CN110348005B - Distribution network equipment state data processing method and device, computer equipment and medium - Google Patents

Distribution network equipment state data processing method and device, computer equipment and medium Download PDF

Info

Publication number
CN110348005B
CN110348005B CN201910443912.9A CN201910443912A CN110348005B CN 110348005 B CN110348005 B CN 110348005B CN 201910443912 A CN201910443912 A CN 201910443912A CN 110348005 B CN110348005 B CN 110348005B
Authority
CN
China
Prior art keywords
processing
neural network
network model
distribution network
network equipment
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.)
Active
Application number
CN201910443912.9A
Other languages
Chinese (zh)
Other versions
CN110348005A (en
Inventor
郝方舟
马捷然
陈杰锋
罗林欢
晏小卉
沈超
雷超平
孙奇珍
林翔
Current Assignee (The listed assignees may be inaccurate. Google has not performed a legal analysis and makes no representation or warranty as to the accuracy of the list.)
Guangzhou Power Supply Bureau of Guangdong Power Grid Co Ltd
Original Assignee
Guangzhou Power Supply Bureau of Guangdong Power Grid Co Ltd
Priority date (The priority date 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 date listed.)
Filing date
Publication date
Application filed by Guangzhou Power Supply Bureau of Guangdong Power Grid Co Ltd filed Critical Guangzhou Power Supply Bureau of Guangdong Power Grid Co Ltd
Priority to CN201910443912.9A priority Critical patent/CN110348005B/en
Publication of CN110348005A publication Critical patent/CN110348005A/en
Application granted granted Critical
Publication of CN110348005B publication Critical patent/CN110348005B/en
Active legal-status Critical Current
Anticipated expiration legal-status Critical

Links

Images

Classifications

    • G—PHYSICS
    • G06—COMPUTING; CALCULATING OR COUNTING
    • G06F—ELECTRIC DIGITAL DATA PROCESSING
    • G06F16/00—Information retrieval; Database structures therefor; File system structures therefor
    • G06F16/30—Information retrieval; Database structures therefor; File system structures therefor of unstructured textual data
    • G06F16/35—Clustering; Classification
    • G—PHYSICS
    • G06—COMPUTING; CALCULATING OR COUNTING
    • G06F—ELECTRIC DIGITAL DATA PROCESSING
    • G06F18/00—Pattern recognition
    • G06F18/20—Analysing
    • G06F18/21—Design or setup of recognition systems or techniques; Extraction of features in feature space; Blind source separation
    • G06F18/214—Generating training patterns; Bootstrap methods, e.g. bagging or boosting
    • G—PHYSICS
    • G06—COMPUTING; CALCULATING OR COUNTING
    • G06F—ELECTRIC DIGITAL DATA PROCESSING
    • G06F40/00—Handling natural language data
    • G06F40/20—Natural language analysis
    • G06F40/279—Recognition of textual entities
    • G06F40/289—Phrasal analysis, e.g. finite state techniques or chunking
    • G—PHYSICS
    • G06—COMPUTING; CALCULATING OR COUNTING
    • G06Q—INFORMATION AND COMMUNICATION TECHNOLOGY [ICT] SPECIALLY ADAPTED FOR ADMINISTRATIVE, COMMERCIAL, FINANCIAL, MANAGERIAL OR SUPERVISORY PURPOSES; SYSTEMS OR METHODS SPECIALLY ADAPTED FOR ADMINISTRATIVE, COMMERCIAL, FINANCIAL, MANAGERIAL OR SUPERVISORY PURPOSES, NOT OTHERWISE PROVIDED FOR
    • G06Q10/00—Administration; Management
    • G06Q10/06—Resources, workflows, human or project management; Enterprise or organisation planning; Enterprise or organisation modelling
    • G06Q10/063—Operations research, analysis or management
    • G—PHYSICS
    • G06—COMPUTING; CALCULATING OR COUNTING
    • G06Q—INFORMATION AND COMMUNICATION TECHNOLOGY [ICT] SPECIALLY ADAPTED FOR ADMINISTRATIVE, COMMERCIAL, FINANCIAL, MANAGERIAL OR SUPERVISORY PURPOSES; SYSTEMS OR METHODS SPECIALLY ADAPTED FOR ADMINISTRATIVE, COMMERCIAL, FINANCIAL, MANAGERIAL OR SUPERVISORY PURPOSES, NOT OTHERWISE PROVIDED FOR
    • G06Q50/00—Systems or methods specially adapted for specific business sectors, e.g. utilities or tourism
    • G06Q50/06—Electricity, gas or water supply
    • 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
    • H02J13/00—Circuit arrangements for providing remote indication of network conditions, e.g. an instantaneous record of the open or closed condition of each circuitbreaker in the network; Circuit arrangements for providing remote control of switching means in a power distribution network, e.g. switching in and out of current consumers by using a pulse code signal carried by the network

Abstract

The invention relates to a distribution network equipment state data processing method, a distribution network equipment state data processing device, computer equipment and a distribution network equipment state data processing medium. The method for processing the state data of the distribution network equipment comprises the following steps: acquiring historical state data of the distribution network equipment; performing word segmentation processing on the historical state data to obtain a plurality of words; performing text clustering processing, keyword calculation processing and association analysis processing on the multiple participles; establishing a neural network model according to the output parameters of the correlation analysis processing, and performing data training on the neural network model; updating the weight of the neural network model according to the output parameters of the data training; and monitoring the state information of the distribution network equipment according to the updated neural network model, and overhauling the distribution network equipment when the state of the distribution network equipment is detected to be in an abnormal state according to the state information. Compared with the traditional periodic detection, the method and the device can monitor and overhaul the state of the distribution network equipment more accurately and timely, and the monitored logic factors are richer.

