A power grid operation situation reasoning method and system based on an echo state network

Abstract

The present application belongs to the technical field of power systems, and relates to a power grid operation situation reasoning method and system based on an echo state network, aiming to improve the accuracy and efficiency of power grid fault prediction. The method builds an echo state network model, automatically adjusts the network hyperparameters using Bayesian optimization techniques to improve model performance. At the same time, the characteristic values calculated by wavelet packet decomposition are input into the model to realize the prediction of the power grid state and the identification of the fault location. The system includes an echo network construction module, a Bayesian hyperparameter optimization module, a feature decomposition calculation module, and an output reasoning optimization module. The present application also relates to a computer device and a computer readable storage medium for implementing the above method. The technical solution improves the accuracy of power grid fault detection and the efficiency of power grid operation situation reasoning through automatic hyperparameter adjustment and systematic verification, supports fast power supply recovery after power grid failure, and has important practical value and broad application prospects.

Description

TECHNICAL FIELD

[0001] The application belongs to the technical field of power systems, and particularly relates to a power grid operation situation reasoning method based on an echo state network. BACKGROUND

[0002] The echo state network (ESN) prediction is widely applied and has a memory capability. The reservoir of the ESN can save information of previous states, which enables it to memorize and model the past input data, so as to make better prediction on future input. The ESN can effectively process high-dimensional data. The random connection in the reservoir produces a large number of nonlinear mappings, which helps to efficiently reduce the dimension or extract features of the input data. For the value of the hyperparameter, the improved ESN is obtained by using Bayesian optimization, so as to adapt to the demand of the power grid operation situation reasoning.

[0003] Most of the existing distributed power grids are radial and have many line branches. In actual operation, there will be problems such as grounding fault and overload. These faults are long-term difficult problems, and the maintenance process is complicated and time-consuming, difficult to implement disposal, and not conducive to the rapid recovery of the power grid. Therefore, it is necessary to construct a fault identification model of the operation situation, and use technologies such as echo state network to reason the fault situation of the power grid operation, so as to ensure the stable operation of the power grid.

[0004] The present application proposes a power grid operation situation reasoning method based on Bayesian optimization echo state network, which is used for fault situation reasoning of the power grid. The technical scheme constructs an echo state network based on Bayesian optimization, proposes a pre-judgment method of power grid fault, and innovatively designs a fault situation reasoning model based on the echo state network, reasons the identification problem of the fault, and supports the rapid recovery of power supply of the power grid fault. SUMMARY

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

[0006] To solve the above technical problems, the present application provides the following technical scheme. A power grid operation situation reasoning method based on an echo state network comprises: constructing an echo state network model;

[0007] The hyperparameters are found by Bayesian optimization, the evaluation index of the performance of the echo state network model is established by sampling according to the prior information through a collection function, and an optimization algorithm with a Gaussian process as a probability agent model is adopted;

[0008] The characteristic values of the wavelet packet coefficients obtained by decomposition calculation are input into the model to obtain predicted values, and the characteristic values of the output reasoning are obtained.

[0009] The echo state network based on Bayesian optimization is applied.

[0010] As a preferred scheme of the power grid operation situation reasoning method based on an echo state network, the echo state network model comprises an input layer, a reserve pool and an output layer.

[0011] As a preferred scheme of the power grid operation situation reasoning method based on an echo state network, the hyperparameter is found by Bayesian optimization, and the representation is as follows:

[0012] y=f(x)+∈

[0013] wherein x represents a hyperparameter data set, y represents an observation value, and ∈ represents a noise with a mean of 0 and a variance of a 2 .

