Intelligent power distribution network abnormity monitoring method and system

Abstract

The invention relates to the technical field of mode recognition, in particular to an intelligent power distribution network abnormity monitoring method and system, and the method comprises the following steps: collecting the current, voltage and temperature readings of transformer monitoring and power distribution line analysis based on the real-time data of a power distribution network, carrying out the time sequence segmentation, eliminating the noise, and generating a real-time data set of the power distribution network. According to the invention, through refining processing and efficient analysis of the real-time data of the intelligent power distribution network, the monitoring and exception handling capabilities of the power grid are significantly enhanced. The detailed time sequence processing and noise elimination ensure the quality and accuracy of a data set, and a solid foundation is provided for data analysis. And by adopting a gridding data structure, the processing of complex data is simplified, and the processing speed and the data management efficiency are improved. The abnormal mode is effectively identified through intensity analysis, the fault occurrence rate of the power grid is remarkably reduced, and the safety and the operation efficiency are improved. And through combination of abnormal clustering analysis and risk assessment, the abnormal state of the power grid is accurately predicted and identified.

Description

Technical Field

[0001] The present invention relates to the field of pattern recognition technology, and in particular to a method and system for monitoring abnormalities in a smart distribution network. Background Art

[0002] Pattern recognition is a computer science technology that involves the technical field of automatically identifying regularities and features from data. This technology is often used in many fields such as image processing, speech recognition, and text analysis. In the pattern recognition process, it usually includes four steps: data preprocessing, feature extraction, model training, and pattern recognition. By using various algorithms, such as machine learning algorithms, neural networks, statistical methods, etc., pattern recognition can effectively identify meaningful patterns from large amounts of data to provide support for decision-making.

[0003] Among them, the abnormal monitoring method of the intelligent distribution network refers to the use of pattern recognition technology to monitor and analyze the operating data in the distribution network in order to detect and handle abnormal situations in a timely manner. The purpose of this method is mainly to improve the reliability and safety of the power grid, optimize power grid management, reduce the occurrence of failures and accidents, and thus ensure the stability and efficiency of power supply. By analyzing the operating data of the power grid, the abnormal monitoring system can predict and identify potential fault points and take measures in advance to avoid large-scale power outages.

[0004] Existing technologies are deficient in real-time processing and abnormal response of high-dimensional and complex data. These technologies usually lack effective data structure simplification and efficient abnormality identification mechanisms, resulting in slow processing speed, untimely response, and increased risks in grid operation. For example, failure to quickly identify and respond to anomalies may cause faults to affect a wider area and threaten grid stability. In addition, the deficiencies of traditional methods in preprocessing and feature extraction may lead to insufficient sensitivity and accuracy of the monitoring system, making it impossible to effectively predict and prevent potential large-scale power outages, further affecting the reliability and economic benefits of the grid. Summary of the invention

[0005] The purpose of the present invention is to solve the shortcomings of the prior art and to propose a method and system for abnormal monitoring of a smart distribution network.

[0006] In order to achieve the above object, the present invention adopts the following technical solution: a method for monitoring abnormality of a smart distribution network, comprising the following steps:

[0007] S1: Based on the real-time data of the distribution network, the current, voltage, and temperature readings of transformer monitoring and distribution line analysis are collected, and time series are segmented to eliminate noise and generate a real-time data set of the distribution network;

[0008] S2: Based on the real-time data set of the distribution network, a grid-based simplified data structure is adopted to partition and classify the current and voltage data of each grid point, record the current and voltage measurement values ​​of the grid point, perform data density analysis in the local area, and generate grid density information;

[0009] S3: extracting core points based on the grid density information, constructing clusters by expanding the core points outward, processing cluster boundaries, checking the standard deviation and mean of each cluster, marking abnormal clusters, and generating clusters and abnormal cluster markers;

[0010] S4: Based on the clustering and abnormal cluster marking, combined with the current operation data of the power grid components, the abnormal probability is calculated, compared with the risk threshold adjusted in real time, the current risk level is evaluated, and the abnormal state prediction and threshold judgment are generated.

[0011] As a further solution of the present invention, the steps of acquiring the real-time data set of the distribution network are specifically as follows:

[0012] S111: collect real-time monitoring data from the distribution network, including current, voltage, and temperature readings of transformers and distribution lines, and generate a preliminary monitoring data set;

[0013] S112: Based on the preliminary monitoring data set, perform time series segmentation, apply a moving average method to each data point in the time series for denoising, and obtain processed time series data;

[0014] S113: Using the processed time series data, apply the variance calculation formula:

[0015]

[0016] Quantify the discrete degree of data points to obtain real-time monitoring data set of distribution network;

[0017] Among them, x i Represents multiple data point values, μ represents the data average, and n represents the total number of data.

