CN109493249B - Analysis method of electricity consumption data on multiple time scales - Google Patents
Analysis method of electricity consumption data on multiple time scales Download PDFInfo
- Publication number
- CN109493249B CN109493249B CN201811308693.5A CN201811308693A CN109493249B CN 109493249 B CN109493249 B CN 109493249B CN 201811308693 A CN201811308693 A CN 201811308693A CN 109493249 B CN109493249 B CN 109493249B
- Authority
- CN
- China
- Prior art keywords
- power consumption
- user
- users
- clustering
- similarity
- 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
Links
- 230000005611 electricity Effects 0.000 title claims abstract description 48
- 238000004458 analytical method Methods 0.000 title abstract description 26
- 239000013598 vector Substances 0.000 claims abstract description 62
- 238000000034 method Methods 0.000 claims abstract description 37
- 230000003442 weekly effect Effects 0.000 claims abstract description 31
- 239000011159 matrix material Substances 0.000 claims abstract description 20
- 230000002159 abnormal effect Effects 0.000 claims description 22
- 238000013135 deep learning Methods 0.000 claims description 5
- 238000012545 processing Methods 0.000 description 14
- 230000006399 behavior Effects 0.000 description 10
- 238000007405 data analysis Methods 0.000 description 7
- 230000008569 process Effects 0.000 description 7
- 230000007774 longterm Effects 0.000 description 6
- 230000008859 change Effects 0.000 description 5
- 238000004422 calculation algorithm Methods 0.000 description 4
- 230000000694 effects Effects 0.000 description 3
- 238000005265 energy consumption Methods 0.000 description 3
- 238000000605 extraction Methods 0.000 description 3
- 238000007781 pre-processing Methods 0.000 description 3
- 238000004364 calculation method Methods 0.000 description 2
- 238000006243 chemical reaction Methods 0.000 description 2
- 230000006872 improvement Effects 0.000 description 2
- 238000002360 preparation method Methods 0.000 description 2
- 230000000750 progressive effect Effects 0.000 description 2
- 230000009467 reduction Effects 0.000 description 2
- 238000012216 screening Methods 0.000 description 2
- 238000007796 conventional method Methods 0.000 description 1
- 238000010586 diagram Methods 0.000 description 1
- 238000005516 engineering process Methods 0.000 description 1
- 230000003203 everyday effect Effects 0.000 description 1
- 239000000284 extract Substances 0.000 description 1
- 230000006870 function Effects 0.000 description 1
- 238000011835 investigation Methods 0.000 description 1
- 238000012986 modification Methods 0.000 description 1
- 230000004048 modification Effects 0.000 description 1
- 238000003062 neural network model Methods 0.000 description 1
- 238000012567 pattern recognition method Methods 0.000 description 1
- 238000003672 processing method Methods 0.000 description 1
Images
Classifications
-
- 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
- Y02—TECHNOLOGIES OR APPLICATIONS FOR MITIGATION OR ADAPTATION AGAINST CLIMATE CHANGE
- Y02D—CLIMATE CHANGE MITIGATION TECHNOLOGIES IN INFORMATION AND COMMUNICATION TECHNOLOGIES [ICT], I.E. INFORMATION AND COMMUNICATION TECHNOLOGIES AIMING AT THE REDUCTION OF THEIR OWN ENERGY USE
- Y02D10/00—Energy efficient computing, e.g. low power processors, power management or thermal management
Landscapes
- Business, Economics & Management (AREA)
- Health & Medical Sciences (AREA)
- Engineering & Computer Science (AREA)
- Economics (AREA)
- Public Health (AREA)
- Water Supply & Treatment (AREA)
- General Health & Medical Sciences (AREA)
- Human Resources & Organizations (AREA)
- Marketing (AREA)
- Primary Health Care (AREA)
- Strategic Management (AREA)
- Tourism & Hospitality (AREA)
- Physics & Mathematics (AREA)
- General Business, Economics & Management (AREA)
- General Physics & Mathematics (AREA)
- Theoretical Computer Science (AREA)
- Management, Administration, Business Operations System, And Electronic Commerce (AREA)
- Information Retrieval, Db Structures And Fs Structures Therefor (AREA)
- Supply And Distribution Of Alternating Current (AREA)
Abstract
The application discloses an analysis method of electricity consumption data on a multi-time scale, which comprises the following steps: forming a vector by taking the electricity consumption data of a user as a unit, and outputting a hidden layer feature vector by constructing an RBM (radial basis function) model; clustering hidden layer characteristic vectors to obtain different weekly power consumption mode clusters; calculating the similarity among clusters of different weekly power consumption modes; according to all weekly electricity consumption mode clustering results of the same user, constructing a yearly electricity consumption data vector of the user; generating a user similarity matrix according to the similarity between clusters of different weekly power consumption modes and the annual power consumption data vectors of all users; and clustering the annual power consumption data vectors of all the users according to the user similarity matrix to obtain different annual power consumption mode clusters of the users. By applying the method and the device, the analysis of the power utilization data can be conveniently carried out from a multi-time scale.
