A smart maintenance method for smart distribution network

Abstract

The present invention provides an intelligent maintenance method for an intelligent distribution network, which belongs to the technical field of intelligent distribution networks, and includes: dividing the topological structure of the distribution network to obtain a plurality of sub-structures, extracting the sub-fault data of each line section on the sub-structure from the historical fault database, and obtaining the first array of the corresponding line section based on each fault type; according to the execution function of the first sub-structure and the distribution type, and in combination with the first array, weights are set for different fault types, and the fault labels are set and calibrated in combination with the symbol combination of the first array of the corresponding line section under different fault types; multi-dimensional detection is performed on each line section in real time according to a preset detection method to obtain measurement data, and the section loss coefficient of the corresponding line section under each detection index is obtained; the section loss coefficient and the detection index are compared with the corresponding calibrated line to determine the fault maintenance plan corresponding to the first sub-structure. Timeliness and efficiency of line maintenance are ensured.

Description

Technical Field

[0001] The present invention relates to the technical field of smart distribution networks, and in particular to an intelligent maintenance method for smart distribution networks. Background Art

[0002] The distribution network is composed of overhead lines, cables, towers, distribution transformers, disconnectors, reactive power compensators and some ancillary facilities. It plays an important role in distributing electric energy in the power grid. Since there are many types of equipment involved in the distribution network and they are all implemented based on the power grid lines, it is particularly important to perform real-time detection of the lines.

[0003] The existing method of circuit maintenance is to manually perform regular maintenance on a certain section of the line. However, it is unknown whether there is any abnormality in the line. It can only be discovered through fixed-point inspections at regular intervals. In other words, fault discovery and maintenance are not timely. If maintenance cannot be carried out in time, the line will be paralyzed.

[0004] Therefore, the present invention proposes an intelligent maintenance method for a smart distribution network. Summary of the invention

[0005] The present invention provides an intelligent maintenance method for an intelligent distribution network, which is used to provide a basis for subsequent effective maintenance by dividing the distribution network to facilitate block management, and obtain a first array for each route segment based on a historical fault database, and obtain a segment loss coefficient by weight setting and real-time detection of data of each line in the distribution network, thereby ensuring the accuracy of the fault maintenance plan, thereby ensuring the timeliness and efficiency of the maintenance line.

[0006] The present invention provides an intelligent maintenance method for an intelligent distribution network, comprising:

[0007] Step 1: dividing the topological structure of the distribution network to obtain a plurality of first substructures, wherein the division is achieved according to the regional functions of the distribution network;

[0008] Step 2: extracting sub-fault data of each line section on the first sub-structure from the historical fault database, and dividing the sub-fault data according to the fault type to obtain a first array of the corresponding line section based on each fault type;

[0009] Step 3: according to the execution function and the power distribution type of the first substructure, and in combination with the first array under the corresponding fault type for each line segment in the first substructure, weights are set for different fault types, and in combination with the symbol combination of the first array under different fault types for the corresponding line segment, fault labels are set and marked on the corresponding line segment;

[0010] Step 4: Perform multi-dimensional detection on each line section in real time according to a preset detection method to obtain measurement data, and extract data from the measurement data according to the detection index under each dimension to obtain the section loss coefficient of the corresponding line section under each detection index;

[0011] Step 5: Compare the section loss coefficient and the detection index with the corresponding calibrated line, determine the fault maintenance plan corresponding to the first sub-structure and send it to the maintenance personnel terminal for maintenance reminder.

[0012] Preferably, obtaining a first array of the corresponding line segment based on each fault type includes:

[0013] Dividing the sub-fault data according to the fault type to obtain a number of divided data;

[0014] The number of fault occurrences of the corresponding fault type, the average value of the fault coverage length under all fault occurrences, the variance of the fault coverage length under all fault occurrences, the average value of the duration of the fault under all fault occurrences, and the variance of the duration of the fault under all fault occurrences are extracted from each partitioned data, and a first array for the corresponding fault type is generated.

