An intelligent optimization method for the path of live working on distribution networks

Abstract

The present invention relates to the technical field of data processing, and particularly relates to an intelligent optimization method for the live working path of a distribution network; obtaining similar local extreme points according to a data sequence; obtaining the extreme value credibility according to the data difference characteristics between the similar local extreme points and adjacent data points; obtaining similar credible extreme points according to the extreme value credibility; obtaining marked extreme points according to the change characteristics of the similar credible extreme points; obtaining a target data segment and target extreme points according to the distribution characteristics of the marked extreme points; obtaining the weight coefficient of the target extreme points and the final optimized extreme value according to the target data segment. The present invention obtains abnormal data through the Isolation Forest algorithm according to the optimized extreme value, improves the accuracy of obtaining abnormal data, obtains the degree of abnormality according to the abnormal data and optimizes the live working path, thereby improving the optimization effect of the live working path and the stability of the power grid operation.

Description

Technical Field

[0001] The present invention relates to the technical field of data processing, and particularly relates to an intelligent optimization method for the live working path of a distribution network. Background Art

[0002] In the operation of a power system, there may be a situation where multiple grid nodes are abnormal simultaneously. In order to repair the grid nodes quickly and efficiently, it is necessary to plan the live working paths of multiple grid nodes. While ensuring that the overall working path is relatively short during the maintenance process, grid nodes with a greater degree of abnormality can be repaired preferentially to ensure the stable operation of the power grid; therefore, the optimization of the live working path is particularly important.

[0003] During the optimization process of the live working path of a distribution network, the degree of abnormality of grid nodes plays a decisive role in the path optimization result, and the accuracy of the analysis of the degree of abnormality affects the accuracy of the path optimization. The Isolation Forest algorithm is an existing method for detecting abnormal data and can detect abnormal data in a dataset; thus, the Isolation Forest can be used to detect abnormal data in the time-series data for monitoring the operating state of grid nodes, so as to judge the degree of abnormality of grid nodes. In the Isolation Forest algorithm, each isolation tree is constructed by randomly selecting features and randomly selecting a splitting value between the maximum and minimum values of the corresponding features; however, the time-series data for monitoring the operating state of grid nodes has strong local characteristics. The traditional Isolation Forest algorithm is not sensitive enough to the change of local density when selecting the splitting value between the maximum and minimum values, making it difficult for the Isolation Forest to capture local characteristics, resulting in the mixing of abnormal samples and normal samples, making it difficult to detect abnormal data, and leading to low accuracy in analyzing the degree of abnormality of grid nodes, resulting in poor optimization effect of the live working path. Summary of the Invention

[0004] In order to solve the above technical problem that the low accuracy of analyzing the degree of abnormality of grid nodes by the Isolation Forest algorithm leads to a poor optimization effect of the live working path, the purpose of the present invention is to provide an intelligent optimization method for the live working path of a distribution network, and the specific technical solution adopted is as follows:

[0005] Obtain a data sequence of influencing factors for optimizing the working path of grid nodes, and the path length of grid nodes; obtain similar local extreme points according to the data fluctuation characteristics of the data sequence; obtain the extreme credibility of similar local extreme points according to the change characteristics between the similar local extreme points and adjacent data points and the data difference characteristics between the similar local extreme points.

[0006] Obtain the same - type credible extreme points and the sequence of credible extreme points according to the extreme - value credibility of the same - type local extreme points; obtain the marked extreme points according to the change characteristics between the same - type credible extreme points in the sequence of credible extreme points; obtain the target data segment and the target extreme points according to the distribution characteristics of the marked extreme points in the data sequence;

[0007] Obtain the weight coefficient of the target extreme points according to the correlation characteristics between the target data segments; obtain the optimized extreme value according to the target extreme points and the weight coefficient; obtain the abnormal data of the data sequence through the Isolation Forest algorithm based on the optimized extreme value;

[0008] Obtain the abnormal degree of the influencing factors according to the distribution characteristics of the abnormal data; optimize the live - working path according to the abnormal degree.

[0009] Further, the step of obtaining the same - type local extreme points according to the data fluctuation characteristics of the data sequence includes:

[0010] The same - type local extreme points include the same - type local maximum points and the same - type local minimum points;

[0011] For any data point in the data sequence, when the any data point is greater than the values of two adjacent data points, the any data point is used as a local maximum point, and all local maximum points are used as the same - type local maximum points; when the any data point is less than the values of two adjacent data points, the any data point is used as a local minimum point, and all local minimum points are used as the same - type local minimum points.

