CN114597886A - Power distribution network operation state evaluation method based on interval type two fuzzy clustering analysis - Google Patents
Power distribution network operation state evaluation method based on interval type two fuzzy clustering analysis Download PDFInfo
- Publication number
- CN114597886A CN114597886A CN202111468189.3A CN202111468189A CN114597886A CN 114597886 A CN114597886 A CN 114597886A CN 202111468189 A CN202111468189 A CN 202111468189A CN 114597886 A CN114597886 A CN 114597886A
- Authority
- CN
- China
- Prior art keywords
- data
- cluster
- power distribution
- clustering
- distribution network
- Prior art date
- Legal status (The legal status is an assumption and is not a legal conclusion. Google has not performed a legal analysis and makes no representation as to the accuracy of the status listed.)
- Pending
Links
Images
Classifications
-
- H—ELECTRICITY
- H02—GENERATION; CONVERSION OR DISTRIBUTION OF ELECTRIC POWER
- H02J—CIRCUIT ARRANGEMENTS OR SYSTEMS FOR SUPPLYING OR DISTRIBUTING ELECTRIC POWER; SYSTEMS FOR STORING ELECTRIC ENERGY
- H02J3/00—Circuit arrangements for ac mains or ac distribution networks
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06F—ELECTRIC DIGITAL DATA PROCESSING
- G06F18/00—Pattern recognition
- G06F18/20—Analysing
- G06F18/23—Clustering techniques
- G06F18/232—Non-hierarchical techniques
- G06F18/2321—Non-hierarchical techniques using statistics or function optimisation, e.g. modelling of probability density functions
- G06F18/23213—Non-hierarchical techniques using statistics or function optimisation, e.g. modelling of probability density functions with fixed number of clusters, e.g. K-means clustering
-
- H—ELECTRICITY
- H02—GENERATION; CONVERSION OR DISTRIBUTION OF ELECTRIC POWER
- H02J—CIRCUIT ARRANGEMENTS OR SYSTEMS FOR SUPPLYING OR DISTRIBUTING ELECTRIC POWER; SYSTEMS FOR STORING ELECTRIC ENERGY
- H02J2203/00—Indexing scheme relating to details of circuit arrangements for AC mains or AC distribution networks
- H02J2203/10—Power transmission or distribution systems management focussing at grid-level, e.g. load flow analysis, node profile computation, meshed network optimisation, active network management or spinning reserve management
-
- Y—GENERAL TAGGING OF NEW TECHNOLOGICAL DEVELOPMENTS; GENERAL TAGGING OF CROSS-SECTIONAL TECHNOLOGIES SPANNING OVER SEVERAL SECTIONS OF THE IPC; TECHNICAL SUBJECTS COVERED BY FORMER USPC CROSS-REFERENCE ART COLLECTIONS [XRACs] AND DIGESTS
- Y04—INFORMATION OR COMMUNICATION TECHNOLOGIES HAVING AN IMPACT ON OTHER TECHNOLOGY AREAS
- Y04S—SYSTEMS INTEGRATING TECHNOLOGIES RELATED TO POWER NETWORK OPERATION, COMMUNICATION OR INFORMATION TECHNOLOGIES FOR IMPROVING THE ELECTRICAL POWER GENERATION, TRANSMISSION, DISTRIBUTION, MANAGEMENT OR USAGE, i.e. SMART GRIDS
- Y04S10/00—Systems supporting electrical power generation, transmission or distribution
- Y04S10/50—Systems or methods supporting the power network operation or management, involving a certain degree of interaction with the load-side end user applications
Landscapes
- Engineering & Computer Science (AREA)
- Data Mining & Analysis (AREA)
- Physics & Mathematics (AREA)
- Theoretical Computer Science (AREA)
- Computer Vision & Pattern Recognition (AREA)
- Artificial Intelligence (AREA)
- Bioinformatics & Cheminformatics (AREA)
- Bioinformatics & Computational Biology (AREA)
- Life Sciences & Earth Sciences (AREA)
- Evolutionary Biology (AREA)
- Evolutionary Computation (AREA)
- General Engineering & Computer Science (AREA)
- General Physics & Mathematics (AREA)
- Probability & Statistics with Applications (AREA)
- Power Engineering (AREA)
- Supply And Distribution Of Alternating Current (AREA)
Abstract
The invention relates to a method for evaluating the running state of a power distribution network based on interval type two fuzzy clustering analysis, which is characterized in that a data imbalance factor is brought into a cluster center updating calculation function through an interval type two c mean value fuzzy clustering method based on the local fuzzy measurement of a boundary area, so that the cluster center is related to not only a membership function of an unbalanced data set, but also the imbalance among clusters; the data samples in the boundary area are subjected to calculation analysis and clustering processing through an interval two-type fuzzy c-means clustering algorithm, an optimized clustering center updating function considering the local fuzzy measurement of the boundary area is introduced, and the clustering effect of the interval two-type fuzzy c-means clustering method on the unbalanced operation data of the power distribution network is improved. The method and the device use the improved clustering algorithm for analyzing the unbalanced data frequently alarmed by the power distribution network, evaluate the running state of the power distribution network, judge the types of the alarming faults and predict the possibility of the future alarming of the power distribution network.
