CN113011530A - Intelligent ammeter fault prediction method based on multi-classifier fusion - Google Patents
Intelligent ammeter fault prediction method based on multi-classifier fusion Download PDFInfo
- Publication number
- CN113011530A CN113011530A CN202110472639.XA CN202110472639A CN113011530A CN 113011530 A CN113011530 A CN 113011530A CN 202110472639 A CN202110472639 A CN 202110472639A CN 113011530 A CN113011530 A CN 113011530A
- Authority
- CN
- China
- Prior art keywords
- fault
- classifier
- data
- samples
- classifiers
- 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.)
- Granted
Links
- 238000000034 method Methods 0.000 title claims abstract description 43
- 230000004927 fusion Effects 0.000 title claims abstract description 17
- 239000011159 matrix material Substances 0.000 claims abstract description 30
- 238000012706 support-vector machine Methods 0.000 claims abstract description 16
- 230000002159 abnormal effect Effects 0.000 claims abstract description 13
- 238000009826 distribution Methods 0.000 claims abstract description 11
- 238000013528 artificial neural network Methods 0.000 claims abstract description 9
- 238000007637 random forest analysis Methods 0.000 claims abstract description 8
- 238000012545 processing Methods 0.000 claims abstract description 7
- 238000005070 sampling Methods 0.000 claims abstract description 7
- 238000004422 calculation algorithm Methods 0.000 claims abstract description 4
- 238000012549 training Methods 0.000 claims description 17
- 238000012360 testing method Methods 0.000 claims description 13
- 230000005611 electricity Effects 0.000 claims description 7
- 238000011156 evaluation Methods 0.000 claims description 6
- 238000010276 construction Methods 0.000 claims description 5
- 238000004364 calculation method Methods 0.000 claims description 4
- 238000010187 selection method Methods 0.000 claims description 4
- 239000000126 substance Substances 0.000 claims description 2
- 238000007781 pre-processing Methods 0.000 claims 1
- 238000010586 diagram Methods 0.000 abstract description 3
- 230000006870 function Effects 0.000 description 6
- 238000012423 maintenance Methods 0.000 description 5
- 238000011160 research Methods 0.000 description 3
- 238000004891 communication Methods 0.000 description 2
- 238000011161 development Methods 0.000 description 2
- 230000009286 beneficial effect Effects 0.000 description 1
- 230000005540 biological transmission Effects 0.000 description 1
- 238000004140 cleaning Methods 0.000 description 1
- 238000003066 decision tree Methods 0.000 description 1
- 230000007547 defect Effects 0.000 description 1
- 238000005516 engineering process Methods 0.000 description 1
- 230000010354 integration Effects 0.000 description 1
- 238000010801 machine learning Methods 0.000 description 1
- 238000004519 manufacturing process Methods 0.000 description 1
- 238000007500 overflow downdraw method Methods 0.000 description 1
- 238000012216 screening Methods 0.000 description 1
- 238000003860 storage Methods 0.000 description 1
- 238000012795 verification Methods 0.000 description 1
Images
Classifications
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06F—ELECTRIC DIGITAL DATA PROCESSING
- G06F18/00—Pattern recognition
- G06F18/20—Analysing
- G06F18/24—Classification techniques
- G06F18/241—Classification techniques relating to the classification model, e.g. parametric or non-parametric approaches
- G06F18/2411—Classification techniques relating to the classification model, e.g. parametric or non-parametric approaches based on the proximity to a decision surface, e.g. support vector machines
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06F—ELECTRIC DIGITAL DATA PROCESSING
- G06F18/00—Pattern recognition
- G06F18/20—Analysing
- G06F18/24—Classification techniques
- G06F18/243—Classification techniques relating to the number of classes
- G06F18/24323—Tree-organised classifiers
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06N—COMPUTING ARRANGEMENTS BASED ON SPECIFIC COMPUTATIONAL MODELS
- G06N3/00—Computing arrangements based on biological models
- G06N3/02—Neural networks
- G06N3/04—Architecture, e.g. interconnection topology
- G06N3/045—Combinations of networks
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06N—COMPUTING ARRANGEMENTS BASED ON SPECIFIC COMPUTATIONAL MODELS
- G06N3/00—Computing arrangements based on biological models
- G06N3/02—Neural networks
- G06N3/08—Learning methods
-
- 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
- G06Q10/04—Forecasting or optimisation specially adapted for administrative or management purposes, e.g. linear programming or "cutting stock problem"
-
- 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—Information and communication technology [ICT] specially adapted for implementation of business processes of specific business sectors, e.g. utilities or tourism
- G06Q50/06—Energy or water supply
-
- 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)
- Theoretical Computer Science (AREA)
- Physics & Mathematics (AREA)
- Data Mining & Analysis (AREA)
- Business, Economics & Management (AREA)
- General Physics & Mathematics (AREA)
- Life Sciences & Earth Sciences (AREA)
- Artificial Intelligence (AREA)
- Economics (AREA)
- Health & Medical Sciences (AREA)
- Evolutionary Computation (AREA)
- General Engineering & Computer Science (AREA)
