Automatic quasi-synchronization paralleling method and device based on prediction model
By combining the gray prediction model and the BP neural network model, the prediction of emergency power generation vehicles parallelized at the same time is optimized, and the problems of insufficient adaptive learning ability and large full-cycle sampling errors are solved, and efficient and accurate judgment of closing moments is achieved.
Patent Information
- Application Number
- CN202510654026.6
- Authority / Receiving Office
- CN · China
- Patent Type
- Applications(China)
- Current Assignee / Owner
- Filing Date
- 2025-05-21
- Publication Date
- 2025-08-22
AI Technical Summary
In the parallel prediction of emergency power generation vehicles during the prior art, the adaptive learning ability of the gray prediction model is limited, there is a lot of room for improvement in neural network prediction performance, and traditional algorithms require full-cycle sampling, resulting in large errors.
Combining the gray prediction model and the BP neural network model, a prediction model is constructed using the finite phase angle difference and slip angle frequency data, and network parameters are optimized through the backpropagation algorithm of the BP neural network to predict the parallel closing phase angle.
It achieves efficient and accurate prediction performance, can provide accurate judgment basis at the moment of closing, avoid errors, and improves the accuracy and efficiency of emergency power generation vehicles paralleling at the same time.
Smart Images
Figure CN120528017A_ABST
Abstract
Description
Technical Field
[0001] The present invention relates to the field of synchronous paralleling of emergency power generation vehicles, and in particular to an automatic quasi-synchronous paralleling method and device based on a prediction model. Background Art
[0002] Traditional synchronous parallel algorithms, such as the Fourier transform algorithm and the recursive least squares algorithm, require full-cycle sampling (synchronous sampling) of the signal. From the perspective of signal processing theory, full-cycle sampling can ensure the integrity and accuracy of the sampled data, thereby ensuring that the algorithm accurately extracts and analyzes the signal features. Otherwise, large errors will be generated. With the promotion of prediction models, it is explored how to construct a feasible prediction model based on limited phase angle difference and slip angle frequency data, supported by the determination of closing conditions. Currently, in the field of automatic quasi-synchronous parallel prediction, methods such as time series analysis, wavelet transform, and gray prediction model have been widely used and have achieved initial and significant results.
[0003] Grey prediction models are favored for their high computational efficiency and predictive accuracy. However, they have limitations in adaptive learning of sequences. In contrast, neural network prediction models iteratively process input information and utilize adaptive mechanisms to adjust internal weights and thresholds, enabling a close fit to arbitrary functional relationships and significantly improving prediction accuracy. Although neural networks demonstrate good fitting capabilities when trained on discrete data, their predictive performance still has significant room for improvement. Therefore, a prediction method with strong adaptive learning capabilities and the potential to significantly improve prediction results is urgently needed. Summary of the Invention
[0004] The purpose of the present invention is to overcome the defects of the above-mentioned prior art and to provide an automatic quasi-synchronous paralleling method and device based on a prediction model.
[0005] The purpose of the present invention can be achieved by the following technical solutions:
[0006] An automatic quasi-synchronous paralleling method and device based on a prediction model, comprising the following steps:
[0007] S01. Obtaining time series data on operating parameters of the emergency generator and the power grid system;
[0008] S02. Use the grey prediction model to estimate the original data sequence to obtain a new sequence data set;
[0009] S03. Use the BP neural network model to optimize and analyze the new sequence data set to predict the phase angle before parallel closing;
[0010] S04. Determine the parallel closing time based on the prediction results.
[0011] Furthermore, the operating parameter time series data includes phase angle difference data and slip frequency data.
[0012] Furthermore, the grey prediction model used in step S02 can construct a feasible prediction model with limited phase angle difference and slip angle frequency data, and be supported by the determination of closing conditions.
[0013] Furthermore, the BP neural network model prediction process is used in step S03, including network construction, network training and network prediction.
[0014] Furthermore, the parallel closing moment in step S04 is based on the leading phase angle calculated according to the prediction result of the slip angle frequency, which is used as the basis for determining the closing moment.
[0015] A computer device includes a processor and a memory, wherein the memory is used to store a computer program and the processor is used to execute the computer program. The processor is characterized in that the processor is used to execute the computer program to perform the above method.
[0016] Compared with the prior art, the present invention has the following advantages:
[0017] 1) The present invention uses a grey prediction model to estimate the original phase angle difference data and slip frequency data sequence to obtain a preliminary prediction value.
[0018] 2) Based on the grey model prediction, the present invention optimizes and adjusts the weight matrix and neuron threshold in the network through the back propagation algorithm of the BP neural network, accelerates the training convergence process of the network, and ultimately achieves high efficiency and high precision of the prediction performance.
[0019] 3) The present invention can calculate the leading phase angle at the next moment based on the predicted value of the slip angular frequency, and use this as the basis for judging the closing condition and determining the closing moment. BRIEF DESCRIPTION OF THE DRAWINGS
[0020] Figure 1 It is a flowchart of the implementation of an automatic quasi-synchronous parallel method based on a prediction model.
[0021] Figure 2 It is a grey neural network prediction model
[0022] Figure 3 It is the principle of grey neural network prediction quasi-synchronous control DETAILED DESCRIPTION
[0023] The present invention is described in detail below with reference to the accompanying drawings and specific embodiments. This embodiment is implemented based on the technical solution of the present invention, and provides a detailed implementation method and specific operation process, but the protection scope of the present invention is not limited to the following embodiments.
