CN115758255B - Power consumption abnormal behavior analysis method and device under fusion model - Google Patents
Power consumption abnormal behavior analysis method and device under fusion model Download PDFInfo
- Publication number
- CN115758255B CN115758255B CN202310035095.XA CN202310035095A CN115758255B CN 115758255 B CN115758255 B CN 115758255B CN 202310035095 A CN202310035095 A CN 202310035095A CN 115758255 B CN115758255 B CN 115758255B
- Authority
- CN
- China
- Prior art keywords
- data
- standard
- historical
- electricity consumption
- model
- 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
- 238000004458 analytical method Methods 0.000 title claims abstract description 87
- 230000004927 fusion Effects 0.000 title claims abstract description 34
- 206010000117 Abnormal behaviour Diseases 0.000 title abstract description 15
- 230000005611 electricity Effects 0.000 claims abstract description 325
- 230000002159 abnormal effect Effects 0.000 claims abstract description 126
- 238000007477 logistic regression Methods 0.000 claims abstract description 23
- 238000002372 labelling Methods 0.000 claims abstract description 16
- 238000004140 cleaning Methods 0.000 claims abstract description 13
- 238000000034 method Methods 0.000 claims description 30
- 238000012549 training Methods 0.000 claims description 21
- 230000006872 improvement Effects 0.000 claims description 20
- 239000011159 matrix material Substances 0.000 claims description 19
- 238000004422 calculation algorithm Methods 0.000 claims description 8
- 238000005457 optimization Methods 0.000 claims description 8
- 238000012216 screening Methods 0.000 claims description 3
- 238000012163 sequencing technique Methods 0.000 claims description 3
- 230000009466 transformation Effects 0.000 claims description 3
- 230000010354 integration Effects 0.000 claims 1
- 230000005856 abnormality Effects 0.000 abstract description 10
- 238000005516 engineering process Methods 0.000 abstract description 3
- 230000006399 behavior Effects 0.000 description 17
- 230000006870 function Effects 0.000 description 11
- 238000004364 calculation method Methods 0.000 description 7
- 238000013473 artificial intelligence Methods 0.000 description 4
- 230000008569 process Effects 0.000 description 4
- 238000003066 decision tree Methods 0.000 description 2
- 238000010586 diagram Methods 0.000 description 2
- 238000000605 extraction Methods 0.000 description 2
- 230000007246 mechanism Effects 0.000 description 2
- 230000004913 activation Effects 0.000 description 1
- 238000004891 communication Methods 0.000 description 1
- 238000004590 computer program Methods 0.000 description 1
- 238000013075 data extraction Methods 0.000 description 1
- 230000000694 effects Effects 0.000 description 1
- 230000014509 gene expression Effects 0.000 description 1
- 238000012417 linear regression Methods 0.000 description 1
- 239000000463 material Substances 0.000 description 1
- 238000012986 modification Methods 0.000 description 1
- 230000004048 modification Effects 0.000 description 1
- 238000012544 monitoring process Methods 0.000 description 1
- 238000012545 processing Methods 0.000 description 1
- 238000010845 search algorithm Methods 0.000 description 1
- 238000012706 support-vector machine Methods 0.000 description 1
Images
Classifications
-
- 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
- Supply And Distribution Of Alternating Current (AREA)
- Management, Administration, Business Operations System, And Electronic Commerce (AREA)
Abstract
The invention relates to a power consumption abnormality analysis technology, and discloses a power consumption abnormality behavior analysis method under a fusion model, which comprises the following steps: carrying out data cleaning and data labeling on the historical electricity consumption data to obtain standard historical electricity consumption data; extracting first abnormal data corresponding to the standard historical electricity consumption data, and extracting second abnormal data corresponding to the standard historical electricity consumption data; extracting first standard data corresponding to the first optimized data and second standard data corresponding to the second abnormal data by using a preset primary regression model; generating standard analysis data according to the first standard data and the second standard data, and iteratively updating the primary regression model into a logistic regression model according to the standard analysis data and the standard historical electricity consumption data; and acquiring real-time electricity utilization data, and analyzing abnormal electricity utilization data corresponding to the real-time electricity utilization data by utilizing a logistic regression model. The invention also provides an electricity consumption abnormal behavior analysis device under the fusion model. The invention can improve the efficiency of power consumption abnormality analysis.
Description
Technical Field
The invention relates to the technical field of electricity consumption abnormality analysis, in particular to an electricity consumption abnormality behavior analysis method and device under a fusion model.
Background
Along with the promotion of modern processes, the demand of people for electric power is also increasing, and the power supply mechanism further expands the electric power deployment of each place, so that more electric potential hazards are brought, and in order to ensure the electricity safety, abnormal behavior analysis is required to be carried out on electricity data.
The existing electricity abnormal behavior analysis technology is mostly based on the electricity abnormal recognition of manual analysis. For example, based on human experience, analyzing the electricity data which does not accord with the previous electricity data, judging the type of the abnormal behavior according to the electricity value, in practical application, the electricity abnormal recognition based on manual analysis has higher requirements on the experience of staff, and the manual analysis consumes time and labor, so that the requirement of timely monitoring and timely finding potential safety hazards is difficult to meet, and the analysis efficiency of the electricity abnormal behavior is possibly lower.
Disclosure of Invention
The invention provides a method and a device for analyzing abnormal electricity consumption behavior under a fusion model, and mainly aims to solve the problem of low efficiency of analyzing abnormal electricity consumption behavior.