- Strategic Management (AREA)
- General Health & Medical Sciences (AREA)
- Human Resources & Organizations (AREA)
- Software Systems (AREA)
- Computer Vision & Pattern Recognition (AREA)
- Biophysics (AREA)
- Molecular Biology (AREA)
- Computing Systems (AREA)
- Mathematical Physics (AREA)
- Marketing (AREA)
- Computational Linguistics (AREA)
- General Business, Economics & Management (AREA)
- Bioinformatics & Computational Biology (AREA)
- Evolutionary Biology (AREA)
- Biomedical Technology (AREA)
- Bioinformatics & Cheminformatics (AREA)
- Tourism & Hospitality (AREA)
- Game Theory and Decision Science (AREA)
- Operations Research (AREA)
- Quality & Reliability (AREA)
- Entrepreneurship & Innovation (AREA)
- Development Economics (AREA)
- Public Health (AREA)
- Water Supply & Treatment (AREA)
- Primary Health Care (AREA)
- Management, Administration, Business Operations System, And Electronic Commerce (AREA)
Abstract
A fault prediction method of an intelligent ammeter based on multi-classifier fusion is characterized in that a normal distribution completion and box diagram method is adopted to fill missing values and replace abnormal values of an original data set aiming at the characteristics of large scale, high dimensionality, complex structure, error and abnormal data of the fault data of the intelligent ammeter; eliminating redundant and irrelevant characteristics by calculating a correlation coefficient between the characteristic attribute and the fault type to form a characteristic subset; and a mixed sampling strategy of oversampling a few samples and undersampling a plurality of samples is constructed, and the problem of unbalanced fault data is solved. Calculating the accuracy of processing the fault data of the intelligent electric meter by a Support Vector Machine (SVM), a BP neural network and a random forest algorithm, and constructing a confusion matrix representing the performance of each classifier; and (3) considering the recognition capability of each classifier for different fault types, distributing weights to each classifier, further constructing a multi-classifier decision function, and taking the weights and the largest class as the fault prediction result of the sample.
Description
Technical Field
The invention relates to a fault prediction method for an intelligent electric meter, in particular to a fault prediction method for an intelligent electric meter based on multi-classifier fusion.
Background
The intelligent electric meter is used as an important component device of the electricity utilization information acquisition system, undertakes the tasks of electric energy acquisition and metering transmission, the coverage rate of the construction of the electricity utilization information acquisition system is continuously expanded along with the development of the current society and the improvement of the regional economic level, and the faults of the intelligent electric meter are remarkably characterized by burst property, difficult recurrence, complexity and multifacetability. In addition, because the sources of the intelligent electric meters are different, and original equipment and manufacturing processes selected by multiple suppliers in China are different, the fault types of the installed intelligent electric meters are different, when a fault occurs, a maintenance worker is required to quickly maintain the intelligent electric meters, and in actual operation, the operation and maintenance system cannot judge the specific fault to cause untimely maintenance of the fault electric meters. How to determine which kind of trouble has taken place for the smart electric meter improves smart electric meter's maintenance efficiency, becomes a key problem of ammeter maintenance now.
Aiming at the problem of fault prediction of the intelligent electric meter, the traditional method is based on a model, a knowledge drive or a probabilistic reasoning method, and aims at developing qualitative reasoning in the aspects of a failure physical model of the intelligent electric meter and the like. With the development of big data technology, some researchers have proposed a method based on data driving, and have adopted machine learning algorithms such as support vector machine, decision tree, neural network and the like to develop relevant research. Each classifier presents different advantages and disadvantages when processing different types of data samples: the SVM is suitable for classification of large-scale and high-dimensionality data sets, but needs prior knowledge and has different kernel function selection standards; the BP neural network has larger fault-tolerant capability, but has slow convergence speed and is easy to overfit; the random forest has high tolerance to abnormal values and noise, is not easy to generate an overfitting phenomenon, and has low classification accuracy on small data or low-dimensional data.
In summary, in consideration of the characteristics of large scale, high dimensionality, complex structure, error and abnormal data of the fault data of the intelligent electric meter, a method for hybrid integration of multiple models is adopted to combine the advantages of multiple classifiers, and a method for predicting the fault of the intelligent electric meter based on the fusion of the multiple classifiers is provided.
