CN114611396B - Line loss analysis method based on big data - Google Patents
Line loss analysis method based on big data Download PDFInfo
- Publication number
- CN114611396B CN114611396B CN202210254768.6A CN202210254768A CN114611396B CN 114611396 B CN114611396 B CN 114611396B CN 202210254768 A CN202210254768 A CN 202210254768A CN 114611396 B CN114611396 B CN 114611396B
- Authority
- CN
- China
- Prior art keywords
- data
- line loss
- feature extraction
- parameter
- classification result
- 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
Images
Classifications
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06F—ELECTRIC DIGITAL DATA PROCESSING
- G06F30/00—Computer-aided design [CAD]
- G06F30/20—Design optimisation, verification or simulation
- G06F30/27—Design optimisation, verification or simulation using machine learning, e.g. artificial intelligence, neural networks, support vector machines [SVM] or training a model
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06F—ELECTRIC DIGITAL DATA PROCESSING
- G06F17/00—Digital computing or data processing equipment or methods, specially adapted for specific functions
- G06F17/10—Complex mathematical operations
- G06F17/16—Matrix or vector computation, e.g. matrix-matrix or matrix-vector multiplication, matrix factorization
-
- 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
- G06F18/00—Pattern recognition
- G06F18/20—Analysing
- G06F18/22—Matching criteria, e.g. proximity measures
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06F—ELECTRIC DIGITAL DATA PROCESSING
- G06F18/00—Pattern recognition
- G06F18/20—Analysing
- G06F18/24—Classification techniques
-
- 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
-
- 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
- G06F—ELECTRIC DIGITAL DATA PROCESSING
- G06F2119/00—Details relating to the type or aim of the analysis or the optimisation
- G06F2119/10—Noise analysis or noise optimisation
-
- 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)
- General Physics & Mathematics (AREA)
- Evolutionary Computation (AREA)
- General Engineering & Computer Science (AREA)
- Artificial Intelligence (AREA)
- Life Sciences & Earth Sciences (AREA)
- Mathematical Physics (AREA)
- Computer Vision & Pattern Recognition (AREA)
- Software Systems (AREA)
- Evolutionary Biology (AREA)
- Bioinformatics & Computational Biology (AREA)
- Bioinformatics & Cheminformatics (AREA)
- Computing Systems (AREA)
- Computational Linguistics (AREA)
- Computational Mathematics (AREA)
- Molecular Biology (AREA)
- Biophysics (AREA)
- Biomedical Technology (AREA)
- Health & Medical Sciences (AREA)
- Pure & Applied Mathematics (AREA)
- Mathematical Optimization (AREA)
- Mathematical Analysis (AREA)
- General Health & Medical Sciences (AREA)
- Geometry (AREA)
- Computer Hardware Design (AREA)
- Medical Informatics (AREA)
- Algebra (AREA)
- Databases & Information Systems (AREA)
- Management, Administration, Business Operations System, And Electronic Commerce (AREA)
- Supply And Distribution Of Alternating Current (AREA)
Abstract
The invention provides a line loss analysis method, which is characterized in that data to be analyzed is input into a line loss identification network model, and the line loss identification network model comprises an input layer, a shallow layer feature extraction module, a first classifier, a deep layer feature extraction module and a second classifier; inputting a first feature obtained by shallow feature extraction into a first classifier to obtain a first classification result, wherein the first classification result is the probability of normal or abnormal line loss, and when the first classification result is abnormal, inputting the first feature into a deep feature extraction module to perform feature extraction to obtain a second feature and inputting the second feature into a second classifier to obtain a second classification result, wherein the second classification result comprises the line loss rate and a corresponding classification label, so that the efficiency is greatly improved; when the model is trained, the historical similar day data and the adjacent day data are comprehensively considered for data restoration, so that the accuracy of the model is greatly improved.
Description
Technical Field
The invention belongs to the field of electric energy loss of a power grid, and particularly relates to a line loss analysis method based on big data.
Background
With the rapid development of national economy and the continuous deepening of economic system reform of socialist market, the construction of resource-saving and environment-friendly society becomes the focus of people's attention. This puts new and higher demands on the line loss management work in the power system. However, a large amount of power loss phenomena still commonly exist in the power supply management of the current power system due to the line loss management problem, so how to take active and effective measures to improve the line loss management level in the power system and reduce the power consumption is an important problem to be solved urgently by the current related power management departments.