[0015] Preferably, weights are set for different fault types, including:

[0016] Retrieving the initial weight of each fault type based on the first sub-structure from the power distribution-type-weight database;

[0017] Sorting the line weights of each line section involved in the first substructure by size, and sorting the first arrays of each line section in the first substructure based on the same fault type according to the size sorting result, to construct an analysis matrix;

[0018] Determine a first discrete degree of each first remaining discrete point by taking the first element in the column vector as a first reference point, and determine a second discrete degree of each second remaining discrete point by taking the element with the largest element value in the column vector as a second reference point;

[0019] According to the first discrete degree and the second discrete degree, a relative stability coefficient is set for each remaining discrete point except two basic points, and a reference discrete point is locked according to a maximum coefficient among the relative stability coefficients;

[0020] When the first reference point, the second reference point, and the third reference point can form a triangle, obtaining a first value based on each boundary point of the triangle and a first value of the centroid of the triangle, and averaging all the first values ​​to obtain a first average value;

[0021] When the first reference point, the second reference point, and the third reference point cannot form a triangle, any two points are connected to obtain a first straight line, and the second value of each point on the first straight line is obtained in turn, and all the second values ​​are averaged to obtain a second average value;

[0022] Based on the average value obtained from each column vector, an average array of the corresponding fault type is constructed. At the same time, the frontmost vector in each column of the analysis matrix corresponding to the fault type is locked. The sum of the line weights of the line segments corresponding to the values ​​is calculated, and a reference weight array corresponding to the fault type is constructed, where N represents the total number of line segments involved in the first substructure;

[0023] According to the average array and the reference weight array of different fault types involved in the same first sub-structure, and combined with the initial weight of each fault type, the type weights of different fault types are obtained;

[0024] ;

[0025] in, represents the type weight of the j1th fault type; represents the failure coefficient of the j1th failure type; Indicates the total number of fault types; M2 indicates the number of the first sub-structure; represents the value of the k1th element in the average array for the j1th fault type under the u1th first substructure; represents the reference weight of the k1th element in the reference weight array for the j1th fault type under the u1th first substructure; represents the maximum value of the k1th element in the average array involved in all the first sub-structures based on the j1th fault type; represents the average value of the k1th element in the average array involved in all the first sub-structures based on the j1th fault type; represents the variance of the value of the k1th element in the average array involved in the remaining structures except the u1th first sub-structure based on the j1th fault type; represents the initial weight of the j1th fault type.

[0026] Preferably, the relative stability coefficient is set, including:

[0027] ;

[0028] in, Respectively represent the values ​​of the elements corresponding to the first reference point, the remaining discrete points, and the second reference point; Respectively represent the first discrete degree and the second discrete degree corresponding to the remaining discrete points; Represents the relative stability coefficient corresponding to the remaining discrete points.

[0029] Preferably, the method further comprises: constructing a fault label, comprising:

[0030] Match the first array of each fault type in the corresponding line segment with the type-array-symbol comparison table to obtain a unique symbol of the corresponding fault type, and combine the unique symbols of all fault types in the same line segment to obtain a symbol combination;

[0031] The type weight of each fault type is assigned to the unique symbol corresponding to the symbol combination to obtain the fault label.

[0032] Preferably, after extracting data from the measurement data according to the detection indicators in each dimension, the following steps are included:

[0033] The conventional filtering method is used to filter the extracted data under each detection index respectively;

[0034] Determine the reference value of the filtered data based on preset judgment conditions;

[0035] ;

[0036] in, Indicates the reference value of the corresponding filtered data; Indicates the number of preset judgment conditions; represents the judgment weight of the i1th preset judgment condition; Indicates the satisfaction coefficient of the extracted data before filtering based on the i1th preset judgment condition; Indicates the satisfaction coefficient of the filtered data based on the i1th preset judgment condition; Represents the data before filtering With the filtered data Semantic similarity coefficient based on the i1th preset judgment condition; Indicates all The variance of Respectively represent the judgment function, when hour, The value of is 0; when hour, The value of is 1; when hour, The value of is 1; when hour, The value of is 0;

[0037] The filtered data whose reference value is greater than the preset value is retained;

[0038] Otherwise, switch the data filtering mode to re-filter the extracted data under the corresponding detection index.

[0039] Preferably, obtaining the section loss coefficient of the corresponding section line under each detection index includes:

[0040] Retrieving a data analysis model matching the detection indicator from an indicator-analysis database;

[0041] Input the filtered data under the corresponding detection index into the corresponding data analysis model to obtain the corresponding section loss coefficient.