[0014] As a preferred scheme of the power grid operation situation reasoning method based on an echo state network, the optimization algorithm using the Gaussian process as a probability agent model is represented as follows:

[0015] The distribution of the Gaussian process can be represented by a mean function m() and a covariance function k():

[0016] f(x)~GP(m(x),k(x,x'))

[0017]

[0018] f is a function containing hyperparameters, K is a covariance matrix, and X 1：t is a hyperparameter set data obtained by the first t optimizations. t

[0019] The probability agent model distribution obtained by the first t optimizations and the t+1 optimization is as follows:

[0020]

[0021] k t+1 = [k(x t+1 , x1), …, k(x t+1 , xt)] T

[0022] The prediction distribution is obtained as follows:

[0023]

[0024] ​As a preferred embodiment of the power grid operation status reasoning method based on echo state network described in this invention, the predicted value obtained by inputting the feature value of wavelet packet coefficients obtained by decomposition calculation into the model includes the following: the three-phase status values ​​of A, B, and C in the output feature values ​​observe whether the three-phase power supply is normal or faulty; the fault type identifies whether it is a transverse fault or a longitudinal fault; and the location feature value can identify four uniformly distributed distances.

[0025] As a preferred embodiment of the power grid operation status reasoning method based on echo state networks described in this invention, the application of the Bayesian optimized echo state network includes,

[0026] Based on the constructed echo state network model, the dataset is divided into training set, optimization set, validation set, and test set. The hyperparameter set of the state echo network is initialized, and the hyperparameter set is optimized using Bayesian methods. The optimized dataset and the initialized hyperparameter set are input, and the probabilistic improved acquisition function is selected. The probabilistic surrogate model adopts a Gaussian model.

[0027] As a preferred embodiment of the power grid operation status reasoning method based on echo state networks described in this invention, the application of the Bayesian optimized echo state network further includes...

[0028] Based on the dataset and the surrogate model, the distribution p(f) of the objective function value for this iteration is calculated. * |X 1：t x t+1 Maximize the acquisition function to find the optimal hyperparameter set x for this iteration. t+1 , using x t+1 Calculate the objective function value f * , will data (x t+1 f * Add it to the optimization dataset, t = t + 1;

[0029] If the number of iterations exceeds 500, the hyperparameter set will be optimized and output; otherwise, the optimization will be performed again.

[0030] Input the optimal hyperparameter set and construct the echo state network model;

[0031] Select training set data, update the reserve pool state; train and output the weight matrix;

[0032] For the trained model, input validation data to validate the model and analyze its effectiveness;

[0033] To test the model, input the test set into the model, compare the output results with the dataset, and test the accuracy of detecting power grid fault types and fault distances.

[0034] As a preferred scheme of the power grid operation situation reasoning system based on an echo state network, the scheme comprises an echo network construction module, a Bayesian hyperparameter optimization module, a feature decomposition calculation module and an output reasoning optimization module.

[0035] A computer device comprises a memory and a processor, and the memory stores a computer program, wherein the processor implements the steps of any one of the power grid operation situation reasoning methods based on an echo state network when executing the computer program.

[0036] A computer readable storage medium stores a computer program, wherein the computer program is executed by a processor to implement the steps of any one of the power grid operation situation reasoning methods based on an echo state network.

[0037] The beneficial effects of the present application are as follows:

[0038] (1) Improve prediction accuracy: use wavelet packet eigenvalues to predict power grid state and fault location.

[0039] (2) Optimize model performance: automatically adjust network hyperparameters through Bayesian optimization.

[0040] (3) Data set rational distribution: ensure the comprehensiveness of model training and verification.

[0041] (4) Automatic hyperparameter adjustment: reduce manual intervention and improve optimization efficiency.

[0042] (5) Systematic verification: ensure model reliability through training, verification and testing.

[0043] (6) Modular design: facilitate system maintenance and upgrade.

[0044] (7) Hardware and software integration: facilitate actual deployment and application. BRIEF DESCRIPTION OF DRAWINGS

[0045] In order to more clearly illustrate the technical solutions of the embodiments of the present application, the following will briefly introduce the drawings needed to be used in the embodiment description. Obviously, the drawings in the following description are only some embodiments of the present application, and for those skilled in the art, other drawings can also be obtained without creative labor on the basis of these drawings.

