement the simulation analysis method for implementing the overvoltage suppression method for switching transformers as proposed in the above embodiment.

[0184] This embodiment further provides a storage medium on which a computer program is stored. When the program is executed by a processor, the simulation analysis method for implementing the method for suppressing overvoltage of a switching transformer as proposed in the above embodiment is implemented.

[0185] The storage medium proposed in this embodiment and the simulation analysis method for implementing the method for suppressing overvoltage of switching transformers proposed in the above embodiment belong to the same inventive concept. The technical details not described in detail in this embodiment can be referred to the above embodiment, and this embodiment has the same beneficial effects as the above embodiment.

[0186] Through the above description of the implementation mode, the technicians in the relevant field can clearly understand that the present invention can be implemented by means of software and necessary general hardware, and of course can also be implemented by hardware, but in many cases the former is a better implementation mode. Based on such an understanding, the technical solution of the present invention is essentially or the part that contributes to the prior art can be embodied in the form of a software product, and the computer software product can be stored in a computer-readable storage medium, such as a computer floppy disk, a read-only memory (ROM), a random access memory (RAM), a flash memory (FLASH), a hard disk or an optical disk, etc., including a number of instructions for a computer device (which can be a personal computer, a server, or a network device, etc.) to execute the methods of various embodiments of the present invention.

[0187] In summary, the present invention improves the quality and reliability of the initial data set by collecting multi-source data and performing high-quality preprocessing. Through dynamic feature extraction and fusion, full use is made of time series characteristics and system topology information. The multi-scale overvoltage prediction model constructed based on the improved LSTM and wavelet transform significantly improves the prediction accuracy. The adaptive overvoltage suppression strategy generation algorithm takes into account many factors and improves the suppression effect. The multi-objective optimization and decision-making method achieves a balance between suppression effect, economic cost and system stability. Through simulation verification and iterative optimization, the applicability and reliability of the method are further improved.

[0188] Example 2

[0189] Referring to Tables 1 to 3, an embodiment of the present invention provides a simulation analysis method for a method for suppressing overvoltage of a switching transformer. In order to verify its beneficial effects, an exemplary application scenario and a comparative scheme are provided for scientific demonstration.

[0190] First, we selected a typical 110kV substation as the simulation object, which contains two main transformers and multiple outgoing lines. We collected one year of historical data, including parameters such as voltage, current, active power, reactive power, as well as transformer switching operation records and related environmental data.

[0191] We divide the data set into a training set (80%) and a test set (20%), and apply the method proposed in the present invention to perform simulation analysis. At the same time, we select a traditional threshold-based overvoltage suppression method for comparison.

[0192] Table 1 shows the comparison of statistical features before and after data preprocessing:

[0193] Table 1 Comparison of statistical characteristics before and after data preprocessing

[0194] Statistical characteristics Before pretreatment After preprocessing Mean 110.5kV 110.2kV Standard Deviation 3.2kV 2.8kV Maximum 128.7kV 118.5kV Minimum 98.2kV 102.5kV Missing rate 2.3％ 0％

[0195] As can be seen from Table 1, through data preprocessing, we effectively removed outliers, filled missing values, and made the data distribution more concentrated.

[0196] Table 2 shows the performance comparison of the multi-scale overvoltage prediction models:

[0197] Table 2 Performance comparison of overvoltage prediction models

[0198] Performance Indicators Traditional LSTM Method of the present invention RMS E(kV) 2.15 1.37 MAE(kV) 1.82 1.12 MAP E (%) 1.65 1.02 <![CDATA[R 2 ]]> 0.87 0.94

[0199] It can be seen from Table 2 that the multi-scale overvoltage prediction model proposed in the present invention is superior to the traditional LSTM model in various performance indicators, and the prediction accuracy is significantly improved.

[0200] Table 3 shows the comparison of overvoltage suppression effects:

[0201] Table 3 Comparison of overvoltage suppression effects

[0202] Evaluation Metrics Traditional methods Method of the present invention Overvoltage peak value (pu) 1.28 1.15 Overvoltage duration (ms) 35 22 System stability margin (%) 15.3 23.7 Economic cost (10,000 yuan / year) 85.6 72.3

[0203] It can be seen from Table 3 that compared with the traditional method, the method of the present invention has significant improvements in overvoltage suppression effect, system stability and economy, etc. The overvoltage peak value is reduced by 10.16%, the duration is shortened by 37.14%, the system stability margin is increased by 54.9%, and the annual economic cost is reduced by 15.54%.