Description

Distribution network equipment state data processing method and device, computer equipment and medium
Technical Field
The invention relates to the field of computer data processing, in particular to a distribution network equipment state data processing method and device, computer equipment and a computer storage medium.
Background
In the operation process of the power system, an administrator needs to periodically detect the state information of the power distribution network so as to find and solve the abnormality of the state of the distribution network equipment as soon as possible and ensure the normal operation of the power system.
With the development of a power grid and the gradual improvement of the power supply reliability requirements of users, the traditional periodic detection mode cannot meet the operation and maintenance requirements of a power distribution network, and has the defects of low efficiency, low accuracy and over-subjective detection.
Disclosure of Invention
Therefore, it is necessary to provide a method and a device for processing state data of distribution network equipment, aiming at the problems of low efficiency, low accuracy and too subjective detection in the conventional distribution network periodic maintenance.
A distribution network equipment state data processing method comprises the following steps: acquiring historical state data of the distribution network equipment; performing word segmentation processing on the historical state data to obtain a plurality of words; performing text clustering processing, keyword calculation processing and association analysis processing on the multiple participles; establishing a neural network model according to the output parameters of the correlation analysis processing, and performing data training on the neural network model; updating the weight of the neural network model according to the output parameters of the data training; and monitoring the state information of the distribution network equipment according to the updated neural network model, and overhauling the distribution network equipment when the state of the distribution network equipment is detected to be in an abnormal state according to the state information.
The method for processing the state data of the distribution network equipment is based on the neural network algorithm, optimizes the neural network algorithm according to the running mechanism of the distribution network equipment, and applies the optimized algorithm to the state parameter monitoring of the distribution network equipment. Compared with the traditional periodic detection mode, the method has the advantages that the data obtained by monitoring the state of the power equipment is more accurate and timely, the monitored logic factors are richer, and the state of the distribution network equipment can be monitored efficiently and conveniently.
In one embodiment, before performing text clustering processing, keyword calculation processing, and association analysis processing on the plurality of segmented words, the method further includes: filtering stop words in the multiple participles according to a preset stop word library; performing text clustering processing, keyword calculation processing and association analysis processing on the plurality of participles, wherein the text clustering processing, the keyword calculation processing and the association analysis processing comprise the following steps: and performing text clustering processing, keyword calculation processing and association analysis processing on the filtered multiple participles.
In one embodiment, after filtering the plurality of segmented words according to a preset word-stopping library, the method further includes: inquiring whether a preset word segmentation library contains filtered word segmentation or not; and if the preset word segmentation library does not contain the filtered word segmentation, updating the filtered word segmentation into the word segmentation library.
In one embodiment, after filtering the plurality of segmented words according to a preset word-stopping library, the method further includes: receiving a correction instruction; modifying the filtered participles according to the modification instruction; performing text clustering processing, keyword calculation processing and association analysis processing on the plurality of participles, wherein the text clustering processing, the keyword calculation processing and the association analysis processing comprise the following steps: and performing text clustering processing, keyword calculation processing and association analysis processing on the corrected multiple participles.
In one embodiment, the text clustering process, the keyword calculation process, and the association analysis process are performed on the plurality of segmented words, and the method includes: performing text clustering processing on the multiple word segments to obtain at least one category; respectively calculating keywords in each category; and performing association analysis processing on the keywords in each category.
In one embodiment, creating a neural network model according to the output parameters of the correlation analysis process, and performing data training on the neural network model includes: initializing at least one parameter of a preset neural network model; calculating hidden layer output of the neural network model according to at least one parameter and preset training data; and calculating the output layer output of the neural network model according to the hidden layer output.
In one embodiment, updating the weights of the neural network model according to the output parameters of the data training includes: inputting training data into a neural network model, and recording output of an output layer of the neural network model; calculating an error between the output of the output layer and the desired output value; and updating the weights of the input layer to the hidden layer of the neural network model according to the error.
In one embodiment, a distribution network device status data processing apparatus is provided, which includes: the acquisition module is used for acquiring historical state data of the distribution network equipment; the word segmentation module is used for carrying out word segmentation processing on the historical state data to obtain a plurality of words; the cluster analysis module is used for carrying out text cluster processing, keyword calculation processing and association analysis processing on the multiple participles; the training module is used for creating a neural network model according to the output parameters of the correlation analysis processing and carrying out data training on the neural network model; the updating module is used for updating the weight of the neural network model according to the output parameters of the data training; and the monitoring module is used for monitoring and detecting the state information of the distribution network equipment according to the updated neural network model, and overhauling the distribution network equipment when the state of the distribution network equipment is detected to be in an abnormal state according to the state information.
The distribution network equipment state data processing device is based on the neural network algorithm, optimizes the neural network algorithm according to the running mechanism of the distribution network equipment, and applies the optimized algorithm to the state parameter monitoring of the distribution network equipment. Compared with the traditional periodic detection mode, the device can monitor the state of the distribution network equipment more accurately and timely according to the data obtained by monitoring the state of the power equipment, and meanwhile, the monitored logic factors are richer, so that the state of the distribution network equipment can be monitored conveniently and efficiently.
In one embodiment, a computer device is provided, comprising a memory, a processor and a computer program stored on the memory and executable on the processor, the steps of the method of any of the embodiments of the invention being implemented when the computer program is executed by the processor.
In one embodiment, a computer-readable storage medium is provided, on which a computer program is stored, which, when being executed by a processor, carries out the steps of the method of any of the embodiments of the invention.
Drawings
Fig. 1 is a schematic flow chart of a distribution network device status data processing method according to an embodiment of the present invention;
fig. 2 is a schematic flow chart of a distribution network device status data processing method according to another embodiment of the present invention;
fig. 3 is a schematic flow chart of a distribution network device status data processing method according to another embodiment of the present invention;
fig. 4 is a schematic structural diagram of a distribution network device status data processing apparatus according to an embodiment of the present invention;
fig. 5 is a schematic structural diagram of a cluster analysis module in the distribution network device status data processing apparatus according to an embodiment of the present invention;
fig. 6 is a schematic structural diagram of a training module in a distribution network device status data processing apparatus according to an embodiment of the present invention;