In order to achieve the above object, the present invention provides a method for analyzing abnormal behavior of electricity under a fusion model, including:
acquiring historical electricity utilization data, and performing data cleaning and data labeling on the historical electricity utilization data to obtain standard historical electricity utilization data;
extracting first abnormal data corresponding to the standard historical electricity consumption data by using a preset vector machine model, and extracting second abnormal data corresponding to the standard historical electricity consumption data by using a preset gradient lifting model;
weighting and optimizing the first abnormal data into first optimized data by using a preset primary regression model, extracting first standard data corresponding to the first optimized data by using the vector machine model, weighting and optimizing the second abnormal data into second optimized data by using the primary regression model, and extracting second standard data corresponding to the second optimized data by using the gradient lifting model;
generating standard analysis data according to the first standard data and the second standard data by using the primary regression model, calculating a loss value of the primary regression model according to the standard analysis data and the standard historical electricity consumption data, and performing iterative updating on the primary regression model by using the loss value to obtain a logistic regression model, wherein the calculating the loss value of the primary regression model according to the standard analysis data and the standard historical electricity consumption data comprises the following steps:
Counting the data length of the standard analysis data;
calculating a loss value of the primary regression model according to the data length, the standard analysis data and the standard historical electricity consumption data by using the following regression loss value algorithm:
wherein ,
it is meant that the loss value is such that,
refers to the data length, and the data length of the standard historical electricity consumption data of the data length of the standard analysis data is consistent,
refers to the first
The number of the two-dimensional space-saving type,
refers to the first of the standard historical electricity consumption data
The tag of the individual data is used to determine,
is a sign of a logarithmic function,
refers to the first of the standard analysis data
A tag for the individual data;
and acquiring real-time electricity utilization data, and analyzing abnormal electricity utilization data corresponding to the real-time electricity utilization data by using the vector machine model, the gradient lifting model and the logistic regression model.
Optionally, the performing data cleaning and data labeling on the historical electricity consumption data to obtain standard historical electricity consumption data includes:
sequencing the historical electricity utilization data according to a time sequence to obtain time sequence historical electricity utilization data;
screening out offside data and messy code data from the time sequence historical electricity utilization data to obtain primary historical electricity utilization data;
Performing time sequence clustering on the primary historical electricity consumption data to obtain a plurality of historical clustering data sets;
filling missing data in the primary historical electricity consumption data according to the historical cluster data group to obtain secondary historical electricity consumption data;
and marking the abnormal electricity consumption data in the secondary historical electricity consumption data to obtain standard historical electricity consumption data.
Optionally, the performing time-series clustering on the primary historical electricity consumption data to obtain a plurality of historical clustering data sets includes:
splitting the primary historical electricity utilization data into a plurality of electricity utilization data groups, randomly selecting central data of each electricity utilization data group, and calculating first data difference values between each data in the primary historical electricity utilization data and the central data of each electricity utilization data group;
grouping each data in the primary historical electricity utilization data according to the first data difference value and a nearby principle to obtain a plurality of standard electricity utilization data sets;
calculating average value data corresponding to each standard electricity utilization data set, and calculating second data difference values between each data in each standard electricity utilization data set and the corresponding average value data;
and grouping the data in the primary historical electricity utilization data again according to the second data difference value and the nearby principle so as to repeatedly and iteratively calculate the mean value data of each group of data, thereby obtaining a plurality of historical clustering data groups corresponding to the primary historical electricity utilization data.
Optionally, the filling the missing data in the primary historical electricity consumption data according to the historical cluster data set to obtain secondary historical electricity consumption data includes:
selecting missing data in the primary historical electricity utilization data one by one as target missing data, and taking a historical clustering data set corresponding to the target missing data as a target historical clustering data set;
and replacing the target missing data by using the mean value data corresponding to the target historical cluster data group until the target missing data is the last missing data in the primary historical electricity utilization data, and taking the updated primary historical electricity utilization data as secondary historical electricity utilization data.
Optionally, the extracting, by using a preset vector machine model, first abnormal data corresponding to the standard historical electricity consumption data includes:
extracting features of the standard historical electricity utilization data by using a preset vector machine model to obtain a first electricity feature set;
performing feature classification on the first electric feature set by using a classifier of the vector machine model to obtain a plurality of first electric feature sets;
calculating feature correlation values corresponding to the first electric feature groups respectively, and calculating a covariance matrix according to all the feature correlation values;
And carrying out linear transformation on the covariance matrix to obtain a hyperplane matrix, and calculating first abnormal data corresponding to the standard historical electricity consumption data according to the hyperplane matrix and the covariance matrix.
Optionally, the extracting the second abnormal data corresponding to the standard historical electricity consumption data by using a preset gradient lifting model includes:
extracting features of the standard historical electricity utilization data by using a preset gradient lifting model to obtain a second electricity utilization feature set;
selecting learners in the gradient lifting model one by one as target learners, and calculating initial characteristic data corresponding to the second electricity utilization characteristic set by using the target learners;
calculating initial feature residual errors corresponding to the target learner according to the standard historical electricity consumption data and the initial feature data;
and replacing the initial characteristic data by the initial characteristic residual, returning to the step of selecting learners in the gradient lifting model one by one as target learners until the target learners are the last learners in the gradient lifting model, and collecting all the initial characteristic residual into second abnormal data corresponding to the standard historical electricity consumption data.
Optionally, the weighting optimization of the first abnormal data into first optimized data by using a preset primary regression model includes:
calculating the primary probability density of the first abnormal data by using a preset primary regression model;
calculating an expected improvement value of the first anomaly data based on the primary probability density;
and selecting a maximum expected value from the expected improvement values, and taking the maximum expected value as the first optimization data.
Optionally, the calculating the expected improvement value of the first anomaly data according to the primary probability density includes:
extracting a first probability density and a second probability density from the primary probability density;
calculating an expected improvement value of the first anomaly data from the first and second probability densities using an expected improvement value formula:
wherein ,
it is meant that the desired improvement value,
refers to the probability that the loss value is less than a preset loss threshold,
it is meant that the loss threshold value is set,
means that the loss value of the initial regression model is
The probability of the time-of-day,
refers to loss value of
The integral of the time-of-day,
is the sign of the integral and,
refers to the second probability density of the first probability, 
It is meant that the first probability density,
is a proportional sign.
Optionally, the analyzing abnormal electricity consumption data corresponding to the real-time electricity consumption data by using the vector machine model, the gradient lifting model and the logistic regression model includes:
calculating a first real-time characteristic corresponding to the real-time electricity utilization data by using the vector machine model;
calculating a second real-time characteristic corresponding to the real-time electricity utilization data by using the gradient lifting model;
and analyzing the first real-time feature and the second real-time feature by using the logistic regression model to obtain abnormal electricity utilization data.