[0012] Further, the step of obtaining the extreme - value credibility of the same - type local extreme points according to the change characteristics between the same - type local extreme points and adjacent data points, and the data difference characteristics between the same - type local extreme points includes:

[0013] Construct a two - dimensional rectangular coordinate system for the data sequence. For any local maximum point among the same - type local maximum points; calculate the average value of the absolute values of the straight - line slopes between the any local maximum point and two adjacent data points in the two - dimensional rectangular coordinate system and perform a negative - correlation mapping to obtain the local smoothness of the any local maximum point;

[0014] Calculate the average value of the absolute values of the differences between the any local maximum point and two adjacent local maximum points and perform a negative - correlation mapping to obtain the extreme - value difference characteristic value of the any local maximum point; calculate the product of the local smoothness and the extreme - value difference characteristic value to obtain the extreme - value credibility of the any local maximum point.

[0015] Further, the step of obtaining the same - type credible extreme points and the sequence of credible extreme points according to the extreme - value credibility of the same - type local extreme points includes:

[0016] The same kind of credible extreme points include the same kind of credible maximum points and the same kind of credible minimum points; the sequence of credible extreme points includes the sequence of credible maximum points and the sequence of credible minimum points;

[0017] When the extreme value credibility of the local maximum point exceeds the preset credible threshold, the local maximum point serves as the same kind of credible maximum point; all the same kind of credible maximum points are constructed into the sequence of credible maximum points in the order of the data sequence; when the extreme value credibility of the local minimum point exceeds the preset credible threshold, the local minimum point serves as the same kind of credible minimum point; all the same kind of credible minimum points are constructed into the sequence of credible minimum points in the order of the data sequence.

[0018] Further, the step of obtaining the marked extreme points according to the variation characteristics between the same kind of credible extreme points in the sequence of credible extreme points includes:

[0019] The marked extreme points include marked maximum points and marked minimum points;

[0020] For any same kind of credible maximum point, calculate the absolute value of the slope of the straight line between the any same kind of credible maximum point and the adjacent same kind of credible maximum point in the two-dimensional rectangular coordinate system to obtain the fluctuation characterization value of the any same kind of credible maximum point; when the fluctuation characterization value exceeds the preset fluctuation threshold, the any same kind of credible maximum point serves as the reference maximum point;

[0021] For any reference maximum point, calculate the ratio of the number of reference maximum points and the same kind of credible maximum points in the preset neighborhood window of the any reference maximum point in the sequence of credible maximum points to obtain the neighborhood occupancy ratio of the any reference maximum point; when the neighborhood occupancy ratio exceeds the preset occupancy threshold, all the same kind of credible maximum points in the preset neighborhood window of the any reference maximum point serve as the marked maximum points.

[0022] Further, the step of obtaining the target data segment and the target extreme points according to the distribution characteristics of the marked extreme points in the data sequence includes:

[0023] Obtain the intersection data subsequence of any adjacent marked maximum points and any adjacent marked minimum points in the data sequence as the target data segment; the target extreme points include target maximum points and target minimum points, and all the same kind of local maximum points in the target data segment serve as the target maximum points, and all the same kind of local minimum points in the target data segment serve as the target minimum points.

[0024] Further, the step of obtaining the weight coefficient of the target extreme points according to the correlation characteristics between the target data segments includes:

[0025] For any target data segment in the data sequence, calculate the average value of the dynamic time warping distance between the any target data segment and other target data segments after normalization, and use it as the weight coefficient of any target extreme point in the any target data segment.

[0026] Further, the step of obtaining the optimized extreme value according to the target extreme point and the weight coefficient includes:

[0027] The optimized extreme values include optimized maximum values and optimized minimum values;

[0028] Calculate the average value of the product of the target maximum value point in the data sequence and the corresponding weight coefficient to obtain the optimized maximum value of the data sequence; calculate the average value of the product of the target minimum value point in the data sequence and the corresponding weight coefficient to obtain the optimized minimum value of the data sequence.

[0029] Further, the step of obtaining the abnormal degree of the influencing factor according to the distribution characteristics of the abnormal data includes:

[0030] Calculate the average value of the absolute value of the difference between the abnormal data in the data sequence and the preset normal value and normalize it to obtain the first abnormal coefficient of the influencing factor; calculate the average value of the time distance between adjacent abnormal data in the data sequence and perform a negative correlation mapping to obtain the second abnormal coefficient of the influencing factor; calculate the product of the first abnormal coefficient and the second abnormal coefficient to obtain the abnormal degree of the influencing factor.