[0042] Preferably, the preset detection methods correspond one to one with the fault types;

[0043] Each detection indicator corresponds to a preset detection method.

[0044] Preferably, determining a fault maintenance plan corresponding to the first sub-structure includes:

[0045] Determine whether the first ratio of the section loss coefficient corresponding to the detection index to the corresponding preset threshold is greater than 1 / 2,

[0046] If so, retain the section loss coefficient corresponding to the detection indicator and the sub-label in the fault label of the fault type corresponding to the detection indicator obtained based on the comparison result;

[0047] Otherwise, only the section loss coefficient corresponding to the detection index is retained;

[0048] The retained results under all detection indicators under each line segment are converted into result vectors respectively, and at the same time, a connection vector is constructed according to the connection relationship between each line segment in the first substructure;

[0049] All result vectors and connection vectors in the corresponding first substructure are input into the vector analysis model to obtain the corresponding fault maintenance plan.

[0050] Compared with the prior art, the present invention has the following beneficial effects:

[0051] By dividing the distribution network, it is convenient to carry out block management to provide a basis for subsequent effective maintenance. The first array for each route is obtained based on the historical fault database. The section loss coefficient is obtained by weight setting and real-time detection of the data of each line in the distribution network to ensure the accuracy of the fault maintenance plan, thereby ensuring the timeliness and efficiency of the maintenance line.

[0052] Other features and advantages of the present invention will be described in the following description, and partly become apparent from the description, or understood by practicing the present invention. The purpose and other advantages of the present invention can be realized and obtained by the structures particularly pointed out in the written description and the accompanying drawings.

[0053] The technical solution of the present invention is further described in detail below through the accompanying drawings and embodiments. BRIEF DESCRIPTION OF THE DRAWINGS

[0054] The accompanying drawings are used to provide a further understanding of the present invention and constitute a part of the specification. Together with the embodiments of the present invention, they are used to explain the present invention and do not constitute a limitation of the present invention. In the accompanying drawings:

[0055] Figure 1 The present invention is a flowchart of an intelligent maintenance method for an intelligent distribution network according to an embodiment of the present invention. DETAILED DESCRIPTION

[0056] The preferred embodiments of the present invention are described below in conjunction with the accompanying drawings. It should be understood that the preferred embodiments described herein are only used to illustrate and explain the present invention, and are not used to limit the present invention.

[0057] The present invention provides an intelligent maintenance method for an intelligent distribution network. Figure 1 As shown, including:

[0058] Step 1: dividing the topological structure of the distribution network to obtain a plurality of first substructures, wherein the division is achieved according to the regional functions of the distribution network;

[0059] Step 2: extracting sub-fault data of each line section on the first sub-structure from the historical fault database, and dividing the sub-fault data according to the fault type to obtain a first array of the corresponding line section based on each fault type;

[0060] Step 3: according to the power distribution type of the first substructure and in combination with the first array under the corresponding fault type for each line segment in the first substructure, weights are set for different fault types, and in combination with the symbol combination of the first array under different fault types for the corresponding line segment, fault labels are set and marked on the corresponding line segment;

[0061] Step 4: Perform multi-dimensional detection on each line section in real time according to a preset detection method to obtain measurement data, and extract data from the measurement data according to the detection index under each dimension to obtain the section loss coefficient of the corresponding line section under each detection index;

[0062] Step 5: Compare the section loss coefficient and the detection index with the corresponding calibrated line, determine the fault maintenance plan corresponding to the first sub-structure and send it to the maintenance personnel terminal for maintenance reminder.

[0063] Preferably, the preset detection methods correspond one to one with the fault types;

[0064] Each detection indicator corresponds to a preset detection method.

[0065] In this embodiment, the topological structure of the distribution network is preset, and the topological structure is the connection relationship structure of each line in the distribution network.

[0066] In this embodiment, the functions of different areas in the distribution network are different. For example, area A is used for long-distance transmission, area B is used for short-distance transmission, and area C is used for medium-distance transmission.

[0067] In this embodiment, the historical fault database includes historical fault information of different lines in the distribution network, such as line short circuit fault, line disconnection fault, line insulation fault, line resistance fault, line high temperature fault, etc., and the types are short circuit type, disconnection type, insulation type, high temperature type, etc.