In order to solve the above problems, the present invention further provides an electricity consumption abnormality behavior analysis device under a fusion model, the device comprising:
the data labeling module is used for acquiring historical electricity utilization data, and carrying out data cleaning and data labeling on the historical electricity utilization data to obtain standard historical electricity utilization data;
the primary characteristic module is used for extracting first abnormal data corresponding to the standard historical electricity utilization data by using a preset vector machine model, and extracting second abnormal data corresponding to the standard historical electricity utilization data by using a preset gradient lifting model;
The secondary feature module is used for weighting and optimizing the first abnormal data into first optimized data by using a preset primary regression model, extracting first standard data corresponding to the first optimized data by using the vector machine model, weighting and optimizing the second abnormal data into second optimized data by using the primary regression model, and extracting second standard data corresponding to the second optimized data by using the gradient lifting model;
the model training module is configured to generate standard analysis data according to the first standard data and the second standard data by using the primary regression model, calculate a loss value of the primary regression model according to the standard analysis data and the standard historical electricity consumption data, and iteratively update the primary regression model by using the loss value to obtain a logistic regression model, where the calculating the loss value of the primary regression model according to the standard analysis data and the standard historical electricity consumption data includes: counting the data length of the standard analysis data; calculating a loss value of the primary regression model according to the data length, the standard analysis data and the standard historical electricity consumption data by using the following regression loss value algorithm:
wherein ,
it is meant that the loss value is such that,
refers to the data length, and the data length of the standard historical electricity consumption data of the data length of the standard analysis data is consistent,
refers to the first
The number of the two-dimensional space-saving type,
refers to the first of the standard historical electricity consumption data
The tag of the individual data is used to determine,
is a sign of a logarithmic function,
refers to the first of the standard analysis data
A tag for the individual data;
and the anomaly analysis module is used for acquiring real-time electricity utilization data and analyzing the anomaly electricity utilization data corresponding to the real-time electricity utilization data by utilizing the vector machine model, the gradient lifting model and the logistic regression model.
According to the embodiment of the invention, the historical electricity data is obtained, the historical electricity data is subjected to data cleaning and data marking to obtain the standard historical electricity data, so that the data noise of the training set data can be reduced, missing data is filled according to time sequence characteristics, the accuracy of the training set data is improved, the abnormal electricity data is determined through the data marking, the subsequent loss value calculation is convenient, the problem of dimension disasters and nonlinearity separability can be solved by extracting the first abnormal data corresponding to the standard historical electricity data through a preset vector machine model, the complexity in calculation is reduced, the complexity of the model can be limited by extracting the second abnormal data corresponding to the standard historical electricity data through a preset gradient lifting model, the occurrence of overfitting is prevented, and the model training speed is accelerated;
The first abnormal data are weighted and optimized into first optimized data by utilizing a preset primary regression model, first standard data corresponding to the first optimized data are extracted by utilizing the vector machine model, second abnormal data are weighted and optimized into second optimized data by utilizing the primary regression model, second standard data corresponding to the second optimized data are extracted by utilizing the gradient lifting model, the training parameters of the vector machine model and the gradient lifting model can be subjected to prior optimization by utilizing the primary regression model, so that the model training efficiency is improved, the efficiency of electricity utilization abnormal analysis is further improved, standard analysis data are generated by utilizing the primary regression model according to the first standard data and the second standard data, the model fusion of the vector machine model and the gradient lifting model can be realized, the accuracy of electricity utilization abnormal behavior analysis is further improved, abnormal electricity utilization data corresponding to the real-time electricity utilization data can be analyzed by utilizing the vector machine model, the gradient lifting model and the logic regression model, and the abnormal electricity utilization efficiency can be ensured, and the accuracy of the abnormal electricity utilization analysis of the abnormal data can be ensured. Therefore, the method and the device for analyzing the abnormal electricity consumption behavior under the fusion model can solve the problem of low efficiency of analyzing the abnormal electricity consumption behavior.
Drawings
FIG. 1 is a flow chart of a method for analyzing abnormal electricity consumption behavior under a fusion model according to an embodiment of the present invention;
FIG. 2 is a flowchart illustrating a method for extracting first exception data according to an embodiment of the present invention;
FIG. 3 is a flowchart illustrating a first optimized data extraction process according to an embodiment of the present invention;
FIG. 4 is a functional block diagram of an apparatus for analyzing abnormal behavior of electricity under a fusion model according to an embodiment of the present invention;
the achievement of the objects, functional features and advantages of the present invention will be further described with reference to the accompanying drawings, in conjunction with the embodiments.
Detailed Description
It should be understood that the specific embodiments described herein are for purposes of illustration only and are not intended to limit the scope of the invention.
The embodiment of the application provides a power consumption abnormal behavior analysis method under a fusion model. The execution main body of the electricity consumption abnormal behavior analysis method under the fusion model comprises, but is not limited to, at least one of a server side, a terminal and the like which can be configured to execute the method provided by the embodiment of the application. In other words, the analysis method of abnormal electricity consumption behavior under the fusion model may be executed by software or hardware installed in a terminal device or a server device, where the software may be a blockchain platform. The service end includes but is not limited to: a single server, a server cluster, a cloud server or a cloud server cluster, and the like. The server may be an independent server, or may be a cloud server that provides cloud services, cloud databases, cloud computing, cloud functions, cloud storage, network services, cloud communications, middleware services, domain name services, security services, content delivery networks (Content Delivery Network, CDN), and basic cloud computing services such as big data and artificial intelligence platforms.
Referring to fig. 1, a flow chart of a method for analyzing abnormal electricity consumption behavior under a fusion model according to an embodiment of the invention is shown. In this embodiment, the method for analyzing abnormal electricity consumption behavior under the fusion model includes:
s1, acquiring historical electricity consumption data, and performing data cleaning and data labeling on the historical electricity consumption data to obtain standard historical electricity consumption data.
In the embodiment of the invention, the historical electricity consumption data refers to electricity consumption data of a mechanism or equipment to be detected in a past period of time, wherein the electricity consumption data comprises a starting time stamp, an ending time stamp, total electricity consumption, current, voltage and other data of electricity consumption.
In the embodiment of the present invention, the performing data cleaning and data labeling on the historical electricity consumption data to obtain standard historical electricity consumption data includes:
sequencing the historical electricity utilization data according to a time sequence to obtain time sequence historical electricity utilization data;
screening out offside data and messy code data from the time sequence historical electricity utilization data to obtain primary historical electricity utilization data;
performing time sequence clustering on the primary historical electricity consumption data to obtain a plurality of historical clustering data sets;
filling missing data in the primary historical electricity consumption data according to the historical cluster data group to obtain secondary historical electricity consumption data;
And marking the abnormal electricity consumption data in the secondary historical electricity consumption data to obtain standard historical electricity consumption data.
