CN115616283A - Abnormal electricity utilization detection method based on neural network - Google Patents
Abnormal electricity utilization detection method based on neural network Download PDFInfo
- Publication number
- CN115616283A CN115616283A CN202211153799.9A CN202211153799A CN115616283A CN 115616283 A CN115616283 A CN 115616283A CN 202211153799 A CN202211153799 A CN 202211153799A CN 115616283 A CN115616283 A CN 115616283A
- Authority
- CN
- China
- Prior art keywords
- data
- abnormal
- neural network
- electricity utilization
- feature extraction
- Prior art date
- Legal status (The legal status is an assumption and is not a legal conclusion. Google has not performed a legal analysis and makes no representation as to the accuracy of the status listed.)
- Pending
Links
Images
Classifications
-
- G—PHYSICS
- G01—MEASURING; TESTING
- G01R—MEASURING ELECTRIC VARIABLES; MEASURING MAGNETIC VARIABLES
- G01R22/00—Arrangements for measuring time integral of electric power or current, e.g. electricity meters
- G01R22/06—Arrangements for measuring time integral of electric power or current, e.g. electricity meters by electronic methods
-
- 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
- G06N3/049—Temporal neural networks, e.g. delay elements, oscillating neurons or pulsed inputs
-
- 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
Landscapes
- Engineering & Computer Science (AREA)
- Physics & Mathematics (AREA)
- Theoretical Computer Science (AREA)
- General Physics & Mathematics (AREA)
- Data Mining & Analysis (AREA)
- Evolutionary Computation (AREA)
- Artificial Intelligence (AREA)
- Biomedical Technology (AREA)
- Biophysics (AREA)
- Computational Linguistics (AREA)
- Health & Medical Sciences (AREA)
- Life Sciences & Earth Sciences (AREA)
- General Health & Medical Sciences (AREA)
- Molecular Biology (AREA)
- Computing Systems (AREA)
- General Engineering & Computer Science (AREA)
- Mathematical Physics (AREA)
- Software Systems (AREA)
- Power Engineering (AREA)
- Testing Or Calibration Of Command Recording Devices (AREA)
Abstract
The invention discloses an abnormal electricity utilization detection method based on a neural network, which is characterized in that a real-time electricity utilization data source of a user side is obtained through a meter based on the user side, and the data source is analyzed and judged. The abnormal electricity utilization detection method based on the neural network not only improves the efficiency of detecting electricity utilization abnormity, but also improves the accuracy of detecting electricity utilization abnormity, and has remarkable advantages in solving abnormal electricity utilization behaviors such as electricity and electricity leakage and the like which relate to various complex phenomena and various factors by preprocessing data, combining a neural network detection model and utilizing a mode of combining detection and a price method, reduces the complexity of manual modeling, realizes automatic training learning and modeling of a system, achieves the purpose of quickly and accurately positioning abnormal electricity utilization suspected users, and provides convenience for monitoring abnormal electricity utilization behaviors such as electricity and electricity leakage.
Description
Technical Field
The invention relates to the technical field of abnormal power utilization detection of a power grid, in particular to a method for detecting abnormal power utilization based on a neural network.
Background
Electric power is energy using electric energy as power, the discovery and application of electric power has raised the second industrialized climax, and has become one of three scientific and technological revolution which occur in the world since 18 th century of human history, the life of people is changed by science from this science and technology, a large-scale electric power system appearing in 20 th century is one of the most important achievements in the science history of human engineering, and the electric power system is an electric power production and consumption system which is composed of links of power generation, power transmission, power transformation, power distribution, power utilization and the like, and converts the primary energy in the nature into electric power through a mechanical energy device, and then supplies the electric power to each user through the power transmission, the power transformation and the power distribution.
With the continuous improvement of the informatization degree of an electric power system and the rapid increase of the data quantity of distribution and utilization electricity, various devices and systems have a large amount of data to be processed, and meanwhile, due to various communication faults, equipment faults, power grid fluctuation, abnormal power utilization behaviors of users and the like, a phenomenon of large amount of data abnormity occurs, the abnormal data influence the accuracy, completeness, self-consistency and dynamic property of electric energy data, but important event information of the power grid is also stored, so that an algorithm suitable for large-scale power utilization data mining is researched, an effective abnormity discovery model is established for analyzing, identifying and processing abnormal power utilization information, the method has important significance for analyzing and mining event information in the power industry and the development of an intelligent power grid, and an on-site detection method is mostly adopted in the early stage for detecting the equipment faults and the abnormal power utilization of the users, namely a technical person goes to a power utilization site to perform investigation, the processing method consumes manpower and material resources, has low efficiency and poor effect, and simultaneously, and is not beneficial to the management of the power industry.