fig. 7 is a schematic structural diagram of a computer device according to an embodiment of the present invention.
Detailed Description
In order to make the objects, technical solutions and advantages of the present invention more apparent, the present invention is described in further detail below with reference to the accompanying drawings and embodiments. It should be understood that the specific embodiments described herein are merely illustrative of the invention and are not intended to limit the invention.
In one embodiment, a method for processing network device status data is provided, which includes the following steps: acquiring historical state data of the distribution network equipment; performing word segmentation processing on the historical state data to obtain a plurality of words; performing text clustering processing, keyword calculation processing and association analysis processing on the multiple participles; establishing a neural network model according to the output parameters of the correlation analysis processing, and performing data training on the neural network model; updating the weight of the neural network model according to the output parameters of the data training; and monitoring the state information of the distribution network equipment according to the updated neural network model, and overhauling the distribution network equipment when the state of the distribution network equipment is detected to be in an abnormal state according to the state information.
In one embodiment, a distribution network device state data processing method is provided. As shown in fig. 1, the method for processing status data of distribution network devices includes the following steps:
s101, obtaining historical state data of the distribution network equipment.
The method comprises the steps of obtaining historical state data of at least one type of distribution network equipment in the power system.
In one embodiment, historical state data of at least one type of distribution network equipment in a power system under different conditions is obtained. For example, historical state data of each distribution network device under different conditions of climate, environment, equipment age, load condition and the like is obtained. The historical state data includes fault data.
And S103, performing word segmentation processing on the historical state data to obtain a plurality of words.
In this step, the character sequence included in the history state data is segmented into a plurality of meaningful words by word segmentation processing. The historical state data comprises fault description information, and the fault description information is segmented into a plurality of segmented words after word segmentation processing.
In one embodiment, the word segmentation processing is carried out on the historical state data through an Ansj word segmentation algorithm.
And S105, performing text clustering processing, keyword calculation processing and association analysis processing on the plurality of participles.
The plurality of participles are divided into at least one category through text clustering processing. And respectively carrying out keyword calculation on the participles in each category according to a preset keyword algorithm to obtain at least one keyword in each category. And performing association analysis processing on at least one keyword in the same category, and calculating the association degree or the support degree among a plurality of keywords in the same category.
In one embodiment, after the text clustering process, if there is a participle that does not belong to any one of the categories, the participle is filtered as a noise object.
In one embodiment, as shown in fig. 2, step S105 includes:
s1051, carrying out text clustering processing on the plurality of participles to obtain at least one category.
In one embodiment, the plurality of participles are subjected to text clustering processing through a density-based clustering algorithm to obtain at least one category, and the at least one category is connected with any two participles in the category in terms of density.
Wherein, the neighborhood radius is set as E, and the threshold of the core object is set as M. Then the sphere area with neighborhood radius E as the radius becomes the E neighborhood of the object with the particular object as the center of the sphere. The definition of the neighborhood radius E can be characterized by distance, cosine similarity, Word2Vec, and the like. For example, this step can be characterized by cosine similarity. And if the number of the objects in the neighborhood of the specific object E is more than or equal to M, the object is called a core object or a core point.
In one embodiment, after a core point is determined, starting from the core point, the core point is continuously expanded to an area with reachable density, so that the maximum set of density connection is obtained, that is, a maximized area including a core object and a boundary object is obtained, and any two points in the area are connected in density.
Wherein, density connection means that: given a set of objects D, if there is an object o belonging to D, making both objects p and q reachable from o with respect to E and M densities, then for objects p to q are connected with respect to E and M densities. The density can reach to mean that: given a set of objects D, if there is one object chain p1, p2, p3,., pn, p1 ═ q, pn ═ p, for pi belonging to D, i belonging to 1 to n, p (i +1) is directly density reachable from pi with respect to E and M, then object p is said to be density reachable from object q with respect to E and M. The direct density can be reached by: given a set of objects D, if object p is within E neighborhood of q, and q is a core object, then object p is said to be directly density reachable from object q.
Given an object set D, if an object q exists in a core object p, but the q object itself is not a core object, q is called a boundary object. Given a set of objects D, if object o is neither a core object nor a boundary object, then o is referred to as a noise object.
In one embodiment, after any core point is determined, all density-connected data points from the core point are found. And traversing all core points in the neighborhood of the core point, and searching points connected with the data point density until no data point which can be expanded exists, wherein the boundary nodes of the finally clustered clusters are all non-core data points. In one embodiment, core points that are not clustered are found and the above steps are repeated until there are no new core points in the dataset. Data points in the data set that are not contained in any cluster are identified as outliers, i.e., noise objects.
S1052, respectively calculating keywords in each category.
And calculating the keywords in each category according to the statistical information of the words in each category. For each category, determining a candidate word set of the category through preprocessing, quantifying the score of each candidate word by adopting a characteristic value, and determining a keyword of the category from the candidate words according to the score.
Optionally, the feature value is a feature value based on a word frequency (TF) and an inverse document probability (IDF). Alternatively, the feature value is a feature value based on document position information. Alternatively, the above feature value is a feature value calculated based on a word span.
And S1053, performing association analysis processing on the keywords in each category.
Wherein, this step includes: and calculating the support degree among the keywords in each category. The support degree refers to the number of times of the occurrence of several associated data in the data set, which accounts for the proportion of the total data set, or the probability of the occurrence of several associated data.
In one embodiment, for any two keywords X and Y in each category, the support S (X, Y) between keyword X and keyword Y is calculated according to the following formula:
Figure BDA0002072973310000071
where num (xy) is the number of times that keyword X and keyword Y occur simultaneously, and num (a) is the total number of keywords.
In one embodiment, for any three keywords X, Y and Z in each category, the support S (X, Y, Z) between keyword X, keyword Y, and keyword Z is calculated according to the following formula:
Figure BDA0002072973310000072
wherein: num (xyz) is the number of times that keyword X, keyword Y, and keyword Z occur simultaneously, and num (a) is the total number of keywords.
And S107, creating a neural network model according to the output parameters of the correlation analysis processing, and performing data training on the neural network model.
In one embodiment, as shown in fig. 3, step S107 includes the following steps:
s1071, initializing at least one parameter of the preset neural network model.
Specifically, the method comprises the following steps: initializing the number n of nodes of an input layer, the number l of nodes of a hidden layer, the number m of nodes of an output layer and the weight omega from the input layer to the hidden layer of a preset neural network modelijWeight ω from hidden layer to output layerjkBias of input layer to hidden layer ajBias of the hidden layer to the output layer bkAnd a learning rate of η; and initializing the excitation function g (x) to
Figure BDA0002072973310000081
S1072, calculating the hidden layer output of the neural network model according to the at least one parameter and preset training data. Wherein, according to