Description
Technical Field
The application relates to a data analysis technology, in particular to an analysis method of electricity consumption data on a multi-time scale.
Background
With the continuous improvement of the informatization degree of the power industry and the rapid popularization of the smart power grid, the power load data presents the trend of diversification and mass, and the power load analysis and the power utilization user side are written as the current focus of attention of the power industry. Common power load analysis mostly refers to single-day or annual power load data, lacks connection and extension in the time dimension, writes to the side of user and also stops simply clustering according to the power consumption index, and abnormal user investigation is often realized by the manual work. How to analyze the power load from multiple time dimensions, grasp more accurate power consumption patterns of users and automatically screen abnormal outlier users becomes a problem which needs to be solved at present.
The invention patent application with publication number CN107220906A discloses a multi-time scale electricity consumption abnormity analysis method based on an electricity consumption acquisition system, in the method, the multi-time scale comes from the multi-scale of an acquisition device, namely, a day, month and year data set is processed, so that data with different time scales needs to be processed respectively, indexes are selected respectively for analysis, and the processing complexity is high.
The invention patent application with publication number CN106529707A discloses a load electricity consumption pattern recognition method, which calculates physical indexes of electricity consumption patterns and then performs clustering according to the indexes. Under the processing mode, the selection and the definition of the indexes have great influence on the result, and the processing result is unstable.
As can be seen from the above, the current multi-time scale data analysis is problematic.
Disclosure of Invention
The application provides an analysis method of electricity consumption data on a multi-time scale, which can analyze electricity consumption loads from a multi-time dimension.
In order to achieve the purpose, the following technical scheme is adopted in the application:
a method for analyzing electricity usage data on multiple time scales, comprising:
forming vectors by taking week as a unit for the electricity consumption data of a user, performing deep learning of hidden layer features by constructing an RBM restricted Boltzmann machine model, and outputting hidden layer feature vectors;
clustering the hidden layer characteristic vectors to obtain different weekly power consumption mode clusters; calculating the similarity among clusters of different weekly power consumption modes;
according to all weekly electricity consumption mode clustering results of the same user, constructing a yearly electricity consumption data vector of the user;
generating a user similarity matrix according to the similarity between clusters of different weekly power consumption modes and the annual power consumption data vectors of all users, wherein each point in the user similarity matrix represents the similarity of the annual power consumption data vectors between every two users;
and performing primary clustering on the annual power consumption data vectors of all the users according to the user similarity matrix to obtain different annual power consumption mode clusters of the users.
Preferably, after the annual electricity consumption data vectors of all the users are subjected to primary clustering, the method further comprises the following steps: performing secondary clustering on the DBSCAN in each obtained annual power utilization mode cluster of the user, determining annual power utilization data vectors of users with abnormal outliers, and taking the corresponding users as the abnormal outlier users; and outputting the annual power utilization mode of the user after secondary clustering.