Disclosure of Invention
Aiming at the defects of the prior art, the invention provides the abnormal electricity utilization detection method based on the neural network, which has the advantages of convenience for rapidly finding electricity utilization abnormality and the like, and solves the problems of low efficiency and poor accuracy of the traditional abnormal electricity utilization detection method.
The purpose of the invention is realized by the following technical scheme, which comprises the following steps:
s1, starting: the method comprises the steps that a real-time power utilization data source of a user side is obtained based on a meter of the user side, preliminary judgment is conducted on the data source, users who are unlikely to have abnormal power utilization behaviors are eliminated, and data samples which are likely to have abnormal behaviors are reserved;
s2, data acquisition: collecting the data samples which possibly have the abnormity, and training a data set, wherein the data mainly come from electric energy metering data, operation condition data, event records and other standby data in the electric energy meter and the collection terminal;
s3, data processing: preprocessing the acquired data, wherein the data preprocessing process comprises data cleaning, missing value processing, abnormal value processing, data transformation and data formatting, and is used for standardizing the data and improving the data analysis efficiency;
s4, feature extraction: after the data are preprocessed, extracting the characteristics of the data, removing the data which are impossible to be abnormal in the data, and extracting the abnormal data;
s5, inputting a detection model: inputting the abnormal data into a neural network detection model, and analyzing the data by the detection model;
s6, ending: and carrying out the same mark classification on the abnormal data obtained after analysis so as to facilitate the checking of workers.
Preferably, the data processing mainly includes data cleaning and normalization processing, the data cleaning mainly includes deleting repeated information, filling missing information, correcting error information, and the like, wherein the data normalization processing formula is as follows:
wherein x is sample data, x = [ x ] 1 ，x 2 ，…，x n ]N is the number of samples, mean (x) represents the mean of the calculated samples; max (x) represents the maximum value of the calculated samples; min (x) represents the calculated sample minimum; x' i The processed data are normalized.
Preferably, the feature extraction module FEM is used for feature extraction, and the feature extraction module is formed by vertically connecting a plurality of long and short term memory blocks, that is, the output of the previous long and short term memory block is used as the input of the next long and short term memory block.
Preferably, the feature extraction further includes processing the high-dimensional data, dividing the input high-dimensional data into three sub-data with the same size, and implementing feature extraction of the input high-dimensional data by using a structural recurrent neural network constructed by ten feature extraction modules for the sub-data.
Preferably, the detection model uses a loss function to calculate the loss deviation degree of the current model, and performs feedback adjustment to adjust the parameter settings of each neural network through calculation of the loss function, wherein the loss function used in the method is as follows:
f Loss ＝-mean[y real ln(y class )]
wherein, y real Representing the actual class label, y class And representing a classification result, and optimizing the loss function by using adaptive matrix estimation to obtain a minimized loss function value.
Preferably, the input detection model further comprises result evaluation, the result of the model is evaluated by using a confusion matrix, the abnormal electricity utilization data is defined as a negative class, the normal electricity utilization data is defined as a positive class, and the classification result is represented by the confusion matrix.
Compared with the prior art, the beneficial effects are:
the abnormal electricity utilization detection method based on the neural network can improve the efficiency and accuracy of electricity utilization abnormality detection. Through the preprocessing of data, combine neural network detection model, utilize the mode that detects and valence method combine together, have showing the advantage in solving the case that relates to the unusual power consumption behaviors such as the electricity of stealing electric leakage of multiple complex phenomenon and multifactor, and reduced the complexity of artifical modeling, realized systematic automatic training study and modeling, reach quick accurate abnormal electricity suspicion user of location again, for obtaining to steal the electricity, the unusual power consumption behaviors such as electric leakage provide convenience.
Drawings
In order to more clearly illustrate the embodiments of the present invention or the technical solutions in the prior art, the drawings used in the embodiments or the description of the prior art will be briefly described below, it is obvious that the drawings in the following description are only some embodiments of the present invention, and for those skilled in the art, other drawings can be obtained according to the drawings without creative efforts.
Fig. 1 is a flowchart of an abnormal electricity consumption detection method based on a neural network according to the present invention.
Fig. 2 is a system block diagram of an abnormal electricity usage detection method based on a neural network according to the present invention.