[0018] As a further solution of the present invention, the step of obtaining the grid density information is specifically as follows:

[0019] S211: Based on the real-time data set of the distribution network, a gridding strategy is implemented to simplify the data structure, define the boundary and size of each grid, and generate a gridding data structure;

[0020] S212: Utilizing the grid data structure, extracting current and voltage information of each grid point, recording the measured values, and forming a grid point current and voltage record table;

[0021] S213: Execute data density calculation according to the grid point current and voltage record table, using the formula:

[0022]

[0023] Generate gridded density information;

[0024] Among them, D grid represents the data density of each grid, i represents the current value, v represents the voltage value, N represents the total number of grid points, and w i Represents the weight coefficient of the grid.

[0025] As a further solution of the present invention, the steps of obtaining clustering and abnormal cluster labels are specifically as follows:

[0026] S311: Based on the grid density information, select a grid point with the highest data density, identify the grid point as a core point, and form a core point set;

[0027] S312: using the core point set, constructing clusters by connecting adjacent grid points of similar data density, ensuring that the grid points around each core point meet the density threshold, and generating a preliminary clustering area;

[0028] S313: Based on the preliminary clustering area, the cluster boundary is adjusted, and the boundary grid points are reallocated to define the cluster boundary by comparing the data density of the adjacent grids, thereby completing the processing of the cluster boundary;

[0029] S314: Calculate the standard deviation and mean of the data points within each cluster according to the adjusted cluster boundary data, using the formula:

[0030]

[0031] Mark clusters as abnormal clusters, generate clusters and abnormal cluster labels;

[0032] Among them, SD represents the dispersion of data points within the cluster, which is used to measure the deviation of data points from the mean. The larger the standard deviation, the more dispersed the distribution of data points. ∑ represents the summation operation of all data points in the cluster. N represents the total number of data points in the target cluster. x i represents the actual observed value of the i-th data point in the cluster, and μ represents the average value of all data points in the cluster.

[0033] As a further solution of the present invention, the steps of obtaining abnormal state prediction and threshold judgment are specifically as follows:

[0034] S411: Based on the clustering and abnormal cluster marking data, combined with the operation data of the current power grid components, the abnormal probability of each cluster is calculated to obtain a cluster abnormal probability result;

[0035] S412: Compare the clustering abnormal probability result with the set real-time adjustment risk threshold, determine whether the abnormal probability exceeds the threshold through a logical comparison operation, and generate a risk comparison result;

[0036] S413: Based on the risk comparison result, a quantitative method is used to evaluate the risk level of the current power grid, a mathematical model is applied to calculate the data deviation and abnormal frequency within the cluster, the risk level is evaluated, and a risk level result is generated;

[0037] S414: Combine the risk level result and the cluster abnormal probability result data to predict the abnormal state and perform threshold judgment, using the formula:

[0038]

[0039] Generate abnormal state prediction and threshold judgment;

[0040] Among them, P risk represents the predicted probability of risk occurrence, x represents the abnormality probability obtained from data analysis, θ is the risk threshold, and k is a parameter adjusted according to the system response sensitivity.

[0041] A smart distribution network abnormality monitoring system, the smart distribution network abnormality monitoring system is used to execute the above-mentioned smart distribution network abnormality monitoring method, the system comprises:

[0042] The data acquisition module receives real-time data from transformer monitoring and distribution line analysis, including current, voltage and temperature readings, and performs time series segmentation and noise elimination on the data to obtain a real-time data set;

[0043] The data gridding module classifies and organizes the current and voltage data in a gridded data structure based on the real-time data set, records the measured values ​​of multiple grid points, analyzes the data density in each area, and generates gridded density information;

[0044] The cluster analysis module uses the grid density information to select core points and expand outward to form clusters, process cluster boundaries, analyze the standard deviation and mean of each cluster, mark abnormal clusters in the data pattern, and form clusters and abnormal cluster marks;

[0045] The risk assessment module uses the clustering and abnormal cluster markings, combined with the current operation data of the distribution network, to calculate the probability value of the abnormal cluster, compares it with the set risk threshold, evaluates the current risk level of the system, and generates a risk level assessment result;

[0046] The state prediction module extracts the statistical features of each abnormal cluster from the risk level assessment result, calculates the deviation value from the standard threshold data, performs abnormal state prediction and threshold judgment, and obtains abnormal state prediction results.