Description
Technical Field
The invention belongs to the technical field of power distribution networks, and particularly relates to a power distribution network operation state evaluation method based on interval two-type fuzzy clustering analysis.
Background
At present, the scale of a power distribution network in China is rapidly enlarged, a complex power distribution framework based on the characteristics of distribution, multiple loads, multiple power supplies and the like is formed, the requirement of a terminal user on the reliability of power supply is gradually improved, but under the condition that the power distribution network fails, power failure can cause great economic loss and the social stability is influenced. Therefore, the most important task at present is to accurately and quickly analyze the data of the power distribution network, adopt a reasonable and effective maintenance strategy and ensure the stable operation of a national power supply system. By comprehensively aggregating and analyzing the data of the power distribution network, the operation state of the power distribution network can be effectively evaluated, and the change of the operation state of the power distribution network and potential safety risks can be found.
The power distribution network is an important link for connecting the power transmission network and users, the running state of the power distribution network directly influences the power supply reliability of the power system, a small amount of abnormal data still exist in high-stability normal running data of the power distribution network, and the two kinds of data jointly form unbalanced data of the power distribution network. For the processing of unbalanced data sets, the existing research mainly focuses on the data preprocessing level, that is, the unbalanced data sets are converted into approximately balanced data sets through technologies such as "oversampling" and "undersampling", and then are analyzed by the existing algorithm.
The oversampling technology generates new data samples by continuously interpolating the data samples of the minority class clusters, increases the scale of the minority class clusters and reduces the class cluster imbalance degree; the undersampling technology reduces samples of most clusters and reduces imbalance degree of cluster scale by randomly selecting samples of most clusters. Although the problem of cluster-like size imbalance can be solved to some extent by methods of data preprocessing such as sampling, these methods inevitably lead to data overfitting or information loss.
Disclosure of Invention
The invention aims to overcome the defects of the prior art, provides a distribution network operation state evaluation method based on interval two-type fuzzy clustering analysis, and can be used for listing data with states difficult to distinguish into boundary areas by introducing the concept of a rough set, reminding operation and maintenance personnel to pay attention to the boundary data, and improving the clustering precision and the state evaluation precision.
The technical problem to be solved by the invention is realized by adopting the following technical scheme:
the method for evaluating the running state of the power distribution network based on interval type two fuzzy clustering analysis comprises the following steps:
step 1, collecting unbalance monitoring data of a power distribution station, setting threshold parameters of common fault types in the unbalance monitoring data of the power distribution according to a common fault index system of a power system, and simultaneously using the collected unbalance monitoring data as data samples;
step 2, randomly selecting data in the unbalanced monitoring data as an initial clustering center in the iterative process of the interval type two fuzzy c-means clustering algorithm according to the interval type two fuzzy c-means clustering algorithm, and setting clustering analysis parameters according to the characteristics of historical data;
step 3, calculating Euclidean distances between each data sample in the unbalanced monitoring data and an initial clustering center, dividing the data samples in the unbalanced monitoring data into a lower approximate set or a boundary region of a normal cluster or an abnormal cluster according to the Euclidean distances, and calculating the unbalance degree between the clusters;
step 4, substituting the cluster unbalance degree obtained in the step 3 into an optimized cluster center updating formula based on interval two-type fuzzy cluster analysis to perform iterative calculation, calculating a cluster center, and determining the cluster to which the sample belongs by calculating the membership degree;
step 5, comparing the clustering centers and the clusters obtained by calculation in the step 4 with the clustering centers and the clusters of the previous iteration, counting samples of an approximate set and a boundary area on each cluster if the clustering centers and the clusters are not updated any more, and evaluating the running state of the power distribution network; otherwise, returning to the step 3.
Moreover, the specific implementation method of the step 2 is as follows: and randomly selecting two kinds of data in the unbalanced detection data as the initial clustering centers, wherein one group of data is used as the initial clustering center of the normal cluster of the power distribution network, and the other group of data is used as the initial clustering center of the abnormal cluster of the power distribution network, and setting a distance judgment threshold value and a fuzzy coefficient according to the historical data characteristics of the actual operation record of the power distribution network system.
