Data-driven high-proportion renewable energy power system operation scene identification method

A renewable energy and power system technology, applied in the direction of electrical digital data processing, special data processing applications, information technology support systems, etc., can solve the problems of identifying high-proportion renewable energy power system operating scenarios and their changing laws The effect of improving identification ability and improving operation efficiency

Active Publication Date: 2021-01-26
TSINGHUA UNIV +1
3 Cites 0 Cited by

AI-Extracted Technical Summary

Problems solved by technology

[0006] However, there are no reports on the use of data-driven methods to identify hig...
the structure of the environmentally friendly knitted fabric provided by the present invention; figure 2 Flow chart of the yarn wrapping machine for environmentally friendly knitted fabrics and storage devices; image 3 Is the parameter map of the yarn covering machine
View more

Abstract

The invention provides a data-driven high-proportion renewable energy power system operation scene identification method, and belongs to the technical field of power system operation. The method comprises the following steps: firstly, carrying out refined operation simulation on a to-be-identified power system to obtain a daily operation mode vector corresponding to the system; preprocessing all daily operation mode vectors by using a principal component analysis method to obtain a preprocessed power system operation mode matrix; determining a typical operation scene of the power system through a Kmeans + + algorithm and the compactness index of the operation mode of the power system; and realizing visualization of the operation characteristics by utilizing a t-SNE algorithm to obtain an extreme operation scene. According to the method, a typical scene in planning and operation can be effectively determined by using a data driving method, an extreme operation mode in protection and stability analysis can be rapidly identified, important reference can be provided for planning and operation personnel of a power system, and the analysis capability of the high-proportion renewable energy power system is improved.

Application Domain

Technology Topic

Electricity systemVisualization +7

Image

  • Data-driven high-proportion renewable energy power system operation scene identification method
  • Data-driven high-proportion renewable energy power system operation scene identification method
  • Data-driven high-proportion renewable energy power system operation scene identification method

Examples

  • Experimental program(1)

