Multi-dimensional safety analysis system for power grid dispatching

Abstract

A power grid dispatching multi-dimensional safety analysis system comprises a control center, and a power utilization abnormity analysis module, a weak link analysis module and a distribution network defect analysis module which are connected with the control center, and the power utilization abnormity analysis module is used for analyzing power utilization abnormity; the weak link analysis module is used for analyzing weak links of the power grid network; and the distribution network defect analysis module is used for analyzing distribution network defects. The method can improve the operation stability of the power grid, reduces the operation risk of the power grid, helps related personnel to quickly locate weak links which may cause faults when distribution network faults occur, and improves the detection efficiency.

Description

Technical Field

[0001] The present invention relates to the technical field of power systems, and in particular to a multi-dimensional safety analysis system for power grid dispatching. Background Art

[0002] As a vital pillar for the high-quality development of the national economy, the stability of electricity must be guaranteed. However, with the continuous expansion of my country's power sector and the optimization of its functions, the complexity of power dispatching systems has increased, leading to a significant increase in abnormal power dispatching incidents and service failures, posing a serious threat to the normal power supply and safe operation of the power grid. For example, abnormal power consumption exceeding allocated power and equipment indicators increases the burden on the power grid and can cause fires. Other examples include the 2003 North American blackout, the 2012 Indian blackout, the 2019 Venezuelan blackout, and the 2020 Texas blackout, all of which caused significant losses. While there is currently research on various security aspects, such as abnormal power consumption analysis and weak link analysis, these studies primarily focus on a single aspect of security and fail to examine security from multiple perspectives. Furthermore, current power consumption analysis primarily relies on data from a single scenario, making it less practical. Weak link analysis, on the other hand, primarily involves single-shot analysis without correlating it with grid faults, resulting in poor and inefficient fault analysis. Summary of the Invention

[0003] In response to the above problems, the present invention provides a multi-dimensional safety analysis system for power grid dispatching that improves the stability of power grid operation and reduces the risk of power grid operation.

[0004] The technical solution of the present invention is: a multi-dimensional safety analysis system for power grid dispatching, including a control center and a power consumption anomaly analysis module, a weak link analysis module and a distribution network defect analysis module connected to the control center.

[0005] The power consumption anomaly analysis module is used to analyze power consumption anomalies;

[0006] The weak link analysis module is used to analyze the weak links of the power grid network;

[0007] The distribution network defect analysis module is used to analyze distribution network defects.

[0008] The power consumption anomaly analysis module includes:

[0009] First, standardize the user's electricity consumption data;

[0010] Then, the standardized data is subjected to feature extraction and dimensionality reduction;

[0011] Finally, clustering algorithms are used to perform cluster analysis on the reduced-dimensional electricity consumption data to analyze electricity consumption anomalies.

[0012] During the standardization process, the maximum and minimum normalization method is selected. After finding the maximum and minimum data in the entire electricity consumption data set, the formula transformation is used to make the value range of all data be [0, 1]. The specific formula is as follows:

[0013]

[0014] Among them, x represents the initial data recorded by the measuring equipment, x max and x min They represent the maximum and minimum values ​​in the record respectively, and x′ represents the standardized data, with a value range of [0, 1].

[0015] During feature extraction, the six characteristic values ​​of mean, standard deviation, variance, range, skewness and kurtosis of the specific period characteristics are extracted. The formulas are:

[0016] The average value feature mean extracted by a specific period:

[0017]

[0018] The standard deviation feature std extracted by a specific period:

[0019]

[0020] The variance feature var extracted by a specific period:

[0021]

[0022] The range feature extracted by a specific period:

[0023] range=x max-t -x min-t

[0024] The skew feature extracted by a specific period:

[0025]

[0026] Kurtosis feature kurt extracted by specific period:

[0027]

[0028] In the above formula, the values ​​of i are 1, 7, and 30, which represent the daily, weekly, and monthly cycles respectively. x represents the user's electricity consumption within the daily, weekly, and monthly cycles. n represents the total number of cycles of the user's electricity consumption data. max-t and x min-t Indicates the maximum and minimum electricity consumption of the user in a daily, weekly, or monthly cycle;

[0029] Combining the above features, dimensionality reduction is performed through principal component analysis.

[0030] The DPeaks clustering algorithm is used to perform cluster analysis on the electricity consumption data after dimensionality reduction. For each data point i, the DPeaks clustering algorithm calculates two quantities: the local density ρ of point i and the distance δ between point i and the nearest higher density point, which are:

[0031]

[0032]

[0033] Among them, m is the total number of data points, X(d ij -d c ) is the indicator function, when d ij -d c When x(d ij -d c )=1, otherwise X(d ij -d c )=0,d ij is the Euclidean distance between point i and point j, d c is the cutoff distance, ρ j and ρ i Represent the local density of point i and point j respectively;

[0034] Based on the above formula, the local density ρ and distance δ of each point are calculated, and the cluster center and core point are obtained. Then, for each remaining point, it is divided into the cluster closest to the cluster center.