According to the generation reason, the line loss rate calculation method comprises theoretical line loss and synchronous line loss, the theoretical line loss is the line loss rate obtained according to the topological structure and parameters of the power grid, and the synchronous line loss is calculated based on the collected data of the intelligent electric meter. The method for calculating the synchronous line loss depends on the collection condition of the intelligent electric meter, in order to ensure the accuracy and the integrity of electric meter data, in the prior art, different measures are adopted for repairing through stage mean values, historical similar days, adjacent days, same-day electric quantity of similar distribution areas and abnormal condition classification, however, the repairing through the method needs to classify the condition of interrupted collection, and then different repairing measures are selected, so that the efficiency is low. When the line loss rate is classified, CN113866562A classifies input data through a convolutional neural network to obtain the line loss rate and a corresponding classification label, however, because the user quantity is large, the quantity collected through the smart electric meters is large, the time consumed for classification through the above method is long, when a missing value is repaired, classification analysis is performed through the missing condition of the missing data, and the efficiency is low.
Disclosure of Invention
The invention aims to solve the problems in the prior art and provides a line loss analysis method based on big data so as to improve the efficiency and accuracy of line loss identification.
The invention is realized by the following technical scheme:
the application provides a line loss analysis method based on big data, which comprises the following steps:
acquiring data to be analyzed, wherein the data to be analyzed comprises a transformer resistance parameter, a line impedance parameter, an inductive reactance parameter, a capacity parameter, a power parameter, total station active power, a power factor, a transformer load factor, voltage and current;
carrying out data preprocessing on the data to be analyzed to obtain the data to be analyzed;
inputting the data to be analyzed into a line loss identification network model, wherein the line loss identification network model comprises an input layer, a shallow layer feature extraction module, a first classifier, a deep layer feature extraction module and a second classifier; inputting a first feature obtained by shallow feature extraction into a first classifier to obtain a first classification result, wherein the first classification result is the probability of normal or abnormal line loss, when the first classification result is abnormal, inputting the first feature into a deep feature extraction module to perform feature extraction to obtain a second feature, and inputting the second feature into a second classifier to obtain a second classification result, wherein the second classification result comprises a line loss rate and a corresponding classification label, and the classification label comprises large negative loss, small negative loss, high loss and super large loss;
the line loss identification network model adopts historical data as candidate training samples; performing data preprocessing and manual marking on the candidate training samples to obtain training samples; and constructing a line loss recognition network model, and training the network model based on the training samples.
The data preprocessing of the candidate training samples comprises: data cleaning, data integration, data transformation and data specification.
The data cleaning includes null value processing and noise data processing.
The vacancy value Hij is obtained through the following steps:
based on historical data, a historical data matrix is established
N is the number of historical data, m =10, and each row of data comprises values corresponding to a transformer resistance parameter, a line impedance parameter, an inductive reactance parameter, a capacity parameter, a power parameter, total station active power, a power factor, a transformer load factor, voltage and current;
selecting k historical similar day data with vacant data items Hi to obtain a historical similar day data matrix S, wherein
The historical similar day data is obtained through calculation of a KNN algorithm;
p temporary data entries Hi with blank data entries are selectedObtaining near-day data matrix N from the near-day data, wherein the near-day data matrix is described
The missing term Hij of the empty data entry Hi is obtained by the following formula,
is/are>
And the dsi and dni are respectively the similarity between each piece of data in the historical similar day data matrix and Hi and the similarity between each piece of data in the adjacent day data matrix and Hi, and the similarity can be obtained by Euclidean distance calculation.
The shallow layer feature extraction module comprises a first convolution layer and a second convolution layer.
The deep feature extraction module constructs a deep feature extraction network using N sets of lightweight network structures, where the lightweight network structures are SqueezeNet, N is a positive integer, and preferably, N =4.
Compared with the prior art, the invention judges whether the line loss is abnormal or not through the shallow layer characteristics, and the line loss rate classification is further carried out only when the line loss classification result is abnormal, thereby improving the efficiency; when the model is trained, the accuracy of data restoration is improved by comprehensively considering historical similar day data and adjacent day data and by means of weighted average, and therefore the line loss identification network model is improved.