Example Embodiment

[0054]The present invention proposes a data-driven high-proportion renewable energy power system operation scenario identification method. The following further describes the present invention in detail with reference to the drawings and specific embodiments.
[0055]The present invention provides a data-driven high-proportion renewable energy power system operation scenario identification method, and the overall process is as followsfigure 1 As shown, including the following steps:
[0056]1) Refined operation simulation of the power system to be identified, and obtain the daily operation mode vector of the system;
[0057]Obtain power system load data, unit information, network topology information, renewable resources and their temporal and spatial correlation information from the power system operation planning department (the load data and renewable resource information are hourly data throughout the year), and use TH-DSED Software for fine operation simulation of power system (TH-DSED software copyright information: Kang Chongqing, Zhang Ning, Xia Qing, Xu Qianyao, Du Ershun, Ji Zhen. Power system sequential operation simulation software TH-DSED, software copyright, 2013SR053997.), The output data of the software includes daily traditional unit output data, renewable energy output data, load data and line flow data of the power system;
[0058]The above-mentioned four kinds of data constitute the daily operation mode vector of the power system on the corresponding day:
[0059]
[0060]Among them, g is the dispatching output of traditional units, r is the output of renewable energy units, f is the line power flow, and d is the load data. TsRepresents the number of time periods per day, |g|, |r|, |f|, |d| represents the number of traditional units, renewable energy units, lines, and load nodes, respectively. The dimension of the vector is N-dimensional (where N=(g|+|r|+|f|+|d|)×Ts).
[0061]In the present invention, through this step, the matrix P of M N-dimensional daily operation mode vectors is obtained0=(p1 p2 …PM). Among them, M is the number of days corresponding to the daily operation mode vector.
[0062]M is determined according to the time scale of the analyzed power system, which is the same as the time scale of the input data; N is related to the scale of the analyzed power system. The larger the system scale, the larger the value of N. Tofigure 2 Take the 3-node power system shown as an example. The system includes 3 nodes, 3 lines, 2 thermal power units, 1 wind power unit, and 1 load (ie |g|=2, |r|=1,|f |=3, |d|=1), if the operating mode of the system in a certain level year is analyzed, the operating mode is hourly data, namely Ts=24, then the obtained daily operating mode vector forms a matrix P0The dimension is M=365, N=(2+1+3+1)×24=168.
[0063]2) Preprocess all the daily operation mode vectors obtained in step 1) to obtain the preprocessed power system operation mode matrix;
[0064]Step 1) The matrix composed of the obtained daily operating mode vectors contains massive operating scene data, which often has the problem of bad data or dimensional redundancy, and the obtained scene data needs to be preprocessed. The present invention uses the principal component analysis PCA algorithm with higher efficiency and easy control of the compression degree to perform dimensionality reduction preprocessing on high-dimensional operation mode data. The basic principle of principal component analysis is to perform linear transformation on the data under the condition of maintaining the characteristics of the largest variance in the sample points to achieve dimensionality reduction.
[0065]Specific steps are as follows:
[0066]2-1) Pair matrix P0All the daily operation mode vectors in China are centralized, and the new matrix obtained after centering is denoted as P;
[0067]
[0068]among them, Is P0The average value of all the daily operating mode vectors in.
[0069]2-2) Calculate the covariance matrix of matrix P Perform eigenvalue decomposition on the covariance matrix Cov to obtain N eigenvalues ​​λ1,λ2,...,λNThe eigenvector h corresponding to each eigenvalue1,h2,...,hN, Where λ1≥λ2≥...≥λN(Where λiIs the i-th eigenvalue, and its corresponding eigenvector is hi);
[0070]2-3) from h1,h2,...,hNTake the first K eigenvectors to form a matrix H=(h1 h2... hK). Among them, the vector dimension K of the operating mode after compression can be determined by the compression coefficient θ0(Take 0.01 in this embodiment) Determine, where θ0=0 means no compression:
[0071]
[0072]2-4) The power system operation mode matrix P′=(p′1 p′2…P′M):
[0073]P′=HTP
[0074]P′ is the power system operating scene data after preprocessing to remove redundant information, p′iIt is the i-th column vector of the linear change matrix P'(representing the daily operation mode vector after preprocessing), and the dimension of P'is M*K, and P'is the data basis for subsequent data-driven algorithm analysis.
[0075]3) Cluster the preprocessed power system operation mode matrix, and extract the typical operation scenarios of the power system; the specific steps are as follows:
[0076]3-1) Set the initial number of clusters, L=2;
[0077]3-2) For the current L, randomly select a daily operating mode vector in P′ as the first cluster center
[0078]3-3) For each current non-cluster center's daily operating mode vector p′ in P′i, Calculate the vector to all selected cluster centers The minimum distance of:
[0079]
[0080]Among them, j is the number of selected cluster centers, 1≤j
[0081]For all diRespectively normalize, select a daily operation mode vector as the new cluster center with probability S from the daily operation mode vectors of all current non-cluster centers through random simulation. Among them, the expression of S is:
[0082]
[0083]Where n is the number of daily operating mode vectors in P′ that have not been selected as cluster centers.