Detailed Description
The embodiments of the present disclosure are described in detail below with reference to the accompanying drawings.
The embodiments of the present disclosure are described below with specific examples, and other advantages and effects of the present disclosure will be readily apparent to those skilled in the art from the disclosure in the specification. It is to be understood that the described embodiments are merely illustrative of some, and not restrictive, of the embodiments of the disclosure. The disclosure may be carried into practice or applied to various other specific embodiments, and various modifications and changes may be made in the details within the description and the drawings without departing from the spirit of the disclosure. It is to be noted that the features in the following embodiments and examples may be combined with each other without conflict. All other embodiments, which can be derived by a person skilled in the art from the embodiments disclosed herein without making any creative effort, shall fall within the protection scope of the present disclosure.
Referring to fig. 1-2, in the present embodiment, a method for detecting abnormal electricity consumption based on a neural network includes the following steps:
s1, starting: acquiring a real-time power consumption data source of a user side based on a meter of the user side, preliminarily judging the data source, eliminating users who are unlikely to have abnormal power consumption behaviors, and reserving data samples which are likely to have abnormality;
s2, data acquisition: collecting the data samples which possibly have the abnormity, and training a data set, wherein the data mainly come from electric energy metering data, operation condition data, event records and other standby data in the electric energy meter and the collection terminal;
s3, data processing: preprocessing the acquired data, wherein the data preprocessing process comprises data cleaning, missing value processing, abnormal value processing, data transformation and data formatting, and is used for standardizing the data and improving the data analysis efficiency;
s4, feature extraction: after data preprocessing is finished, feature extraction is carried out on the data, abnormal data which cannot exist in the data are removed again, then abnormal data in the data are extracted, a feature extraction module FEM is adopted for feature extraction, the feature extraction module is formed by vertically connecting a plurality of long and short-term memory blocks, namely the output of the last long and short-term memory block is used as the input of the next long and short-term memory block, the feature extraction also comprises the steps of processing high-dimensional data, dividing the input high-dimensional data into three subdata with the same size, and aiming at the subdata, the feature extraction of the input high-dimensional data is realized by utilizing a structural recurrent neural network constructed by ten feature extraction modules;
s5, inputting a detection model: inputting the abnormal data into a neural network detection model, analyzing the data by the detection model, and further evaluating the result by inputting the detection model, wherein the result of the model is evaluated by using a confusion matrix, the abnormal electricity utilization data is defined as a negative class, the normal electricity utilization data is defined as a positive class, and the classification result is represented by the confusion matrix;
s6, ending: and carrying out the same mark classification on the abnormal data obtained after the analysis so as to facilitate the examination of workers.
The data processing mainly comprises data cleaning and normalization processing, the data cleaning mainly comprises deleting repeated information, filling missing information, correcting error information and the like, and the data normalization processing formula is as follows:
wherein x is sample data, x = [ x ] 1 ，x 2 ，…，x n ]N is the number of samples, and mean (x) represents the average value of the calculated samples; max (x) represents the maximum value of the calculated samples; min (x) represents the calculated sample minimum; x' i To normalize the processed data, the processed data is summarized.
Meanwhile, the detection model calculates the loss deviation degree of the current model by using a loss function, performs feedback regulation by calculating the loss function, and adjusts the parameter setting of each neural network, wherein the loss function used in the method is as follows:
f Loss ＝-meqn[y real ln(y class )]
wherein, y real Representing the actual class label, y class Representing a classification result, optimizing the loss function by using adaptive matrix estimation to obtain a minimized loss function value, and adopting a neural network detection model for the detection model, wherein the neural network detection model comprises a calculation module used for calculating the loss deviation degree of the current model, and a historical data acquisition module used for extracting historical electricity utilization data of a preset time length according to the current electricity utilization time period based on a meter at the user side, and the historical electricity utilization data comprises historical electricity utilization power, historical air temperature corresponding to the historical electricity utilization power and whether the electricity utilization power is in a power utilization peak period; the marking module is used for preprocessing the historical electricity utilization data, marking the preprocessed historical electricity utilization data based on the expert database, wherein the marking content comprises an electricity stealing mark and an electricity leakage mark; the historical data matrix construction module is used for constructing a historical electricity utilization data matrix based on the marked historical electricity utilization data, and each element in the historical electricity utilization data matrix is historical electricity utilization power and corresponding mark content; the dividing module is used for dividing all elements in the historical electricity utilization data matrix into a training data set and a testing data set; the training module is used for substituting the training data set into a deep learning algorithm for training, and taking the marked content as output to obtain a deep neural network model; a test module for substituting the test data set into the deep neural network model to obtain a test result, judging whether the test result is in an acceptable range, and if not, adjusting the networkAnd (5) carrying out iterative training again on the parameters, and if the parameters are within an acceptable range, obtaining a corresponding neural network detection model.