Drawings
Fig. 1 is a model structure of a line loss identification network model.
Detailed Description
In order to make the technical means, the creation characteristics, the achievement purposes and the functions of the invention clearer and easier to understand, the invention is further explained by combining the drawings and the detailed description:
the application provides a line loss analysis method based on big data, which comprises the following steps:
acquiring data to be analyzed, wherein the data to be analyzed comprises a transformer resistance parameter, a line impedance parameter, an inductive reactance parameter, a capacity parameter, a power parameter, total station active power, a power factor, a transformer load factor, voltage and current;
carrying out data preprocessing on the data to be analyzed to obtain the data to be analyzed;
inputting the data to be analyzed into a line loss identification network model, wherein the line loss identification network model comprises an input layer, a shallow layer feature extraction module, a first classifier, a deep layer feature extraction module and a second classifier; inputting a first feature obtained by shallow feature extraction into a first classifier to obtain a first classification result, wherein the first classification result is the probability of normal or abnormal line loss, when the first classification result is abnormal, inputting the first feature into a deep feature extraction module to perform feature extraction to obtain a second feature, and inputting the second feature into a second classifier to obtain a second classification result, wherein the second classification result comprises a line loss rate and a corresponding classification label, and the classification label comprises large negative loss, small negative loss, high loss and super large loss;
the line loss identification network model adopts historical data as candidate training samples; performing data preprocessing and manual marking on the candidate training samples to obtain training samples; and constructing a line loss recognition network model, and training the network model based on the training samples.
The data preprocessing of the candidate training samples comprises: data cleaning, data integration, data transformation and data specification.
The data cleaning includes null value processing and noise data processing.
The data preprocessing technology can improve the quality of data, thereby being beneficial to improving the precision and the performance of subsequent data processing.
There are attributes that for some reason are not recorded, and these null values must be processed.
The loss of line loss data can be classified into two types, mechanical and artificial. The mechanical reason is data loss caused by data collection or storage errors caused by mechanical reasons, such as data storage failure, memory damage and mechanical failure, which causes data collection failure in a certain period of time. The artificial reason is that data is missing due to subjective errors of people, historical limitations or deliberate hiding operations, for example, a line loss on-site transcriber mistakenly records data during data recording.
From a mathematical statistical point of view, missing data can also be classified as completely random, and non-random.
Common vacancy value processing includes: the method for filling the vacancy values is simple but quite unreliable, and the vacancy values are filled in the following mode.
The vacancy value Hij is obtained through the following steps:
based on historical data, a historical data matrix is established
N is the number of historical data, m =10, and each row of data comprises values corresponding to a transformer resistance parameter, a line impedance parameter, an inductive reactance parameter, a capacity parameter, a power parameter, total station active power, a power factor, a transformer load factor, voltage and current;
selecting k pieces of historical similar day data with vacant data items Hi to obtain a historical similar day data matrix S, wherein the historical similar day data matrix S is obtained
The historical similar day data is obtained through calculation of a KNN algorithm;
selecting p adjacent day data with vacant data items Hi to obtain an adjacent day data matrix N, wherein the adjacent day data matrix is defined as N
The missing items Hij of the missing data entries Hi may be calculated by means of weighted average, however, considering the similarity of the data in the near days and other factors, the missing items Hij of the missing data entries Hi may also be obtained by the following method:
the missing term Hij of the empty data entry Hi is obtained by the following formula,
is/are>
And the dsi and the dni are respectively the similarity between each piece of data in the historical similar day data matrix and the Hi and the similarity between each piece of data in the adjacent day data matrix and the Hi, and the similarity can be obtained by Euclidean distance calculation.
The shallow layer feature extraction module comprises a first convolution layer and a second convolution layer.
The deep feature extraction module constructs a deep feature extraction network using N sets of lightweight network structures, the lightweight network structure being SqueezeNet, N being a positive integer, preferably, N =4.
Finally, it should be noted that the above-mentioned technical solution is only one embodiment of the present invention, and it will be apparent to those skilled in the art that various modifications and variations can be easily made based on the application method and principle of the present invention disclosed, and the method is not limited to the above-mentioned specific embodiment of the present invention, so that the above-mentioned embodiment is only preferred, and not restrictive.