[0068] In this embodiment, each line section has corresponding sub-fault data, and the sub-fault data is extracted from a historical fault database and can be directly extracted.

[0069] In this embodiment, the first array is related to the number of abnormalities, the fault coverage length, and the duration of the fault.

[0070] In this embodiment, the power distribution type is related to the power load range of the power distribution. The greater the power load, the more stringent the corresponding fault detection needs to be.

[0071] In this embodiment, each fault type under the corresponding segment line has its corresponding first array, and each array can be represented by a unique symbol, so the unique symbols under different fault types are combined to obtain a symbol combination. For example, there are 3 fault types, and fault type 1 is represented by %R, fault type 2 is represented by *7, and fault type 3 is represented by &1. At this time, the symbol combination is: %R*7&1.

[0072] In this embodiment, the fault label includes: a unique symbol for each fault type and a type weight of the corresponding fault type. It should be noted that the label is set to reduce the space capacity occupied by the array, thereby avoiding inefficient processing caused by data redundancy while reducing the space occupation.

[0073] In this embodiment, the preset detection method is related to transmission current detection, insulation performance detection, line resistance detection, high temperature detection, detection of whether the line is disconnected, etc.

[0074] In this embodiment, each preset detection method corresponds to a detection dimension, and there is a detection indicator under each detection dimension, and the detection indicators are current detection indicators, insulation performance detection indicators, line resistance detection indicators, high temperature detection indicators, whether the line is disconnected detection indicators, etc.

[0075] In this embodiment, the result obtained after data detection is performed in a relevant manner is the extracted data under the corresponding detection index. Because the data are collected by synchronous detection or asynchronous detection using different preset detection methods, the data obtained are all together. Therefore, relevant data is obtained by extraction according to the detection index.

[0076] In this embodiment, the section loss coefficient refers to the value of the line anomaly existing under the corresponding detection index.

[0077] In this embodiment, the purpose of the comparison is to combine the section loss coefficient with the historical faults for analysis, so as to further ensure the accuracy of the maintenance plan and prevent problems before they occur.

[0078] The beneficial effects of the above technical solution are: by dividing the distribution network, it is convenient to carry out block management to provide a basis for subsequent effective maintenance, and based on the historical fault database, a first array for each route is obtained, and the section loss coefficient is obtained by weight setting and real-time detection of the data of each line in the distribution network, so as to ensure the accuracy of the fault maintenance plan, thereby ensuring the timeliness and efficiency of the maintenance line.

[0079] The present invention provides an intelligent maintenance method for an intelligent distribution network, which obtains a first array of corresponding segment lines based on each fault type, including:

[0080] Dividing the sub-fault data according to the fault type to obtain a number of divided data;

[0081] The number of fault occurrences of the corresponding fault type, the average value of the fault coverage length under all fault occurrences, the variance of the fault coverage length under all fault occurrences, the average value of the duration of the fault under all fault occurrences, and the variance of the duration of the fault under all fault occurrences are extracted from each partitioned data, and a first array for the corresponding fault type is generated.

[0082] In this embodiment, the first array = {number of fault occurrences, average value of fault coverage length, variance of fault coverage length, average value of duration of fault, variance of duration of fault}, and there are 5 elements in the first array.

[0083] In this embodiment, the fault coverage length refers to the length of the line abnormality. For example, when a line resistance abnormality is detected, the length of the line resistance is L1, and the time period of the line group abnormality is 2 hours.

[0084] The beneficial effect of the above technical solution is: by dividing the sub-fault data, the divided data of each line section is obtained, and then the first array is formed from three aspects: the number of fault occurrences, the fault coverage length and the duration of the fault.