Preferably, the hidden layer feature vectors are clustered by the KMeans clustering method.
Preferably, the generating the user similarity matrix includes:
and for any two users, determining the similarity between the weekly power consumption modes of the two users according to the similarity between the clusters of the different weekly power consumption modes, and counting the sum of the similarity of each week as the similarity of the annual power consumption data vectors of the two users.
Preferably, annual electricity consumption data vectors of all users are primarily clustered through a KMeans clustering method.
Preferably, the performing quadratic clustering includes: and when secondary clustering is carried out, optimizing according to a preset outlier index to ensure that the outlier index is optimal.
Preferably, the outlier indicator is
Where Cin represents the average of the distances from all points to other points in the class, Cout represents the average of the distances from all points to each point in other classes, and cer represents the minimum of the average distances from outliers to all points in each class.
According to the technical scheme, the vector is formed by taking the electricity consumption data of the user as a unit, the deep learning of the hidden layer feature is carried out by constructing the RBM restricted Boltzmann machine model, and the hidden layer feature vector is output. And through learning of the hidden layer feature vector, feature extraction and data dimension reduction are realized so as to simplify subsequent processing. Next, clustering hidden layer feature vectors to obtain different weekly power consumption mode clusters; calculating the similarity among clusters of different weekly power consumption modes; and constructing a yearly electricity consumption data vector of the user according to all weekly electricity consumption mode clustering results of the same user. And the annual power consumption vector statistics is carried out on the power consumption habits of the user in each week, and the subsequent clustering process is simplified on the basis of keeping the short-term power consumption mode. Generating a user similarity matrix according to the similarity between clusters of different weekly power consumption modes and the annual power consumption data vectors of all users; and according to the user similarity matrix, clustering the annual power consumption data vectors of all the users to obtain different annual power consumption mode clusters and abnormal outlier user clusters of the users. By the mode, the user power load analysis is realized from the time dimension of the week and the year by utilizing the one-time process, the short-term week power utilization mode and the long-term year power utilization behavior of the user are obtained simultaneously, the abnormal outlier user is screened out after the user is subjected to class division, and the clustering processing is not required to be carried out by forming vectors by all data of the user in the whole year.
Drawings
FIG. 1 is a schematic overall concept flow diagram of a method for analyzing power consumption data on multiple time scales in the present application;
fig. 2 is a schematic flow chart of a method for analyzing power consumption data on multiple time scales according to the present application.
Detailed Description
For the purpose of making the objects, technical means and advantages of the present application more apparent, the present application will be described in further detail with reference to the accompanying drawings.
In order to extract and analyze the user electricity utilization modes and classify and screen the users, the method for deeply clustering the electricity utilization loads considering different time scales is designed and realized, and an RBM limited Boltzmann machine feature learning algorithm is combined. The whole idea flow of the application is shown in fig. 1.
As can be seen from fig. 1, the method for analyzing the power consumption load data in the present application can implement deep clustering of the power consumption load on multiple time scales, and specifically includes three stages, namely short-term power consumption mode clustering, power consumption scale conversion, long-term user clustering and abnormal user analysis. Short-term power utilization pattern clustering extracts and divides the user cycle power utilization patterns from the angles of cycle, amplitude, peak-valley difference and the like, so that the analysis of power utilization load under a short time scale is realized; the electricity utilization scale conversion combines a short time scale and a long time scale, converts a typical electricity utilization mode into annual electricity utilization behavior change of a user, and prepares for user analysis under the next long time scale; and the long-term user clustering and the abnormal user analysis realize the division of the users, acquire the electricity utilization characteristics of typical users and screen out abnormal electricity utilization users. In the whole process, only 24-point electricity utilization data of the user all the year is input, and short-term electricity utilization modes, long-term electricity utilization behavior changes and abnormal electricity utilization users are output through progressive learning analysis from a short term to a long term layer by layer.