[0046] Figure 1 A general echo state network model schematic diagram of a power grid operation situation reasoning method based on an echo state network provided by an embodiment of the present application. DETAILED DESCRIPTION

[0047] In order to make the above objectives, characteristics and advantages of the present application more obvious and easy to understand, the specific embodiments of the present application will be described in detail below with reference to the accompanying drawings. Obviously, the described embodiments are part of the embodiments of the present application, rather than all the embodiments. Based on the embodiments in the present application, all other embodiments obtained by those skilled in the art without creative labor should belong to the protection scope of the present application.

[0048] Embodiment 1

[0049] Reference Figure 1 For the first embodiment of the present application, the embodiment provides a power grid operation situation reasoning method based on an echo state network, as shown in formula (1), comprising: Figure 1

[0050] In the embodiment of the present application, the echo state network model

[0051] The echo state network (ESN) is composed of an input layer, a reservoir, and an output layer, as shown in formula (2). Figure 1

[0052] The input layer shows that u(n) is an input parameter, and the number of neurons K. Win is the input weight matrix of the input layer to the reservoir.

[0053] The reservoir x(n) is the internal neuron state, and the number is N. The reservoir neuron stores historical sequence information, and the connection state is random and fixed. W is the internal weight matrix of the reservoir.

[0054] The output layer y(n) is the output state, and the number of neurons is L. W out is the output weight matrix. W back is the connection weight matrix of the output layer to the reservoir. W in , W, W back are generated before training and will not change during training.

[0055] The reservoir state update equation is:

[0056] x(n+1)=f n (W in u(n+1)+Wx(n)+W back y(n)) (1)

[0057] f n is the activation function of the reservoir neuron, usually selected as the Sigmoid function.

[0058] The network output equation is:

[0059] y(n+1)=f out (W out ​​(u(n+1), x(n+1))) (2)

[0060] fout is the activation function of the output layer neuron. Only W is updated during training out .

[0061] 5.2 Bayesian Optimization

[0062] Bayesian optimization is an algorithm to effectively find hyperparameters with good generalization ability. It samples according to the prior information through the acquisition function, establishes the evaluation index of the performance of the echo state network model, and uses the optimization algorithm with Gaussian process as the probability agent model. x is the hyperparameter data set, and y is the observed value:

[0063] y = f(x) + ∈. ∈ is independent and identically distributed, with mean 0 and variance a2 noise.

[0064] The distribution of the Gaussian process can be represented by the mean function m() and the covariance function k():

[0065] f(x) ~ GP(m(x), k(x, x')) (3)

[0066]

[0067] f is a function containing hyperparameters, K is the covariance matrix, which is calculated by the hyperparameter set data X 1：t = {x1, …, xt} obtained by the first t optimizations. t

[0068] The probability agent model distribution obtained by the first t and t+1 optimizations is as follows:

[0069]

[0070] k t+1 = [k(x t+1 , x1), …, k(x t+1 , xt)] T (6)

[0071] The predicted distribution is obtained:

[0072]

[0073] 5.3 Power grid fault situation reasoning model

[0074] The characteristic values of the wavelet packet coefficients obtained by decomposition calculation are input into the above model to obtain the predicted value, and further obtain the characteristic values of the output reasoning.

[0075] ​The output feature values ​​show whether the three-phase power supply is normal or faulty based on the three-phase status values ​​of A, B, and C. The fault type can be identified as a transverse or longitudinal fault, and the location feature values ​​can identify four uniformly distributed distances.

[0076] Table 1-1 Power Grid Fault Identification Model

[0077]

[0078] 5.4 Application of Bayesian Optimized Echo State Network

[0079] Step 1: Construct a state echo network based on the principles in section 5.1, and divide the dataset into training, optimization, validation, and test sets in a 5:2:2:1 ratio. Initialize the state echo network hyperparameter set using Bayesian optimization. Input the optimization dataset and the initialized hyperparameter set, select the probabilistic improved acquisition function, and use a Gaussian model for the probabilistic surrogate model. Perform 500 iterations.