According to the abnormal electricity utilization detection method based on the neural network, six steps of starting, data acquisition, data processing, feature extraction, inputting the detection model and ending are adopted to detect abnormal electricity utilization, so that the efficiency of electricity utilization abnormality detection is improved, and the accuracy of electricity utilization abnormality detection is also improved.
The method comprises the steps of acquiring a real-time power consumption data source of a user side through a meter based on the user side, primarily judging the data source, eliminating users who are unlikely to have power consumption abnormal behaviors, reserving data samples which are likely to have abnormalities, collecting the data samples which are likely to have abnormalities, training a data set, normalizing data and improving data analysis efficiency, after the data are preprocessed, extracting characteristics of the data according to the data, eliminating the data which are unlikely to have abnormalities in the data again, extracting abnormal data in the data, inputting the abnormal data into a neural network detection model, analyzing the data through the detection model, and classifying the abnormal data obtained after analysis by using the same mark so as to facilitate checking of working personnel, so that the power consumption abnormality detection efficiency is improved, and the power consumption abnormality detection accuracy is also improved.
Through the preprocessing of data, combine neural network detection model, utilize the mode that detects and valence method combine together, have showing the advantage in solving the case that relates to the unusual power consumption behaviors such as the electricity of stealing electric leakage of multiple complex phenomenon and multifactor, and reduced the complexity of artifical modeling, realized systematic automatic training study and modeling, reach quick accurate abnormal electricity suspicion user of location again, for obtaining to steal the electricity, the unusual power consumption behaviors such as electric leakage provide convenience.
The abnormal electricity utilization detection method based on the neural network can detect and obtain data with abnormal electricity utilization modes from actual user data, and comprises the following steps: successfully constructing a user abnormal electricity utilization mode detection model by using a TensorFlow framework; a plurality of feature extraction modules are utilized to construct a feature extraction network based on a structure recurrent neural network; a data feature matching neural network is constructed by utilizing a three-layer fully-connected neural network, a model is compared with a support vector machine and a BP neural network, and the experimental result shows that the abnormal electricity utilization detection method based on the neural network can find abnormal data in an electricity utilization mode and has higher accuracy.
The foregoing is directed to embodiments of the present invention, substantially all of which have been shown and described, and the equivalents of the features and advantages of the invention, including the method and apparatus for performing the method and apparatus of the invention, as well as the method and apparatus of the invention, as described herein, may be practiced otherwise than as specifically described herein.
The above description is for the purpose of illustrating embodiments of the invention and is not intended to limit the invention, and it will be apparent to those skilled in the art that any modification, equivalent replacement, or improvement made without departing from the spirit and principle of the invention shall fall within the protection scope of the invention.
Claims (6)
1. An abnormal electricity utilization detection method based on a neural network is characterized by comprising the following steps:
s1, starting: the method comprises the steps that a real-time power utilization data source of a user side is obtained based on a meter of the user side, the data source is judged, users without abnormal power utilization behaviors are eliminated, and abnormal data samples are reserved;
s2, data acquisition: collecting abnormal data samples, and training a data set, wherein the data comprises electric energy metering data, operation condition data and event record standby data in an electric energy meter and a collection terminal;
s3, data processing: preprocessing the acquired data, wherein the data preprocessing process comprises data cleaning, missing value processing, abnormal value processing, data transformation and data formatting;
s4, feature extraction: extracting the characteristics of the preprocessed data, removing the data which is unlikely to have abnormity from the data, and extracting the abnormal data;
s5, inputting a detection model: inputting the abnormal data into a neural network detection model, and analyzing the data;
s6, ending: and performing same label classification on the abnormal data obtained after analysis.
2. The method according to claim 1, wherein the data processing is mainly a cleaning and normalization processing of data, the cleaning of data is mainly deleting repeated information, filling missing information and correcting error information, and the data normalization processing formula is as follows:
wherein x is sample data, x = [ x ] 1 ，x 2 ，…，x n ]N is the number of samples, and mean (x) represents the average value of the calculated samples; max (x) represents the maximum value of the calculated samples; min (x) represents the calculated sample minimum; x' i The processed data are normalized.