[0031] Further, the step of optimizing the live working path according to the abnormal degree includes:

[0032] Calculate the ratio of the abnormal degree of the influencing factor of the power grid node to the path length and normalize it to obtain the maintenance coefficient of the power grid node; optimize the live working path of the distribution network according to the maintenance coefficient.

[0033] The present invention has the following beneficial effects:

[0034] In the embodiments of the present invention, obtaining similar local extreme points can preliminarily determine the data points with anomalies or noises in the data sequence; obtaining the extreme value credibility can analyze whether the similar local extreme points are noise data, and then obtain similar credible extreme points and a credible extreme point sequence, improving the accuracy of obtaining the final abnormal data. Obtaining the marked extreme points can characterize the target data segments and target extreme points where abnormal conditions may exist in the data sequence, and then can accurately obtain the local abnormal feature range in the data sequence. Obtaining the weight coefficient can further determine the accuracy of the target extreme point in characterizing the abnormal features; enabling more accurate acquisition of abnormal data based on the optimized extreme points through the isolation forest, improving the accuracy of abnormal data detection and the accuracy of calculating the abnormal degree of power grid nodes, and ultimately improving the optimization effect of the live working path and the stability of power grid operation. BRIEF DESCRIPTION OF THE DRAWINGS

[0035] To more clearly illustrate the technical solutions and advantages in the embodiments of the present invention or the prior art, the following will briefly introduce the drawings required for the description of the embodiments or the prior art. Obviously, the following described drawings are only some embodiments of the present invention. For those of ordinary skill in the art, without creative efforts, other drawings can be obtained based on these drawings.

[0036] Figure 1 It is a flowchart of an intelligent optimization method for a live working path in a distribution network provided by an embodiment of the present invention. DETAILED DESCRIPTION OF THE EMBODIMENTS

[0037] To further elaborate on the technical means and effects adopted by the present invention to achieve the intended invention purpose, the following, in combination with the drawings and preferred embodiments, details the specific implementation manners, structures, features, and effects of an intelligent optimization method for a live working path in a distribution network proposed according to the present invention. In the following description, different "one embodiment" or "another embodiment" do not necessarily refer to the same embodiment. In addition, the specific features, structures, or characteristics in one or more embodiments can be combined in any suitable form.

[0038] Unless otherwise defined, all technical and scientific terms used herein have the same meaning as commonly understood by those of ordinary skill in the technical field to which the present invention belongs.

[0039] The following specifically describes the specific solution of an intelligent optimization method for a live working path in a distribution network provided by the present invention with reference to the drawings.

[0040] Please refer to Figure 1 , which shows a flowchart of an intelligent optimization method for a live working path in a distribution network provided by an embodiment of the present invention. The method includes the following steps:

[0041] Step S1: Obtain the data sequence of the influencing factors for optimizing the operation path of power grid nodes, i.e., the path length of power grid nodes; obtain the same-type local extreme points according to the data fluctuation characteristics of the data sequence; obtain the extreme value credibility of the same-type local extreme points according to the change characteristics between the same-type local extreme points and adjacent data points and the data difference characteristics between the same-type local extreme points.

[0042] In the embodiment of the present invention, the implementation scenario is the intelligent optimization of the live working path of the distribution network; the optimization of the power grid operation path depends on the geographical location of the power grid nodes and the abnormal degree of the power grid. On the premise of ensuring a relatively short overall live working path, the power grid nodes with a higher abnormal degree are preferentially repaired. Since the geographical location of the power grid nodes is fixed, the abnormal degree of the power grid nodes mainly affects the effect of path optimization. First, obtain the data sequence of the influencing factors for optimizing the operation path of power grid nodes. The influencing factors refer to the characteristics that can represent the operation state of the power system, such as characteristics like current, voltage, and temperature, etc.; in the embodiment of the present invention, voltage is used as the influencing factor for analysis; obtain the time sequence of the voltage of each power grid node as the data sequence for analysis. It should be noted that the analysis process for other characteristics is the same, and the implementer can determine the influencing factors according to the implementation scenario. Further, obtain the path lengths from the maintenance site to different power grid nodes to facilitate the final path optimization.