Further, the step 3 includes the steps of:
step 3.1, calculating a first Euclidean distance between each data sample in the unbalanced monitoring data set and the normal clustering center in the step 2 and a second Euclidean distance between each data sample and the abnormal clustering center in the step 2, judging the size of the first Euclidean distance and the second Euclidean distance, and acquiring the ratio of a larger numerical value to a smaller numerical value of the first Euclidean distance and the second Euclidean distance;
step 3.2, comparing the ratio obtained in the step 3.1 with a distance judgment threshold, and if the ratio is greater than the distance judgment threshold, dividing the data sample into a lower approximate set of the corresponding class cluster with a smaller Euclidean distance; otherwise, dividing the image into boundary areas;
step 3.3, respectively calculating the ratio of the sample number of the upper approximate set in the normal cluster and the abnormal cluster to the sample number of all the upper approximate sets in the imbalance monitoring data to obtain the imbalance degree f between the normal cluster and the abnormal cluster:
wherein the content of the first and second substances,for the number of approximate set samples on the minority class clusters of the cross class clusters in the current loop iteration,the number of samples of the approximation set on most class clusters that are cross class clusters.
Moreover, the specific implementation method for calculating the clustering center in the step 4 is as follows: substituting the unbalance degree calculated in the step 3 into an optimized clustering center v based on the interval two-type fuzzy clustering analysisi:
Wherein v isiCluster center for ith iteration, ωlTo approximate the weighting factor, ωbIs a lower approximate weighting coefficient, f is the degree of imbalance, m is the blur coefficient, XijIs a data sample, xjIs a sample, aijIs a fuzzy membership degree of two types, iCin order to approximate the region data set for the next time,is a boundary region data set.
Moreover, the specific implementation method for calculating the belonged cluster of the membership grade determination sample in the step 4 is as follows:
wherein, muijIs the degree of membership,is muijThe upper degree of membership of (a) is,μ ijis muijLower degree of membership, distance djiCluster center v for the ith iterationiAnd sample xjA distance d between themziRepresenting cluster center v of the ith iterationiAnd sample data sample xzThe distance between the clusters, k is the number of the clusters,C iin order to approximate the region data set for the next time,is a boundary region data set; for sample xjRelative to class CiFuzzy membership degree of (2)Comprises the following steps:
μ i(xj)=min{μij(m1),μij(m2)}
wherein, muij(m1) And muij(m2) When the fuzzy coefficient m is equal to m1And m ═ m2When xjRelative to class CiIs measured by the linear fuzzy membership, sample xjTo cluster CiFinal degree of membership beta ofijComprises the following steps:
wherein N isiIs a cluster CiN is the total number of samples, based on the final degree of membership betaijAnd determining the class cluster to which all sample data belongs.
Further, the step 5 includes the steps of:
step 5.1, performing iterative update calculation on a clustering center according to the dividing and clustering result of the data samples in the step 3;
step 5.2, judging whether the clustering center is updated, if the clustering center is not updated, performing step 5.3, otherwise, returning to step 3;
step 5.3, counting approximate set samples under the normal cluster, determining that the samples belong to normal data, marking 0 for the data, and indicating that the data correspond to the power distribution network operation system and no fault of the type occurs; counting approximate set samples under the abnormal cluster, determining that the samples belong to fault samples, and marking the samples with '1' to indicate that the type of fault occurs in the power distribution network operation system; the boundary area samples are counted and marked with a "2" indicating that the power distribution network operating system is likely to have the type of fault in the future.
The invention has the advantages and positive effects that:
1. according to the interval type c mean value fuzzy clustering method based on the boundary region local fuzzy measurement, data unbalance factors are brought into a cluster center updating calculation function, so that the cluster center is related to a membership function of an unbalanced data set and is related to the unbalance among clusters; the data samples in the boundary area are subjected to calculation analysis and clustering processing through an interval two-type fuzzy c-means clustering algorithm, and an optimized clustering center updating function considering the local fuzzy measurement of the boundary area is introduced, so that the clustering effect of the interval two-type fuzzy c-means clustering method on the unbalanced operation data of the power distribution network is improved. The improved aggregation clustering algorithm is used for analyzing the unbalanced data of the frequent alarm of the power distribution network and evaluating the running state of the power distribution network, so that the fault type of the alarm can be judged, and the possibility of the future alarm of the power distribution network can be predicted.