The analytical method in the present application is described in detail below. FIG. 2 is a flow chart showing the analysis method of the present application. As shown in fig. 2, the method includes:
The electricity consumption data of the user is input in a week unit, that is, the electricity consumption data of the user is formed into a vector in a week unit. The total amount of data input comprises M-week electricity consumption data of N users. And if the annual data analysis of the user is finally needed, the power utilization data of each week of the year of the user needs to be input. It can be seen that, in the present application, the annual power consumption data analysis can be finally realized, but the original input does not need to be input in units of annual power consumption data, but the power consumption data is input in units of weeks, so that the analysis is performed from a short-term perspective, and the processing complexity is reduced.
After the data is obtained, the dirty data is cleaned through data preprocessing. The specific preprocessing is the same as the conventional method, such as data completion and other operations.
And after the data preprocessing is finished, constructing an RBM restricted Boltzmann machine model, and performing a learning process of hidden layer characteristics. The RBM is a limited Boltzmann machine, is an unsupervised neural network model, and is constructed in a process of deep learning in an iterative mode. Specifically, the RBM is composed of a visible layer and a hidden layer, which correspond to the original input data and the hidden layer characteristics, respectively. The whole network is trained by reconstructing input data as much as possible at the output end, when the algorithm is converged in iteration or reaches a preset iteration round, the algorithm is stopped, the output result is the learned hidden layer feature at the moment, the hidden layer feature is a more powerful and ordered expression of the input data, the hidden layer feature can be regarded as a high degree of extraction and analysis of the input data, and the method has very good functions of autonomous feature extraction and data dimension reduction. Through learning the electricity utilization data of the user in the week unit in the step, the cycle and fluctuation characteristics which are difficult to find manually are expected to be obtained, and the accuracy of the next analysis is improved.
Different clusters are obtained through clustering processing on the hidden layer characteristics, and correspond to a typical 7-day power consumption mode, so that power consumption load analysis under a short time scale is completed. In the clustering process, a distance-based clustering method, such as the KMeans clustering method, is generally used.
After the processing of steps 201-202, the user data characteristics of 7 days are extracted, the classification processing is completed, and data preparation is carried out for the subsequent long-time scale data analysis.
In this step, according to the 7-day clustering result obtained in step 202, the annual power consumption data of the user are labeled, and an annual power consumption behavior model of the user is constructed. Specifically, the clustering results of the annual weekly power consumption patterns of the same user are combined together to form a vector (hereinafter referred to as an annual power consumption data vector of the user) representing the annual power consumption behavior model of the user, so that the annual power consumption behavior model of the corresponding user is obtained corresponding to all users.
It should be noted that, since the annual energy consumption data vector uses the clustering result in the foregoing step 202 as the input data, each data point represents the short-term energy consumption pattern of the current week, that is, the annual energy consumption data vector in this step is not a vector directly formed by the original annual data, but a vector processed by the step 201 and the step 203. Meanwhile, the annual power consumption data vector is formed, so that the annual power consumption data analysis based on the annual power consumption data vector is compared with the analysis by directly forming the vector by using the original annual power consumption data of the user, the short-term power consumption mode information is reserved, and the processing complexity is greatly reduced.
In addition, preparation is made for similarity calculation of user annual data, and in the step, the similarity between weekly power consumption pattern clusters is calculated by using 7-day clustering results. The specific similarity calculation method adopts the existing method, and is not described herein again.
And step 204, generating a user similarity matrix according to the similarity among the clusters of different weekly power consumption modes and the annual power consumption data vectors of all users.
And comparing every two annual data vectors of all users according to the similarity between clusters of different weekly power consumption modes and the annual power consumption data vectors of all users to obtain a user similarity matrix. Each point in the user similarity matrix represents the similarity of annual power consumption data vectors between every two users.
Specifically, the specific way of generating the user similarity matrix may be: for any two users A and B, according to the similarity between different weekly electricity utilization mode clusters, determining the similarity between the weekly electricity utilization modes of the users A and B, and counting the sum of the weekly similarities as the similarity of the annual electricity utilization data vectors of the users A and B.