Claims (6)
1. A method of analyzing line loss based on big data, the method comprising: acquiring data to be analyzed, wherein the data to be analyzed comprises a transformer resistance parameter, a line impedance parameter, an inductive reactance parameter, a capacity parameter, a power parameter, total station active power, a power factor, a transformer load factor, voltage and current;
carrying out data preprocessing on the data to be analyzed to obtain the data to be analyzed;
inputting the data to be analyzed into a line loss identification network model, wherein the line loss identification network model comprises an input layer, a shallow layer feature extraction module, a first classifier, a deep layer feature extraction module and a second classifier; inputting a first feature obtained by shallow feature extraction into a first classifier to obtain a first classification result, wherein the first classification result is the probability of normal or abnormal line loss, when the first classification result is abnormal, inputting the first feature into a deep feature extraction module to perform feature extraction to obtain a second feature, and inputting the second feature into a second classifier to obtain a second classification result, wherein the second classification result comprises a line loss rate and a corresponding classification label, and the classification label comprises large negative loss, small negative loss, high loss and super large loss;
the line loss identification network model adopts historical data as candidate training samples; performing data preprocessing and manual marking on the candidate training samples to obtain training samples; constructing a line loss recognition network model, and training the network model based on the training samples;
the pre-processing the data of the candidate training samples comprises: clearing data;
the data cleaning comprises vacancy value processing, wherein vacancy values Hij are obtained through the following steps:
based on historical data, a historical data matrix is established
N is the number of historical data, m =10, and each row of data comprises values corresponding to a transformer resistance parameter, a line impedance parameter, an inductive reactance parameter, a capacity parameter, a power parameter, total station active power, a power factor, a transformer load factor, voltage and current;
selecting k pieces of historical similar day data with vacant data items Hi to obtain a historical similar day data matrix S, wherein the historical similar day data matrix S is obtained
The historical similar day data is obtained through calculation of a KNN algorithm;
selecting p adjacent-day data with vacant data items Hi to obtain an adjacent-day data matrix N, wherein the adjacent-day data matrix is defined as
The missing term Hij of the empty data entry Hi is obtained by the following formula,
the above-mentioned
And the dsi and the dni are respectively the similarity between each piece of data in the historical similar day data matrix and the Hi and the similarity between each piece of data in the adjacent day data matrix and the Hi, and the similarity can be obtained by Euclidean distance calculation.
2. The method for analyzing line loss based on big data as claimed in claim 1, wherein the pre-processing the candidate training samples further comprises: data integration, data transformation and data specification.
3. The method for analyzing line loss based on big data as claimed in claim 2, wherein the data cleaning further comprises noise data processing.
4. The method for analyzing line loss based on big data as claimed in claim 1, wherein the shallow feature extraction module comprises a first convolutional layer and a second convolutional layer.
5. The big-data-based line loss analysis method according to claim 1, wherein the deep feature extraction module constructs the deep feature extraction network using N sets of lightweight network structures, wherein the lightweight network structures are SqueezeNet and N is a positive integer.
6. The big-data based line loss analysis method according to claim 5, wherein the deep feature extraction module uses N sets of lightweight network structures to construct the deep feature extraction network, preferably N =4.