Disclosure of Invention
The invention provides a fault prediction method of an intelligent electric meter based on multi-classifier fusion, aiming at the defects of the intelligent electric meter in fault prediction research in the prior art. According to the method, aiming at the characteristic that the fault data structure of the intelligent ammeter is complex, missing value filling and abnormal value replacement are carried out on an original data set, redundancy and irrelevant characteristics are eliminated, a characteristic subset is formed, a mixed sampling strategy is constructed, and the problem of unbalanced fault data is solved; constructing a confusion matrix representing the performance of classifiers such as a Support Vector Machine (SVM), a BP neural network, a random forest algorithm and the like, distributing weights for all the classifiers, further constructing a multi-classifier decision function, and taking the weights and the maximum class as a fault prediction result of the sample.
1. Analyzing fault data information of the intelligent ammeter, and processing missing values and abnormal values of ammeter fault data acquired from the electricity utilization information acquisition system;
2. calculating a correlation coefficient between each characteristic attribute and a fault type by adopting a characteristic selection method, and eliminating the characteristic attribute with small correlation with the fault type to form a characteristic subset;
3. constructing a mixed sampling method, and eliminating data imbalance characteristics by adopting oversampling on a few samples and adopting an undersampling mode on a plurality of samples;
4. dividing a data set into a training group and a testing group according to the proportion of 8:3, respectively inputting samples of the training group and the testing group into three classifiers, namely a Support Vector Machine (SVM), a BP neural network and a random forest, training and testing, counting the prediction result of each classifier, and constructing a confusion matrix according to the prediction result in a statistical probability mode;
5. combining the recognition capability of each classifier on different fault types, providing a weight distribution method, constructing an evaluation matrix, and calculating to obtain a weight coefficient corresponding to each classifier;
6. and constructing a decision function containing the weight coefficient of each classifier, inputting the preprocessed training group and test group data into each classifier model for training and prediction, adding the weights of the classifiers with the same prediction result, taking the weight and the largest class as the prediction result of the fault sample, taking the prediction result of the fault sample as a label corresponding to each fault type, and determining which fault occurs to the ammeter according to the label.
In step 1, the fault data information of the smart meter includes, but is not limited to: ammeter fault type, ammeter serial number that dispatches from the factory, power supply unit, fault status display data: voltage, current, chip manufacturer, communication interface, filing date, verification date, delivery date and unit; meter mounting date, fault and meter dismounting date, equipment dismounting time and SG186 electricity collecting reading; battery age, battery voltage, number of uncappings, etc.
The failure types are roughly as follows: appearance faults, metering performance faults, storage unit faults, processing unit faults, display unit faults, control unit faults, power unit faults, communication unit faults, clock unit faults, software faults, and other faults.
Due to manual statistics errors and the like, the data set comprises problems of multiple sample duplication, data loss and data abnormity. And performing fault category screening and sample duplicate removal on the original data set is a data cleaning operation necessary for model construction.
Marking by determining whether a data unit is empty, missing, for a time series X [ X ]1,x2...xn]The mean value mu and the variance sigma of the missing data are calculated by a method of complementing according to normal distribution, and data conforming to the normal distribution are generated:
where N (μ, σ) represents a normal distribution with a mean μ and a variance σ. The box diagram is adopted to judge abnormal values, and the following ranges are defined:
[Q1-1.5IQR,Q3+1.5IQR]
IQR＝Q3-Q1
wherein, Q1 and Q3 respectively represent the first quartile and the third quartile of the data set, IQR is the threshold judgment range, and data which do not meet the threshold judgment range are all considered as abnormal values and need to be deleted or replaced.
In the step 2, redundant features and irrelevant features exist in a plurality of data attributes, a feature selection method is needed to select the data attribute most beneficial to model construction, and a feature subset is formed by removing the feature attribute with small relevance to the fault type.
Assume that a fault data set is of the form D { (x)1,y1),(x2,y2)...(xnum,ynum) In which xiFeature attribute information representing the ith sample, with a dimension N, num 1,2,3iAnd representing the fault type of the ith sample, the correlation coefficient between each characteristic attribute and the fault type in the ith sample can be represented as:
ri＝[ρ1,ρ2...ρk...ρN]
where ρ iskRepresenting the correlation coefficient between the kth characteristic attribute and the fault type, riAnd representing a correlation coefficient set under the ith fault type, wherein the calculation process is as follows:
wherein λ iskRepresenting the kth feature.
By removing the characteristic attributes with small correlation with the fault types, redundant characteristics and irrelevant characteristics are eliminated, and a characteristic subset is formed.