The user similarity matrix generated in the step is used for carrying out user clustering and analysis under the following long-time scale.
And step 205, performing primary clustering on the annual power consumption data vectors of all the users according to the user similarity matrix to obtain different annual power consumption mode clusters of the users.
And after the user similarity matrix is obtained, clustering the user behavior data according to the matrix to obtain a user division result. The clustering process may adopt a KMeans clustering method. Therefore, annual power utilization mode division of different users can be obtained.
So far, the most basic data analysis method flow in the present application is finished.
Further, to implement automatic screening of users with abnormal outliers, it is preferable that the following processing is further included:
The basic classification of the user is achieved by the processing of step 205. In this step, improved secondary clustering is performed on each sub-user cluster to determine abnormal outlier users.
Specifically, during secondary clustering, an outlier index can be predefined, and clustering parameters of each subclass are automatically optimized through grid search, so that the outlier index is optimal; and finally, analyzing the clustering result, outputting more accurate typical user power consumption modes and abnormal outlier users, and performing side writing on the users in a long time scale. The outlier index can be selected according to actual needs, for example, the distance between the annual power consumption data vector of the outlier and other data vectors.
Specifically, the common indexes for measuring the clustering effect, such as contour coefficients, landed indexes, mutual information, and the like, cannot consider the influence of abnormal outliers on the clustering result, and the clustering result is invalid when only one type exists. To address this issue, the outlier Index Discrete Index can be defined as follows:
where Cin represents the average of the distances from all points to other points in the class, Cout represents the average of the distances from all points to each point in other classes, and cer represents the minimum of the average distances from outliers to all points in each class. The larger the outlier index is, the farther the abnormal point is from the clustering sub-clusters, the higher the cohesion of the clustering sub-clusters is, the farther the clusters are, and the better clustering effect is achieved.
The above is the specific implementation of the analysis method for the user electricity consumption data in the application. Through the processing, the user electricity utilization data can be analyzed from a multi-time scale, the processing complexity is reduced, and meanwhile, the screening of abnormal outlier users can be automatically realized.
Compared with the application mentioned in the background part, the processing method of the application has the following advantages:
1. the method and the device start from 24 points of data of a user every day, and analyze the weekly power utilization mode and the annual power utilization behavior change, and are progressive one by one. Meanwhile, a deep clustering mode is adopted, different time scales are combined, and comprehensive analysis is performed to form uniform output. Besides, the method and the device can obtain the typical weekly power consumption mode and the annual behavior change model of the user. The treatment effects cannot be achieved by the invention patent application with the publication number of CN 107220906A;
2. according to the method and the device, additional domain knowledge is not needed, the 24-point power consumption of the user is fully mined, and the short-term and long-term trends and the change rules of the 24-point power consumption are obtained through deep clustering. The invention patent application with the publication number of CN106529707A is analyzed through a density clustering and gravitation searching algorithm, deep features are learned through an RBM hidden variable model, deep clustering based on outlier indexes is carried out through time scale change, and therefore, the power utilization mode can be identified, user clusters can be divided, and abnormal users can be screened.
The above description is only for the purpose of illustrating the preferred embodiments of the present invention and is not to be construed as limiting the invention, and any modifications, equivalents, improvements and the like made within the spirit and principle of the present invention should be included in the scope of the present invention.
Claims (7)
1. A method for analyzing power consumption data on multiple time scales is characterized by comprising the following steps:
forming vectors by taking week as a unit for the electricity consumption data of a user, performing deep learning of hidden layer features by constructing an RBM restricted Boltzmann machine model, and outputting hidden layer feature vectors;
clustering the hidden layer characteristic vectors to obtain different weekly power consumption mode clusters; calculating the similarity among clusters of different weekly power consumption modes;
according to all weekly electricity consumption mode clustering results of the same user, constructing a yearly electricity consumption data vector of the user;
generating a user similarity matrix according to the similarity between clusters of different weekly power consumption modes and the annual power consumption data vectors of all users, wherein each point in the user similarity matrix represents the similarity of the annual power consumption data vectors between every two users;
and performing primary clustering on the annual power consumption data vectors of all the users according to the user similarity matrix to obtain different annual power consumption mode clusters of the users.