[0043] The detection of abnormal data in the data sequence can be performed through the existing isolation forest algorithm. When constructing an isolation tree in the isolation forest algorithm, a splitting value is randomly selected between the maximum value and the minimum value in the data set for splitting processing; due to the local characteristics of the data sequence for monitoring power grid nodes, the data is abnormal within a local time range, but its abnormal characteristics are weak within the overall long time range and are difficult to be detected, resulting in the splitting value selected by the isolation forest algorithm between the maximum value and the minimum value being not sensitive enough to the change of local density and difficult to capture the local characteristics, thus reducing the detection accuracy of abnormal data. Therefore, it is necessary to improve the isolation forest algorithm, obtain the local extreme values in the data sequence, obtain the weight of each local extreme point according to the distribution characteristics of the local extreme values, and then obtain the optimized extreme values, so as to improve the ability of the isolation forest to capture local characteristics and improve the detection accuracy of abnormal data.

[0044] First, obtain all the extreme points in the data sequence, and obtain the same-type local extreme points according to the data fluctuation characteristics of the data sequence, specifically including: the same-type local extreme points include the same-type local maximum points and the same-type local minimum points; for any data point in the data sequence, when the any data point is greater than the values of the two adjacent data points on the left and right, the any data point is used as a local maximum point, and all local maximum points are used as the same-type local maximum points; when the any data point is less than the values of the two adjacent data points on the left and right, the any data point is used as a local minimum point, and all local minimum points are used as the same-type local minimum points; thus, all the same-type local extreme points in the data sequence are obtained.

[0045] Further, after obtaining all the same-type local extreme points, if there are noise data among them, it will greatly interfere with the subsequent judgment results of the isolation forest algorithm. Therefore, it is necessary to judge the same-type local extreme points to obtain the non-noise extreme points; obtain the extreme value credibility of the same-type local extreme points according to the change characteristics between the same-type local extreme points and the adjacent data points and the data difference characteristics between the same-type local extreme points.

[0046] Preferably, in an embodiment of the present invention, obtaining the extreme value credibility includes: constructing a two-dimensional rectangular coordinate system for the data sequence, for any local maximum point among the same-type local maximum points; calculating the average value of the absolute values of the straight-line slopes between the any local maximum point and the two adjacent data points on the left and right in the two-dimensional rectangular coordinate system and performing a negative correlation mapping to obtain the local smoothness of the any local maximum point; when the absolute value of the straight-line slope between the any local maximum point and the two adjacent data points on the left and right is larger, it means that the distribution is more discrete, and the any local maximum point is more likely to be a noise data point, and the value of the local smoothness is smaller. Calculate the average value of the absolute values of the differences between the any local maximum point and the two adjacent local maximum points on the left and right and perform a negative correlation mapping to obtain the extreme value difference characteristic value of the any local maximum point; when the difference between the any local maximum point and the two nearest local maximum points on the left and right is larger, it means that the value of the any local maximum point is more abnormal and is more likely to be a noise data point, and the extreme value difference characteristic value is smaller. Calculate the product of the local smoothness and the extreme value difference characteristic value to obtain the extreme value credibility of the any local maximum point; when the extreme value credibility is smaller, the any local maximum point is more likely to be noise data. It should be noted that the calculation method of the extreme value credibility of the local minimum point is the same as that of the local maximum point and will not be elaborated. The formula for obtaining the extreme value credibility of the local maximum point includes:

[0047]

[0048] In the formula, F represents the extreme value credibility of the local maximum point, exp() represents the exponential function with the natural constant as the base, |K1| represents the absolute value of the slope of the straight line between the local maximum point and the left adjacent data point, and |K2| represents the absolute value of the slope of the straight line between the local maximum point and the right adjacent data point. represents the local smoothness of the local maximum point, |H1| represents the absolute value of the numerical difference between the local maximum point and the left adjacent local maximum point, and |H2| represents the absolute value of the numerical difference between the local maximum point and the right adjacent local maximum point. represents the extreme value difference eigenvalue of the local maximum point.

[0049] Step S2: Obtain the same-kind credible extreme points and the credible extreme point sequence according to the extreme value credibility of the same-kind local extreme points; obtain the marked extreme points according to the change characteristics between the same-kind credible extreme points in the credible extreme point sequence; and obtain the target data segment and the target extreme points according to the distribution characteristics of the marked extreme points in the data sequence.