2. According to the invention, on the basis of considering the local fuzzy measurement of the boundary region, the cluster center updating formula of the interval two-type fuzzy c-means algorithm is optimized, so that the adverse effect of the boundary region occupied by most clusters on the clustering effect of few clusters can be reduced, the cluster center of small-scale clusters is always maintained at a more ideal position, the phenomenon that data originally belonging to most clusters are mistakenly divided into at least a plurality of clusters can be inhibited, the data characteristics of the few clusters can be better kept, therefore, the clustering performance of the algorithm on unbalanced data can be improved, and the accuracy and rapidity of data aggregation of a power distribution network can be actually improved; not only has higher academic research significance, but also has strong engineering application value.
Drawings
FIG. 1 is a diagram of an evaluation architecture of the present invention;
FIG. 2 is a flow chart of an evaluation method of the present invention;
FIG. 3 is a diagram of a dimension-reduced data space distribution diagram according to the present invention
FIG. 4 is a diagram of the conventional fuzzy c-means clustering results
FIG. 5 is a two-section fuzzy c-means clustering result chart according to the present invention.
Detailed Description
The present invention will be described in further detail with reference to the accompanying drawings.
The method for evaluating the running state of the power distribution network based on the interval type two fuzzy clustering analysis is shown in fig. 1 and 2 and comprises the following steps:
step 1, collecting unbalance monitoring data of a power distribution station, setting threshold parameters of common fault types in the unbalance monitoring data of the power distribution according to a common fault index system of a power system, and taking the collected unbalance monitoring data as data samples.
According to a common fault index system of the power system, different types of abnormal fault data compared with normal operation data of the power distribution network are screened from unbalanced power monitoring data, and threshold parameters required by all faults are analyzed. Common alarm information of the power distribution system is shown in table 1, and alarm thresholds of each monitored variable of the power distribution system are shown in table 2.
Table 1 common alarm list for power distribution system
TABLE 2 commonly used monitoring variables for distribution systems
Step 2, applying an interval type two fuzzy c-means algorithm, randomly selecting two groups of data in the unbalanced monitoring data as initial clustering centers in an interval type two fuzzy c-means algorithm iteration process, and setting parameters related to a clustering analysis algorithm according to historical data characteristics; the data of the initial clustering centers are selected randomly, namely two groups of data in the unbalanced monitoring data are selected randomly, one group of data is used as the initial clustering centers of the normal state clusters of the power distribution network, and the other group of data is used as the initial clustering centers of the abnormal clusters of the power distribution network; and setting a distance judgment threshold value according to the historical data characteristics of the actual operation records of the power distribution network systemAnd a blurring coefficient m1=2,m2=10。
Step 3, calculating Euclidean distances between each sample in the unbalanced data and an initial clustering center, dividing the data samples in the unbalanced monitoring data into a lower approximate set or a boundary region of a normal or abnormal cluster according to the Euclidean distances, and calculating the unbalance degree between the clusters, wherein the method comprises the following steps:
step 3.1, calculating a first Euclidean distance between each data sample in the unbalanced data set and a normal cluster center and a second Euclidean distance between each data sample in the unbalanced data set and an abnormal cluster center, judging the size of the first Euclidean distance and the size of the second Euclidean distance, and acquiring the ratio of a larger numerical value to a smaller numerical value of the first Euclidean distance and the second Euclidean distance;
step 3.2, comparing the obtained ratio with a distance judgment threshold, and if the ratio is greater than the distance judgment threshold, dividing the data sample into a lower approximate set of the corresponding class cluster of the smaller Euclidean distance in the first Euclidean distance and the second Euclidean distance; otherwise, dividing the image into boundary areas;
step 3.3, respectively calculating the ratio of the sample number of the upper approximate set in the normal cluster and the abnormal cluster to the sample number of all the upper approximate sets in the imbalance monitoring data to obtain the imbalance degree f between the normal cluster and the abnormal cluster:
wherein the content of the first and second substances,for the number of approximate set samples under the minority class cluster of the cross class cluster in the current loop iteration,the number of samples of the approximation set under most class clusters that are cross class clusters.
Step 4, substituting the cluster unbalance degree calculated in the step 3 into the optimized clustering center viUpdating formula to iteratively calculate clustering center vi:
Wherein v isiCluster center for ith iteration, ωlTo approximate the weighting factor, ωbIs a lower approximate weighting coefficient, f is the degree of imbalance, m is the blur coefficient, XijAs data samples, xjIs a sample, aijIs a fuzzy membership degree of two types, iCin order to approximate the region data set for the next time,is a boundary region data set.
Aiming at the influence of unbalanced class cluster scale on the clustering result of the interval type two fuzzy c-means algorithm, the invention firstly provides a clustering center updating method based on interval type two fuzzy clustering analysis. And optimizing an iterative center updating formula under the condition of considering the imbalance of the boundary area samples, and inhibiting the phenomenon that the clustering center of the minority class is shifted due to the traction of the boundary area data. In the optimized cluster center updating formula, the larger the unbalance degree of the data set is, the smaller the iteration center coefficient f of the boundary area is, and the smaller the contribution weight of the boundary area part to the cluster center is, so that the aim of inhibiting the cluster center from deviating to the direction of most clusters is fulfilled.