In the step 3, the sample data amounts of the unbalanced data in different categories are different, so that the classification accuracy is easily reduced. Therefore, the method for oversampling a few samples and undersampling a majority of samples is considered to solve the problem of data imbalance.
Assuming that the data set comprises M kinds of characteristic attributes in total, counting the number U of samples corresponding to each characteristic attribute in the data setjM, calculating an average Mean _ U of the number of samples corresponding to the M kinds of characteristic attributes:
if the number of the samples corresponding to the j-th class characteristic attribute is smaller than the Mean _ U, adopting an oversampling mode, otherwise adopting an undersampling mode, wherein the calculation formula of the number of the sampled samples is as follows:
wherein, UjRepresenting a sample of fault data, U ', before sampling'jAnd the Mean _ U represents the average value of the number of samples corresponding to the M characteristic attributes.
In step 4, the confusion matrix is used to measure the judgment ability of the classifier for each fault. Aiming at the Support Vector Machine (SVM), BP neural network and 3 random forest classifiers adopted by the invention, assuming that the fault types of the intelligent electric meter are n, the dimension of the confusion matrix is n multiplied by n, and is expressed as:
wherein the content of the first and second substances,
representing the statistical probability that the kth classifier correctly identified the sample fault type,
and the statistical probability that the Kth classifier wrongly identifies the ith fault as the jth fault is shown, and the value of K is 1,2 and 3.
In the formula, mlThe number of samples included in the ith fault type is shown, and num _ j represents the number of recognition results output by the classifier. CM (compact message processor)kThe fault is a confusion matrix of the Kth classifier, and the ith fault, the jth fault and the l fault are three types under n types of faults which are independent and not influenced mutually.
In said step 5, according to the confusion matrix CMkAnd distributing weights for each classifier according to the number of samples corresponding to each fault type.
Construction of evaluation matrix EMK(n × 1) is as follows:
EMK＝CMK·A·M
in the formula, A is a coefficient matrix of n multiplied by n, diagonal elements of the coefficient matrix are all 1, and non-diagonal elements of the coefficient matrix are all-1; m ═ M1,m2...ml...mn]TRepresenting a fault sample matrix, mlIndicating the number of samples contained under the i-th fault type.
The weight coefficient lambda corresponding to the Kth classifierKExpressed as:
in the step 6, X is usedKRepresenting the output result of the Kth classifier, comprehensively considering the weight coefficients of the classifiers, adding the weights of the classifiers with the same classification result, and storing the weights in the matrix BjThe method comprises the following steps:
Bj＝λp+λq,Xp＝Xq
in the formula, Xp＝XqDenotes that the output of the p-th and q-th classifiers are the same, lambdapAnd λqRepresenting the weight coefficients of the p-th classifier and the q-th classifier, and constructing a decision function F as follows:
F＝max(Bj),j＝1,2...t
t represents the matrix BjAnd taking the weight and the maximum category as the prediction result of the fault sample.
Drawings
FIG. 1 is a flow chart of a fault prediction method of an intelligent ammeter based on multi-classifier fusion according to the invention;
FIG. 2 is a schematic diagram of a multi-classifier fusion model.
Detailed Description
The invention is further described with reference to the following drawings and detailed description.
As shown in fig. 1, the process of the intelligent electric meter fault prediction method based on multi-classifier fusion of the present invention is as follows:
1. analyzing fault data information and fault types of the intelligent ammeter, and processing missing values and abnormal values of ammeter fault data acquired from the electricity utilization information acquisition system;
2. calculating a correlation coefficient between each characteristic attribute and a fault type by adopting a characteristic selection method, and eliminating the characteristic attribute with small correlation with the fault type to form a characteristic subset;
3. constructing a mixed sampling method, and eliminating data imbalance characteristics by adopting oversampling on a few samples and adopting an undersampling mode on a plurality of samples;
4. dividing a data set into a training group and a testing group according to a ratio of 8:3, respectively inputting samples of the training group and the testing group into three classifiers, namely a Support Vector Machine (SVM), a BP neural network and a random forest, training and testing, counting prediction results of the classifiers, and constructing a confusion matrix of the prediction results in a statistical probability mode;
5. combining the recognition capability of each classifier on different fault types, providing a weight distribution method, constructing an evaluation matrix, and calculating to obtain a weight coefficient corresponding to each classifier;
6. a decision function containing the weight coefficient of each classifier is constructed, the preprocessed training group and test group data are input into each classifier model for training and prediction, the weights of the classifiers with the same prediction result are added, the weight and the largest class are taken as the prediction result of the fault sample, the prediction result of the fault sample can be taken as a label corresponding to various fault types, and the fault of the ammeter can be determined according to the label, as shown in fig. 2.