[0050] After obtaining the extreme value credibility of the same-kind local extreme points, it is possible to determine whether the same-kind local extreme points are noise data. Therefore, the same-kind credible extreme points and the credible extreme point sequence are obtained according to the extreme value credibility of the same-kind local extreme points, which specifically includes: the same-kind credible extreme points include the same-kind credible maximum points and the same-kind credible minimum points; the credible extreme point sequence includes the credible maximum point sequence and the credible minimum point sequence. When the extreme value credibility of the local maximum point exceeds the preset credible threshold, this local maximum point is used as the same-kind credible maximum point; all the same-kind credible maximum points are constructed into a credible maximum point sequence in the order of the data sequence; when the extreme value credibility of the local minimum point exceeds the preset credible threshold, this local minimum point is used as the same-kind credible minimum point; all the same-kind credible minimum points are constructed into a credible minimum point sequence in the order of the data sequence; in the embodiment of the present invention, the preset credible threshold is 0.3, and the implementer can determine it by himself according to the implementation scenario.

[0051] Furthermore, during the normal operation of the power grid, the data fluctuation characteristics of the data sequence are small, while when an abnormality occurs, the data fluctuation characteristics are obvious, and the difference between adjacent same-kind local extreme points is large. Therefore, the greater the abnormality degree of the corresponding data segment, the greater the importance of the extreme points in this type of data segment to the final isolation forest algorithm for obtaining abnormal data. Therefore, the marked extreme points can be obtained according to the change characteristics between the same-kind credible extreme points in the credible extreme point sequence.

[0052] Preferably, in the embodiments of the present invention, the step of obtaining the marked extreme points includes: The marked extreme points include marked maximum points and marked minimum points; for any same-kind credible maximum point, calculate the absolute value of the slope of the straight line between this any same-kind credible maximum point and the adjacent same-kind credible maximum point in the two-dimensional rectangular coordinate system to obtain the fluctuation characterization value of this any same-kind credible maximum point; in the embodiments of the present invention, calculate the absolute value of the slope of the straight line with the previous adjacent same-kind credible maximum point for this any same-kind credible maximum point. When the fluctuation characterization value exceeds the preset fluctuation threshold, it means that the fluctuation feature of this any same-kind credible maximum point is relatively large, and this any same-kind credible maximum point is used as the reference maximum point; analyze and judge the abnormal features of the data segment near it based on this reference maximum point. For any reference maximum point, calculate the ratio of the number of reference maximum points and same-kind credible maximum points in the preset neighborhood window of this any reference maximum point in the sequence of credible maximum points to obtain the neighborhood occupancy ratio of this any reference maximum point; in the embodiments of the present invention, the preset neighborhood window is a window centered on the reference maximum point with a data length of 7, which can be determined by the implementer according to the implementation scenario. If the reference maximum point does not meet the condition of this preset neighborhood window at the edge of the sequence of credible maximum points, then calculate with the preset neighborhood window closest to the window center of this reference maximum point. When the neighborhood occupancy ratio is larger, it means that the number of other reference maximum points near this reference maximum point is more, and the fluctuation feature of this data segment is more obvious; when the neighborhood occupancy ratio exceeds the preset occupancy threshold, all the same-kind credible maximum points in the preset neighborhood window of this any reference maximum point are used as marked maximum points, and the marked maximum points mean the data points with more obvious abnormal fluctuations; in the embodiments of the present invention, the preset occupancy threshold is 0.6, which can be determined by the implementer according to the implementation scenario. It should be noted that the steps for obtaining the marked minimum points and the marked maximum points are the same; the marked extreme points can characterize the positions of abnormal fluctuations in the data sequence, so the maximum value and the minimum value required in the isolation forest can be optimized according to the distribution characteristics of the marked extreme points.

[0053] Further, after obtaining the marked extreme points, it is necessary to analyze the data segments with abnormal fluctuations according to the positions of the marked extreme points in the data sequence, and then analyze the weights of the extreme points according to the data segments to achieve the purpose of optimizing the maximum and minimum values; therefore, the target data segments and the target extreme points can be obtained according to the distribution characteristics of the marked extreme points in the data sequence.

[0054] Preferably, in an embodiment of the present invention, obtaining the target data segment and the target extreme points includes: obtaining, in the data sequence, the intersection data subsequence of any adjacent marked maximum points and any adjacent marked minimum points as the target data segment; the target data segment reflects the data segment with relatively obvious fluctuations in the data sequence, and the abnormal data in the target data segment appears with a relatively high frequency. The target extreme points include target maximum points and target minimum points. All the same-type local maximum points in the target data segment are used as the target maximum points, and all the same-type local minimum points in the target data segment are used as the target minimum points; by obtaining the target extreme points in different target data segments, the local abnormal characteristics in the power grid operation process can be determined, which is convenient for the isolation forest algorithm to capture the local abnormal characteristics.