[0035] The calculation formulas for the benchmark density and benchmark distance of each cluster are as follows:

[0036]

[0037] in, and are the reference density and reference distance of the cth cluster respectively, α and β are empirical constants, and N (c) is the total number of data points contained in the cth cluster, M (c) is the set of data points contained in the cth cluster; if the local density ρ of a data point i Less than the reference density ρ0 and the distance δ i If it is greater than the reference distance δ0, the point is an abnormal point, indicating that the electricity consumption data is abnormal.

[0038] The weak link analysis module includes:

[0039] First, the edge betweenness is selected as the vulnerability index of the weak link of the power grid, and its formula is:

[0040]

[0041] Among them, E represents the edge betweenness, S j_ij is the number of shortest paths between nodes j_i and j_j, S j_ij (l) represents the number of shortest paths passing through line l, and N1 is the total number of lines;

[0042] Then, the edge betweenness index is used to calculate the vulnerability of each grid component, and the grid components are sorted by vulnerability, and the component with the highest vulnerability is selected as the first key grid component. This component is then removed from the grid, and the grid topology is updated. The vulnerability of the remaining grid components is calculated in the new grid topology, and the grid components are sorted by vulnerability. The grid component with the highest vulnerability is selected as the second key grid component. This process is iterated K times to obtain a set of K grid components with the highest vulnerability, which serves as the weak link of the power grid network.

[0043] The distribution network defect analysis module includes:

[0044] Combined with the existing distribution network defect reports, correlation analysis is performed to explore the correlation between weak links and defects. For each defect, the key words of the defect report are extracted, and combined with the status of the weak power grid components of the distribution network, an association record is formed. This association record is used as input and input into the frequent pattern growth algorithm for fault correlation analysis to form association rules between weak links and faults.

[0045] In operation, the present invention analyzes abnormal electricity consumption of users and weak links in the distribution network based on artificial intelligence and data mining algorithms, and performs correlation analysis based on distribution network defects and weak links, thereby improving the stability of power grid operation and reducing power grid operation risks. When a distribution network failure occurs, it helps relevant personnel to quickly locate the weak links that may cause the failure, thereby improving detection efficiency. BRIEF DESCRIPTION OF THE DRAWINGS

[0046] Figure 1 This is a principle block diagram of the present invention. DETAILED DESCRIPTION

[0047] like Figure 1 As shown, the present invention provides a multi-dimensional safety analysis system for power grid dispatching, including a control center and a power consumption anomaly analysis module, a weak link analysis module and a distribution network defect analysis module connected to the control center.

[0048] The power consumption anomaly analysis module is used to analyze power consumption anomalies;

[0049] The weak link analysis module is used to analyze the weak links of the power grid network;

[0050] The distribution network defect analysis module is used to analyze distribution network defects.

[0051] Since the three modules, namely the power consumption anomaly analysis module, the weak link analysis module and the distribution network defect analysis module, are all related to the quality of power grid dispatching, they can improve the safety of power grid operation, avoid abnormal power consumption, focus on the quality of the power grid itself, and help with rapid defect analysis.

[0052] The present invention transmits analysis of power consumption anomalies, weak links, etc. to a control center, facilitating monitoring and improving the safety of the power grid.

[0053] Power consumption anomaly analysis module: First, the user's power consumption data is standardized. The present invention uses the maximum and minimum normalization method to standardize the original data. After finding the maximum and minimum data in the entire power consumption data set, the formula is transformed so that the value range of all data is [0, 1]. The specific formula is as follows:

[0054]

[0055] Among them, x represents the initial data recorded by the measuring equipment, x max and x min Respectively represent the maximum and minimum values ​​in the record, x′ represents the normalized data, and its value range is [0, 1]. Through this method, the data value range can be limited, the data quality can be improved, and the accuracy of abnormal power consumption analysis can be improved;

[0056] Then, the standardized data is subjected to feature extraction. The present invention extracts six characteristic values ​​of the daily, weekly, and monthly periodic features, namely, the mean value, standard deviation, variance, range, skewness, and kurtosis. The formulas are:

[0057] The average value feature mean extracted by a specific period:

[0058]

[0059] The standard deviation feature std extracted by a specific period:

[0060]

[0061] The variance feature var extracted by a specific period:

[0062]

[0063] The range feature extracted by a specific period:

[0064] range=x max-t -x min-t

[0065] The skew feature extracted by a specific period:

[0066]

[0067] Kurtosis feature kurt extracted by specific period:

[0068]

[0069] In the above formula, the values ​​of i are 1, 7, and 30, representing the cycles of day, week, and month respectively. i Indicates the user's electricity consumption in a period of day, week, month, etc., n represents the total number of cycles of the user's electricity consumption data, x max-t and x min-t Indicates the maximum and minimum electricity consumption of the user in a daily, weekly, monthly, etc. period;

[0070] Combining the above characteristics, principal component analysis is used to reduce the dimensionality of the original electricity consumption data and remove useless and duplicate information, providing a better basis for subsequent abnormal electricity consumption analysis.