Figure BDA0002072973310000082
Calculating hidden layer output H of the neural network modelj。
S1073, calculating the output layer output of the neural network model according to the hidden layer output. Wherein, according to
Figure BDA0002072973310000083
Calculating the output layer output O of the neural network modelk。
And S109, updating the weight of the neural network model according to the output parameters of the data training.
In one embodiment, step S109 includes: inputting training data into the neural network model, and recording the output of the output layer of the neural network model. Calculating an error between the output of the output layer and a desired output value; for example, according to
Figure BDA0002072973310000084
Calculating the error between the output of the output layer and the desired output value, wherein OkFor the output layer output of the neural network model, YKIs a desired output value; wherein, i is 1 … n, j is 1 … l, and k is 1 … m. And updating the weights from the input layer to the hidden layer of the neural network model according to the error.
In one embodiment, the weights of the input layer to the hidden layer of the neural network model are updated conditioned on the error function reaching a minimum value. Specifically, the above-described weights are updated according to the following formula:
Figure BDA0002072973310000091
wherein e isk＝Yk-Ok。
And S111, monitoring the state information of the distribution network equipment according to the updated neural network model, and overhauling the distribution network equipment when the state of the distribution network equipment is detected to be in an abnormal state according to the state information.
The neural network model with the updated weight is applied to a distribution network system and used for detecting state information of distribution network equipment, including monitoring the running state, risk information and the like of the distribution network equipment. And when the state of the distribution network equipment is detected to be in an abnormal state according to the state information, the distribution network equipment is overhauled.
The method for processing the state data of the distribution network equipment is based on the neural network algorithm, optimizes the neural network algorithm according to the running mechanism of the distribution network equipment, and applies the optimized algorithm to the state parameter monitoring of the distribution network equipment. Compared with the traditional periodic detection mode, the method has the advantages that the data obtained by monitoring the state of the power equipment is more accurate and timely, the monitored logic factors are richer, and the state of the distribution network equipment can be monitored efficiently and conveniently. The embodiment of the invention improves the power supply reliability and the maintenance efficiency of the distribution network to a greater extent, and plays a great help role in performance support and operation protection of distribution network equipment.
In an embodiment, before step S105, the method for processing status data of distribution network devices further includes the following steps: and filtering stop words in the plurality of segmented words according to a preset stop word library. The stop word refers to a word which does not need to be processed, and can be a punctuation mark or a common tone word, an adverb, a connecting word and the like. In this embodiment, a word stop library is preset, and the word stop library includes at least one stop word. After step S103, each obtained participle is compared with the word-stopping library, and if the obtained participle exists in the word-stopping library, the participle is filtered out. At this time, step S105 includes: and performing text clustering processing, keyword calculation processing and association analysis processing on the filtered multiple participles. Therefore, the calculation amount can be reduced, and the calculation efficiency can be improved.
In an embodiment, after filtering stop words in the plurality of segmented words according to a preset stop word library, the method for processing the state data of the distribution network device further includes the following steps: inquiring whether a preset word segmentation library contains filtered word segmentation or not; and if the preset word segmentation library does not contain the filtered word segmentation, updating the filtered word segmentation into the word segmentation library. Therefore, the number of samples in the word segmentation library can be increased, the word segmentation library is improved, and the accuracy of subsequent word segmentation is improved.
In order to improve the accuracy of data processing, in an embodiment, after filtering the plurality of participles according to a preset stop word library, the method for processing the state data of the distribution network device further includes the following steps: receiving a correction instruction; correcting the filtered participles according to the correction instruction; accordingly, step S105 includes: and performing text clustering processing, keyword calculation processing and association analysis processing on the corrected multiple participles. Therefore, when the automatic word segmentation result is not accurate enough, the word segmentation result is corrected according to the correction instruction so as to ensure the accuracy of word segmentation and further ensure the accuracy of subsequent data processing. Furthermore, the word segmentation library can be updated according to the corrected word segmentation, so that the accuracy of word segmentation of the system is improved.
In one embodiment, as shown in fig. 4, a distribution network device status data processing apparatus 40 is provided, which includes an obtaining module 401, a word segmentation module 402, a cluster analysis module 403, a training module 404, an updating module 405, and a monitoring module 406, where: the obtaining module 401 is configured to obtain historical state data of the distribution network device; the word segmentation module 402 is configured to perform word segmentation processing on the historical state data to obtain a plurality of words; the cluster analysis module 403 is configured to perform text clustering, keyword calculation, and association analysis on the multiple segmented words; the training module 404 is configured to create a neural network model according to the output parameters of the association analysis processing, and perform data training on the neural network model; the updating module 405 is configured to update the weight of the neural network model according to the output parameter of the data training; the monitoring module 406 is configured to monitor the state information of the distribution network device according to the updated neural network model, and repair the distribution network device when detecting that the state of the distribution network device is in an abnormal state according to the state information.
The distribution network equipment state data processing device is based on the neural network algorithm, optimizes the neural network algorithm according to the running mechanism of the distribution network equipment, and applies the optimized algorithm to the state parameter monitoring and maintenance of the distribution network equipment. Compared with the traditional periodic detection mode, the device can monitor the state of the distribution network equipment more accurately and timely according to the data obtained by monitoring the state of the power equipment, and meanwhile, the monitored logic factors are richer, so that the state of the distribution network equipment can be monitored conveniently and efficiently. The embodiment of the invention improves the power supply reliability and the maintenance efficiency of the distribution network to a greater extent, and plays a great help role in performance support and operation protection of distribution network equipment.
In one embodiment, in order to reduce the amount of computation and improve the computation efficiency, the distribution network device state data processing apparatus further includes a filtering module, where the filtering module is configured to filter stop words in the multiple participles according to a preset stop word library; at this time, the cluster analysis module 403 is further configured to perform text clustering, keyword calculation, and association analysis on the filtered multiple segmented words.
In one embodiment, in order to perfect a segmentation word library, the distribution network device state data processing apparatus further includes a query module, configured to query whether a preset segmentation word library includes filtered segmentation words; the word segmentation module 402 is further configured to update the filtered word segments into the word segmentation library when the preset word segmentation library does not contain the filtered word segments. Therefore, the number of samples in the word segmentation library can be increased, the word segmentation library is improved, and the accuracy of subsequent word segmentation is improved.