[0055] Step S3, obtaining the weight coefficients of the target extreme points according to the relevant characteristics between the target data segments; obtaining the optimized extreme values according to the target extreme points and the weight coefficients; obtaining the abnormal data of the data sequence through the isolation forest algorithm according to the optimized extreme values.

[0056] According to the daily electricity consumption characteristics, it can be known that the data sequence of the power grid operation has periodicity, resulting in a certain periodicity in the abnormal conditions of the influencing factors of the power grid nodes, while the abnormal conditions caused by noise factors are relatively random; in order to further determine whether the target extreme points can represent the abnormal conditions in the power grid operation process, the weight coefficients of the target extreme points can be obtained according to the relevant characteristics between the target data segments, which specifically includes: for any target data segment in the data sequence, calculating the average value of the dynamic time warping distance between the any target data segment and other target data segments after negative correlation mapping as the weight coefficient of any target extreme point in the any target data segment; it should be noted that the dynamic time warping distance is obtained through the existing DTW algorithm, and the specific calculation steps will not be elaborated. When the change characteristics of the two sequences are more similar and the sequence values are closer, the dynamic time warping distance is smaller. Therefore, when the data change characteristics of the any target data segment and other target data segments are more similar, the greater the possibility that the any target data segment is abnormally operating and the smaller the possibility of being caused by noise; the greater the weight coefficient of any target extreme point in the any target data segment participating in optimizing the maximum and minimum values in the isolation forest. The formula for obtaining the weight coefficient includes:

[0057]

[0058] In the formula, R represents the weight coefficient of any target extreme point in the target data segment, B represents the number of other target data segments, D b represents the dynamic time warping distance between the target data segment and the b-th other target data segment, and exp() represents the exponential function with the natural constant as the base.

[0059] Further, after obtaining the weight coefficients of the target extreme points in the target data segment, the optimized extreme values can be obtained based on the target extreme points and the weight coefficients, specifically including: The optimized extreme values include optimized maximum values and optimized minimum values; calculate the average value of the product of the target maximum value points and the corresponding weight coefficients in the data sequence to obtain the optimized maximum value of the data sequence; calculate the average value of the product of the target minimum value points and the corresponding weight coefficients in the data sequence to obtain the optimized minimum value of the data sequence. The obtained optimized extreme values can more accurately characterize the abnormal characteristics in local time periods during the power grid operation process, avoiding the inaccuracy of directly determining abnormal data based on the maximum and minimum values in the entire data sequence according to the isolation forest algorithm, which may lead to the difficulty in distinguishing local abnormal data from normal data.

[0060] After obtaining the optimized extreme values, the abnormal data of the data sequence can be obtained according to the optimized extreme values through the isolation forest algorithm. It should be noted that the isolation forest algorithm belongs to the prior art, and the specific calculation steps will not be elaborated here; construct isolation trees based on the optimized extreme values and the characteristics of the data sequence and calculate the anomaly scores, and finally obtain the abnormal data in the data sequence. The abnormal data obtained through the optimized extreme values can more accurately find the operation abnormal data in different time periods of the data sequence, improve the accuracy of anomaly detection, and facilitate the final optimization effect of the live working path.

[0061] Step S4, obtain the abnormal degree of the influencing factors according to the distribution characteristics of the abnormal data; optimize the live working path according to the abnormal degree.

[0062] After obtaining the abnormal data in the data sequence, the abnormal degree of the influencing factors can be obtained according to the distribution characteristics of the abnormal data, specifically including: calculate the average value of the absolute value of the difference between the abnormal data and the preset normal value in the data sequence and normalize it to obtain the first abnormal coefficient of the influencing factor; the larger the first abnormal coefficient, the greater the data difference and the greater the abnormal degree. Calculate the average value of the time distances between adjacent abnormal data in the data sequence and perform a negative correlation mapping to obtain the second abnormal coefficient of the influencing factor; the time distance refers to the time difference between adjacent abnormal data points; the shorter the time difference, the more frequent the abnormal situation and the greater the abnormal degree. Calculate the product of the first abnormal coefficient and the second abnormal coefficient to obtain the abnormal degree of the influencing factor.