[0085] The present invention provides an intelligent maintenance method for an intelligent distribution network, which sets weights for different fault types, including:

[0086] Retrieving the initial weight of each fault type based on the first sub-structure from the power distribution-type-weight database;

[0087] Sorting the line weights of each line section involved in the first sub-structure by size, and sorting the first arrays of each line section in the first sub-structure based on the same fault type according to the size sorting result, to construct an analysis matrix;

[0088] Determine a first discrete degree of each first remaining discrete point by taking the first element in the column vector as a first reference point, and determine a second discrete degree of each second remaining discrete point by taking the element with the largest element value in the column vector as a second reference point;

[0089] According to the first discrete degree and the second discrete degree, a relative stability coefficient is set for each remaining discrete point except two basic points, and a reference discrete point is locked according to a maximum coefficient among the relative stability coefficients;

[0090] When the first reference point, the second reference point, and the third reference point can form a triangle, obtaining a first value based on each boundary point of the triangle and a first value of the centroid of the triangle, and averaging all the first values ​​to obtain a first average value;

[0091] When the first reference point, the second reference point, and the third reference point cannot form a triangle, any two points are connected to obtain a first straight line, and the second value of each point on the first straight line is obtained in turn, and all the second values ​​are averaged to obtain a second average value;

[0092] Based on the average value obtained from each column vector, an average array of the corresponding fault type is constructed. At the same time, the frontmost vector in each column vector of the analysis matrix corresponding to the fault type is locked. The sum of the line weights of the line segments corresponding to the values ​​is calculated, and a reference weight array corresponding to the fault type is constructed, wherein N represents the total number of line segments involved in the first substructure;

[0093] According to the average array and the reference weight array of different fault types involved in the same first sub-structure, and combined with the initial weight of each fault type, the type weights of different fault types are obtained;

[0094] ;

[0095] in, represents the type weight of the j1th fault type; represents the failure coefficient of the j1th failure type; Indicates the total number of fault types; M2 indicates the number of the first sub-structure; represents the value of the k1th element in the average array for the j1th fault type under the u1th first substructure; represents the reference weight of the k1th element in the reference weight array for the j1th fault type under the u1th first substructure; represents the maximum value of the k1th element in the average array involved in all the first sub-structures based on the j1th fault type; represents the average value of the k1th element in the average array involved in all the first sub-structures based on the j1th fault type; represents the variance of the value of the k1th element in the average array involved in the remaining structures except the u1th first sub-structure based on the j1th fault type; represents the initial weight of the j1th fault type.

[0096] Preferably, setting a relative stability coefficient includes:

[0097] ;

[0098] in, Respectively represent the values ​​of the elements corresponding to the first reference point, the remaining discrete points, and the second reference point; Respectively represent the first discrete degree and the second discrete degree corresponding to the remaining discrete points; Represents the relative stability coefficient corresponding to the remaining discrete points.

[0099] In this embodiment, the power distribution-type-weight database includes the power distribution type and initial weights for different fault types encountered based on the power distribution type, and is pre-stored.

[0100] In this embodiment, the line weights are pre-set. It should be noted that before the topological structure of the distribution network is built, the weights of the lines involved are known, reflecting the importance of different lines, and the sum of the weights of all lines is 1.

[0101] In this embodiment, the analysis matrix = .

[0102] In this embodiment, the first discrete degree ,in, Indicates The fitting constant obtained by fitting all discrete points to the reference point; the second degree of discreteness ,in, Indicates is the fitting constant obtained after fitting all discrete points with the reference point; it should be noted that different reference points correspond to different fitting results, which belongs to the prior art.

[0103] In this embodiment, after the corresponding column vector is plotted, the ordinate corresponds to the value of the element of the corresponding column vector, and thus the boundary of the triangle or the value of the point on the straight line can be directly determined.

[0104] In this embodiment, the average value of the corresponding column vector is the first average value or the second average value.

[0105] In this embodiment, there is an average array for each fault type under the same first sub-structure.

[0106] In this embodiment, average array={average values ​​respectively related to 5 elements}.

[0107] In this embodiment, the reference weight array={the sum of the line weights corresponding to the 5 elements respectively}.

[0108] The beneficial effect of the above technical solution is: by constructing an analysis matrix for each fault type to calculate the discrete degree of the remaining discrete points, and then locking the reference point, and combining the first reference point and the second reference point to perform analysis under different connection structures, and then obtain the average array and reference weight data of different fault types, and combine with the initial weight to provide a data basis for calculating the weight of the fault type, ensuring that it is in line with the current operating conditions and more representative, indirectly ensuring the accuracy of the maintenance plan.