In an embodiment, in order to improve accuracy of data processing, the distribution network device status data processing apparatus further includes: the receiving module is used for receiving a correction instruction; the correction module is used for correcting the filtered word segmentation according to the correction instruction; at this time, the cluster analysis module 403 is further configured to perform text clustering, keyword calculation, and association analysis on the plurality of modified segmented words. Therefore, when the automatic word segmentation result is not accurate enough, the word segmentation result is corrected according to the correction instruction so as to ensure the accuracy of word segmentation and further ensure the accuracy of subsequent data processing. Furthermore, the word segmentation library can be updated according to the corrected word segmentation, so that the accuracy of word segmentation of the system is improved.
In one embodiment, as shown in fig. 5, the cluster analysis module 403 includes a clustering unit 4031, a first calculating unit 4032 and an association analysis unit 4033, where the clustering unit 4031 is configured to perform text clustering on the multiple segmented words to obtain at least one category; the first calculation unit 4032 is used for calculating keywords in each category respectively; the association analysis unit 4033 is used to perform association analysis processing on the keywords in each category.
In one embodiment, the association analysis unit 4033 is further configured to calculate a support degree between the keywords in each category.
In one embodiment, the association analysis unit 4033 is further configured to calculate a support degree S (X, Y) between the keyword X and the keyword Y according to the following formula for any two keywords X and Y in each category:
Figure BDA0002072973310000121
where num (xy) is the number of times that keyword X and keyword Y occur simultaneously, and num (a) is the total number of keywords.
In one embodiment, the association analysis unit 4033 is further configured to calculate the support degree S (X, Y, Z) between the keyword X, the keyword Y and the keyword Z for any three keywords X, Y and Z in each category according to the following formula:
Figure BDA0002072973310000122
where num (xyz) is the number of times that keyword X, keyword Y, and keyword Z appear simultaneously, and num (a) is the total number of keywords.
In one embodiment, as shown in fig. 6, the training module 404 includes an initialization unit 4041, a second calculation unit 4042, and a third calculation unit 4043, where the initialization unit 4041 is configured to initialize at least one parameter of a preset neural network model; the second calculating unit 4042 is configured to calculate a hidden layer output of the neural network model according to the at least one parameter and preset training data; the third calculating unit 4043 is configured to calculate an output layer output of the neural network model according to the hidden layer output.
In one embodiment, the initialization unit 4041 is further configured to initialize the number of nodes of the input layer of the preset neural network model to n, the number of nodes of the hidden layer to l, the number of nodes of the output layer to m, and the weight ω from the input layer to the hidden layerijWeight ω from hidden layer to output layerikBias of input layer to hidden layer ajBias of the hidden layer to the output layer bkAnd a learning rate of η; and initializing the excitation function g (x) to
Figure BDA0002072973310000123
In one embodiment, the second computing unit 4042 is configured to compute based on
Figure BDA0002072973310000131
Computing a hidden layer output H of the neural network modelj。
In one embodiment, the third computing unit 4043 is configured to calculate the data according to
Figure BDA0002072973310000132
Computing an output layer output O of the neural network modelk。
In one embodiment, the update module includes: the device comprises a recording unit, a fourth calculating unit and a weight updating unit, wherein the recording unit is used for inputting training data into the neural network model and recording the output of an output layer of the neural network model; the fourth calculating unit is used for calculating the error between the output layer output and the expected output value; and the weight updating unit is used for updating the weight from the input layer to the hidden layer of the neural network model according to the error.
In one embodiment, as shown in fig. 7, there is provided a computer device comprising a memory, a processor and a computer program stored on the memory and executable on the processor, wherein the processor when executing the computer program implements the steps of: acquiring historical state data of the distribution network equipment; performing word segmentation processing on the historical state data to obtain a plurality of words; performing text clustering processing, keyword calculation processing and association analysis processing on the multiple participles; establishing a neural network model according to the output parameters of the correlation analysis processing, and performing data training on the neural network model; updating the weight of the neural network model according to the output parameters of the data training; and monitoring the state information of the distribution network equipment according to the updated neural network model, and overhauling the distribution network equipment when the state of the distribution network equipment is detected to be in an abnormal state according to the state information.
In one embodiment, before performing the text clustering process, the keyword calculation process, and the association analysis process on the plurality of segmented words, the processor executes the computer program to further implement the following steps: filtering stop words in the multiple participles according to a preset stop word library; performing text clustering processing, keyword calculation processing and association analysis processing on the plurality of participles, wherein the text clustering processing, the keyword calculation processing and the association analysis processing comprise the following steps: and performing text clustering processing, keyword calculation processing and association analysis processing on the filtered multiple participles.
In one embodiment, after filtering the above-mentioned multiple segmented words according to a preset word-stopping library, the processor executes the computer program to further implement the following steps: inquiring whether a preset word segmentation library contains filtered word segmentation or not; and if the preset word segmentation library does not contain the filtered word segmentation, updating the filtered word segmentation into the word segmentation library.
In one embodiment, after filtering the above-mentioned multiple segmented words according to a preset word-stopping library, the processor executes the computer program to further implement the following steps: receiving a correction instruction; modifying the filtered participles according to the modification instruction; performing text clustering processing, keyword calculation processing and association analysis processing on the plurality of participles, wherein the text clustering processing, the keyword calculation processing and the association analysis processing comprise the following steps: and performing text clustering processing, keyword calculation processing and association analysis processing on the corrected multiple participles.
In one embodiment, the text clustering process, the keyword calculation process, and the association analysis process are performed on the plurality of segmented words, and the method includes: performing text clustering processing on the multiple word segments to obtain at least one category; respectively calculating keywords in each category; and performing association analysis processing on the keywords in each category.
In one embodiment, the association analysis processing is performed on the keywords in each category, and includes: and calculating the support degree among the keywords in each category.
In one embodiment, calculating the support between the keywords in each category comprises: for any two keywords X and Y in each category, the support S (X, Y) between the keyword X and the keyword Y is calculated according to the following formula:
Figure BDA0002072973310000141
where num (xy) is the number of times that keyword X and keyword Y occur simultaneously, and num (a) is the total number of keywords.
In one embodiment, calculating the support between the keywords in each category comprises: for any three keywords X, Y and Z in each category, a support degree S (X, Y, Z) between keyword X, keyword Y, and keyword Z is calculated according to the following formula:
Figure BDA0002072973310000142
where num (xyz) is the number of times that keyword X, keyword Y, and keyword Z appear simultaneously, and num (a) is the total number of keywords.
In one embodiment, creating a neural network model according to the output parameters of the correlation analysis process, and performing data training on the neural network model, includes: initializing at least one parameter of a preset neural network model; calculating hidden layer output of the neural network model according to at least one parameter and preset training data; and calculating the output layer output of the neural network model according to the hidden layer output.