[0080] Step 2: Calculate the possible distribution p(f) of the objective function value for this iteration based on the dataset from Step 1 and the surrogate model. * |X 1：t x t+1 Maximize the acquisition function to find the optimal hyperparameter set xt+1 for this iteration, and use xt+1 to calculate the objective function value f. * This set of data (x) t+1 f * Add it to the optimization dataset, t = t + 1;

[0081] Step 3: If the number of iterations in Step 2 exceeds 500, optimize the hyperparameter set output; otherwise, repeat Step 2.

[0082] Step 4: Input the optimal hyperparameter set from Step 3 to construct the echo state network model. The input layer is a column vector consisting of 39 feature values, the reservoir contains 39 neurons, and the output layer contains a column vector of 5 elements.

[0083] Step 5: Select training set data and update the reserve pool state from Step 4; train the output weight matrix Wout;

[0084] Step 6: For the model trained in Step 4, input validation data to validate the model and analyze its effectiveness;

[0085] Step 7: Test the model by inputting the test set into the model and comparing the output results with the dataset to test its accuracy in detecting power grid fault types and fault distances.

[0086] Example 2

[0087] The second embodiment of the present invention differs from the first embodiment in that:

[0088] If the functions are implemented in the form of software function units and sold or used as independent products, they can be stored in a computer readable storage medium. Based on this understanding, the technical solutions of the present application or the parts of the present application that essentially contribute to the prior art or the parts of the technical solutions of the present application can be embodied in the form of a software product. The computer software product is stored in a storage medium and includes a plurality of instructions for causing a computer device (which can be a personal computer, a server, or a network device, etc.) to execute all or part of the steps of the methods described in the various embodiments of the present application. The aforementioned storage medium includes: a U disk, a mobile hard disk, a read-only memory (ROM, Read-Only Memory), a random access memory (RAM, Random Access Memory), a magnetic disk or an optical disk, and various media that can store program codes.

[0089] The logic and / or steps represented in the flowcharts or otherwise described herein, for example, can be considered as a list of executable instructions for implementing logic functions, which can be specifically embodied in any computer readable medium for use by or in conjunction with an instruction execution system, device or apparatus, such as a computer-based system, a system including a processor or other system that can fetch and execute instructions from the instruction execution system, device or apparatus. For the purpose of this specification, the "computer readable medium" can be any device that can contain, store, communicate, propagate or transport programs for use by or in conjunction with the instruction execution system, device or apparatus, or in conjunction with these instruction execution systems, devices or apparatus.

[0090] More specific examples (a non-exhaustive list) of the computer readable medium include the following: an electrical connection having one or more wires (electrical devices), a portable computer diskette (magnetic devices), a random access memory (RAM), a read-only memory (ROM), an erasable programmable read-only memory (EPROM or flash memory), an optical fiber device, and a portable compact disc read-only memory (CDROM). In addition, the computer readable medium can even be paper or other suitable medium on which the program can be printed, because the program can be electronically obtained, for example, by optical scanning of the paper or other medium, followed by editing, interpreting or processing, if necessary, in other suitable ways, to be electronically obtained and then stored in the computer memory.

[0091] It should be understood that various parts of the present application can be implemented in hardware, software, firmware, or a combination thereof. In the above embodiments, a plurality of steps or methods can be implemented in software or firmware stored in a memory and executed by a suitable instruction execution system. For example, if implemented in hardware, and as in another embodiment, it can be implemented using any or a combination of the following technologies, which are known in the art: discrete logic circuitry having logic gates for implementing logic functions on data signals, application specific integrated circuits having appropriate combinational logic gates, programmable gate arrays (PGA), field programmable gate arrays (FPGA), etc.

[0092] Embodiment 3

[0093] In a third embodiment of the present application, the embodiment provides an echo state network-based power grid operation situation reasoning system, characterized in that: comprising an echo network construction module, a Bayesian hyperparameter optimization module, a feature decomposition calculation module, and an output reasoning optimization module.

[0094] The echo network construction module constructs an echo state network model.

[0095] The Bayesian hyperparameter optimization module finds hyperparameters through Bayesian optimization, samples according to prior information through a collection function, establishes an evaluation index of the performance of the echo state network model, and adopts an optimization algorithm with a Gaussian process as a probabilistic surrogate model.