[0084]Therefore, the daily operating mode vector with a larger d value has a higher probability of being selected as the new cluster center.
[0085]3-4) Repeat steps 3-3) until all cluster centers corresponding to the current L are selected.
[0086]3-5) Calculate the power system operation mode tightness CP index corresponding to the current number of clusters L. The power system operation mode tightness CP is defined as the average distance from each operation mode in each category to the corresponding cluster center. The expression is as follows:
[0087]
[0088]Among them, L represents the number of clusters, ΩjRepresents the set of j-th operating modes after clustering, Represents the cluster centers of the j-th operation mode in P′ obtained in step 2-4), and the tightness can be used to describe the dispersion of the operation mode in a certain classification mode.
[0089]3-6) Let L=L+1;
[0090]Repeat steps 3-2) to 3-5) until the change rate of the tightness index value corresponding to the current L relative to the last tightness index value corresponding to the current L is less than the set tightness change threshold δ (in the present invention) δ is taken as 1%), then the current L is the optimal number of clusters and is recorded as L*; L*Corresponding cluster center This is the typical operating scenario of the power system.
[0091]Generally speaking, the compactness decreases with the increase of the number of clusters. When the increase in the number of unit clusters makes the compactness change less than the set compactness change threshold δ (in the present invention, δ is taken as 1%), the compactness reaches Saturation, the number of clusters corresponding to the saturation point can be considered as the optimal number of clusters in the system.image 3 It is a schematic diagram of the change trend of the tightness index with the number of clusters. It can be seen that as the number of clusters L increases, the system tightness index decreases.*After the tightness index tends to be unchanged (change is less than δ), so the optimal number of clusters is L*.
[0092]Through the above standard k-means++ algorithm (step 3-2) to step 3-4)), all operating mode vectors in P′ can be clustered according to the given number of clustering centers, and the clustering results obtained are calculated for compactness index , Take the saturation point L*As the optimal number of typical scenes, the corresponding cluster center in the clustering result This is a typical operating scenario.
[0093]4) Dimensionality reduction and visual analysis of operating characteristics to obtain extreme operating scenarios of the power system;
[0094]The present invention extracts the main characteristics of the operation mode of the power system through the dimensionality reduction visualization algorithm t-SNE, which can map the high-dimensional data P′ after step 2) preprocessing to a 2-dimensional space to obtain a corresponding 2-dimensional matrix Realize the visualization of high-dimensional nonlinear operation mode of power system, the above process can beFigure 4To illustrate, the dimension of each daily operating mode vector after preprocessing is K-dimensional, and the t-SNE algorithm can convert each daily operating mode vector pi'Reduce the dimensionality to a 2-dimensional space to obtain the corresponding 2-dimensional daily operating mode vector (Figure 4The dimension reduction dimension 1 and dimension reduction dimension 2) in the dimensionality reduction dimension are the specific linear combination of the original dimension, and the daily operation vector after the dimension reduction is visually displayed. The boundary operation mode vector in the visualization result is the power system The extreme operating scenarios that need to be analyzed in protection and stability analysis.
[0095]Step 3) can determine typical operating scenarios in planning and operation, step 4) can quickly identify extreme operating scenarios in protection and stability analysis, and the results of steps 3) and 4) are the identification results of the power system operation mode.
the structure of the environmentally friendly knitted fabric provided by the present invention; figure 2 Flow chart of the yarn wrapping machine for environmentally friendly knitted fabrics and storage devices; image 3 Is the parameter map of the yarn covering machine
Login to view more

PUM

no PUM

Description & Claims & Application Information

We can also present the details of the Description, Claims and Application information to help users get a comprehensive understanding of the technical details of the patent, such as background art, summary of invention, brief description of drawings, description of embodiments, and other original content. On the other hand, users can also determine the specific scope of protection of the technology through the list of claims; as well as understand the changes in the life cycle of the technology with the presentation of the patent timeline. Login to view more.
the structure of the environmentally friendly knitted fabric provided by the present invention; figure 2 Flow chart of the yarn wrapping machine for environmentally friendly knitted fabrics and storage devices; image 3 Is the parameter map of the yarn covering machine
Login to view more

Similar technology patents

Classification and recommendation of technical efficacy words

  • Improve identification
  • Improve operational efficiency

Tape cartridge format identification in a single reel tape handling device

InactiveUS6918554B2Improve identificationReduce damagePicture changing apparatusAccessories for auxillary signalsEngineeringMagnetic tape
Owner:QUANTUM CORP
Who we serve
  • R&D Engineer
  • R&D Manager
  • IP Professional
Why Eureka
  • Industry Leading Data Capabilities
  • Powerful AI technology
  • Patent DNA Extraction
Social media
Try Eureka
PatSnap group products