2. The method of claim 1, wherein after initially clustering the annual electricity consumption data vectors for all users, the method further comprises: performing secondary clustering on the DBSCAN in each obtained annual power utilization mode cluster of the user, determining annual power utilization data vectors of users with abnormal outliers, and taking the corresponding users as the abnormal outlier users; and outputting the annual power utilization mode of the user after secondary clustering.
3. The method according to claim 1 or 2, characterized in that the hidden layer feature vectors are clustered by the KMeans clustering method.
4. The method of claim 1 or 2, wherein the generating the user similarity matrix comprises:
and for any two users, determining the similarity between the weekly power consumption modes of the two users according to the similarity between the clusters of the different weekly power consumption modes, and counting the sum of the similarity of each week as the similarity of the annual power consumption data vectors of the two users.
5. The method according to claim 1 or 2, characterized in that the annual electricity consumption data vectors of all users are primarily clustered by the KMeans clustering method.
6. The method of claim 2, wherein performing DBSCAN quadratic clustering comprises: and when secondary clustering is carried out on the DBSCAN, optimizing according to the preset outlier index to ensure that the outlier index is optimal.
7. The method of claim 6, wherein the outlier indicator is
Where Cin represents the average of the distances from all points to other points in the class, Cout represents the average of the distances from all points to each point in other classes, and cer represents the minimum of the average distances from outliers to all points in each class.
Priority Applications (1)
Application Number | Priority Date | Filing Date | Title |
---|---|---|---|
CN201811308693.5A CN109493249B (en) | 2018-11-05 | 2018-11-05 | Analysis method of electricity consumption data on multiple time scales |
Applications Claiming Priority (1)
Application Number | Priority Date | Filing Date | Title |
---|---|---|---|
CN201811308693.5A CN109493249B (en) | 2018-11-05 | 2018-11-05 | Analysis method of electricity consumption data on multiple time scales |
Publications (2)
Publication Number | Publication Date |
---|---|
CN109493249A CN109493249A (en) | 2019-03-19 |
CN109493249B true CN109493249B (en) | 2021-11-12 |
Family
ID=65693789
Family Applications (1)
Application Number | Title | Priority Date | Filing Date |
---|---|---|---|
CN201811308693.5A Active CN109493249B (en) | 2018-11-05 | 2018-11-05 | Analysis method of electricity consumption data on multiple time scales |
Country Status (1)
Country | Link |
---|---|
CN (1) | CN109493249B (en) |
Families Citing this family (1)
Publication number | Priority date | Publication date | Assignee | Title |
---|---|---|---|---|
CN112801821A (en) * | 2021-02-25 | 2021-05-14 | 国网山东省电力公司经济技术研究院 | Engineering construction team portrait method and system based on big data analysis |
Citations (5)
Publication number | Priority date | Publication date | Assignee | Title |
---|---|---|---|---|
CN104572940A (en) * | 2014-12-30 | 2015-04-29 | 中国人民解放军海军航空工程学院 | Automatic image annotation method based on deep learning and canonical correlation analysis |
CN106296465A (en) * | 2016-08-23 | 2017-01-04 | 四川大学 | A kind of intelligent grid exception electricity consumption behavioral value method |
CN107220906A (en) * | 2017-05-31 | 2017-09-29 | 国网上海市电力公司 | Multiple Time Scales multiplexing electric abnormality analysis method based on electricity consumption acquisition system |
CN108345908A (en) * | 2018-02-10 | 2018-07-31 | 武汉轻工大学 | Sorting technique, sorting device and the storage medium of electric network data |
CN108460410A (en) * | 2018-02-08 | 2018-08-28 | 合肥工业大学 | Electricity consumption mode identification method and system, the storage medium of citizen requirement side |
-
2018