In detail, the historical electricity data is sequenced according to the time sequence, so that the time sequence characteristics of the historical electricity data can be conveniently extracted by obtaining the time sequence historical electricity data, subsequent time sequence clustering is facilitated, and the accuracy of filling the missing data is further improved.
In detail, regular expressions or character string search algorithms can be utilized to screen offside data and messy code data from the time sequence historical electricity consumption data to obtain primary historical electricity consumption data, wherein the offside data is data which is far larger than normal data or far smaller than data which exceeds a data value range of the normal data in the time sequence historical electricity consumption data, and the messy code data is data with character errors in the recording process, such as @ or # and the like.
In detail, the performing time sequence clustering on the primary historical electricity consumption data to obtain a plurality of historical clustering data sets includes:
splitting the primary historical electricity utilization data into a plurality of electricity utilization data groups, randomly selecting central data of each electricity utilization data group, and calculating first data difference values between each data in the primary historical electricity utilization data and the central data of each electricity utilization data group;
Grouping each data in the primary historical electricity utilization data according to the first data difference value and a nearby principle to obtain a plurality of standard electricity utilization data sets;
calculating average value data corresponding to each standard electricity utilization data set, and calculating second data difference values between each data in each standard electricity utilization data set and the corresponding average value data;
and grouping the data in the primary historical electricity utilization data again according to the second data difference value and the nearby principle so as to repeatedly and iteratively calculate the mean value data of each group of data, thereby obtaining a plurality of historical clustering data groups corresponding to the primary historical electricity utilization data.
In detail, the randomly selecting the central data of each electricity consumption data group refers to randomly selecting one data for each electricity consumption data group as the central data corresponding to the electricity consumption data group; the first data difference value refers to the data difference between each data in the primary historical electricity utilization data and the central data, and the second data difference value refers to the data difference between each data in the primary historical material data and the mean value data; and calculating the average value data corresponding to each standard electricity utilization data set by using an average value algorithm.
In detail, the grouping of the data in the primary historical electricity consumption data again according to the second data difference value and the nearby principle to repeatedly and iteratively calculate the mean value data of each group of data, and the obtaining of the plurality of clustering data groups corresponding to the primary historical electricity consumption data refers to repeatedly and iteratively grouping the primary historical electricity consumption data according to the data difference between the mean value data and each data, calculating the grouped mean value data, and taking the grouped result as the clustering data group until the data difference value between the mean value data before and after iteration is smaller than a preset iteration threshold value.
In the embodiment of the present invention, the filling the missing data in the primary historical electricity consumption data according to the historical cluster data set to obtain secondary historical electricity consumption data includes:
selecting missing data in the primary historical electricity utilization data one by one as target missing data, and taking a historical clustering data set corresponding to the target missing data as a target historical clustering data set;
and replacing the target missing data by using the mean value data corresponding to the target historical cluster data group until the target missing data is the last missing data in the primary historical electricity utilization data, and taking the updated primary historical electricity utilization data as secondary historical electricity utilization data.
Specifically, the missing data refers to data which is not recorded in the primary historical electricity consumption data or is data which is left after the offside data and the messy code data are removed before, and the electricity consumption abnormal data refers to partial data in the secondary historical electricity consumption data corresponding to the alarm time of the temperature sensor or partial data in the secondary historical electricity consumption data corresponding to the abnormal time marked by a user.
In the embodiment of the invention, the historical electricity consumption data is obtained, and is subjected to data cleaning and data labeling to obtain the standard historical electricity consumption data, so that the data noise of the training set data can be reduced, the missing data is filled according to the time sequence characteristics, the accuracy of the training set data is improved, the abnormal electricity consumption data is determined through the data labeling, and the subsequent loss value calculation is convenient.
S2, extracting first abnormal data corresponding to the standard historical electricity consumption data by using a preset vector machine model, and extracting second abnormal data corresponding to the standard historical electricity consumption data by using a preset gradient lifting model.
In the embodiment of the present invention, referring to fig. 2, the extracting, by using a preset vector machine model, first abnormal data corresponding to the standard historical electricity consumption data includes:
S21, extracting features of the standard historical electricity utilization data by using a preset vector machine model to obtain a first electricity feature set;
s22, performing feature classification on the first electric feature set by using a classifier of the vector machine model to obtain a plurality of first electric feature sets;
s23, respectively calculating the characteristic correlation values corresponding to the first electric characteristic groups, and calculating a covariance matrix according to all the characteristic correlation values;
s24, performing linear transformation on the covariance matrix to obtain a hyperplane matrix, and calculating first abnormal data corresponding to the standard historical electricity consumption data according to the hyperplane matrix and the covariance matrix.
In detail, the characteristic extraction can be performed on the standard historical power consumption data by utilizing convolution check in the vector machine model to obtain a first power characteristic set; the classifier may be a linear classifier supporting a vector machine (support vector machines, SVM for short).
Specifically, the radial basis function may be used to calculate the feature correlation values corresponding to the first electrical feature groups, where calculating the covariance matrix according to all the feature correlation values refers to calculating the covariance matrix according to all the feature correlation values by using a covariance formula; the covariance matrix may be linearly transformed using a householder transform to obtain a hyperplane matrix.
In the embodiment of the present invention, the extracting the second abnormal data corresponding to the standard historical electricity consumption data by using a preset gradient lifting model includes:
extracting features of the standard historical electricity utilization data by using a preset gradient lifting model to obtain a second electricity utilization feature set;
selecting learners in the gradient lifting model one by one as target learners, and calculating initial characteristic data corresponding to the second electricity utilization characteristic set by using the target learners;
calculating initial feature residual errors corresponding to the target learner according to the standard historical electricity consumption data and the initial feature data;
and replacing the initial characteristic data by the initial characteristic residual, returning to the step of selecting learners in the gradient lifting model one by one as target learners until the target learners are the last learners in the gradient lifting model, and collecting all the initial characteristic residual into second abnormal data corresponding to the standard historical electricity consumption data.
In detail, the convolution in the gradient lifting model may be used to check the standard historical electricity data to perform feature extraction, so as to obtain a second electricity feature set, the learner refers to a decision tree model, the decision tree is a very commonly used classification method, and is a supervised learning, and the step of integrating all the initial feature residuals into second abnormal data corresponding to the standard historical electricity data refers to adding all the initial feature residuals to obtain the second abnormal data.