[0047] Compared with the prior art, the advantages and positive effects of the present invention are:

[0048] In the present invention, the monitoring and exception handling capabilities of the power grid are significantly enhanced through the refined processing and efficient analysis of the real-time data of the smart distribution network. Meticulous time series processing and noise elimination ensure the quality and accuracy of the data set, providing a solid foundation for data analysis. The use of a grid data structure simplifies the processing of complex data, improves the processing speed and data management efficiency. Through density analysis, abnormal patterns are effectively identified, the occurrence rate of power grid failures is significantly reduced, and safety and operating efficiency are improved. The combination of abnormal clustering analysis and risk assessment can accurately predict and identify abnormal power grid conditions, optimize response strategies, ensure stable power supply, and improve the overall operational quality of the power grid. BRIEF DESCRIPTION OF THE DRAWINGS

[0049] Figure 1 It is a schematic diagram of the workflow of the present invention;

[0050] Figure 2 A flow chart of the steps for acquiring a real-time data set of a power distribution network according to the present invention;

[0051] Figure 3 A flowchart of the steps for obtaining grid density information of the present invention;

[0052] Figure 4 A flow chart of the steps for obtaining clustering and abnormal cluster marking of the present invention;

[0053] Figure 5 The figure is a flow chart of the acquisition steps of abnormal state prediction and threshold judgment of the present invention. DETAILED DESCRIPTION

[0054] In order to make the purpose, technical solution and advantages of the present invention more clearly understood, the present invention is further described in detail below in conjunction with the accompanying drawings and embodiments. It should be understood that the specific embodiments described herein are only used to explain the present invention and are not intended to limit the present invention.

[0055] In the description of the present invention, it should be understood that the terms "length", "width", "up", "down", "front", "back", "left", "right", "vertical", "horizontal", "top", "bottom", "inside", "outside" and the like indicate positions or positional relationships based on the positions or positional relationships shown in the drawings, and are only for the convenience of describing the present invention and simplifying the description, rather than indicating or implying that the device or element referred to must have a specific orientation, be constructed and operated in a specific orientation, and therefore cannot be understood as limiting the present invention. In addition, in the description of the present invention, "multiple" means two or more, unless otherwise clearly and specifically defined.

[0056] Embodiment 1

[0057] See also Figure 1 The present invention provides a technical solution: a method for monitoring abnormalities in a smart distribution network, comprising the following steps:

[0058] S1: Based on the real-time data of the distribution network, the current, voltage, and temperature readings of transformer monitoring and distribution line analysis are collected, and time series are segmented to eliminate noise and generate a real-time data set of the distribution network;

[0059] S2: Based on the real-time data set of the distribution network, a grid-based simplified data structure is used to partition and classify the current and voltage data of each grid point, record the current and voltage measurement values ​​of the grid point, perform data density analysis in the local area, and generate grid density information;

[0060] S3: Extract core points based on grid density information, build clusters by expanding the core points outward, process cluster boundaries, check the standard deviation and mean of each cluster, mark abnormal clusters, and generate clusters and abnormal cluster labels;

[0061] S4: Based on clustering and abnormal cluster marking, combined with the current operation data of power grid components, the abnormal probability is calculated, compared with the risk threshold adjusted in real time, the current risk level is evaluated, and abnormal state prediction and threshold judgment are generated.

[0062] The real-time data set of the distribution network includes transformer monitoring data, current data, voltage data, and temperature data. The grid density information includes the current and voltage measurement values ​​of each grid point and the data density of the local area. The clustering and abnormal cluster markings include the cluster standard deviation, cluster mean, and abnormal cluster identification. The abnormal state prediction and threshold judgment include abnormal probability and risk level assessment.

[0063] See also Figure 2 ,The specific steps for obtaining the real-time data set of the distribution network are:

[0064] S111: collect real-time monitoring data from the distribution network, including current, voltage, and temperature readings of transformers and distribution lines, and generate a preliminary monitoring data set;

[0065] Among them, current, voltage and temperature readings are collected from the distribution network, according to the formula Calculate the grid status index R.

[0066] In the formula, I represents current, U represents voltage, and T represents temperature.