[0024] like Figure 1 As shown, the steps of the automatic quasi-synchronous paralleling method based on the prediction model in this embodiment include:
[0025] S01. Obtaining time series data on operating parameters of the emergency generator and the power grid system;
[0026] Specifically, in the field of automatic quasi-synchronous parallel prediction, we explore how to build a feasible prediction model based on limited phase angle difference and slip angle frequency data, supported by the judgment of closing conditions, so as to achieve advanced control and avoid missing the best closing time.
[0027] S02. Use the grey prediction model to estimate the original data sequence to obtain a new sequence data set;
[0028] Specifically, the original sequence x i (n)(including the phase angle difference θ i and slip frequency ω i ) Input the GM(1,1) model, construct the grey differential equation, and generate a new sequence x by solving the equation i+1 (n), whose specific form is shown in formula (1). Then, the generated sequence x i+1 (n) Input the neural network model, perform iterative training, and continuously optimize the network parameters until convergence is reached.
[0029]
[0030] Simplifying (1) we can get formula (2):
[0031]
[0032] S03. Use the BP neural network model to optimize and analyze the new sequence data set to predict the phase angle before parallel closing;
[0033] Specifically, the prediction process of the grey neural network combination model includes three main steps: network construction, network training and network prediction. Its network topology is as follows: Figure 2 As shown, assume that the LC layer contains m neurons, represented by G = [g1, g2, ..., g m ], the threshold matrix of the LC layer is g g =[g g1 ,g g2 ,…,g gm ], the threshold of the LD layer is g y The weight between the LA layer and the LB layer is denoted as λ 11 , the weight matrix between the LB layer and the LC layer is λ2=[λ 21 ,λ 22 ,…,λ2m ] T , the weight matrix between the LC layer and the LD layer is λ3=[λ 31 ,λ 32 ,…,λ 3m ] T .
[0034] Specifically, based on the parameters output by the grey theory, the initial weights and thresholds of the neural network input side can be further obtained.
[0035]
[0036] The activation function used is shown below.
[0037]
[0038] Finally, the prediction information of the proposed model can be expressed as follows:
[0039] y i+1 =f(f(x i+1 (n)·λ1+β k )·λ2+β y ) (6)
[0040] Set the maximum number of network training times to N. During the network training process, when the loss function L converges to a minimum value, or its value is lower than a pre-set threshold, it means that the network has reached the optimal state. In this optimal state, the prediction result y i+1 The difference between the true value and the true value is the smallest, and the two are closest.
[0041] S04. Determine the parallel closing time based on the prediction results.
[0042] Specifically, in the process of implementing synchronous switching operation, we use the initial angle θ i and sliding angle frequency ω i The original data is used to estimate the next time point θ i+1 and ω i+1 The new data are input into the neural network structure to realize cyclic learning. The maximum number of learning cycles is set to N = 1000, and the threshold of the loss function is set to 0.01. i+1 and y ωi+1 Then, according to the predicted value of slip angle frequency y ωi+1 Calculate the leading phase angle θ at the next moment yi+1 , which is used as the basis for judging the closing conditions. Then the grey neural network predicts the quasi-synchronous control as follows Figure 3 shown.
[0043] Specifically, when it is determined that the optimal closing time is between the current calculation point i and the next calculation point i+1, the change of the slip angle frequency and the phase angle can be analyzed by formula (7), so as to estimate the time required from the current calculation point to reach the optimal closing time.
[0044]
[0045] θ i+1 represents the traditional predicted phase angle difference, θ yi+1 represents the Echizen phase angle,
[0046] The delay t yi Then the closing command can be issued.
[0047] The above description is merely a specific embodiment of the present invention, but the scope of protection of the present invention is not limited thereto. Anyone skilled in the art can easily conceive of various prediction algorithms within the technical scope disclosed in the present invention, and such modifications or substitutions are intended to be within the scope of protection of the present invention. Therefore, the scope of protection of the present invention shall be subject to the scope of protection of the claims.
Claims
1. An automatic quasi-synchronous parallel method and device based on a prediction model, characterized in that the steps include: S01. Obtaining time series data on operating parameters of the emergency generator and the power grid system; S02. Use the grey prediction model to estimate the original data sequence to obtain a new sequence data set; S03. Use the BP neural network model to optimize and analyze the new sequence data set to predict the phase angle before parallel closing; S04. Determine the parallel closing time based on the prediction results.
2. The automatic quasi-synchronous parallel method based on the prediction model according to claim 1 is characterized in that ,The operating parameter time series data includes phase angle difference data and slip frequency data.
3. The automatic quasi-synchronous parallel method based on the prediction model according to claim 1 is characterized in that ,The gray prediction model used in the step S02 can construct a feasible ,prediction model with limited phase angle difference and slip angular frequency data, and ,is supported by the determination of closing conditions.
4. The automatic quasi-synchronous parallel method based on the prediction model according to claim 1 is characterized in that ,In step S03, the BP neural network model prediction process is used, ,including network construction, network training and network prediction.
5. The automatic quasi-synchronous parallel method based on the prediction model according to claims 1 to 4, characterized in that The parallel closing moment in step S04 is the leading phase angle calculated based on the prediction result of the slip angle frequency, which is used as the basis for judging the closing moment.
6. A computer device comprising a processor and a memory, wherein the memory is used to store a computer program and the processor is used to execute the computer program, wherein: The processor is configured to execute the computer program to perform the method according to any one of claims 1 to 5.