[0109] The present invention provides an intelligent maintenance method for an intelligent distribution network, further comprising: constructing a fault tag, including:

[0110] Match the first array of each fault type in the corresponding line segment with the type-array-symbol comparison table to obtain a unique symbol of the corresponding fault type, and combine the unique symbols of all fault types in the same line segment to obtain a symbol combination;

[0111] The type weight of each fault type is assigned to the unique symbol corresponding to the symbol combination to obtain the fault label.

[0112] In this embodiment, the type-array-symbol comparison table includes different fault types, the first array and the symbols matching therewith, and is pre-stored and can be used directly.

[0113] The beneficial effect of the above technical solution is that through the combination of unique symbols and type weights, it is easy to obtain fault labels, save space, and facilitate subsequent rapid retrieval and use of historical faults.

[0114] The present invention provides an intelligent maintenance method for an intelligent distribution network, which extracts data from the measurement data according to the detection index under each dimension, and comprises:

[0115] The conventional filtering method is used to filter the extracted data under each detection index respectively;

[0116] Determine the reference value of the filtered data based on preset judgment conditions;

[0117] ;

[0118] in, Indicates the reference value of the corresponding filtered data; Indicates the number of preset judgment conditions; represents the judgment weight of the i1th preset judgment condition; Indicates the satisfaction coefficient of the extracted data before filtering based on the i1th preset judgment condition; Indicates the satisfaction coefficient of the filtered data based on the i1th preset judgment condition; Represents the data before filtering With the filtered data Semantic similarity coefficient based on the i1th preset judgment condition; Indicates all The variance of Respectively represent the judgment function, when hour, The value of is 0; when hour, The value of is 1; when hour, The value of is 1; when hour, The value of is 0;

[0119] The filtered data whose reference value is greater than the preset value is retained;

[0120] Otherwise, switch the data filtering mode to re-filter the extracted data under the corresponding detection index.

[0121] In this embodiment, the conventional filtering method refers to a statistical filtering method, and the switched data filtering method is a model-based data filtering, both of which belong to the prior art.

[0122] In this embodiment, the preset judgment conditions include semantic judgment, data value size judgment, etc.

[0123] The beneficial effects of the above technical solution are: by filtering the extracted data, data redundancy is effectively avoided to reduce the accuracy of the analysis results, and the reference mechanism is determined by pre-set judgment conditions to effectively ensure that the filtered data is still valuable.

[0124] The present invention provides an intelligent maintenance method for an intelligent distribution network, which obtains the section loss coefficient of a corresponding section line under each detection index, including:

[0125] Retrieving a data analysis model matching the detection indicator from an indicator-analysis database;

[0126] Input the filtered data under the corresponding detection index into the corresponding data analysis model to obtain the corresponding section loss coefficient.

[0127] In this embodiment, the indicator-analysis database includes a data analysis model that matches different detection indicators, and the data analysis model is obtained by training a neural network model based on the detection data under the indicator and the abnormal analysis results of the detection data as samples. Therefore, the abnormal analysis coefficient can be directly obtained as the section loss coefficient.

[0128] The beneficial effect of the above technical solution is that by retrieving the model from the database and analyzing the data by the model, it is convenient to directly obtain the section loss coefficient, which provides convenience for subsequent acquisition of solutions.

[0129] The present invention provides an intelligent maintenance method for an intelligent distribution network, which determines a fault maintenance plan corresponding to a first sub-structure, comprising:

[0130] Determine whether the first ratio of the section loss coefficient corresponding to the detection index to the corresponding preset threshold is greater than 1 / 2,

[0131] If so, retain the section loss coefficient corresponding to the detection indicator and the sub-label in the fault label of the fault type corresponding to the detection indicator obtained based on the comparison result;

[0132] Otherwise, only the section loss coefficient corresponding to the detection index is retained;

[0133] The retained results under all detection indicators under each line segment are converted into result vectors respectively, and at the same time, a connection vector is constructed according to the connection relationship between each line segment in the first substructure;

[0134] All result vectors and connection vectors in the corresponding first substructure are input into the vector analysis model to obtain the corresponding fault maintenance plan.

[0135] In this embodiment, the preset threshold values ​​of different detection indicators are different. For example, the preset threshold value of detection indicator 1 is 0.2, and the preset threshold value of detection indicator 2 is 0.3. At this time, the section loss coefficient under detection indicator 1 is 0.5, and the section loss coefficient under detection indicator 2 is 0.6. At this time, the first ratio for detection indicator 1 is 0.5 / 0.2, and the first ratio for detection indicator 2 is 0.6 / 0.3.