In the embodiment of the invention, the problem of dimension disaster and nonlinearity separability can be solved by extracting the first abnormal data corresponding to the standard historical electricity data by using the preset vector machine model, the complexity in calculation is reduced, the complexity of the model can be limited by extracting the second abnormal data corresponding to the standard historical electricity data by using the preset gradient lifting model, the occurrence of overfitting is prevented, and the model training speed is accelerated.
And S3, weighting and optimizing the first abnormal data into first optimized data by using a preset primary regression model, extracting first standard data corresponding to the first optimized data by using the vector machine model, weighting and optimizing the second abnormal data into second optimized data by using the primary regression model, and extracting second standard data corresponding to the second optimized data by using the gradient lifting model.
In an embodiment of the present invention, referring to fig. 3, the weighting optimization of the first abnormal data into first optimized data by using a preset primary regression model includes:
s31, calculating the primary probability density of the first abnormal data by using a preset primary regression model;
S32, calculating an expected improvement value of the first abnormal data according to the primary probability density;
s33, selecting a maximum expected value from the expected improvement values, and taking the maximum expected value as the first optimization data.
In detail, the primary regression model is a linear regression model, and the first anomaly data can be regarded as random variables, and the parameters are optimized by posterior calculation of model parameters.
In detail, the calculating the primary probability density of the first abnormal data by using a preset primary regression model includes: selecting a loss threshold value of the primary regression model as a target threshold value according to the first abnormal data, calculating a first probability density of the first abnormal data when the first probability density is larger than or equal to the target threshold value, calculating a second probability density of the first abnormal data when the first probability density is smaller than the target threshold value, and integrating the first probability density and the second probability density into a primary probability density.
In detail, the calculating the expected improvement value of the first anomaly data according to the primary probability density includes:
extracting a first probability density and a second probability density from the primary probability density;
Calculating an expected improvement value of the first anomaly data from the first and second probability densities using an expected improvement value formula:
wherein ,
it is meant that the desired improvement value,
refers to the probability that the loss value is less than a preset loss threshold,
it is meant that the loss threshold value is set,
means that the loss value of the initial regression model is
The probability of the time-of-day,
refers to loss value of
The integral of the time-of-day,
is the sign of the integral and,
refers to the second probability density of the first probability,
it is meant that the first probability density,
is a proportional sign.
In detail, by calculating the expected improvement value of the first abnormal data according to the first probability density and the second probability density by using the expected improvement value formula, quick model parameter tuning can be realized, and the model training speed is further improved.
In detail, the method for extracting the first standard data corresponding to the first optimized data by using the vector machine model is consistent with the method for extracting the first abnormal data corresponding to the standard historical electricity consumption data by using the preset vector machine model in the step S2, and will not be described herein.
Specifically, the method for weighting and optimizing the second abnormal data to second optimized data by using the primary regression model is consistent with the method for weighting and optimizing the first abnormal data to first optimized data by using the preset primary regression model in the step S3, which is not described herein.
In detail, the method for extracting the second standard data corresponding to the second optimized data by using the gradient lifting model is consistent with the step of extracting the second abnormal data corresponding to the standard historical electricity consumption data by using the preset gradient lifting model in the step S2, which is not described herein.
In the embodiment of the invention, the first abnormal data is weighted and optimized into the first optimized data by utilizing the preset primary regression model, the first standard data corresponding to the first optimized data is extracted by utilizing the vector machine model, the second abnormal data is weighted and optimized into the second optimized data by utilizing the primary regression model, and the second standard data corresponding to the second optimized data is extracted by utilizing the gradient lifting model, so that the training parameters of the vector machine model and the gradient lifting model can be optimized a priori by utilizing the primary regression model, the model training efficiency is improved, and the power consumption abnormality analysis efficiency is improved.
S4, generating standard analysis data according to the first standard data and the second standard data by using the primary regression model, calculating a loss value of the primary regression model according to the standard analysis data and the standard historical electricity consumption data, and performing iterative updating on the primary regression model by using the loss value to obtain a logistic regression model.
In the embodiment of the present invention, the method for generating the standard analysis data according to the first standard data and the second standard data by using the primary regression model is consistent with the method for weighting and optimizing the first abnormal data into the first optimized data by using the preset primary regression model in the step S3, which is not described herein.
In the embodiment of the present invention, the calculating the loss value of the primary regression model according to the standard analysis data and the standard historical electricity consumption data includes:
counting the data length of the standard analysis data;
calculating a loss value of the primary regression model according to the data length, the standard analysis data and the standard historical electricity consumption data by using the following regression loss value algorithm:
wherein ,
it is meant that the loss value is such that,
refers to the data length, and the data length of the standard historical electricity consumption data of the data length of the standard analysis data is consistent,
refers to the first
The number of the two-dimensional space-saving type,
refers to the first of the standard historical electricity consumption data
The tag of the individual data is used to determine,
is a sign of a logarithmic function,
refers to the first of the standard analysis data
A tag for the data.
In detail, the regression loss value algorithm is utilized to calculate the loss value of the primary regression model according to the data length, the standard analysis data and the standard historical electricity consumption data, so that the saturation phenomenon of an activation function can be prevented, the convergence speed is also improved, and the model training efficiency is further improved.
In detail, the step of iteratively updating the primary regression model by using the loss value to obtain a logistic regression model includes: judging whether the loss value is smaller than a preset loss threshold value or not; if not, optimizing parameters in the primary regression model according to the loss value, and returning to the step of weighting and optimizing the first abnormal data into first optimized data by using a preset primary regression model; if yes, the primary regression model with updated parameters is used as a logistic regression model.
In the embodiment of the invention, the primary regression model is utilized to generate the standard analysis data according to the first standard data and the second standard data, so that the model fusion of the vector machine model and the gradient lifting model can be realized, and the accuracy of analysis of abnormal power consumption behaviors is further improved.
S5, acquiring real-time electricity utilization data, and analyzing abnormal electricity utilization data corresponding to the real-time electricity utilization data by using the vector machine model, the gradient lifting model and the logistic regression model.
In the embodiment of the invention, the real-time electricity data refers to electricity data needed to be subjected to analysis of abnormal electricity behaviors, and the abnormal electricity data refers to corresponding electricity data when abnormal electricity behaviors possibly occur.