[0204] Through the above simulation analysis results, we can conclude that the simulation analysis method of the switching transformer overvoltage suppression method proposed in the present invention is significantly superior to the traditional method in data preprocessing, overvoltage prediction and suppression effect. This method can effectively improve the safety and stability of the power system, while reducing economic costs, and has important practical application value.

[0205] 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 preferred embodiments, those skilled in the art should understand that the technical solutions of the present invention may be modified or replaced by equivalents without departing from the spirit and scope of the technical solutions of the present invention, which should all be included in the scope of the claims of the present invention.

Claims

1. A simulation analysis method for switching transformer overvoltage suppression method, characterized in that: include, Collecting and preprocessing multi-source data; the multi-source data includes real-time voltage data, historical voltage data, transformer parameters, system topology information, environmental factors and load data; Extract and fuse features from preprocessed multi-source data; Based on the fused features, an overvoltage prediction model is constructed, and based on the prediction results, an overvoltage suppression generation algorithm is established; Based on the suppression strategy generated by the overvoltage suppression generation algorithm, a multi-objective optimization method is used to obtain and implement the optimal suppression strategy, and strategy evaluation and iterative optimization are performed according to the implementation results.

2. The simulation analysis method for the switching transformer overvoltage suppression method according to claim 1, characterized in that: The preprocessing includes removing outliers, filling in gaps and standardizing. The removal of outliers using the moving median method is shown below: x′ i =median(x i-k ,…,x i ,…,x i+k ) Among them, x′ i is the processed voltage data, x i is the original collected voltage data, k is the window size selected according to the transformer switching frequency; The multivariate interpolation method used to fill the gap is shown in the following formula: in, is the estimated missing voltage value, f is the interpolation function trained based on historical data, x1, x2, x n are the known voltage values ​​and related environmental parameters adjacent to the missing data time; The standardization uses the Z-score standardization method as shown below: Among them, z is the standardized voltage data, x is the original or processed voltage data, μ is the mean of the historical voltage data, and σ is the standard deviation of the historical voltage data.

3. The simulation analysis method for the switching transformer overvoltage suppression method according to claim 2, characterized in that: The feature extraction and fusion includes: Perform time-frequency joint analysis on the preprocessed voltage data to extract voltage features; Extract system topology features; fusing the voltage characteristics and the system topology characteristics; The time-frequency joint analysis is shown in the following formula: Among them, WPT j,k (z) is the wavelet packet coefficient of voltage data, j is the decomposition scale, k is the node number, h k (n) is the wavelet filter coefficient, ψ(z) is the selected wavelet function, n is the discrete time index, representing the data sampling point, j / 2 is the normalization factor, 2 j z represents the scale transformation of the input signal z; The extraction of system topological features using graph convolutional network is shown in the following formula: H (l+1) =σ(D- 1 / 2 AD- 1 / 2 H (l) W (l) ) Among them, H (l) is the node feature matrix of the lth layer, which represents the state characteristics of each node in the power system at the lth layer, H (l+1) is the node feature matrix of the l+1th layer, A is the system topology adjacency matrix, reflecting the connection relationship between the transformer and other devices, D is the degree matrix, used for normalization, W (l) is the weight matrix, σ is the activation function, D -1 / 2 AD -1 / 2 is a symmetric normalized Laplace matrix used to smooth features; The voltage characteristics and the system topology characteristics are combined as shown in the following formula: Among them, α i is the attention weight of the i-th feature, e i is the attention score, a is the attention function, W q and W k is the learnable parameter matrix, h i Input features, including wavelet packet coefficients WPT j,k (z) and the features output by the last layer of the graph convolutional network, n is the total number of features; exp is an exponential function used to convert the score into a positive value; is a normalization factor that ensures that the sum of all weights is 1.

4. The simulation analysis method for the switching transformer overvoltage suppression method according to claim 3 is characterized in that: The input feature F at time step t t Inputting the overvoltage prediction model to obtain a prediction result, and performing multi-scale decomposition and reintegration on the prediction result to obtain a final prediction result; The overvoltage prediction model is shown in the following formula: i t =σ(W xi x t +W hi h t-1 +b i ) f t =σ(W xf x t +W hf h t-1 +b f ) o t =σ(W xo x t +W ho h t-1 +b o ) h t =o t ☉tanh(c t )+r t Among them, i t 、f t , o t They are input gate, forget gate and output gate, which are used to control the flow of information at different time scales during the switching process; c t is the cell state, storing the long-term voltage change trend; h t It is in hidden state, indicating the voltage prediction result at the current moment; is the candidate unit state; g t is the input gate; r t is the residual connection; Wxi, W xf W xo W xc W hi W hf W ho W hc is the weight matrix; b i 、b f 、b o 、b c is the bias term; σ is the sigmoid function; tanh is the hyperbolic tangent function; ⊙ represents the Hadamard product; The multi-scale decomposition is shown in the following formula: Among them, h(t) is the hidden state sequence output by LSTM, which represents the predicted voltage sequence, D j (t) is the detail coefficient, A J (t) is the approximation coefficient, J is the number of decomposition levels; The integration is shown below: Among them, y is the final predicted voltage value, y i is the prediction result of the i-th scale, w i is the fusion weight.