[0136] In this embodiment, the sub-labels reserved are the reserved type weights of the corresponding fault types and the first array.

[0137] In this embodiment, the result vector = {retained results under each detection indicator}.

[0138] In this embodiment, the connection vector={connection relationship between different segments of lines}, wherein the connection relationship refers to whether line 01 and line 02 are connected in series or in parallel.

[0139] In this embodiment, the vector analysis model is obtained by training the neural network model based on samples of different result vectors and connection vectors, as well as maintenance suggestions for vector combinations. Therefore, a fault maintenance plan can be directly obtained. For example, line 01 has poor insulation and needs to be replaced as a whole.

[0140] The beneficial effect of the above technical solution is: comparing the section loss coefficient with a preset threshold to analyze whether the section loss coefficient under different sizes is retained by itself or retained together with the historical label results, thereby ensuring the reliability and timeliness of line maintenance.

[0141] Obviously, those skilled in the art can make various changes and modifications to the present invention without departing from the spirit and scope of the present invention. Thus, if these modifications and variations of the present invention fall within the scope of the claims of the present invention and their equivalents, the present invention is also intended to include these modifications and variations.

Claims

1. An intelligent maintenance method for an intelligent distribution network, characterized in that: include: Step 1: dividing the topological structure of the distribution network to obtain a plurality of first substructures, wherein the division is achieved according to the regional functions of the distribution network; Step 2: extracting sub-fault data of each line section on the first sub-structure from the historical fault database, and dividing the sub-fault data according to the fault type to obtain a first array of the corresponding line section based on each fault type; Step 3: according to the power distribution type of the first substructure and in combination with the first array under the corresponding fault type for each line segment in the first substructure, weights are set for different fault types, and in combination with the symbol combination of the first array under different fault types for the corresponding line segment, fault labels are set and marked on the corresponding line segment; Step 4: Perform multi-dimensional detection on each line section in real time according to a preset detection method to obtain measurement data, and extract data from the measurement data according to the detection index under each dimension to obtain the section loss coefficient of the corresponding line section under each detection index; Step 5: Compare the section loss coefficient and the detection index with the corresponding calibrated line, determine the fault maintenance plan corresponding to the first sub-structure and send it to the maintenance personnel terminal for maintenance reminder; determine the fault maintenance plan corresponding to the first sub-structure, including: Determine whether the first ratio of the section loss coefficient corresponding to the detection index to the corresponding preset threshold is greater than 1 / 2, If so, retain the section loss coefficient corresponding to the detection indicator and the sub-label in the fault label of the fault type corresponding to the detection indicator obtained based on the comparison result; Otherwise, only the section loss coefficient corresponding to the detection index is retained; The retained results under all detection indicators under each line segment are converted into result vectors respectively, and at the same time, a connection vector is constructed according to the connection relationship between each line segment in the first substructure; All result vectors and connection vectors in the corresponding first substructure are input into the vector analysis model to obtain the corresponding fault maintenance plan.

2. The intelligent maintenance method for the intelligent distribution network according to claim 1, characterized in that: The first array of the corresponding segment line based on each fault type is obtained, including: Dividing the sub-fault data according to the fault type to obtain a number of divided data; The number of fault occurrences of the corresponding fault type, the average value of the fault coverage length under all fault occurrences, the variance of the fault coverage length under all fault occurrences, the average value of the duration of the fault under all fault occurrences, and the variance of the duration of the fault under all fault occurrences are extracted from each partitioned data, and a first array for the corresponding fault type is generated.