Determining the belonged cluster of the sample by calculating the membership:
wherein, muijIs the degree of membership,is muijThe upper degree of membership of (a) is,μ ijis muijLower degree of membership, distance djiCluster center v for the ith iterationiAnd sample xjA distance d between themziRepresenting cluster center v of the ith iterationiAnd sample data sample xzThe distance between the clusters, k is the number of the clusters, iCin order to approximate the region data set for the next time,is a boundary region data set.
μ i(xj)=min{μij(m1),μij(m2)}
wherein, muij(m1) And muij(m2) When the fuzzy coefficient m is equal to m1And m ═ m2When xjRelative to class CiIs measured by the linear fuzzy membership, sample xjTo cluster CiFinal degree of membership beta ofijComprises the following steps:
wherein N isiAre like cluster CiN is the total number of samples, based on the final degree of membership betaijAnd determining the class cluster to which all sample data belongs.
Step 5, comparing the clustering centers and the clusters obtained by calculation in the step 4 with the clustering centers and the clusters of the previous iteration, if the clustering centers and the clusters are not updated any more, counting samples of approximate sets and boundary areas on the clusters, and evaluating the running state of the power distribution network; otherwise, returning to the step 3.
Step 5.1, performing iterative update calculation on a clustering center according to the dividing and clustering result of the data samples in the step 3;
step 5.2, judging whether the clustering center is updated, if the clustering center is not updated, performing step 5.3, otherwise, returning to step 3;
step 5.3, counting approximate set samples under the normal cluster, determining that the samples belong to normal data, marking 0 for the data, and indicating that the data correspond to the power distribution network operation system and no fault of the type occurs; counting approximate set samples under the abnormal cluster, determining that the samples belong to fault samples, and marking the samples with '1' to indicate that the type of fault occurs in the power distribution network operation system; the boundary area samples are counted and marked with a "2" indicating that the power distribution network operating system is likely to have the type of fault in the future.
The invention adopts an interval two-type fuzzy c-means algorithm to divide the running state of the power distribution network into a normal state and an abnormal state. The threshold corresponding to each alarm type in the table 1 is extracted from the original data, and then the interval type two fuzzy c-means algorithm based on the unbalanced data clustering proposed by the invention is used for carrying out clustering analysis on the power distribution unbalanced monitoring data. And (4) performing iterative update calculation on the clustering center according to the clustering result of the data sample in the step (3). If the cluster center is not updated any more, counting samples of the lower approximation set and the boundary area in the corresponding cluster, and evaluating the running state of the power distribution network: counting approximate set samples under the normal cluster, determining that the samples belong to normal data, marking the data with '0', and indicating that the data does not have the type of fault corresponding to the power distribution network operation system; counting approximate set samples under the fault cluster, determining that the samples belong to fault samples, and marking the samples with '1' to indicate that the type of fault occurs in the power distribution network operation system; the boundary area samples are counted and marked with a "2" indicating that the power distribution network operating system is likely to have the type of fault in the future. And if the cluster center is continuously updated, returning to the step 3.
Expressed by the formula: let data set U, U ═ Xz1, a, N, dividing a data object set U into k class clusters; initializing a cluster center viDistance judgment threshold valueAnd a blurring coefficient m1,m2。
For each object XzCalculating XzTo respective cluster center point viOf Euclidean distance dijSelect o ═ j | djz=min({diz}) and i ═ 1,. k, ifThenAnd isOtherwise xz∈ jCAnd, for all the clusters of the class,wherein d isizAs the centre of clustering viTo data object XzDistance of dj,zAs the centre of clustering vj'To data object XzDistance of d, djzAs the cluster center vjTo data object XzO and o 'are sets of data j and j' respectively satisfying the condition,in order to determine the threshold value for the distance, jCandrespectively, a boundary region of a coarse set, a lower approximation region and an upper approximation region.
Computing an approximation set on each class clusterNumber of samples | CjAnd if the cluster does not change any more, terminating the algorithm, otherwise, continuing to update the iteration.