- 2018-11-05 CN CN201811308693.5A patent/CN109493249B/en active Active
Patent Citations (5)
Publication number | Priority date | Publication date | Assignee | Title |
---|---|---|---|---|
CN104572940A (en) * | 2014-12-30 | 2015-04-29 | 中国人民解放军海军航空工程学院 | Automatic image annotation method based on deep learning and canonical correlation analysis |
CN106296465A (en) * | 2016-08-23 | 2017-01-04 | 四川大学 | A kind of intelligent grid exception electricity consumption behavioral value method |
CN107220906A (en) * | 2017-05-31 | 2017-09-29 | 国网上海市电力公司 | Multiple Time Scales multiplexing electric abnormality analysis method based on electricity consumption acquisition system |
CN108460410A (en) * | 2018-02-08 | 2018-08-28 | 合肥工业大学 | Electricity consumption mode identification method and system, the storage medium of citizen requirement side |
CN108345908A (en) * | 2018-02-10 | 2018-07-31 | 武汉轻工大学 | Sorting technique, sorting device and the storage medium of electric network data |
Non-Patent Citations (1)
Title |
---|
A Fused Load Curve Clustering Algorithm Based on Wavelet Transform;Zigui Jiang等;《IEEE TRANSACTIONS ON INDUSTRIAL INFORMATICS》;20180531;第14卷(第5期);1856-1865 * |
Also Published As
Publication number | Publication date |
---|---|
CN109493249A (en) | 2019-03-19 |
Similar Documents
Publication | Publication Date | Title |
---|---|---|
CN111444236B (en) | Mobile terminal user portrait construction method and system based on big data | |
CN107122375B (en) | Image subject identification method based on image features | |
Zandkarimi et al. | A generic framework for trace clustering in process mining | |
Berti-Equille | Learn2clean: Optimizing the sequence of tasks for web data preparation | |
Vercamer et al. | Predicting consumer load profiles using commercial and open data | |
CN106250461A (en) | A kind of algorithm utilizing gradient lifting decision tree to carry out data mining based on Spark framework | |
CN110148023A (en) | The electric power integral Method of Commodity Recommendation and system that logic-based returns | |
CN104035996A (en) | Domain concept extraction method based on Deep Learning | |
CN112330078B (en) | Power consumption prediction method and device, computer equipment and storage medium | |
CN112926645B (en) | Electricity stealing detection method based on edge calculation | |
CN112819299A (en) | Differential K-means load clustering method based on center optimization | |
CN111177216A (en) | Association rule generation method and device for behavior characteristics of comprehensive energy consumer | |
CN111008726A (en) | Class image conversion method in power load prediction | |
CN108427756A (en) | Personalized query word completion recommendation method and device based on same-class user model | |
CN113052225A (en) | Alarm convergence method and device based on clustering algorithm and time sequence association rule | |
CN110457472A (en) | The emotion association analysis method for electric business product review based on SOM clustering algorithm | |
CN113094448B (en) | Analysis method and analysis device for residence empty state and electronic equipment | |
CN109493249B (en) | Analysis method of electricity consumption data on multiple time scales | |
Yu et al. | A novel discriminative dictionary pair learning constrained by ordinal locality for mixed frequency data classification | |
CN111090679B (en) | Time sequence data representation learning method based on time sequence influence and graph embedding | |
CN113034193A (en) | Working method for modeling of APP2VEC in wind control system | |
Hamidzadeh et al. | Clustering data stream with uncertainty using belief function theory and fading function | |
CN116450827A (en) | Event template induction method and system based on large-scale language model | |
CN112148942A (en) | Business index data classification method and device based on data clustering | |
Wålinder | Evaluation of logistic regression and random forest classification based on prediction accuracy and metadata analysis |
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 |