[0071] Days and months represent data for a certain period of time and a longer period of stability, respectively. Weeks lie between the two and can help observe situations within a certain period of time and over a long period of time. Through the six eigenvalues, multi-type and multi-faceted data can be provided, making the results more accurate when performing cluster analysis later.

[0072] Finally, the DPeaks clustering algorithm is used to perform cluster analysis on the electricity consumption data after dimensionality reduction. For each data point i, the DPeaks clustering algorithm calculates two quantities: the local density p of point i and the distance δ between point i and the nearest higher density point, which are defined as follows:

[0073]

[0074] Among them, m is the total number of data points, x(d ij -d c ) is the indicator function, when d ij -d c <0, X(d ij -d c )=1, otherwise X(d ij -d c )=0,d ij is the Euclidean distance between point i and point j, d c is the cutoff distance, which is used to control the size of the neighborhood. The distance between the points with the highest density is the maximum distance between any two points. j and ρ i Represent the local density of point i and point j respectively.

[0075] Based on the above formula, the local density ρ and distance δ of each point can be calculated, and the cluster center and core point can be obtained. Then, for each remaining point, it is divided into the cluster closest to the cluster center.

[0076] The calculation formulas for the benchmark density and benchmark distance of each cluster are as follows:

[0077]

[0078] in, and are the base density and base distance of the cth cluster respectively, α and β are empirical constants with values ​​of 0.4 and 3 respectively, N (c) is the total number of data points contained in the cth cluster, M (c) is the set of data points contained in the cth cluster. If the local density ρ of a data point i Less than the reference density ρ0 and the distance δ i If the distance is greater than the reference distance δ0, the point is an outlier, indicating that the power consumption data is abnormal. By analyzing power consumption anomalies, the stability of power grid operation can be improved and the risk of power grid operation can be reduced.

[0079] This method extracts six eigenvalues, which is less than the conventional method, thus improving the analysis accuracy. Furthermore, the clustering algorithm selected is DPeaks, which can identify data clusters of any shape and achieve global optimization, making it particularly suitable for power data analysis.

[0080] Weak link analysis module: First, edge betweenness is selected as the vulnerability indicator of the weak link of the power grid. For each edge of the power grid network, edge betweenness is defined as the ratio of the number of paths passing through the edge among all the shortest paths in the network to the number of shortest paths in the entire network. The formula is:

[0081]

[0082] Where E represents the edge betweenness, S j_ij is the number of shortest paths between nodes j_i and j_j, S j_ij (l) represents the number of shortest paths passing through line l, and N1 is the total number of lines.

[0083] The edge betweenness metric then calculates the vulnerability of each grid component, ranks the components by vulnerability, and selects the component with the highest vulnerability as the first key grid component. This component is then removed from the grid, and the grid topology is updated. The remaining grid components in the new topology are then calculated and ranked by vulnerability, with the component with the highest vulnerability selected as the second key grid component. This process is iterated K times to obtain a set of K grid components with the highest vulnerability, which are then identified as the weak links of the grid network. Weak link analysis shifts from a one-time analysis to an iterative, multiple-step analysis, improving accuracy. This weak link analysis helps personnel focus on these weak links in their work and improve grid stability.

[0084] The distribution network defect analysis module conducts correlation analysis based on existing distribution network defect reports, exploring the relationship between weak links and defects. For each defect, it extracts key words from the defect report and combines them with the status of weak network components to form a correlation record. This correlation record is then fed into the frequent pattern growth (FP-growth) algorithm for fault correlation analysis, generating association rules between weak links and faults in the form of "[Fault: XXXX] => [Weak Network Component: YYYY]." This defect analysis helps personnel quickly locate the weak link that may have caused the fault when a distribution network fault occurs, improving detection efficiency.

[0085] The above description is only a preferred specific embodiment of the present invention, but the scope of protection of the present invention is not limited thereto. Any technician familiar with the technical field, within the technical scope disclosed by the present invention, who makes equivalent replacements or changes based on the technical solution and inventive concept of the present invention, should be covered by the scope of protection of the present invention.