According to the method for evaluating the running state of the power distribution network based on the interval two-type fuzzy clustering analysis, a relevant experiment is carried out on unbalanced monitoring data of a certain power distribution station:
according to the method for evaluating the running state of the power distribution network, the thought of a rough set is fully utilized, and unbalanced data of the power distribution network are clustered into majority normal data and minority abnormal data; if the group of monitoring data is determined to belong to the lower approximate set of the few types of abnormal data, determining the group of data as the occurred fault, and performing real-time alarm processing; if the group of data belongs to the boundary area of the minority class cluster, the group of data belongs to the possible faults, and prediction alarm is carried out, so that the state evaluation of the whole power distribution network operation system is achieved. The state evaluation method can determine the time and the place of the occurred fault, can predict the possible fault of the data in the abnormal boundary area, and is more advanced compared with the traditional state diagnosis method. The method provided by the invention can be deployed in a monitoring analysis machine room of a hub power distribution station, and can be used for analyzing various monitoring data of the power distribution room in time and assisting operation and maintenance personnel to find state changes and potential safety hazards in time.
If the detection data of a certain low-voltage distribution equipment is shown in table 3:
TABLE 3 example of certain Low Voltage distribution Equipment monitoring data
In order to more intuitively show the effect of the algorithm, the Principal feature projection in the data is found out by Principal Component Analysis (PCA), noise and redundancy are eliminated, and the 6-dimensional sample data selected in table 3 is reduced to a 3-dimensional space, and the result is shown in fig. 3, wherein the pattern "star" represents a few types of abnormal data, and the pattern "□" represents a majority types of safe operation data.
The experimental results are as follows:
the effect graphs of the conventional fuzzy c-means algorithm and the interval type-II c-means algorithm are shown in FIG. 4 and FIG. 5, respectively. In the figure, patterns "four" and "□" represent samples that are clustered correctly; the pattern "+" represents a sample originally belonging to an approximate region under the majority class cluster which is wrongly divided into the minority class cluster; the pattern "o" represents samples divided into boundary regions.
According to the experimental test results of 20 groups of data under the state evaluation model provided by the invention, although the traditional fuzzy c-means clustering method introduces the concept of rough concentration about upper and lower approximation and divides some unpredictable samples into the boundary space, the algorithm does not consider that the cluster size imbalance can influence the clustering result, the clustering effect is not ideal for the data aggregation clustering effect of cluster size imbalance, and the results of the state evaluation of the power distribution network are not ideal because 2 groups of most cluster samples in the clustering result of the traditional fuzzy c-means clustering algorithm are clustered by mistake at least into several clusters and 5 groups of most cluster samples are clustered by mistake into the boundary space. The method for evaluating the running state of the power distribution network based on the interval two-type fuzzy clustering algorithm, which is provided in the section, optimizes the cluster center updating formula on the basis of considering the imbalance of the cluster scales, and the improved algorithm is more suitable for the aggregate clustering analysis of the unbalanced running data of the power distribution system than the traditional algorithm. In the clustering result of the interval type two fuzzy c-means clustering method, only one group of most cluster data is clustered by errors to at least a few clusters, and one group of most cluster data is clustered by errors to at least a few cluster boundary regions. Therefore, the evaluation method provided by the invention can effectively perform cluster analysis on the unbalanced monitoring data generated by the power distribution network equipment, and then perform power distribution network operation state evaluation, thereby effectively improving the speed of power distribution network system operation alarm and the accuracy of fault prediction.
It should be emphasized that the embodiments described herein are illustrative rather than restrictive, and thus the present invention is not limited to the embodiments described in the detailed description, but also includes other embodiments that can be derived from the technical solutions of the present invention by those skilled in the art.
Claims (6)
1. The method for evaluating the running state of the power distribution network based on interval two-type fuzzy clustering analysis is characterized by comprising the following steps of: the method comprises the following steps:
step 1, collecting unbalance monitoring data of a power distribution station, setting threshold parameters of common fault types in the unbalance monitoring data of the power distribution according to a common fault index system of a power system, and simultaneously using the collected unbalance monitoring data as data samples;
step 2, randomly selecting data in the unbalanced monitoring data as an initial clustering center in the iterative process of the interval type two fuzzy c-means clustering algorithm according to the interval type two fuzzy c-means clustering algorithm, and setting clustering analysis parameters according to the characteristics of historical data;
step 3, calculating Euclidean distances between each data sample in the unbalanced monitoring data and an initial clustering center, dividing the data samples in the unbalanced monitoring data into a lower approximate set or a boundary region of a normal cluster or an abnormal cluster according to the Euclidean distances, and calculating the unbalanced degree between the clusters;
step 4, substituting the cluster unbalance degree obtained in the step 3 into an optimized cluster center updating formula based on interval two-type fuzzy cluster analysis to perform iterative calculation, calculating a cluster center, and determining the cluster to which the sample belongs by calculating the membership degree;
step 5, comparing the clustering centers and the clusters obtained by calculation in the step 4 with the clustering centers and the clusters of the previous iteration, counting samples of an approximate set and a boundary area on each cluster if the clustering centers and the clusters are not updated any more, and evaluating the running state of the power distribution network; otherwise, returning to the step 3.