[0067] Detailed explanation of the formula and the process of formula calculation and derivation:

[0068] Calculating the grid status index R involves converting the current I, voltage U, and temperature T into a single quantitative index. The formula uses the square root of the sum of squares to synthesize the parameters and provide an overall grid status assessment. Set the current to 50A, the voltage to 220V, and the temperature to 25°C within the target time. According to the formula, first calculate the sum of the squares of the multiple parameters:

[0069] I 2 =2500,U 2 =48400,T 2 =625

[0070] Then add the values ​​together:

[0071] I 2 +U 2 +T 2 =2500+48400+625=51525

[0072] Finally, take the square root to get R:

[0073]

[0074] The results show that the grid status index R is 227.0, which reflects the grid status at the current monitoring point. The stability and security of the grid can be judged by the size of R.

[0075] S112: Based on the preliminary monitoring data set, perform time series segmentation, apply a moving average method to each data point in the time series for denoising, and obtain processed time series data;

[0076] Among them, the moving average method is applied to each data point in the time series for denoising, according to the formula Calculate the moving average MA k .

[0077] In the formula, x j represents a data point in the time series, k represents the size of the moving window, and i is the index of the current data point.

[0078] Detailed explanation of the formula and the process of formula calculation and derivation:

[0079] The moving average method is used to smooth time series data and reduce the impact of noise. In the formula, MA k This represents the moving average calculated under the condition that the window size is k. Set the selected window size to 3 and have data points: 10, 20, 30, 40, 50. First, select the first sequence in the window:

[0080] x1=10, x2=20, x3=30

[0081] Calculate the moving average according to the formula:

[0082]

[0083] As the window moves, calculate the mean of the next series:

[0084] x2=20, x3=30, x4=40

[0085]

[0086] Until the end of the data point, the average of the data points in the window is calculated each time, which helps smooth the time series and provide a more stable basis for data analysis.

[0087] S113: Using the processed time series data, apply the variance calculation formula:

[0088]

[0089] Quantify the discrete degree of data points to obtain real-time monitoring data set of distribution network;

[0090] Among them, x i Represents multiple data point values, μ represents the data average, and n represents the total number of data.

[0091] formula:

[0092]

[0093] Detailed explanation of the formula and the process of formula calculation and derivation:

[0094] The formula is used to calculate the variance of a data point, where σ 2 represents the variance, x i is the value of the data point, μ is the mean of the data points, and n is the total number of data points. The formula reflects the degree to which the data point deviates from the mean.

[0095] There are five data points: 10, 20, 30, 40, 50. First, calculate the mean μ of the data points:

[0096]

[0097] Then, we calculate the square of the difference between each data point and the mean:

[0098] (x1-μ) 2 =(10-30) 2 =400

[0099] (x2-μ) 2 =(20-30) 2 =100

[0100] (x3-μ) 2 =(30-30) 2 =0

[0101] (x4-μ) 2 =(40-30) 2 =100

[0102] (x5-μ) 2 =(50-30) 2 =400

[0103] Add the values:

[0104]

[0105] Finally, we calculate the variance by dividing the sum by the number of data points n:

[0106]

[0107] The results show that the variance is 200, which reflects the dispersion of the data points relative to their mean. A larger variance indicates more dispersed data points.

[0108] See also Figure 3 , the specific steps for obtaining grid density information are:

[0109] S211: Based on the real-time data set of the distribution network, a gridding strategy is implemented to simplify the data structure, define the boundary and size of each grid, and generate a gridded data structure;

[0110] In this step, a gridding strategy is implemented based on the real-time data set of the distribution network to simplify the data structure and define the boundaries and size of each grid. In order to quantify the location and boundaries of each grid, the following reference formula is used: Calculate the center position of each grid cell.

[0111] Where P represents the center position of the grid, x k represents the position coordinates of the points in the grid, K represents the total number of points in the grid,

[0112] Detailed explanation of the formula and the process of formula calculation and derivation:

[0113] In order to locate the center of each grid, it is necessary to calculate the average position coordinates of all points in the grid. Let's set an example where the grid includes three points with position coordinates x1=1, x2=2, and x3=3. Applying the above formula, the calculation process is as follows:

[0114]

[0115] Indicates that the center of the grid is at coordinate 2.

[0116] S212: using the grid data structure, extracting the current and voltage information of each grid point, recording the measured values, and forming a grid point current and voltage record table;

[0117] In this step, the current and voltage information of each grid point is extracted from the defined grid data structure. In order to calculate the average current and voltage of multiple grid points, the following reference formula is used: Among them, I i and V i Represent the current and voltage values ​​of each grid point respectively, and N represents the number of grid points.