3. The method as claimed in claim 1, wherein the feature extraction is performed by using a feature extraction module FEM, the feature extraction module is composed of a plurality of vertically connected long-short term memory blocks, that is, the output of the previous long-short term memory block is used as the input of the next long-short term memory block.
4. The method of claim 1, wherein the feature extraction further comprises processing the high-dimensional data, dividing the input high-dimensional data into three sub-data with the same size, and implementing feature extraction of the input high-dimensional data by using a structural recurrent neural network constructed by ten feature extraction modules for the sub-data.
5. The method of claim 1, wherein the detection model calculates the degree of loss deviation of the current model using a loss function, and performs feedback adjustment by calculating the loss function, and adjusting the parameter settings of each neural network, where the loss function is as follows:
f Loss ＝-mean[y real ln(y class) ]
wherein, y real Representing the actual class label, y class Representing the classification result.
6. The method of claim 1, wherein the input test model further comprises a result evaluation, wherein the result of the model is evaluated using a confusion matrix, wherein the abnormal electricity consumption data is defined as a negative class, the normal electricity consumption data is defined as a positive class, and the classification result is represented by the confusion matrix.
Priority Applications (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN202211153799.9A CN115616283A (en) | 2022-09-21 | 2022-09-21 | Abnormal electricity utilization detection method based on neural network |
Applications Claiming Priority (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN202211153799.9A CN115616283A (en) | 2022-09-21 | 2022-09-21 | Abnormal electricity utilization detection method based on neural network |
Publications (1)
| Publication Number | Publication Date |
|---|---|
| CN115616283A true CN115616283A (en) | 2023-01-17 |
Family
ID=84858873
Family Applications (1)
| Application Number | Title | Priority Date | Filing Date |
|---|---|---|---|
| CN202211153799.9A Pending CN115616283A (en) | 2022-09-21 | 2022-09-21 | Abnormal electricity utilization detection method based on neural network |
Country Status (1)
| Country | Link |
|---|---|
| CN (1) | CN115616283A (en) |
-
2022
- 2022-09-21 CN CN202211153799.9A patent/CN115616283A/en active Pending
Similar Documents
| Publication | Publication Date | Title |
|---|---|---|
| CN110518880B (en) | Photovoltaic power station state diagnosis method and device | |
| CN115170000B (en) | Remote monitoring method and system based on electric energy meter communication module | |
| CN112418687B (en) | User electricity utilization abnormity identification method and device based on electricity utilization characteristics and storage medium | |
| CN116028887B (en) | Analysis method of continuous industrial production data | |
| CN114386884B (en) | Lean evaluation method for power grid dispatching operation | |
| CN116125204A (en) | Fault prediction system based on power grid digitization | |
| CN112926686B (en) | BRB and LSTM model-based power consumption anomaly detection method and device for big power data | |
| CN108335233A (en) | A kind of intelligent grid metric data processing system based on big data technology | |
| CN110930051A (en) | Big data analysis-based resident demand response potential analysis system and method | |
| CN115186935B (en) | Electromechanical device nonlinear fault prediction method and system | |
| CN115616283A (en) | Abnormal electricity utilization detection method based on neural network | |
| CN116015209A (en) | Photovoltaic power station component performance management method, system, device and storage medium | |
| CN115564180A (en) | Power network reliability assessment method based on big data analysis | |
| CN115392663A (en) | Data acquisition and processing method based on big data | |
| CN117540225B (en) | Distributed ups system consistency assessment system and method based on DBSCAN clustering | |
| CN117723917B (en) | Monitoring application method based on optical fiber extrinsic Fabry-Perot type ultrasonic sensor | |
| CN117114254B (en) | Power grid new energy abnormal data monitoring method and system | |
| CN117371623B (en) | Electric energy meter running state early warning method and system | |
| CN109613370B (en) | Load parameter identification system based on smart electric meter data | |
| CN117634892A (en) | Power metering data security risk assessment method and system | |
| CN117665442A (en) | Testing method and system for electric intelligent internet of things lockset | |
| CN116933201A (en) | Method and system for identifying illegal electricity utilization behavior of low-voltage charging pile | |
| CN116579557A (en) | Electricity selling management system for predicting electric quantity and method for predicting electric quantity | |
| CN117455724A (en) | Building energy consumption analysis method and system based on multistage network nodes | |
| CN111105610A (en) | Novel 24-hour acquisition terminal field monitoring equipment and implementation method thereof |
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 |