2. The method for evaluating the operating state of the power distribution network based on the interval type two fuzzy clustering analysis according to claim 1, wherein: the specific implementation method of the step 2 comprises the following steps: and randomly selecting two kinds of data in the unbalanced detection data as the initial clustering centers, wherein one group of data is used as the initial clustering center of the normal cluster of the power distribution network, and the other group of data is used as the initial clustering center of the abnormal cluster of the power distribution network, and setting a distance judgment threshold value and a fuzzy coefficient according to the historical data characteristics of the actual operation record of the power distribution network system.
3. The method for evaluating the operating state of the power distribution network based on the interval type two fuzzy clustering analysis according to claim 1, wherein: the step 3 comprises the following steps:
step 3.1, calculating a first Euclidean distance between each data sample in the unbalanced monitoring data set and the normal clustering center in the step 2 and a second Euclidean distance between each data sample and the abnormal clustering center in the step 2, judging the size of the first Euclidean distance and the second Euclidean distance, and acquiring the ratio of a larger numerical value to a smaller numerical value of the first Euclidean distance and the second Euclidean distance;
step 3.2, comparing the ratio obtained in the step 3.1 with a distance judgment threshold, and if the ratio is greater than the distance judgment threshold, dividing the data sample into a lower approximate set of the corresponding class cluster with a smaller Euclidean distance; otherwise, dividing the image into boundary areas;
step 3.3, respectively calculating the ratio of the sample number of the upper approximate set in the normal cluster and the abnormal cluster to the sample number of all the upper approximate sets in the imbalance monitoring data to obtain the imbalance degree f between the normal cluster and the abnormal cluster:
4. The method for evaluating the operating state of the power distribution network based on the interval type two fuzzy cluster analysis according to claim 1, wherein the method comprises the following steps: the specific implementation method for calculating the clustering center in the step 4 comprises the following steps: substituting the unbalance degree calculated in the step 3 into an optimized clustering center v based on the interval two-type fuzzy clustering analysisi:
Wherein v isiCluster center for ith iteration, ωlTo approximate the weighting factor, ωbIs a lower approximate weighting coefficient, f is the degree of imbalance, m is the blur coefficient, XijIs a data sample, xjAs a sample, aijIs a fuzzy membership degree of two types, iCin order to approximate the region data set for the next time,is a boundary region data set.
5. The method for evaluating the operating state of the power distribution network based on the interval type two fuzzy clustering analysis according to claim 1, wherein: the specific implementation method for calculating the membership degree to determine the belonged cluster of the sample in the step 4 comprises the following steps:
wherein, muijIs the degree of membership,is muijThe upper degree of membership of (a) is,μ ijis muijLower degree of membership, distance djiCluster center v for the ith iterationiAnd sample xjA distance d between themziRepresenting cluster center v of the ith iterationiAnd sample data sample xzThe distance between the clusters, k is the number of the clusters, iCfor the purpose of the lower approximation area data set,is a boundary region data set; for sample xjRelative toClass CiFuzzy membership degree of (2)Comprises the following steps:
μ i(xj)=min{μij(m1),μij(m2)}
wherein, muij(m1) And muij(m2) When the fuzzy coefficient m is equal to m1And m ═ m2When xjRelative to class CiIs measured in a fuzzy membership metric of one type, sample xjTo cluster CiFinal degree of membership beta ofijComprises the following steps:
wherein N isiIs a cluster CiN is the total number of samples, based on the final degree of membership betaijAnd determining the class cluster to which all sample data belongs.
6. The method for evaluating the operating state of the power distribution network based on the interval type two fuzzy clustering analysis according to claim 1, wherein: the step 5 comprises the following steps:
step 5.1, performing iterative update calculation on a clustering center according to the dividing and clustering result of the data samples in the step 3;
step 5.2, judging whether the clustering center is updated, if the clustering center is not updated, performing step 5.3, otherwise, returning to step 3;
step 5.3, counting samples of an approximate set under the normal cluster, determining that the samples belong to normal data, marking the data with 0, and indicating that the data correspond to a power distribution network operation system and no fault of the type occurs; counting approximate set samples under the abnormal cluster, determining that the samples belong to fault samples, and marking the samples with '1' to indicate that the type of fault occurs in the power distribution network operation system; the boundary area samples are counted and marked with a "2" indicating that the power distribution network operating system is likely to have the type of fault in the future.