[0118] Detailed explanation of the formula and the process of formula calculation and derivation:

[0119] The formula is used to evaluate the average value of current and voltage at the grid point and judge the uniformity of power distribution. The current and voltage values ​​of the three grid points are set to I1 = 10A, V1 = 220V, I2 = 12A, V2 = 230V, I3 = 11A, V3 = 225B. The calculation process of the application formula is as follows:

[0120]

[0121] The results show that the average current and voltage values ​​at these three grid points are 314, 33, providing an assessment of the overall power distribution.

[0122] S213: Execute data density calculation according to the grid point current and voltage record table, using the formula:

[0123]

[0124] Generate gridded density information;

[0125] Among them, D grid represents the data density of each grid, i represents the current value, v represents the voltage value, N represents the total number of grid points, and w i Represents the weight coefficient of the grid.

[0126] formula:

[0127]

[0128] The formula is to square the sum of the current and voltage values ​​of each grid point, multiply it by the corresponding weight coefficient, and finally find the average to obtain a grid data density index that takes into account the importance of location. Assume that there are three grid points, and the current and voltage values ​​and weight coefficients are i1 = 10A, v1 = 220V, w1 = 0.5, i2 = 12A, v2 = 230V, w2 = 1.0, i3 = 11A, v3 = 225V, w3 = 0.75. The calculation process is as follows:

[0129] Calculate the weighted current-voltage index for each grid:

[0130] For grid 1: w1·((10 2 +220 2 ) 0.5 )=0.5·(48400) 0.5 =0.5·220=110

[0131] For grid 2: w2·((12 2 +230 2 ) 0.5 )=1.0·(52944) 0.5 =1.0·230=230

[0132] For grid 3: w3·((11 2 +225 2 ) 0.5 )=0.75·(50626) 0.5 =0.75.225=168.75

[0133] Calculate the average:

[0134]

[0135] The results show that when referring to the geographic or functional importance of the grid, the average weighted current-voltage index for the entire region is 169.58, reflecting the data density of the grid area.

[0136] See also Figure 4 , the specific steps for obtaining clustering and abnormal cluster labels are:

[0137] S311: Based on the grid density information, select the grid points with the highest data density, identify the grid points as core points, and form a core point set;

[0138] Based on the grid density information, the grid points with the highest data density are selected and marked as core points to generate a core point set. The data density is measured according to the formula Calculate the data density D at each grid point.

[0139] In the formula, D represents the data density, N represents the total number of grid points, and d i Represents the data value of the i-th grid point.

[0140] Detailed explanation of the formula and the process of formula calculation and derivation:

[0141] The data density D is calculated by i The average value reflects the average concentration of data in the grid. For example, if a grid area has 5 grid points and the data values ​​are 10, 20, 30, 40, and 50 respectively, the data density is calculated as:

[0142]

[0143] The results show that the average data density in the grid area is 30, which reflects the average distribution of data in the area.

[0144] S312: using the core point set, constructing clusters by connecting adjacent grid points of similar data density, ensuring that the grid points around each core point meet the density threshold, and generating a preliminary clustering area;

[0145] Starting from the core point set, clusters are gradually constructed by connecting adjacent grid points with similar data density. The data fit measurement during the connection process is calculated according to the formula Calculate the goodness of fit S of the data at two grid points.

[0146] In the formula, S represents the data fit, d i and d j Represents the data values ​​of the two grid points being compared, and max(D) is the maximum value of all grid point data.

[0147] Detailed explanation of the formula and the process of formula calculation and derivation:

[0148] The data fit S is calculated as the complement of the ratio of the absolute value of the difference between the data values ​​of two grid points to the maximum value of all data. For example, the data values ​​of the two grid points are 40 and 50 respectively, and the maximum value of the entire data set is 100. The data fit is calculated as:

[0149]

[0150] The results show that the two grid points fit the data values ​​with a degree of fit of 90%, indicating that they belong to the same cluster when the clusters are constructed.

[0151] S313: Based on the preliminary clustering area, the cluster boundary is adjusted, and the boundary grid points are reallocated to define the cluster boundary by comparing the data density of the adjacent grids, thereby completing the processing of the cluster boundary;

[0152] Based on the preliminary clustering, the cluster boundaries are adjusted by comparing the data density of adjacent grids and redistributing the boundary grid points to define the cluster boundaries. The basis for boundary adjustment is as follows: Calculate the average data density C between adjacent grid points.

[0153] In the formula, C represents the average data density at the boundary, d i and d j Represents the data values ​​of adjacent grid points.

[0154] Detailed explanation of the formula and the process of formula calculation and derivation:

[0155] The average data density C at the boundary is used to determine the adjustment of the cluster boundary. For example, if the data value of a grid point is 45 and the data value of the adjacent grid point is 55, the average data density at the boundary is calculated as:

[0156]

[0157] The results show that the average data density of these two grid points on the boundary is 50, which serves as the basis for cluster boundary adjustment.