In one embodiment, initializing at least one parameter of the neural network model comprises: initializing the number n of nodes of an input layer, the number l of nodes of a hidden layer, the number m of nodes of an output layer and the weight omega from the input layer to the hidden layer of a preset neural network modelijWeight ω from hidden layer to output layerjkBias of input layer to hidden layer ajBias of the hidden layer to the output layer bkAnd a learning rate of η; the initial excitation function g (x) is
Figure BDA0002072973310000151
In one embodiment, calculating the hidden layer output of the neural network model according to at least one parameter and preset training data comprises: according to
Figure BDA0002072973310000152
Computing hidden layer output H of neural network modelj。
In one embodiment, computing output layer outputs of a neural network model from the hidden layer outputs comprises: according to
Figure BDA0002072973310000153
Computing output layer output O of a neural network modelk。
In one embodiment, updating the weights of the neural network model according to the output parameters of the data training comprises: inputting training data into a neural network model, and recording output of an output layer of the neural network model; calculating an error between the output of the output layer and the desired output value; and updating the weights of the input layer to the hidden layer of the neural network model according to the error.
In one embodiment, calculating an error between the output layer output and the desired output value comprises: according to
Figure BDA0002072973310000154
Calculating an error between the output of the output layer and the desired output value, wherein OkIs the output layer output of the neural network model, YKIs a desired output value; wherein, i is 1 … n, j is 1 … l, and k is 1 … m.
In one embodiment, a computer-readable storage medium is provided, having a computer program stored thereon, which when executed by a processor, performs the steps of: acquiring historical state data of the distribution network equipment; performing word segmentation processing on the historical state data to obtain a plurality of words; performing text clustering processing, keyword calculation processing and association analysis processing on the multiple participles; establishing a neural network model according to the output parameters of the correlation analysis processing, and performing data training on the neural network model; updating the weight of the neural network model according to the output parameters of the data training; and monitoring the state information of the distribution network equipment according to the updated neural network model, and overhauling the distribution network equipment when the state of the distribution network equipment is detected to be in an abnormal state according to the state information.
In one embodiment, before performing the text clustering process, the keyword calculation process, and the association analysis process on the plurality of segmented words, the computer program further performs the following steps when executed by the processor: filtering stop words in the multiple participles according to a preset stop word library; performing text clustering processing, keyword calculation processing and association analysis processing on the plurality of participles, wherein the text clustering processing, the keyword calculation processing and the association analysis processing comprise the following steps: and performing text clustering processing, keyword calculation processing and association analysis processing on the filtered multiple participles.
In one embodiment, after filtering the above-mentioned multiple segmented words according to a preset word stop library, the computer program when executed by the processor further implements the following steps: inquiring whether a preset word segmentation library contains filtered word segmentation or not; and if the preset word segmentation library does not contain the filtered word segmentation, updating the filtered word segmentation into the word segmentation library.
In one embodiment, after filtering the above-mentioned multiple segmented words according to a preset word stop library, the computer program when executed by the processor further implements the following steps: receiving a correction instruction; modifying the filtered participles according to the modification instruction; performing text clustering processing, keyword calculation processing and association analysis processing on the plurality of participles, wherein the text clustering processing, the keyword calculation processing and the association analysis processing comprise the following steps: and performing text clustering processing, keyword calculation processing and association analysis processing on the corrected multiple participles.
In one embodiment, the text clustering process, the keyword calculation process, and the association analysis process are performed on the plurality of segmented words, and the method includes: performing text clustering processing on the multiple word segments to obtain at least one category; respectively calculating keywords in each category; and performing association analysis processing on the keywords in each category.
In one embodiment, the association analysis processing is performed on the keywords in each category, and includes: and calculating the support degree among the keywords in each category.
In one embodiment, calculating the support between the keywords in each category comprises: for any two keywords X and Y in each category, the support S (X, Y) between the keyword X and the keyword Y is calculated according to the following formula:
Figure BDA0002072973310000171
where num (xy) is the number of times that keyword X and keyword Y occur simultaneously, and num (a) is the total number of keywords.
In one embodiment, calculating the support between the keywords in each category comprises: for any three keywords X, Y and Z in each category, a support degree S (X, Y, Z) between keyword X, keyword Y, and keyword Z is calculated according to the following formula:
Figure BDA0002072973310000172
where num (xyz) is the number of times that keyword X, keyword Y, and keyword Z appear simultaneously, and num (a) is the total number of keywords.
In one embodiment, creating a neural network model according to the output parameters of the correlation analysis process, and performing data training on the neural network model, includes: initializing at least one parameter of a preset neural network model; calculating hidden layer output of the neural network model according to at least one parameter and preset training data; and calculating the output layer output of the neural network model according to the hidden layer output.
In one embodiment, initializing at least one parameter of the neural network model comprises: initializing the number n of nodes of an input layer, the number l of nodes of a hidden layer, the number m of nodes of an output layer and the weight omega from the input layer to the hidden layer of a preset neural network modelijWeight ω from hidden layer to output layerikBias of input layer to hidden layer ajBias of the hidden layer to the output layer bkAnd a learning rate of η; the initial excitation function g (x) is
Figure BDA0002072973310000173
In one embodiment, calculating the hidden layer output of the neural network model according to at least one parameter and preset training data comprises: according to
Figure BDA0002072973310000181
Computing hidden layer output H of neural network modelj。
In one embodiment, according toHidden layer output the output layer output of the computational neural network model, including: according to
Figure BDA0002072973310000182
Computing output layer output O of a neural network modelk。
In one embodiment, updating the weights of the neural network model according to the output parameters of the data training comprises: inputting training data into a neural network model, and recording output of an output layer of the neural network model; calculating an error between the output of the output layer and the desired output value; and updating the weights of the input layer to the hidden layer of the neural network model according to the error.
In one embodiment, calculating an error between the output layer output and the desired output value comprises: according to
Figure BDA0002072973310000183
Calculating an error between the output of the output layer and the desired output value, wherein OkIs the output layer output of the neural network model, YKIs a desired output value; wherein, i is 1 … n, j is 1 … l, and k is 1 … m.
The technical features of the embodiments described above may be arbitrarily combined, and for the sake of brevity, all possible combinations of the technical features in the embodiments described above are not described, but should be considered as being within the scope of the present specification as long as there is no contradiction between the combinations of the technical features.
The above-mentioned embodiments only express several embodiments of the present invention, and the description thereof is more specific and detailed, but not construed as limiting the scope of the invention. It should be noted that, for a person skilled in the art, several variations and modifications can be made without departing from the inventive concept, which falls within the scope of the present invention. Therefore, the protection scope of the present patent shall be subject to the appended claims.