In summary, the invention applies the multi-classifier fusion method to the fault prediction of the intelligent ammeter. According to the method, firstly, fault data of the intelligent electric meter are preprocessed, then, weight coefficients are distributed to all classifiers, a multi-classifier fusion decision function is further constructed, and the method can be effectively applied to fault identification research of the intelligent electric meter.
Claims (7)
1. A fault prediction method of an intelligent ammeter based on multi-classifier fusion is characterized in that data acquired from an electricity consumption information acquisition system are subjected to data preprocessing, and then are sequentially used as training samples and test samples to be input into a multi-classifier model adopting a fusion algorithm for training and prediction, so that fault types corresponding to the data samples are acquired, and the method comprises the following steps:
step 1: analyzing fault data information and fault types of the intelligent ammeter, and processing missing values and abnormal values of ammeter fault data acquired from the electricity utilization information acquisition system;
step 2: calculating a correlation coefficient between each characteristic attribute and a fault type by adopting a characteristic selection method, and eliminating the characteristic attribute with small correlation with the fault type to form a characteristic subset;
and step 3: constructing a mixed sampling method, and eliminating data imbalance characteristics by adopting oversampling on a few samples and adopting an undersampling mode on a plurality of samples;
and 4, step 4: dividing a data set into a training group and a testing group according to a ratio of 8:3, respectively inputting samples of the training group and the testing group into three classifiers, namely a Support Vector Machine (SVM), a BP neural network and a random forest, training and testing, counting prediction results of the classifiers, and constructing a confusion matrix of the prediction results in a statistical probability mode;
and 5: combining the recognition capability of each classifier on different fault types, providing a weight distribution method, constructing an evaluation matrix, and calculating to obtain a weight coefficient corresponding to each classifier;
step 6: and constructing a decision function containing the weight coefficient of each classifier, inputting the preprocessed training group and test group data into each classifier model for training and prediction, adding the weights of the classifiers with the same prediction result, and taking the weight and the largest class as the prediction result of the sample.
2. The method for predicting the fault of the intelligent ammeter based on the fusion of the plurality of classifiers according to claim 1, wherein in the step 1, the method for processing the missing value and the abnormal value of the fault data of the ammeter comprises the following steps:
for a time series X [ X ]1,x2...xn]The mean value mu and the variance sigma of the missing data are calculated by a method of complementing according to normal distribution, and data conforming to the normal distribution are generated:
n (μ, σ) represents a normal distribution with a mean μ and a variance σ, and the abnormal value is determined using a box plot, and is defined as follows:
[Q1-1.5IQR,Q3+1.5IQR]
IQR＝Q3-Q1
wherein, Q1 and Q3 respectively represent the first quartile and the third quartile of the data set, IQR is the threshold judgment range, and data which do not meet the threshold judgment range are all considered as abnormal values and need to be deleted or replaced.
3. The method according to claim 1, wherein in the step 2, the feature subset forming method comprises:
assume aThe fault data set is in the form of D { (x)1,y1),(x2,y2)...(xnum,ynum) In which xiFeature attribute information representing the ith sample, with a dimension N, num 1,2,3iAnd representing the fault type of the ith sample, the correlation coefficient between each characteristic attribute and the fault type in the ith sample can be represented as:
ri＝[ρ1,ρ2...ρk...ρN]
where ρ iskRepresenting the correlation coefficient between the kth characteristic attribute and the fault type, riRepresents the set of correlation coefficients, p, for the ith fault typekThe calculation process of (2) is as follows:
wherein λ iskRepresents the kth feature;
by removing the characteristic attributes with small correlation with the fault types, redundant characteristics and irrelevant characteristics are eliminated, and a characteristic subset is formed.
4. The method for predicting the fault of the intelligent ammeter based on the fusion of the plurality of classifiers as claimed in claim 1, wherein in the step 3, the method for eliminating the unbalanced data comprises the following steps:
assuming that the data set comprises M kinds of characteristic attributes in total, counting the number U of samples corresponding to each characteristic attribute in the data setjM, calculating an average Mean _ U of the number of samples corresponding to the M kinds of characteristic attributes:
if the number of samples corresponding to the j-th class characteristic attribute is smaller than the Mean _ U, adopting an oversampling mode, otherwise adopting an undersampling mode, wherein the calculation formula of the number of the sampled samples is as follows:
wherein, UjRepresenting a sample of fault data, U ', before sampling'jAnd the Mean _ U represents the average value of the number of samples corresponding to the M characteristic attributes.