[0158] S314: Calculate the standard deviation and mean of the data points within each cluster based on the adjusted cluster boundary data, using the formula:

[0159]

[0160] Mark clusters as abnormal clusters, generate clusters and abnormal cluster labels;

[0161] Among them, SD represents the dispersion of data points within the cluster, which is used to measure the deviation of data points from the mean. The larger the standard deviation, the more dispersed the distribution of data points. ∑ represents the summation operation of all data points in the cluster. N represents the total number of data points in the target cluster. x i represents the actual observed value of the i-th data point in the cluster, and μ represents the average value of all data points in the cluster.

[0162] formula:

[0163]

[0164] Detailed explanation of the formula and the process of formula calculation and derivation:

[0165] The standard deviation formula is used to calculate the distribution dispersion of data points within a cluster. For example, a cluster is set to include data points 10, 20, 30, 40, and 50. First, calculate the mean:

[0166]

[0167] The square of the deviation of each data point from the mean is then calculated, summed and divided by the number of data points to get the variance, and finally the square root is taken to get the standard deviation:

[0168]

[0169]

[0170] The results show that the distribution dispersion of data points in the cluster around the mean is 14.14. If the value is significantly higher than the preset threshold, the cluster is marked as an abnormal cluster.

[0171] See also Figure 5 ,The specific steps for obtaining abnormal state prediction and threshold judgment are:

[0172] S411: Based on the clustering and abnormal cluster marking data, combined with the current operation data of the power grid components, the abnormal probability of each cluster is calculated to obtain the cluster abnormal probability result;

[0173] Among them, clustering and abnormal cluster labeling data are used for initial abnormal state evaluation, according to the formula Calculate the probability of abnormal state, where o data Represents the operation data, W data represents the weight parameter and N represents the total number of data points.

[0174] Detailed explanation of the formula and the process of formula calculation and derivation:

[0175] Formula P eval Used to calculate the weighted average based on operational data, expressed as anomaly probability. data Taken from the actual operating data of the power grid components, W data According to the influence or importance of multiple data, N is the total number of data points involved in the calculation. For example, there are 5 data points, and the operation data of each point is 20, 30, 40, 50, 60, and the corresponding weights are 1, 2, 1, 3, 2, then the abnormal state probability is calculated as follows:

[0176]

[0177] The results show that based on the current operating data and the assigned weights, the calculated average abnormal probability is 78, reflecting that the current risk level of the power grid is high under the given weights and data conditions.

[0178] S412: Compare the clustering abnormal probability result with the set real-time adjustment risk threshold, determine whether the abnormal probability exceeds the threshold through a logical comparison operation, and generate a risk comparison result;

[0179] Among them, the calculated abnormal state probability is compared with the risk threshold adjusted in real time, according to the formula Calculate the risk ratio, where P eval represents the probability of abnormal state, T risk Represents the risk threshold.

[0180] Detailed explanation of the formula and the process of formula calculation and derivation:

[0181] Formula C risk It is used to evaluate the ratio of the probability of abnormal state to the set risk threshold, and to assist in judging whether the risk level exceeds the acceptable level. eval is the abnormal probability, and T risk It is a risk threshold set based on the system safety standard. For example, P eval is 78, and the risk threshold T risk If is set to 100, the risk ratio is calculated as follows:

[0182]

[0183] The results show that the current abnormal state probability accounts for 78% of the risk threshold and has not yet reached the preset risk threshold, indicating that the current operating state of the power grid system is within an acceptable risk range.

[0184] S413: Based on the risk comparison results, a quantitative method is used to evaluate the risk level of the current power grid, a mathematical model is applied to calculate the data deviation and abnormal frequency within the cluster, the risk level is evaluated, and a risk level result is generated;

[0185] Among them, the relationship between the comparative analysis of risk ratio and the real-time adjustment of risk threshold is calculated according to the formula Calculate the current risk level, where C risk represents the risk ratio, S base Represents the baseline safety threshold.

[0186] Detailed explanation of the formula and the process of formula calculation and derivation:

[0187] Formula R level Used to convert the risk ratio C risk The system's baseline safety threshold S base Compare and evaluate the current risk level. risk The calculated result is 0.78, while the baseline safety threshold S base is 0.8. The risk level is calculated as follows:

[0188]

[0189] The results show that the current risk ratio accounts for 97.5% of the benchmark safety threshold, indicating that the risk level of the current power grid system is at a high level and close to the safety limit set by the system.