Claims (10)

1. A distribution network equipment state data processing method is characterized by comprising the following steps:
acquiring historical state data of the distribution network equipment;
performing word segmentation processing on the historical state data to obtain a plurality of words;
performing text clustering processing on the multiple participles to obtain at least one category, wherein any two participles in the category are connected in density;
performing keyword calculation processing on each category, wherein the process of the keyword calculation processing is quantified by using a characteristic value;
performing association analysis processing on the keywords in each category;
establishing a neural network model according to the output parameters of the correlation analysis processing, and performing data training on the neural network model;
updating the weight of the neural network model according to the output parameters of the data training;
and monitoring the state information of the distribution network equipment according to the updated neural network model, and overhauling the distribution network equipment when the state of the distribution network equipment is detected to be in an abnormal state according to the state information.
2. The method of claim 1, wherein prior to performing the text clustering process, the keyword calculation process, and the association analysis process on the plurality of segmented words, the method further comprises:
filtering stop words in the multiple participles according to a preset stop word library;
the text clustering processing, the keyword calculation processing and the association analysis processing are carried out on the multiple word segments, and the method comprises the following steps: and performing text clustering processing, keyword calculation processing and association analysis processing on the filtered multiple participles.
3. The method of claim 2, wherein after filtering the plurality of segmented words according to a predetermined stop word bank, the method further comprises:
inquiring whether a preset word segmentation library contains filtered word segmentation or not;
and if the preset word segmentation library does not contain the filtered word segmentation, updating the filtered word segmentation into the word segmentation library.
4. The method of claim 2, wherein after filtering the plurality of segmented words according to a predetermined stop word bank, the method further comprises:
receiving a correction instruction;
correcting the filtered participles according to the correction instruction;
the text clustering processing, the keyword calculation processing and the association analysis processing are carried out on the multiple word segments, and the method comprises the following steps: and performing text clustering processing, keyword calculation processing and association analysis processing on the corrected multiple participles.
5. The method according to any one of claims 1 to 4, wherein performing text clustering processing, keyword calculation processing, and association analysis processing on the plurality of segmented words comprises:
performing text clustering processing on the multiple word segments to obtain at least one category;
respectively calculating keywords in each category;
and performing association analysis processing on the keywords in each category.
6. The method of claim 1, wherein creating a neural network model from the output parameters of the correlation analysis process and performing data training on the neural network model comprises:
initializing at least one parameter of a preset neural network model;
calculating hidden layer output of the neural network model according to the at least one parameter and preset training data;
and calculating the output layer output of the neural network model according to the hidden layer output.
7. The method of claim 1, wherein updating the weights of the neural network model according to the output parameters of the data training comprises:
inputting training data into the neural network model, and recording the output layer output of the neural network model;
calculating an error between the output layer output and a desired output value;
and updating the weight from the input layer to the hidden layer of the neural network model according to the error.
8. A distribution network equipment state data processing device is characterized by comprising:
the acquisition module is used for acquiring historical state data of the distribution network equipment;
the word segmentation module is used for carrying out word segmentation processing on the historical state data to obtain a plurality of words;
the clustering module is used for carrying out text clustering processing on the multiple participles to obtain at least one category, and any two participles in the category are connected in density;
the calculation module is used for performing keyword calculation processing on each category, and the process of the keyword calculation processing is quantified by using a characteristic value;
the association module is used for performing association analysis processing on the keywords in each category;
the training module is used for creating a neural network model according to the output parameters of the correlation analysis processing and carrying out data training on the neural network model;
the updating module is used for updating the weight of the neural network model according to the output parameters of the data training;
and the monitoring module is used for monitoring the state information of the distribution network equipment according to the updated neural network model and overhauling the distribution network equipment when the state of the distribution network equipment is detected to be in an abnormal state according to the state information.
9. A computer device comprising a memory, a processor and a computer program stored on the memory and executable on the processor, characterized in that the steps of the method of any of claims 1 to 7 are implemented when the computer program is executed by the processor.
10. A computer-readable storage medium, on which a computer program is stored, which, when being executed by a processor, carries out the steps of the method of any one of claims 1 to 7.
CN201910443912.9A 2019-05-27 2019-05-27 Distribution network equipment state data processing method and device, computer equipment and medium Active CN110348005B (en)