[0063] After obtaining the abnormality degree of the influencing factors of each power grid node, the live working path can be optimized according to the abnormality degree, which specifically includes: calculating and normalizing the ratio of the abnormality degree of the influencing factors of the power grid node to the path length to obtain the maintenance coefficient of the power grid node; optimizing the live working path of the distribution network according to the maintenance coefficient. When the path length from the maintenance site to the power grid node is shorter and the abnormality degree of the power grid node is greater, emergency maintenance is more required to ensure the stable operation of the power grid system. Therefore, the live working can be sorted according to the maintenance coefficient from large to small to achieve the path optimization of the live working. It should be noted that the implementer can determine the path optimization process by himself according to the implementation scenario according to the abnormality degree, which is not limited here.

[0064] In summary, the embodiment of the present invention provides a method for intelligent optimization of the live working path of a distribution network; obtaining the same-kind local extreme points according to the data sequence; obtaining the extreme value credibility according to the data difference characteristics between the same-kind local extreme points and adjacent data points; obtaining the same-kind credible extreme points according to the extreme value credibility; obtaining the marked extreme points according to the change characteristics of the same-kind credible extreme points; obtaining the target data segment and the target extreme points according to the distribution characteristics of the marked extreme points; obtaining the weight coefficient of the target extreme points and the final optimized extreme value according to the target data segment. The present invention obtains abnormal data through the isolation forest algorithm according to the optimized extreme value, improves the accuracy of obtaining abnormal data, obtains the abnormality degree according to the abnormal data and optimizes the live working path, and improves the optimization effect of the live working path and the stability of the power grid operation.

[0065] It should be noted that: the above sequence of the embodiments of the present invention is only for description and does not represent the advantages and disadvantages of the embodiments. The processes depicted in the accompanying drawings do not necessarily require the specific order or continuous order shown to achieve the desired results. In some embodiments, multitasking and parallel processing are also possible or may be advantageous.

[0066] Each embodiment in this specification is described in a progressive manner, and the same or similar parts between each embodiment can be referred to each other. Each embodiment focuses on the differences from other embodiments.

Claims

1. An intelligent optimization method for the path of live working on the distribution network, characterized in that, The method includes the following steps: Obtain a data sequence of influencing factors for optimizing the operation path of power grid nodes, and the path length of the power grid nodes; the influencing factors include current, voltage, and temperature; obtain homogeneous local extreme points according to the data fluctuation characteristics of the data sequence; obtain the extreme value credibility of the homogeneous local extreme points according to the change characteristics between the homogeneous local extreme points and adjacent data points and the data difference characteristics between the homogeneous local extreme points. Obtain homogeneous credible extreme points and a credible extreme point sequence according to the extreme value credibility of the homogeneous local extreme points; obtain marked extreme points according to the change characteristics between the homogeneous credible extreme points in the credible extreme point sequence; obtain a target data segment and target extreme points according to the distribution characteristics of the marked extreme points in the data sequence. Obtain the weight coefficient of the target extreme point according to the correlation characteristics between the target data segments; obtain the optimized extreme value according to the target extreme point and the weight coefficient; obtain the abnormal data of the data sequence through the isolation forest algorithm according to the optimized extreme value. Obtain the abnormal degree of the influencing factors according to the distribution characteristics of the abnormal data; optimize the live working path according to the abnormal degree. The step of obtaining the extreme value credibility of the homogeneous local extreme points according to the change characteristics between the homogeneous local extreme points and adjacent data points and the data difference characteristics between the homogeneous local extreme points includes: Construct a two-dimensional rectangular coordinate system for the data sequence. For any local maximum point among the homogeneous local maximum points; calculate the average value of the absolute values of the straight-line slopes between the any local maximum point and two adjacent data points in the two-dimensional rectangular coordinate system and perform a negative correlation mapping to obtain the local smoothness of the any local maximum point. Calculate the average value of the absolute values of the differences between the any local maximum point and two adjacent local maximum points and perform a negative correlation mapping to obtain the extreme value difference characteristic value of the any local maximum point; calculate the product of the local smoothness and the extreme value difference characteristic value to obtain the extreme value credibility of the any local maximum point. The step of optimizing the live working path according to the abnormal degree includes: Calculate the ratio of the abnormal degree of the influencing factors of the power grid node to the path length and normalize it to obtain the maintenance coefficient of the power grid node; optimize the distribution network live working path according to the maintenance coefficient.