5. The method for predicting the fault of the smart meter based on the fusion of the multiple classifiers according to claim 1, wherein in the step 4, the confusion matrix is constructed by the following steps:
for 3 classifiers of a Support Vector Machine (SVM), a BP neural network and a random forest, assuming that n fault types exist in the intelligent electric meter, the dimension of a confusion matrix is n × n and is represented as:
wherein the content of the first and second substances,
representing the statistical probability that the kth classifier correctly identified the sample fault type,
representing the statistical probability that the Kth classifier wrongly identifies the ith fault as the jth fault, wherein the value of K is 1,2 and 3;
in the formula, mlThe number of samples contained in the I type fault type is represented, and num _ j represents the number of the identification results output by the classifier; CM (compact message processor)kFor the confusion matrix of the Kth classifier, class iThe faults are three types under n types of faults which are independent from each other and do not influence each other.
6. The method for predicting the fault of the intelligent electric meter based on the fusion of the multiple classifiers according to claim 1, wherein in the step 5, the weight coefficient distribution method of each classifier is as follows:
construction of evaluation matrix EMK(n × 1) is as follows:
EMK＝CMK·A·M
in the formula, A is a coefficient matrix of n multiplied by n, diagonal elements of the coefficient matrix are all 1, and non-diagonal elements of the coefficient matrix are all-1; m ═ M1,m2...ml...mn]TRepresenting a fault sample matrix, mlThe number of samples contained in the ith fault type is represented, and n is the fault type of the intelligent ammeter;
the weight coefficient lambda corresponding to the Kth classifierKExpressed as:
EMKis the evaluation matrix corresponding to the Kth classifier, mlIndicating the number of samples contained under the i-th fault type.
7. The method for predicting the fault of the intelligent electric meter based on the fusion of the multiple classifiers according to claim 1, wherein in the step 6, the decision function is constructed by the following steps:
by XKRepresenting the output result of the Kth classifier, comprehensively considering the weight coefficients of the classifiers, adding the weights of the classifiers with the same classification result, and storing the weights in the matrix BjThe method comprises the following steps:
Bj＝λp+λq,Xp＝Xq
in the formula, Xp＝XqDenotes that the output of the p-th and q-th classifiers are the same, lambdapAnd λqRepresenting the weight coefficients of the p-th classifier and the q-th classifier, and constructing a decision function F as follows:
F＝max(Bj),j＝1,2...t
t represents the matrix BjAnd taking the weight and the maximum category as the classification result of the fault sample.
Priority Applications (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN202110472639.XA CN113011530B (en) | 2021-04-29 | 2021-04-29 | Intelligent ammeter fault prediction method based on multi-classifier fusion |
Applications Claiming Priority (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN202110472639.XA CN113011530B (en) | 2021-04-29 | 2021-04-29 | Intelligent ammeter fault prediction method based on multi-classifier fusion |
Publications (2)
| Publication Number | Publication Date |
|---|---|
| CN113011530A true CN113011530A (en) | 2021-06-22 |
| CN113011530B CN113011530B (en) | 2023-04-07 |
Family
ID=76380620
Family Applications (1)
| Application Number | Title | Priority Date | Filing Date |
|---|---|---|---|
| CN202110472639.XA Active CN113011530B (en) | 2021-04-29 | 2021-04-29 | Intelligent ammeter fault prediction method based on multi-classifier fusion |
Country Status (1)
| Country | Link |
|---|---|
| CN (1) | CN113011530B (en) |
Cited By (6)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN113469235A (en) * | 2021-06-24 | 2021-10-01 | 珠海卓邦科技有限公司 | Water fluctuation abnormality identification method and device, computer device and storage medium |
| CN113568739A (en) * | 2021-07-12 | 2021-10-29 | 北京淇瑀信息科技有限公司 | User resource limit distribution method and device and electronic equipment |
| CN113792825A (en) * | 2021-11-17 | 2021-12-14 | 国网江苏省电力有限公司营销服务中心 | Fault classification model training method and device for electricity information acquisition equipment |
| CN114297054A (en) * | 2021-12-17 | 2022-04-08 | 北京交通大学 | Software defect number prediction method based on subspace mixed sampling |