Claims

1. A multi-dimensional safety analysis system for power grid dispatching, characterized in that: It includes a control center and an abnormal power consumption analysis module, a weak link analysis module and a distribution network defect analysis module connected to the control center. The power consumption anomaly analysis module is used to analyze power consumption anomalies; The weak link analysis module is used to analyze the weak links of the power grid network; The distribution network defect analysis module is used to analyze distribution network defects.

2. A multi-dimensional safety analysis system for power grid dispatching according to claim 1, characterized in that: The power consumption anomaly analysis module includes: First, the user's electricity consumption data is standardized; Then, the standardized data is subjected to feature extraction and dimensionality reduction; Finally, a clustering algorithm is used to perform cluster analysis on the reduced-dimensional electricity consumption data to achieve abnormal analysis of electricity consumption.

3. A multi-dimensional safety analysis system for power grid dispatching according to claim 2, characterized in that: During the standardization process, the maximum and minimum normalization method is selected. After finding the maximum and minimum data in the overall electricity consumption data set, the value range of all data is made [0,1] through formula transformation. The specific formula is as follows: Among them, x represents the initial data recorded by the measuring equipment, x max and x min They represent the maximum and minimum values ​​in the record respectively, and x′ represents the standardized data, with a value range of [0,1].

4. A multi-dimensional safety analysis system for power grid dispatching according to claim 2, characterized in that: When extracting features, six eigenvalues ​​are extracted: mean, standard deviation, variance, range, skewness, and kurtosis according to specific periodic features. The formulas are: The average feature mean extracted by a specific period: The standard deviation feature std extracted by a specific period: The variance feature var extracted by a specific period: The range feature extracted by a specific period: range=x max-t -x min-t The skew feature extracted by a specific period: Kurtosis feature kurt extracted by specific period: In the above formula, the values ​​of i are 1, 7, and 30, representing the cycles of day, week, and month respectively. i Indicates the user's electricity consumption in a daily, weekly, or monthly cycle. n indicates the total number of cycles of the user's electricity consumption data. max-t and x min-t Indicates the maximum and minimum power consumption of the user in a daily, weekly, or monthly cycle; Combining the above characteristics, dimensionality reduction is performed through principal component analysis.

5. A multi-dimensional safety analysis system for power grid dispatching according to claim 4, characterized in that: The DPeaks clustering algorithm is used to perform cluster analysis on the electricity consumption data after dimensionality reduction. For each data i, the DPeaks clustering algorithm calculates two quantities: the local density ρ of point i and the distance δ between point i and the nearest higher density point, which are: Among them, m is the total number of data points, x(d ij -d c ) is the indicator function, when d ij -d c When X(d ij -d c )=1, otherwise X(d ij -d c )=0,d ij is the Euclidean distance between point i and point j, d c is the cutoff distance, ρ j and ρ i Represent the local density of point i and point j respectively; Based on the above formula, the local density ρ and distance δ of each point are calculated, and the cluster center and core point are obtained. Then, for each remaining point, it is divided into the cluster closest to the cluster center. The calculation formulas for the base density and base distance of each cluster are as follows: in, and are the reference density and reference distance of the cth cluster, α and β are empirical constants, and N (c) is the total number of data points contained in the cth cluster, M (c) is the set of data points contained in the cth cluster; if the local density ρ of a data point i Less than the reference density ρ0 and distance δ i If it is greater than the reference distance δ0, the point is an abnormal point, indicating that the electricity consumption data is abnormal.

6. A multi-dimensional safety analysis system for power grid dispatching according to claim 1, characterized in that: The weak link analysis module includes: First, the edge betweenness is selected as the vulnerability index of the weak link of the power grid, and its formula is: Among them, E represents the edge betweenness, S j_ij is the number of shortest paths between nodes j_i and j_j, S j_ij (l) represents the number of shortest paths passing through line l, and N1 is the total number of lines; Then, the edge betweenness index is used to calculate the vulnerability of each power grid component, and the power grid components are sorted by vulnerability, and the component with the highest vulnerability is selected as the first key power grid component; then the power grid component is removed from the power grid, the power grid topology is updated, the vulnerability of the remaining power grid components is calculated in the new power grid topology, and the power grid components are sorted by vulnerability, and the power grid component with the highest vulnerability is selected as the second key power grid component. The process is iterated K times to obtain a set of K power grid components with the highest vulnerability as the weak link of the power grid network.

7. A multi-dimensional safety analysis system for power grid dispatching according to claim 1, characterized in that: The distribution network defect analysis module includes: Combined with the existing distribution network defect reports, correlation analysis is performed to explore the correlation between weak links and defects. For each defect, the key words of the defect report are extracted, and combined with the status of weak power grid components in the distribution network, an association record is formed. This association record is used as input and input into the frequent pattern growth algorithm for fault correlation analysis to form association rules between weak links and faults.

Images