2. The intelligent optimization method for the live working path of the distribution network according to claim 1, characterized in that, The step of obtaining homogeneous local extreme points according to the data fluctuation characteristics of the data sequence includes: The homogeneous local extreme points include homogeneous local maximum points and homogeneous local minimum points. For any data point in the data sequence, when the any data point is greater than the values of two adjacent data points, the any data point is used as a local maximum point, and all local maximum points are used as the homogeneous local maximum points; when the any data point is less than the values of two adjacent data points, the any data point is used as a local minimum point, and all local minimum points are used as the homogeneous local minimum points.

3. The intelligent optimization method for the live working path of the distribution network according to claim 2, characterized in that The step of obtaining homogeneous credible extreme points and a credible extreme point sequence according to the extreme value credibility of the homogeneous local extreme points includes: The like credible extreme points include like credible maximum points and like credible minimum points; the credible extreme point sequence includes a credible maximum point sequence and a credible minimum point sequence; When the extreme credibility of the local maximum point exceeds a preset credible threshold, the local maximum point is used as the like credible maximum point; all the like credible maximum points are constructed into the credible maximum point sequence in the order of the data sequence; when the extreme credibility of the local minimum point exceeds a preset credible threshold, the local minimum point is used as the like credible minimum point; all the like credible minimum points are constructed into the credible minimum point sequence in the order of the data sequence.

4. The intelligent optimization method for the live working path of the distribution network according to claim 3, wherein, The step of obtaining the marked extreme points according to the variation characteristics between the like credible extreme points in the credible extreme point sequence includes: The marked extreme points include marked maximum points and marked minimum points; For any like credible maximum point, calculate the absolute value of the straight-line slope between the any like credible maximum point and the adjacent like credible maximum point in the two-dimensional rectangular coordinate system to obtain the fluctuation characterization value of the any like credible maximum point; when the fluctuation characterization value exceeds a preset fluctuation threshold, the any like credible maximum point is used as the reference maximum point; For any reference maximum point, calculate the ratio of the number of reference maximum points and like credible maximum points in the preset neighborhood window of the any reference maximum point in the credible maximum point sequence to obtain the neighborhood occupancy ratio of the any reference maximum point; when the neighborhood occupancy ratio exceeds a preset occupancy threshold, all the like credible maximum points in the preset neighborhood window of the any reference maximum point are used as the marked maximum points.

5. The intelligent optimization method for the live working path of a distribution network according to claim 4, characterized in that, The step of obtaining the target data segment and the target extreme points according to the distribution characteristics of the marked extreme points in the data sequence includes: Obtain the intersection data subsequence of any adjacent marked maximum points and any adjacent marked minimum points in the data sequence as the target data segment; the target extreme points include target maximum points and target minimum points, and all the like local maximum points in the target data segment are used as the target maximum points, and all the like local minimum points in the target data segment are used as the target minimum points.

6. The intelligent optimization method for the live working path of the distribution network according to claim 1, characterized in that, The step of obtaining the weight coefficient of the target extreme points according to the correlation characteristics between the target data segments includes: For any target data segment in the data sequence, calculate the average value of the dynamic time warping distance between the any target data segment and other target data segments after normalization as the weight coefficient of any target extreme point in the any target data segment.

7. An intelligent optimization method for the live working path of a distribution network according to claim 5, characterized in that, The step of obtaining the optimized extreme values according to the target extreme points and the weight coefficients includes: The optimized extreme values include optimized maximum values and optimized minimum values; Calculate the average value of the product of the target maximum points and the corresponding weight coefficients in the data sequence to obtain the optimized maximum value of the data sequence; calculate the average value of the product of the target minimum points and the corresponding weight coefficients in the data sequence to obtain the optimized minimum value of the data sequence.

8. The intelligent optimization method for the live working path of the distribution network according to claim 1, characterized in that The step of obtaining the abnormality degree of the influencing factor according to the distribution characteristics of the abnormal data includes: Calculating the average value of the absolute value of the difference between the abnormal data and the preset normal value in the data sequence and normalizing it to obtain the first abnormality coefficient of the influencing factor; calculating the average value of the time distances between adjacent abnormal data in the data sequence and performing a negative correlation mapping to obtain the second abnormality coefficient of the influencing factor; calculating the product of the first abnormality coefficient and the second abnormality coefficient to obtain the abnormality degree of the influencing factor.

Images