Priority Applications (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN202210254768.6A CN114611396B (en) | 2022-03-15 | 2022-03-15 | Line loss analysis method based on big data |
Applications Claiming Priority (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN202210254768.6A CN114611396B (en) | 2022-03-15 | 2022-03-15 | Line loss analysis method based on big data |
Publications (2)
| Publication Number | Publication Date |
|---|---|
| CN114611396A CN114611396A (en) | 2022-06-10 |
| CN114611396B true CN114611396B (en) | 2023-03-24 |
Family
ID=81863837
Family Applications (1)
| Application Number | Title | Priority Date | Filing Date |
|---|---|---|---|
| CN202210254768.6A Active CN114611396B (en) | 2022-03-15 | 2022-03-15 | Line loss analysis method based on big data |
Country Status (1)
| Country | Link |
|---|---|
| CN (1) | CN114611396B (en) |
Family Cites Families (12)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| US9349105B2 (en) * | 2013-12-18 | 2016-05-24 | International Business Machines Corporation | Machine learning with incomplete data sets |
| CN109002937A (en) * | 2018-09-07 | 2018-12-14 | 深圳供电局有限公司 | Load Forecasting, device, computer equipment and storage medium |
| CN110766096B (en) * | 2019-10-31 | 2022-09-23 | 北京金山云网络技术有限公司 | Video classification method and device and electronic equipment |
| CN110837888A (en) * | 2019-11-13 | 2020-02-25 | 大连理工大学 | Traffic missing data completion method based on bidirectional cyclic neural network |
| CN111143344A (en) * | 2019-12-31 | 2020-05-12 | 新奥数能科技有限公司 | Completion method and device for time series data missing |
| CN111507412B (en) * | 2020-04-20 | 2021-02-19 | 南京工程学院 | Voltage missing value filling method based on historical data auxiliary scene analysis |
| CN111904411B (en) * | 2020-08-25 | 2022-09-23 | 浙江工业大学 | Multi-lead heartbeat signal classification method and device based on multi-scale feature extraction |
| CN112380998A (en) * | 2020-11-16 | 2021-02-19 | 华北电力大学（保定） | Low-voltage transformer area missing data completion method based on matrix completion |
| CN112598234A (en) * | 2020-12-14 | 2021-04-02 | 广东电网有限责任公司广州供电局 | Low-voltage transformer area line loss abnormity analysis method, device and equipment |
| CN113344927A (en) * | 2021-08-05 | 2021-09-03 | 武汉楚精灵医疗科技有限公司 | Image recognition method and device based on deep learning, server and storage medium |
| CN113743537A (en) * | 2021-09-26 | 2021-12-03 | 东南大学 | Deep sparse memory model-based highway electromechanical system fault classification method |
| CN113866562A (en) * | 2021-09-29 | 2021-12-31 | 国网甘肃省电力公司电力科学研究院 | Line loss identification method for power system distribution room |
-
2022
- 2022-03-15 CN CN202210254768.6A patent/CN114611396B/en active Active
Also Published As
| Publication number | Publication date |
|---|---|
| CN114611396A (en) | 2022-06-10 |
Similar Documents
| Publication | Publication Date | Title |
|---|---|---|
| CN112561156A (en) | Short-term power load prediction method based on user load mode classification | |
| CN106709035A (en) | Preprocessing system for electric power multi-dimensional panoramic data | |
| CN107133652A (en) | Electricity customers Valuation Method and system based on K means clustering algorithms | |
| CN104809658A (en) | Method for rapidly analyzing low-voltage distributing network area line loss | |
| CN111488917A (en) | Garbage image fine-grained classification method based on incremental learning | |
| CN112907222B (en) | Multi-source heterogeneous power grid operation supervision service data fusion method | |
| CN106709816A (en) | Power load abnormal data recognition and modification method based on nonparametric regression analysis | |
| CN108011367A (en) | A kind of Characteristics of Electric Load method for digging based on depth decision Tree algorithms | |
| US20220368131A1 (en) | Capacity configuration method and system of energy storage in microgrid | |
| CN112819299A (en) | Differential K-means load clustering method based on center optimization | |
| CN102930007B (en) | User in large-area power-cuts emergency processing sends a telegram in reply urgency level sorting technique | |
| CN114611396B (en) | Line loss analysis method based on big data | |
| CN113987808A (en) | Electricity user complaint early warning method of feature weighted Bayesian network | |
| CN110414776B (en) | Quick response analysis system for power utilization characteristics of different industries | |
| CN114676931B (en) | Electric quantity prediction system based on data center technology | |
| Zhou et al. | Data-driven solutions for building environmental impact assessment | |
| CN114281846B (en) | New energy power generation prediction method based on machine learning | |
| CN115186882A (en) | Clustering-based controllable load spatial density prediction method | |
| Wang et al. | Analysis of user’s power consumption behavior based on k-means | |
| CN117218424B (en) | Material management method and system in photovoltaic power station construction | |
| CN116739623B (en) | Transaction data tracking analysis method based on blockchain technology | |
| CN116258972B (en) | Deep learning-based rock high-steep slope structural surface extraction method | |
| CN113219303B (en) | Extraction method of single-phase grounding fault data of small-current grounding system | |
| Liu et al. | Analysis of SCADA Data Preprocessing Methods for Wind Power Farms | |
| CN117454099A (en) | Wind turbine abnormal data combined cleaning method |
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 |