Priority Applications (1)

Application Number Priority Date Filing Date Title
CN201910443912.9A CN110348005B (en) 2019-05-27 2019-05-27 Distribution network equipment state data processing method and device, computer equipment and medium

Applications Claiming Priority (1)

Application Number Priority Date Filing Date Title
CN201910443912.9A CN110348005B (en) 2019-05-27 2019-05-27 Distribution network equipment state data processing method and device, computer equipment and medium

Publications (2)

Publication Number Publication Date
CN110348005A CN110348005A (en) 2019-10-18
CN110348005B true CN110348005B (en) 2022-02-01

Family

ID=68174370

Family Applications (1)

Application Number Title Priority Date Filing Date
CN201910443912.9A Active CN110348005B (en) 2019-05-27 2019-05-27 Distribution network equipment state data processing method and device, computer equipment and medium

Country Status (1)

Country Link
CN (1) CN110348005B (en)

Families Citing this family (2)

* Cited by examiner, † Cited by third party
Publication number Priority date Publication date Assignee Title
CN112510699A (en) * 2020-11-25 2021-03-16 国网湖北省电力有限公司咸宁供电公司 Transformer substation secondary equipment state analysis method and device based on big data
CN116207845A (en) * 2022-11-28 2023-06-02 无锡广盈集团有限公司 Automatic monitoring method and system for protecting power equipment

Citations (3)

* Cited by examiner, † Cited by third party
Publication number Priority date Publication date Assignee Title
CN105787809A (en) * 2016-03-10 2016-07-20 国家电网公司 Data-mining-based intelligent checking method for running state of power grid equipment
CN107977710A (en) * 2017-12-21 2018-05-01 南方电网科学研究院有限责任公司 Multiplexing electric abnormality data detection method and device
CN109726246A (en) * 2018-12-13 2019-05-07 贵州电网有限责任公司 One kind being associated with reason retrogressive method with visual power grid accident based on data mining

Family Cites Families (1)

* Cited by examiner, † Cited by third party
Publication number Priority date Publication date Assignee Title
CN103914735B (en) * 2014-04-17 2017-03-29 北京泰乐德信息技术有限公司 A kind of fault recognition method and system based on Neural Network Self-learning

Patent Citations (3)

* Cited by examiner, † Cited by third party
Publication number Priority date Publication date Assignee Title
CN105787809A (en) * 2016-03-10 2016-07-20 国家电网公司 Data-mining-based intelligent checking method for running state of power grid equipment
CN107977710A (en) * 2017-12-21 2018-05-01 南方电网科学研究院有限责任公司 Multiplexing electric abnormality data detection method and device
CN109726246A (en) * 2018-12-13 2019-05-07 贵州电网有限责任公司 One kind being associated with reason retrogressive method with visual power grid accident based on data mining

Also Published As

Publication number Publication date
CN110348005A (en) 2019-10-18

Similar Documents

Publication Publication Date Title
JP7169369B2 (en) Method, system for generating data for machine learning algorithms
CN110288004B (en) System fault diagnosis method and device based on log semantic mining
US20210041862A1 (en) Malfunction early-warning method for production logistics delivery equipment
CN111242323B (en) Active automatic system and method for repairing suboptimal operation of a machine
JP3209163B2 (en) Classifier
CN108427720B (en) System log classification method
CN103513983A (en) Method and system for predictive alert threshold determination tool
CN111914090B (en) Method and device for enterprise industry classification identification and characteristic pollutant identification
US20170004413A1 (en) Data driven classification and data quality checking system
CN105934765B (en) Method for constructing abnormal model from abnormal data
Zheng et al. Fault diagnosis method based on supervised particle swarm optimization classification algorithm
CN110348005B (en) Distribution network equipment state data processing method and device, computer equipment and medium
US20220237407A1 (en) Storage medium, estimation method, and information processing device, relearning program, and relearning method
Zhao et al. Probabilistic remaining useful life prediction based on deep convolutional neural network
CN113449098A (en) Log clustering method, device, equipment and storage medium
US20170004414A1 (en) Data driven classification and data quality checking method
CN113657814A (en) Aviation network risk prediction method and risk grade evaluation method
Sharma et al. A semi-supervised generalized vae framework for abnormality detection using one-class classification
CN113778792B (en) Alarm classifying method and system for IT equipment
CN114637649A (en) Alarm root cause analysis method and device based on OLTP database system
CN114861753A (en) Data classification method and device based on large-scale network
KR20110056046A (en) Method for evaluating performance of tire
Liu et al. AutoSlicer: Scalable Automated Data Slicing for ML Model Analysis
CN111553489B (en) Aero-engine high-pressure rotor fault diagnosis method based on ensemble learning
CN113447813B (en) Fault diagnosis method and equipment for offshore wind generating set

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
TA01 Transfer of patent application right

Effective date of registration: 20200924

Address after: 510620 Tianhe District, Guangzhou, Tianhe South Road, No. two, No. 2, No.

Applicant after: Guangzhou Power Supply Bureau of Guangdong Power Grid Co.,Ltd.

Address before: 510620 Tianhe District, Guangzhou, Tianhe South Road, No. two, No. 2, No.

Applicant before: GUANGZHOU POWER SUPPLY Co.,Ltd.

TA01 Transfer of patent application right
GR01 Patent grant
GR01 Patent grant