5. The simulation analysis method for the switching transformer overvoltage suppression method according to claim 4, characterized in that: The overvoltage suppression generation algorithm generates a suppression strategy based on the final prediction result as shown in the following formula: Among them, y t is the predicted voltage value; t is the actual measured voltage value; E t is the current environment information; a t is the suppression action taken at time t; Q([y t ,z t ,E t ],a t ) is the state-action value function; r t+1 To perform an action, a t is the immediate reward obtained; α is the learning rate; γ is the discount factor; Indicates that in the next state ([y t+1 ,z t+1 ,E t+1 ] and select the action that maximizes the Q value.

6. The simulation analysis method for the switching transformer overvoltage suppression method according to claim 5, characterized in that: The multi-objective optimization method constructs a multi-objective optimization model according to the suppression strategy as shown in the following formula: minF(x)=[f1(x),f2(x),…,f m (x) s.t.g i (x)≤0,i=1,2,…,p h j (x)=0,j=1,2,…,q Where x is the decision variable vector, F(x) is the objective function vector; f1(x) represents the voltage deviation minimization target; f2(x) represents the equipment operation number minimization target, f m (x) including other objectives such as maximizing voltage stability margin; g i (x) is the inequality constraint; h j (x) is an equality constraint, such as the power balance equation; p and q are the number of inequality and equality constraints, respectively.

7. The simulation analysis method for the switching transformer overvoltage suppression method according to claim 5, characterized in that: The overvoltage suppression generation algorithm also includes an optimization control sequence and an adaptive regulation mechanism; The optimized control sequence is shown below: Among them, u t ,…,u t+N-1 is the control sequence from the current time t to the time t+N-1; l([y t+k ,z t+k ,E t+k ],u t+k ) is the stage cost function; V f ([y t+N ,z t+N ,E t+N ]) is the terminal cost function; N is the prediction time domain length; f([y t+k ,z t+k ,E t+k ],u t+k ) is the system dynamic model, describing how the state changes with the control action; t+k+1 ,z t+k+1 ,E t+k+1 are the predicted voltage, actual voltage and environmental information after k steps respectively; u t+k is the control action after k steps; The adaptive regulation mechanism dynamically adjusts the strategy parameters according to the system state and the prediction results of the overvoltage prediction model, as shown in the following formula: Among them, θ t is the strategy parameter at time t, including the network parameters of the Q function and the weight coefficient of MPC; θ t+1 is the updated strategy parameter; η is the learning rate; is the performance index J with respect to parameter θ t The gradient of J is the measure of the current policy in state y t ,z t ,E t The function of the performance.

8. A simulation analysis system for a method of suppressing overvoltage of a switching transformer, characterized in that: include: A data acquisition and processing module is used to acquire and pre-process multi-source data; the multi-source data includes real-time voltage data, historical voltage data, transformer parameters, system topology information, environmental influencing factors and load data; Feature extraction module, used to extract and fuse features from preprocessed multi-source data; The model building module is used to build an overvoltage prediction model based on the fused features, and to establish an overvoltage suppression generation algorithm based on the prediction results; The optimization and iteration module uses a multi-objective optimization method to obtain and implement the optimal suppression strategy based on the suppression strategy generated by the overvoltage suppression generation algorithm, and performs strategy evaluation and iterative optimization according to the implementation results.

9. An electronic device, comprising: Memory and processor; The memory is used to store computer executable instructions, and the processor is used to execute the computer executable instructions. When the computer executable instructions are executed by the processor, the steps of the simulation analysis method of the switching transformer overvoltage suppression method described in any one of claims 1 to 7 are implemented.

10. A computer-readable storage medium storing computer-executable instructions, which, when executed by a processor, implement the steps of the simulation analysis method of the switching transformer overvoltage suppression method according to any one of claims 1 to 7.

Images