3. The intelligent maintenance method for the intelligent distribution network according to claim 1, characterized in that: Set weights for different fault types, including: Retrieving the initial weight of each fault type based on the first sub-structure from the power distribution-type-weight database; Sorting the line weights of each line section involved in the first substructure by size, and sorting the first arrays of each line section in the first substructure based on the same fault type according to the size sorting result, to construct an analysis matrix; Determine a first discrete degree of each first remaining discrete point by taking the first element in the column vector as a first reference point, and determine a second discrete degree of each second remaining discrete point by taking the element with the largest element value in the column vector as a second reference point; According to the first discrete degree and the second discrete degree, a relative stability coefficient is set for each remaining discrete point except two basic points, and a reference discrete point is locked according to a maximum coefficient among the relative stability coefficients; When the first reference point, the second reference point, and the third reference point can form a triangle, obtaining a first value based on each boundary point of the triangle and a first value of the centroid of the triangle, and averaging all the first values ​​to obtain a first average value; When the first reference point, the second reference point, and the third reference point cannot form a triangle, any two points are connected to obtain a first straight line, and the second value of each point on the first straight line is obtained in turn, and all the second values ​​are averaged to obtain a second average value; Based on the average value obtained from each column vector, an average array of the corresponding fault type is constructed. At the same time, the frontmost vector in each column of the analysis matrix corresponding to the fault type is locked. The sum of the line weights of the line segments corresponding to the values ​​is calculated, and a reference weight array corresponding to the fault type is constructed, where N represents the total number of line segments involved in the first substructure; According to the average array and the reference weight array of different fault types involved in the same first sub-structure, and combined with the initial weight of each fault type, the type weights of different fault types are obtained; ； in, represents the type weight of the j1th fault type; represents the fault coefficient of the j1th fault type; Indicates the total number of fault types; M2 indicates the number of the first sub-structure; represents the value of the k1th element in the average array for the j1th fault type under the u1th first substructure; represents the reference weight of the k1th element in the reference weight array for the j1th fault type under the u1th first substructure; represents the maximum value of the k1th element in the average array involved in all the first sub-structures based on the j1th fault type; represents the average value of the k1th element in the average array involved in all the first sub-structures based on the j1th fault type; represents the variance of the value of the k1th element in the average array involved in the remaining structures except the u1th first sub-structure based on the j1th fault type; represents the initial weight of the j1th fault type.

4. The intelligent maintenance method for the intelligent distribution network according to claim 3, characterized in that: Set the relative stability factor, including: ； in, Respectively represent the values ​​of the elements corresponding to the first reference point, the remaining discrete points, and the second reference point; Respectively represent the first discrete degree and the second discrete degree corresponding to the remaining discrete points; Represents the relative stability coefficient corresponding to the remaining discrete points.

5. The intelligent maintenance method for the intelligent distribution network according to claim 1, characterized in that: Also includes: Build a failure tag, including: Match the first array of each fault type in the corresponding line segment with the type-array-symbol comparison table to obtain a unique symbol of the corresponding fault type, and combine the unique symbols of all fault types in the same line segment to obtain a symbol combination; The type weight of each fault type is assigned to the unique symbol corresponding to the symbol combination to obtain the fault label.

6. The intelligent maintenance method for the intelligent distribution network according to claim 1, characterized in that: After extracting the measurement data according to the detection indicators under each dimension, it includes: The conventional filtering method is used to filter the extracted data under each detection index respectively; Determine the reference value of the filtered data based on preset judgment conditions; ； in, Indicates the reference value of the corresponding filtered data; Indicates the number of preset judgment conditions; represents the judgment weight of the i1th preset judgment condition; Indicates the satisfaction coefficient of the extracted data before filtering based on the i1th preset judgment condition; Indicates the satisfaction coefficient of the filtered data based on the i1th preset judgment condition; Represents the data before filtering With the filtered data Semantic similarity coefficient based on the i1th preset judgment condition; Indicates all The variance of Respectively represent the judgment function, when hour, The value of is 0; when hour, The value of is 1; when hour, The value of is 1; when hour, The value of is 0; The filtered data whose reference value is greater than the preset value is retained; Otherwise, switch the data filtering mode to re-filter the extracted data under the corresponding detection index.

7. The intelligent maintenance method for the intelligent distribution network according to claim 1, characterized in that: The section loss coefficient of the corresponding line under each detection index is obtained, including: Retrieving a data analysis model matching the detection indicator from an indicator-analysis database; Input the filtered data under the corresponding detection index into the corresponding data analysis model to obtain the corresponding section loss coefficient.

8. The intelligent maintenance method for the intelligent distribution network according to claim 1, characterized in that: The preset detection methods correspond to the fault types one by one; Each detection indicator corresponds to a preset detection method.

Images