| CN114613120A (en) * | 2022-03-29 | 2022-06-10 | 新奥（中国）燃气投资有限公司 | Remote meter reading abnormity identification method and device |
| CN114925938A (en) * | 2022-07-18 | 2022-08-19 | 武汉格蓝若智能技术有限公司 | Electric energy meter running state prediction method and device based on self-adaptive SVM model |
Citations (14)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| WO2010077830A2 (en) * | 2008-12-15 | 2010-07-08 | Accenture Global Services Gmbh | Power grid outage and fault condition management |
| CN106355030A (en) * | 2016-09-20 | 2017-01-25 | 浙江大学 | Fault detection method based on analytic hierarchy process and weighted vote decision fusion |
| CN106603293A (en) * | 2016-12-20 | 2017-04-26 | 南京邮电大学 | Network fault diagnosis method based on deep learning in virtual network environment |
| CN109376944A (en) * | 2018-11-13 | 2019-02-22 | 国网宁夏电力有限公司电力科学研究院 | The construction method and device of intelligent electric meter prediction model |
| CN109978049A (en) * | 2019-03-25 | 2019-07-05 | 山东劳动职业技术学院（山东劳动技师学院） | A kind of classification prediction technique of big data |
| CN110008276A (en) * | 2019-04-26 | 2019-07-12 | 清湖光旭数据科技(北京)有限公司 | A kind of method, device and equipment detecting ammeter exception |
| CN110634080A (en) * | 2018-06-25 | 2019-12-31 | 中兴通讯股份有限公司 | Abnormal electricity utilization detection method, device, equipment and computer readable storage medium |
| CN110794360A (en) * | 2019-10-22 | 2020-02-14 | 中国电力科学研究院有限公司 | Method and system for predicting fault of intelligent electric energy meter based on machine learning |
| CN110826611A (en) * | 2019-10-30 | 2020-02-21 | 华南理工大学 | Stacking sewage treatment fault diagnosis method based on weighted integration of multiple meta-classifiers |
| CN111142060A (en) * | 2019-12-02 | 2020-05-12 | 国网浙江省电力有限公司 | Self-adaptive threshold adjustment diagnosis method based on improved BP neural network |
| CN111967512A (en) * | 2020-08-07 | 2020-11-20 | 国网江苏省电力有限公司电力科学研究院 | Abnormal electricity utilization detection method, system and storage medium |
| CN112085335A (en) * | 2020-08-10 | 2020-12-15 | 国网上海市电力公司 | Improved random forest algorithm for power distribution network fault prediction |
| CN112084237A (en) * | 2020-09-09 | 2020-12-15 | 广东电网有限责任公司中山供电局 | Power system abnormity prediction method based on machine learning and big data analysis |
| CN112395558A (en) * | 2020-11-27 | 2021-02-23 | 广东电网有限责任公司肇庆供电局 | Improved unbalanced data hybrid sampling method suitable for historical fault data of intelligent electric meter |
-
2021
- 2021-04-29 CN CN202110472639.XA patent/CN113011530B/en active Active
Patent Citations (14)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| WO2010077830A2 (en) * | 2008-12-15 | 2010-07-08 | Accenture Global Services Gmbh | Power grid outage and fault condition management |
| CN106355030A (en) * | 2016-09-20 | 2017-01-25 | 浙江大学 | Fault detection method based on analytic hierarchy process and weighted vote decision fusion |
| CN106603293A (en) * | 2016-12-20 | 2017-04-26 | 南京邮电大学 | Network fault diagnosis method based on deep learning in virtual network environment |
| CN110634080A (en) * | 2018-06-25 | 2019-12-31 | 中兴通讯股份有限公司 | Abnormal electricity utilization detection method, device, equipment and computer readable storage medium |
| CN109376944A (en) * | 2018-11-13 | 2019-02-22 | 国网宁夏电力有限公司电力科学研究院 | The construction method and device of intelligent electric meter prediction model |
| CN109978049A (en) * | 2019-03-25 | 2019-07-05 | 山东劳动职业技术学院（山东劳动技师学院） | A kind of classification prediction technique of big data |
| CN110008276A (en) * | 2019-04-26 | 2019-07-12 | 清湖光旭数据科技(北京)有限公司 | A kind of method, device and equipment detecting ammeter exception |
| CN110794360A (en) * | 2019-10-22 | 2020-02-14 | 中国电力科学研究院有限公司 | Method and system for predicting fault of intelligent electric energy meter based on machine learning |
| CN110826611A (en) * | 2019-10-30 | 2020-02-21 | 华南理工大学 | Stacking sewage treatment fault diagnosis method based on weighted integration of multiple meta-classifiers |
| CN111142060A (en) * | 2019-12-02 | 2020-05-12 | 国网浙江省电力有限公司 | Self-adaptive threshold adjustment diagnosis method based on improved BP neural network |
| CN111967512A (en) * | 2020-08-07 | 2020-11-20 | 国网江苏省电力有限公司电力科学研究院 | Abnormal electricity utilization detection method, system and storage medium |
| CN112085335A (en) * | 2020-08-10 | 2020-12-15 | 国网上海市电力公司 | Improved random forest algorithm for power distribution network fault prediction |
| CN112084237A (en) * | 2020-09-09 | 2020-12-15 | 广东电网有限责任公司中山供电局 | Power system abnormity prediction method based on machine learning and big data analysis |