In the embodiment of the present invention, the analyzing abnormal electricity consumption data corresponding to the real-time electricity consumption data by using the vector machine model, the gradient lifting model and the logistic regression model includes:
calculating a first real-time characteristic corresponding to the real-time electricity utilization data by using the vector machine model;
calculating a second real-time characteristic corresponding to the real-time electricity utilization data by using the gradient lifting model;
and analyzing the first real-time feature and the second real-time feature by using the logistic regression model to obtain abnormal electricity utilization data.
In the embodiment of the invention, the abnormal electricity consumption data corresponding to the real-time electricity consumption data is analyzed by the vector machine model, the gradient lifting model and the logistic regression model, so that the accuracy of the abnormal electricity consumption data can be ensured, and the analysis efficiency of the abnormal electricity consumption behavior is improved.
According to the embodiment of the invention, the historical electricity data is obtained, the historical electricity data is subjected to data cleaning and data marking to obtain the standard historical electricity data, so that the data noise of the training set data can be reduced, missing data is filled according to time sequence characteristics, the accuracy of the training set data is improved, the abnormal electricity data is determined through the data marking, the subsequent loss value calculation is convenient, the problem of dimension disasters and nonlinearity separability can be solved by extracting the first abnormal data corresponding to the standard historical electricity data through a preset vector machine model, the complexity in calculation is reduced, the complexity of the model can be limited by extracting the second abnormal data corresponding to the standard historical electricity data through a preset gradient lifting model, the occurrence of overfitting is prevented, and the model training speed is accelerated;
The first abnormal data are weighted and optimized into first optimized data by utilizing a preset primary regression model, first standard data corresponding to the first optimized data are extracted by utilizing the vector machine model, second abnormal data are weighted and optimized into second optimized data by utilizing the primary regression model, second standard data corresponding to the second optimized data are extracted by utilizing the gradient lifting model, the training parameters of the vector machine model and the gradient lifting model can be subjected to prior optimization by utilizing the primary regression model, so that the model training efficiency is improved, the efficiency of electricity utilization abnormal analysis is further improved, standard analysis data are generated by utilizing the primary regression model according to the first standard data and the second standard data, the model fusion of the vector machine model and the gradient lifting model can be realized, the accuracy of electricity utilization abnormal behavior analysis is further improved, abnormal electricity utilization data corresponding to the real-time electricity utilization data can be analyzed by utilizing the vector machine model, the gradient lifting model and the logic regression model, and the abnormal electricity utilization efficiency can be ensured, and the accuracy of the abnormal electricity utilization analysis of the abnormal data can be ensured. Therefore, the analysis method for the abnormal electricity consumption behavior under the fusion model can solve the problem of low analysis efficiency of the abnormal electricity consumption behavior.
Fig. 4 is a functional block diagram of an apparatus for analyzing abnormal behavior of electricity under a fusion model according to an embodiment of the present invention.
The power consumption abnormality behavior analysis device 100 under the fusion model of the present invention may be installed in an electronic apparatus. According to the implemented functions, the power consumption abnormal behavior analysis device 100 under the fusion model may include a data labeling module 101, a primary feature module 102, a secondary feature module 103, a model training module 104 and an abnormal analysis module 105. The module of the invention, which may also be referred to as a unit, refers to a series of computer program segments, which are stored in the memory of the electronic device, capable of being executed by the processor of the electronic device and of performing a fixed function.
In the present embodiment, the functions concerning the respective modules/units are as follows:
the data labeling module 101 is configured to obtain historical electricity consumption data, perform data cleaning and data labeling on the historical electricity consumption data, and obtain standard historical electricity consumption data;
the primary feature module 102 is configured to extract first abnormal data corresponding to the standard historical electricity consumption data by using a preset vector machine model, and extract second abnormal data corresponding to the standard historical electricity consumption data by using a preset gradient lifting model;
The secondary feature module 103 is configured to weight-optimize the first abnormal data into first optimized data by using a preset primary regression model, extract first standard data corresponding to the first optimized data by using the vector machine model, weight-optimize the second abnormal data into second optimized data by using the primary regression model, and extract second standard data corresponding to the second optimized data by using the gradient lifting model;
the model training module 104 is configured to generate standard analysis data according to the first standard data and the second standard data by using the primary regression model, calculate a loss value of the primary regression model according to the standard analysis data and the standard historical electricity consumption data, and perform iterative update on the primary regression model by using the loss value to obtain a logistic regression model, where the calculating the loss value of the primary regression model according to the standard analysis data and the standard historical electricity consumption data includes: counting the data length of the standard analysis data; calculating a loss value of the primary regression model according to the data length, the standard analysis data and the standard historical electricity consumption data by using the following regression loss value algorithm:
wherein ,
it is meant that the loss value is such that,
refers to the data length and the standard historical electricity data of the data length of the standard analysis dataThe data length is consistent and the data is not uniform,
refers to the first
The number of the two-dimensional space-saving type,
refers to the first of the standard historical electricity consumption data
The tag of the individual data is used to determine,
is a sign of a logarithmic function,
refers to the first of the standard analysis data
A tag for the individual data;
the anomaly analysis module 105 is configured to obtain real-time electricity consumption data, and analyze the anomaly electricity consumption data corresponding to the real-time electricity consumption data by using the vector machine model, the gradient lifting model and the logistic regression model.
In detail, each module in the power consumption abnormality analysis device 100 under the fusion model in the embodiment of the present invention adopts the same technical means as the power consumption abnormality analysis method under the fusion model described in fig. 1 to 3, and can generate the same technical effects, which is not described herein.
In the several embodiments provided in the present invention, it should be understood that the disclosed apparatus, device and method may be implemented in other manners. For example, the above-described apparatus embodiments are merely illustrative, and for example, the division of the modules is merely a logical function division, and there may be other manners of division when actually implemented.
The modules described as separate components may or may not be physically separate, and components shown as modules may or may not be physical units, may be located in one place, or may be distributed over multiple network units. Some or all of the modules may be selected according to actual needs to achieve the purpose of the solution of this embodiment.
In addition, each functional module in the embodiments of the present invention may be integrated in one processing unit, or each unit may exist alone physically, or two or more units may be integrated in one unit. The integrated units can be realized in a form of hardware or a form of hardware and a form of software functional modules.