[0190] S414: Combine the risk level results and cluster abnormal probability result data to predict the abnormal state and perform threshold judgment, using the formula:

[0191]

[0192] Generate abnormal state prediction and threshold judgment;

[0193] Among them, P risk represents the predicted probability of risk occurrence, x represents the abnormality probability obtained from data analysis, θ is the risk threshold, and k is a parameter adjusted according to the system response sensitivity.

[0194] formula:

[0195]

[0196] Parameter settings and sources:

[0197] Abnormal probability x: The average abnormal ratio obtained by clustering the operation data of the power grid components in the abnormal cluster. For example, the average abnormal probability calculated from the data is 0.2.

[0198] Risk threshold θ: The threshold set according to the grid security standard is set to 0.5, which means that when the probability of abnormality exceeds 50%, the system will be marked as high risk.

[0199] Adjustment factor k: The factor k adjusts the model's sensitivity to threshold changes and is set based on the system's response speed and past risk events. A setting of 1 means normal sensitivity.

[0200] Calculation process:

[0201] Set x = 0.2, θ = 0.5, k = 1,

[0202] Calculate x-θ=0.2-0.5=-0.3,

[0203] e -k(x-θ) =e -1×-0.3 ≈e 0.3 ≈1.35

[0204]

[0205] Explanation of numerical results:

[0206] The obtained risk probability P riskIt is about 0.426, indicating that the abnormal risk probability of the current power grid system is 42.6%. The result is less than the risk threshold, indicating that the system is currently in a relatively safe state and does not exceed the set 50% high risk threshold.

[0207] A smart distribution network abnormality monitoring system, which is used to execute the above-mentioned smart distribution network abnormality monitoring method, and the system includes:

[0208] The data acquisition module receives real-time data from transformer monitoring and distribution line analysis, including current, voltage and temperature readings, and performs time series segmentation and noise elimination on the data to obtain a real-time data set;

[0209] The data gridding module classifies and organizes the current and voltage data in a gridded data structure based on real-time data sets, records the measured values ​​of multiple grid points, analyzes the data density in each area, and generates gridded density information;

[0210] The cluster analysis module uses grid density information to select core points and expand them outward to form clusters, process cluster boundaries, analyze the standard deviation and mean of each cluster, mark abnormal clusters in the data pattern, and form clusters and abnormal cluster markers;

[0211] The risk assessment module uses clustering and abnormal cluster marking, combined with the current operation data of the distribution network, to calculate the probability value of abnormal clusters, compare it with the set risk threshold, evaluate the current risk level of the system, and generate risk level assessment results;

[0212] The state prediction module extracts the statistical features of each abnormal cluster from the risk level assessment results, calculates the deviation value from the standard threshold data, predicts the abnormal state and makes threshold judgments, and obtains the abnormal state prediction results.

[0213] The above are only preferred embodiments of the present invention and are not intended to limit the present invention in other forms. Any technician familiar with the profession may use the technical contents disclosed above to change or modify them into equivalent embodiments with equivalent changes and apply them to other fields. However, any simple modification, equivalent change and modification made to the above embodiments based on the technical essence of the present invention without departing from the technical solution of the present invention still falls within the protection scope of the technical solution of the present invention.

Claims

1. A method for monitoring abnormality in a smart distribution network, characterized in that: The following steps are involved: Based on the real-time data of the distribution network, the current, voltage, and temperature readings of transformer monitoring and distribution line analysis are collected, and the time series are segmented to eliminate noise and generate a real-time data set of the distribution network; Based on the real-time data set of the distribution network, a grid-based simplified data structure is adopted to partition and classify the current and voltage data of each grid point, record the current and voltage measurement values ​​of the grid point, perform data density analysis in the local area, and generate grid density information; Extracting core points based on the grid density information, constructing clusters by expanding the core points outward, processing cluster boundaries, checking the standard deviation and mean of each cluster, marking abnormal clusters, and generating clusters and abnormal cluster markers; Based on the clustering and abnormal cluster marking, combined with the current operation data of the power grid components, the abnormal probability is calculated, compared with the risk threshold adjusted in real time, the current risk level is evaluated, and the abnormal state prediction and threshold judgment are generated.