| CN112395558A (en) * | 2020-11-27 | 2021-02-23 | 广东电网有限责任公司肇庆供电局 | Improved unbalanced data hybrid sampling method suitable for historical fault data of intelligent electric meter |
Cited By (9)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN113469235A (en) * | 2021-06-24 | 2021-10-01 | 珠海卓邦科技有限公司 | Water fluctuation abnormality identification method and device, computer device and storage medium |
| CN113469235B (en) * | 2021-06-24 | 2024-04-26 | 珠海卓邦科技有限公司 | Water fluctuation abnormality recognition method and device, computer device and storage medium |
| CN113568739A (en) * | 2021-07-12 | 2021-10-29 | 北京淇瑀信息科技有限公司 | User resource limit distribution method and device and electronic equipment |
| CN113792825A (en) * | 2021-11-17 | 2021-12-14 | 国网江苏省电力有限公司营销服务中心 | Fault classification model training method and device for electricity information acquisition equipment |
| CN114297054A (en) * | 2021-12-17 | 2022-04-08 | 北京交通大学 | Software defect number prediction method based on subspace mixed sampling |
| CN114297054B (en) * | 2021-12-17 | 2023-06-30 | 北京交通大学 | Software defect number prediction method based on subspace mixed sampling |
| CN114613120A (en) * | 2022-03-29 | 2022-06-10 | 新奥（中国）燃气投资有限公司 | Remote meter reading abnormity identification method and device |
| CN114925938A (en) * | 2022-07-18 | 2022-08-19 | 武汉格蓝若智能技术有限公司 | Electric energy meter running state prediction method and device based on self-adaptive SVM model |
| CN114925938B (en) * | 2022-07-18 | 2022-10-04 | 武汉格蓝若智能技术有限公司 | Electric energy meter running state prediction method and device based on self-adaptive SVM model |
Also Published As
| Publication number | Publication date |
|---|---|
| CN113011530B (en) | 2023-04-07 |
Similar Documents
| Publication | Publication Date | Title |
|---|---|---|
| CN113011530B (en) | Intelligent ammeter fault prediction method based on multi-classifier fusion | |
| CN105117602B (en) | A kind of metering device running status method for early warning | |
| CN108052528B (en) | A kind of storage equipment timing classification method for early warning | |
| CN111553590A (en) | Radar embedded health management system | |
| CN113792825B (en) | Fault classification model training method and device for electricity information acquisition equipment | |
| CN111738462B (en) | Fault first-aid repair active service early warning method for electric power metering device | |
| CN107967485A (en) | Electro-metering equipment fault analysis method and device | |
| CN109389145B (en) | Electric energy meter manufacturer evaluation method based on metering big data clustering model | |
| CN106570778A (en) | Big data-based data integration and line loss analysis and calculation method | |
| CN112906764B (en) | Communication safety equipment intelligent diagnosis method and system based on improved BP neural network | |
| CN114065605A (en) | Intelligent electric energy meter running state detection and evaluation system and method | |
| CN117034143B (en) | Distributed system fault diagnosis method and device based on machine learning | |
| CN114266289A (en) | Complex equipment health state assessment method | |
| CN114565038A (en) | Intelligent electric meter fault diagnosis method based on improved capsule network | |
| CN114239734A (en) | Distributed vehicle-mounted health management system | |
| CN114118588A (en) | Peak-facing summer power failure prediction method based on game feature extraction under clustering undersampling | |
| CN115617784A (en) | Data processing system and processing method for informationized power distribution | |
| CN115526258A (en) | Power system transient stability evaluation method based on Spearman correlation coefficient feature extraction | |
| CN117674119A (en) | Power grid operation risk assessment method, device, computer equipment and storage medium | |
| CN111143835B (en) | Non-invasive protection method for business logic of electric power metering system based on machine learning | |
| CN116776209A (en) | Method, system, equipment and medium for identifying operation state of gateway metering device | |
| CN112395558A (en) | Improved unbalanced data hybrid sampling method suitable for historical fault data of intelligent electric meter | |
| CN116714469A (en) | Charging pile health monitoring method, device, terminal and storage medium | |
| CN114167837B (en) | Intelligent fault diagnosis method and system for railway signal system | |
| CN115879046A (en) | Internet of things abnormal data detection method based on improved feature selection and hierarchical model |
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 | ||
| GR01 | Patent grant | ||
| GR01 | Patent grant |