It will be evident to those skilled in the art that the invention is not limited to the details of the foregoing illustrative embodiments, and that the present invention may be embodied in other specific forms without departing from the spirit or essential characteristics thereof.
The present embodiments are, therefore, to be considered in all respects as illustrative and not restrictive, the scope of the invention being indicated by the appended claims rather than by the foregoing description, and all changes which come within the meaning and range of equivalency of the claims are therefore intended to be embraced therein. Any reference signs in the claims shall not be construed as limiting the claim concerned.
The embodiment of the application can acquire and process the related data based on the artificial intelligence technology. Among these, artificial intelligence (Artificial Intelligence, AI) is the theory, method, technique and application system that uses a digital computer or a digital computer-controlled machine to simulate, extend and extend human intelligence, sense the environment, acquire knowledge and use knowledge to obtain optimal results.
Furthermore, it is evident that the word "comprising" does not exclude other elements or steps, and that the singular does not exclude a plurality. A plurality of units or means recited in the system claims can also be implemented by means of software or hardware by means of one unit or means. The terms first, second, etc. are used to denote a name, but not any particular order.
Finally, it should be noted that the above-mentioned embodiments are merely for illustrating the technical solution of the present invention and not for limiting the same, and although the present invention has been described in detail with reference to the preferred embodiments, it should be understood by those skilled in the art that modifications and equivalents may be made to the technical solution of the present invention without departing from the spirit and scope of the technical solution of the present invention.
Claims (8)
1. The method for analyzing the abnormal electricity consumption behavior under the fusion model is characterized by comprising the following steps of:
S1: acquiring historical electricity utilization data, and performing data cleaning and data labeling on the historical electricity utilization data to obtain standard historical electricity utilization data; the historical electricity consumption data comprises a starting time stamp of electricity consumption, an ending time stamp of electricity consumption, total amount of electricity consumption, current magnitude and voltage magnitude;
s2: extracting first abnormal data corresponding to the standard historical electricity consumption data by using a preset vector machine model, and extracting second abnormal data corresponding to the standard historical electricity consumption data by using a preset gradient lifting model;
s3: weighting and optimizing the first abnormal data into first optimized data by using a preset primary regression model, extracting first standard data corresponding to the first optimized data by using the vector machine model, weighting and optimizing the second abnormal data into second optimized data by using the primary regression model, and extracting second standard data corresponding to the second optimized data by using the gradient lifting model;
s4: generating standard analysis data according to the first standard data and the second standard data by using the primary regression model, calculating a loss value of the primary regression model according to the standard analysis data and the standard historical electricity consumption data, and performing iterative updating on the primary regression model by using the loss value to obtain a logistic regression model, wherein the calculating the loss value of the primary regression model according to the standard analysis data and the standard historical electricity consumption data comprises the following steps:
S41: counting the data length of the standard analysis data;
s42: calculating a loss value of the primary regression model according to the data length, the standard analysis data and the standard historical electricity consumption data by using the following regression loss value algorithm:
wherein ,
refers to the loss value,/->
Means that the data length is the same as the standard historical electricity consumption data of the standard analysis data, and +.>
Refers to->
Personal (S)>
Means +.>
Tag of personal data->
Is a logarithmic function sign>
Means +.>
A tag for the individual data;
s5: acquiring real-time electricity utilization data, and analyzing abnormal electricity utilization data corresponding to the real-time electricity utilization data by using the vector machine model, the gradient lifting model and the logistic regression model;
the weighting optimization of the first abnormal data into first optimized data by using a preset primary regression model comprises the following steps:
calculating the primary probability density of the first abnormal data by using a preset primary regression model;
calculating an expected improvement value of the first anomaly data based on the primary probability density;
Selecting a maximum expected value from the expected improvement values, and taking the maximum expected value as the first optimization data;
the calculating the expected improvement value of the first anomaly data based on the primary probability density includes:
extracting a first probability density and a second probability density from the primary probability density;
calculating an expected improvement value of the first anomaly data from the first and second probability densities using an expected improvement value formula:
wherein ,
meaning the desired improvement value, +.>
Means the probability of the loss value being smaller than a preset loss threshold,/for the time of the loss value>
Refers to the loss threshold,/->
Means that the loss value of the primary regression model is +.>
Probability of time->
Means that the loss value is +.>
Integration of time->
Is an integral symbol +.>
Means that said second probability density, +.>
Means that said first probability density, < > is>
Is a proportional sign.
2. The method for analyzing abnormal electricity consumption behavior under a fusion model according to claim 1, wherein the step of performing data cleaning and data labeling on the historical electricity consumption data to obtain standard historical electricity consumption data comprises the steps of:
sequencing the historical electricity utilization data according to a time sequence to obtain time sequence historical electricity utilization data;
Screening out offside data and messy code data from the time sequence historical electricity utilization data to obtain primary historical electricity utilization data;
performing time sequence clustering on the primary historical electricity consumption data to obtain a plurality of historical clustering data sets;
filling missing data in the primary historical electricity consumption data according to the historical cluster data group to obtain secondary historical electricity consumption data;
and marking the abnormal electricity consumption data in the secondary historical electricity consumption data to obtain standard historical electricity consumption data.
3. The method for analyzing abnormal electricity consumption behavior under a fusion model according to claim 2, wherein the performing time-series clustering on the primary historical electricity consumption data to obtain a plurality of historical clustering data sets includes:
splitting the primary historical electricity utilization data into a plurality of electricity utilization data groups, randomly selecting central data of each electricity utilization data group, and calculating first data difference values between each data in the primary historical electricity utilization data and the central data of each electricity utilization data group;
grouping each data in the primary historical electricity utilization data according to the first data difference value and a nearby principle to obtain a plurality of standard electricity utilization data sets;
Calculating average value data corresponding to each standard electricity utilization data set, and calculating second data difference values between each data in each standard electricity utilization data set and the corresponding average value data;
and grouping the data in the primary historical electricity utilization data again according to the second data difference value and the nearby principle so as to repeatedly and iteratively calculate the mean value data of each group of data, thereby obtaining a plurality of historical clustering data groups corresponding to the primary historical electricity utilization data.