Priority Applications (1)
Application Number | Priority Date | Filing Date | Title |
---|---|---|---|
CN202111468189.3A CN114597886A (en) | 2021-12-03 | 2021-12-03 | Power distribution network operation state evaluation method based on interval type two fuzzy clustering analysis |
Applications Claiming Priority (1)
Application Number | Priority Date | Filing Date | Title |
---|---|---|---|
CN202111468189.3A CN114597886A (en) | 2021-12-03 | 2021-12-03 | Power distribution network operation state evaluation method based on interval type two fuzzy clustering analysis |
Publications (1)
Publication Number | Publication Date |
---|---|
CN114597886A true CN114597886A (en) | 2022-06-07 |
Family
ID=81814459
Family Applications (1)
Application Number | Title | Priority Date | Filing Date |
---|---|---|---|
CN202111468189.3A Pending CN114597886A (en) | 2021-12-03 | 2021-12-03 | Power distribution network operation state evaluation method based on interval type two fuzzy clustering analysis |
Country Status (1)
Country | Link |
---|---|
CN (1) | CN114597886A (en) |
Cited By (1)
Publication number | Priority date | Publication date | Assignee | Title |
---|---|---|---|---|
CN116153437A (en) * | 2023-04-19 | 2023-05-23 | 乐百氏(广东)饮用水有限公司 | Water quality safety evaluation and water quality prediction method and system for drinking water source |
-
2021
- 2021-12-03 CN CN202111468189.3A patent/CN114597886A/en active Pending
Cited By (2)
Publication number | Priority date | Publication date | Assignee | Title |
---|---|---|---|---|
CN116153437A (en) * | 2023-04-19 | 2023-05-23 | 乐百氏(广东)饮用水有限公司 | Water quality safety evaluation and water quality prediction method and system for drinking water source |
CN116153437B (en) * | 2023-04-19 | 2023-06-30 | 乐百氏(广东)饮用水有限公司 | Water quality safety evaluation and water quality prediction method and system for drinking water source |
Similar Documents
Publication | Publication Date | Title |
---|---|---|
CN109816031B (en) | Transformer state evaluation clustering analysis method based on data imbalance measurement | |
CN105677791B (en) | For analyzing the method and system of the operation data of wind power generating set | |
CN113344134B (en) | Low-voltage distribution monitoring terminal data acquisition abnormality detection method and system | |
CN108053148B (en) | Efficient fault diagnosis method for power information system | |
CN112911627B (en) | Wireless network performance detection method, device and storage medium | |
CN115982602B (en) | Photovoltaic transformer electrical fault detection method | |
CN109389325B (en) | Method for evaluating state of electronic transformer of transformer substation based on wavelet neural network | |
CN109165242B (en) | Fault diagnosis and early warning method based on entropy sorting and space-time analysis | |
CN111538311B (en) | Flexible multi-state self-adaptive early warning method and device for mechanical equipment based on data mining | |
CN115130578A (en) | Incremental rough clustering-based online evaluation method for state of power distribution equipment | |
CN110995153B (en) | Abnormal data detection method and device for photovoltaic power station and electronic equipment | |
CN110543907A (en) | fault classification method based on microcomputer monitoring power curve | |
CN115454778A (en) | Intelligent monitoring system for abnormal time sequence indexes in large-scale cloud network environment | |
CN109240276B (en) | Multi-block PCA fault monitoring method based on fault sensitive principal component selection | |
CN113551927A (en) | Mechanical equipment fault early warning method and system based on vibration signals | |
CN114244751A (en) | Wireless sensor network anomaly detection method and system | |
CN112464439A (en) | Three-layer data mining-based power system fault type analysis method and device | |
CN110443481B (en) | Power distribution automation terminal state evaluation system and method based on hybrid K-nearest neighbor algorithm | |
CN116295948A (en) | Abnormality detection method, system and storage medium of industrial temperature sensor in large temperature difference environment | |
CN115526258A (en) | Power system transient stability evaluation method based on Spearman correlation coefficient feature extraction | |
CN117318614A (en) | Photovoltaic inverter fault prediction method | |
CN114597886A (en) | Power distribution network operation state evaluation method based on interval type two fuzzy clustering analysis | |
CN113608968B (en) | Power dispatching monitoring data anomaly detection method based on density distance comprehensive decision | |
CN109359742B (en) | Method for generating preventive maintenance period of subway subsystem | |
CN112949735A (en) | Liquid hazardous chemical substance volatile concentration abnormity discovery method based on outlier data mining |
Legal Events
Date | Code | Title | Description |
---|---|---|---|
PB01 | Publication | ||
PB01 | Publication | ||
SE01 | Entry into force of request for substantive examination | ||
SE01 | Entry into force of request for substantive examination |