2. The method for monitoring abnormality of a smart distribution network according to claim 1, characterized in that: The steps for acquiring the real-time data set of the distribution network are specifically as follows: Collect real-time monitoring data from the distribution network, including current, voltage, and temperature readings of transformers and distribution lines, to generate a preliminary monitoring data set; Based on the preliminary monitoring data set, time series segmentation is implemented, and each data point in the time series is subjected to denoising processing by applying a moving average method to obtain processed time series data; Using the processed time series data, the variance calculation formula is applied: Quantify the discrete degree of data points to obtain real-time monitoring data set of distribution network; Among them, x i Represents multiple data point values, μ represents the data average, and n represents the total number of data.

3. The method for monitoring abnormality of a smart distribution network according to claim 2, characterized in that: The steps for obtaining the grid density information are specifically as follows: Based on the real-time data set of the distribution network, a gridding strategy is implemented to simplify the data structure, define the boundary and size of each grid, and generate a gridding data structure; Utilizing the grid data structure, extracting current and voltage information of each grid point, recording the measured values, and forming a grid point current and voltage record table; According to the grid point current and voltage record table, data density calculation is performed using the formula: Generate gridded density information; Among them, D grid represents the data density of each grid, i represents the current value, v represents the voltage value, N represents the total number of grid points, and w i Represents the weight coefficient of the grid.

4. The method for monitoring abnormality of a smart distribution network according to claim 3, characterized in that: The steps of obtaining the clustering and abnormal cluster markings are specifically as follows: Based on the grid density information, select a grid point with the highest data density, identify the grid point as a core point, and form a core point set; Using the core point set, clusters are constructed by connecting adjacent grid points of similar data density, ensuring that the grid points around each core point meet the density threshold, and generating preliminary clustering areas; Based on the preliminary clustering area, the clustering boundary is adjusted, and the boundary grid points are reallocated to define the clustering boundary by comparing the data density of the adjacent grids, thereby completing the processing of the clustering boundary; Based on the adjusted cluster boundary data, the standard deviation and mean of the data points within each cluster are calculated using the formula: Mark clusters as abnormal clusters, generate clusters and abnormal cluster labels; Among them, SD represents the dispersion of data points within the cluster, which is used to measure the deviation of data points from the mean. The larger the standard deviation, the more dispersed the distribution of data points. ∑ represents the summation operation of all data points in the cluster. N represents the total number of data points in the target cluster. x i represents the actual observed value of the i-th data point in the cluster, and μ represents the average value of all data points in the cluster.

5. The method for monitoring abnormality of a smart distribution network according to claim 4, characterized in that: The steps of obtaining the abnormal state prediction and threshold judgment are specifically as follows: Based on the clustering and abnormal cluster labeling data, combined with the current operation data of the power grid components, the abnormal probability of each cluster is calculated to obtain a cluster abnormal probability result; Compare the clustering abnormal probability result with the set real-time adjustment risk threshold, determine whether the abnormal probability exceeds the threshold through a logical comparison operation, and generate a risk comparison result; Based on the risk comparison results, a quantitative method is used to evaluate the risk level of the current power grid, a mathematical model is applied to calculate the data deviation and abnormal frequency within the cluster, the risk level is evaluated, and a risk level result is generated; Combine the risk level result and the cluster abnormal probability result data to predict the abnormal state and perform threshold judgment, using the formula: Generate abnormal state prediction and threshold judgment; Among them, P risk represents the predicted probability of risk occurrence, x represents the abnormality probability obtained from data analysis, γ is the risk threshold, and k is a parameter adjusted according to the system response sensitivity.

6. A smart distribution network abnormality monitoring system, characterized in that: According to the method for monitoring abnormality of a smart distribution network according to any one of claims 1 to 5, the system comprises: The data acquisition module receives real-time data from transformer monitoring and distribution line analysis, including current, voltage and temperature readings, and performs time series segmentation and noise elimination on the data to obtain a real-time data set; The data gridding module classifies and organizes the current and voltage data in a gridded data structure based on the real-time data set, records the measured values ​​of multiple grid points, analyzes the data density in each area, and generates gridded density information; The cluster analysis module uses the grid density information to select core points and expand outward to form clusters, process cluster boundaries, analyze the standard deviation and mean of each cluster, mark abnormal clusters in the data pattern, and form clusters and abnormal cluster marks; The risk assessment module uses the clustering and abnormal cluster markings, combined with the current operation data of the distribution network, to calculate the probability value of the abnormal cluster, compares it with the set risk threshold, evaluates the current risk level of the system, and generates a risk level assessment result; The state prediction module extracts the statistical features of each abnormal cluster from the risk level assessment result, calculates the deviation value from the standard threshold data, performs abnormal state prediction and threshold judgment, and obtains abnormal state prediction results.

Images