[0096] The feature decomposition calculation module utilizes decomposition calculation to obtain eigenvalues of wavelet packet coefficients, inputs the eigenvalues into the model to obtain predicted values, and obtains eigenvalues of output reasoning.

[0097] The output reasoning optimization module is based on the Bayesian optimization echo state network.

[0098] It should be noted that the above embodiments are only used to illustrate the technical solutions of the present application and are not limiting. Although the present application 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 application can be modified or replaced equivalently without departing from the spirit and scope of the technical solutions of the present application, and they should be covered in the scope of the claims of the present application.

Claims

1. An echo state network-based power grid operation situation inference method, characterized in that: The method comprises the following steps: constructing an echo state network model; finding hyperparameters through Bayesian optimization, sampling according to prior information through a collection function, establishing an evaluation index of the performance of the echo state network model, and adopting an optimization algorithm with a Gaussian process as a probability agent model; using characteristic values of wavelet packet coefficients obtained through decomposition calculation to input the model to obtain predicted values, and obtaining characteristic values of output reasoning; applying the Bayesian optimization echo state network; the use of characteristic values of wavelet packet coefficients obtained through decomposition calculation to input the model to obtain predicted values comprises observing three-phase power supply normality or failure through A, B and C three-phase state value in output characteristic values, identifying horizontal failure or vertical failure through fault type identification, and identifying four kinds of uniformly distributed distances through position characteristic values; the application of the Bayesian optimization echo state network further comprises, The distribution of the objective function value for this iteration was calculated based on the dataset and the surrogate model. Maximize the acquisition function to find the optimal hyperparameter set x for this iteration. t+1 , using x t+1 Calculate the objective function value f * , data Add it to the optimization dataset, t=t+1; when the number of iterations exceeds 500, the hyperparameter group is output, otherwise the optimization is performed again; inputting the optimal hyperparameter group to construct the echo state network model; selecting training set data to update the reserve pool state; training the output weight matrix; for the trained model, inputting verification data to verify the model and analyzing the effectiveness; testing the model, inputting the test set into the model, comparing the output results and the data set, and testing the accuracy of the power grid fault type and fault distance.

2. The power grid operational posture inference method based on an echo state network according to claim 1, characterized in that: The echo state network model comprises an input layer, a reserve pool and an output layer.

3. The power grid operational posture inference method based on echo state network of claim 2, wherein: The hyperparameters found through Bayesian optimization are represented as where x represents a set of hyperparameter data, and y represents an observation value, a with mean 0 and variance 2 noise.

4. The power grid operational posture inference method based on echo state network of claim 3, wherein: The optimization algorithm with the Gaussian process as the probability agent model is represented as The distribution of the Gaussian process can be represented by a mean function m() and a covariance function k(). f is a function that includes hyperparameters, K is a covariance matrix, and is the data X of hyperparameter groups obtained from the previous t optimizations 1:t = {x1, x2,..., xt} t = {x1, x2,..., xt} The probability agent model distribution obtained through the first t times and the t+1 times optimization is represented as The predicted distribution is represented as 。 5. The echo state network-based power grid operational posture inference method of claim 4, wherein: The application of the Bayesian optimization echo state network comprises According to the constructed echo state network model, the data set is divided into a training set, an optimization set, a verification set and a test set, the state echo network hyperparameter group is initialized, the Bayesian optimization hyperparameter group is adopted, the optimization data set and the initialized hyperparameter group are input, the probability improved collection function is selected, and the probability agent model adopts the Gaussian model.

6. A system for power grid operation situation reasoning based on the echo state network according to any one of claims 1-5, characterized in that: The method comprises the following steps: 7.A computer device, comprising a memory and a processor, wherein the memory stores a computer program, and the computer device is configured to perform the method according to any one of claims 1-6 when the computer program is executed by the processor. The processor executes the computer program to realize the steps of the method of any one of claims 1 to 5.

8. A computer readable storage medium having stored thereon a computer program, characterized in that, The computer program is executed by the processor to realize the steps of the method of any one of claims 1 to 5.

Images