4. The method for analyzing abnormal electricity consumption behavior under a fusion model according to claim 2, wherein the filling missing data in the primary historical electricity consumption data according to the historical cluster data set to obtain secondary historical electricity consumption data comprises:
selecting missing data in the primary historical electricity utilization data one by one as target missing data, and taking a historical clustering data set corresponding to the target missing data as a target historical clustering data set;
and replacing the target missing data by using the mean value data corresponding to the target historical cluster data group until the target missing data is the last missing data in the primary historical electricity utilization data, and taking the updated primary historical electricity utilization data as secondary historical electricity utilization data.
5. The method for analyzing abnormal electricity consumption behavior under a fusion model according to claim 1, wherein the extracting the first abnormal data corresponding to the standard historical electricity consumption data by using a preset vector machine model comprises:
extracting features of the standard historical electricity utilization data by using a preset vector machine model to obtain a first electricity feature set;
performing feature classification on the first electric feature set by using a classifier of the vector machine model to obtain a plurality of first electric feature sets;
calculating feature correlation values corresponding to the first electric feature groups respectively, and calculating a covariance matrix according to all the feature correlation values;
and carrying out linear transformation on the covariance matrix to obtain a hyperplane matrix, and calculating first abnormal data corresponding to the standard historical electricity consumption data according to the hyperplane matrix and the covariance matrix.
6. The method for analyzing abnormal electricity consumption behavior under a fusion model according to claim 1, wherein the extracting the second abnormal data corresponding to the standard historical electricity consumption data by using a preset gradient lifting model comprises:
extracting features of the standard historical electricity utilization data by using a preset gradient lifting model to obtain a second electricity utilization feature set;
Selecting learners in the gradient lifting model one by one as target learners, and calculating initial characteristic data corresponding to the second electricity utilization characteristic set by using the target learners;
calculating initial feature residual errors corresponding to the target learner according to the standard historical electricity consumption data and the initial feature data;
and replacing the initial characteristic data by the initial characteristic residual, returning to the step of selecting learners in the gradient lifting model one by one as target learners until the target learners are the last learners in the gradient lifting model, and collecting all the initial characteristic residual into second abnormal data corresponding to the standard historical electricity consumption data.
7. The method for analyzing abnormal electricity consumption behavior under a fusion model according to claim 1, wherein the analyzing abnormal electricity consumption data corresponding to the real-time electricity consumption data by using the vector machine model, the gradient lifting model and the logistic regression model comprises:
calculating a first real-time characteristic corresponding to the real-time electricity utilization data by using the vector machine model;
calculating a second real-time characteristic corresponding to the real-time electricity utilization data by using the gradient lifting model;
And analyzing the first real-time feature and the second real-time feature by using the logistic regression model to obtain abnormal electricity utilization data.
8. An electrical anomaly analysis device under a fusion model for implementing the electrical anomaly analysis method under a fusion model as defined in claim 1, wherein the device comprises:
the data labeling module is used for acquiring historical electricity utilization data, and carrying out data cleaning and data labeling on the historical electricity utilization data to obtain standard historical electricity utilization data;
the primary characteristic module is used for extracting first abnormal data corresponding to the standard historical electricity utilization data by using a preset vector machine model, and extracting second abnormal data corresponding to the standard historical electricity utilization data by using a preset gradient lifting model;
the secondary feature module is used for weighting and optimizing the first abnormal data into first optimized data by using a preset primary regression model, extracting first standard data corresponding to the first optimized data by using the vector machine model, weighting and optimizing the second abnormal data into second optimized data by using the primary regression model, and extracting second standard data corresponding to the second optimized data by using the gradient lifting model;
The model training module is configured to generate standard analysis data according to the first standard data and the second standard data by using the primary regression model, calculate a loss value of the primary regression model according to the standard analysis data and the standard historical electricity consumption data, and iteratively update the primary regression model by using the loss value to obtain a logistic regression model, where the calculating the loss value of the primary regression model according to the standard analysis data and the standard historical electricity consumption data includes: counting the data length of the standard analysis data; calculating a loss value of the primary regression model according to the data length, the standard analysis data and the standard historical electricity consumption data by using the following regression loss value algorithm:
wherein ,
refers to the loss value,/->
Means that the data length is the same as the standard historical electricity consumption data of the standard analysis data, and +.>
Refers to->
Personal (S)>
Means +.>
Tag of personal data->
Is a logarithmic function sign>
Means +.>
A tag for the individual data;
And the anomaly analysis module is used for acquiring real-time electricity utilization data and analyzing the anomaly electricity utilization data corresponding to the real-time electricity utilization data by utilizing the vector machine model, the gradient lifting model and the logistic regression model.
Priority Applications (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN202310035095.XA CN115758255B (en) | 2023-01-10 | 2023-01-10 | Power consumption abnormal behavior analysis method and device under fusion model |
Applications Claiming Priority (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN202310035095.XA CN115758255B (en) | 2023-01-10 | 2023-01-10 | Power consumption abnormal behavior analysis method and device under fusion model |
Publications (2)
| Publication Number | Publication Date |
|---|---|
| CN115758255A CN115758255A (en) | 2023-03-07 |
| CN115758255B true CN115758255B (en) | 2023-05-05 |
Family
ID=85348913
Family Applications (1)
| Application Number | Title | Priority Date | Filing Date |
|---|---|---|---|
| CN202310035095.XA Active CN115758255B (en) | 2023-01-10 | 2023-01-10 | Power consumption abnormal behavior analysis method and device under fusion model |
Country Status (1)
| Country | Link |
|---|---|
| CN (1) | CN115758255B (en) |
Families Citing this family (2)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN116090795B (en) * | 2023-03-28 | 2023-07-14 | 佰聆数据股份有限公司 | DNN-based distribution network grid dynamic resource planning method and device |
| CN116154239B (en) * | 2023-04-14 | 2023-11-21 | 湖南省计量检测研究院 | Multi-level implementation-based hydrogen fuel cell energy conversion method and device |
Citations (2)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN107563539A (en) * | 2017-07-24 | 2018-01-09 | 佛山市顺德区中山大学研究院 | Short-term and long-medium term power load forecasting method based on machine learning model |
| CN115034519A (en) * | 2022-07-29 | 2022-09-09 | 广东电网有限责任公司 | Method and device for predicting power load, electronic equipment and storage medium |
-
2023
- 2023-01-10 CN CN202310035095.XA patent/CN115758255B/en active Active
Patent Citations (2)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN107563539A (en) * | 2017-07-24 | 2018-01-09 | 佛山市顺德区中山大学研究院 | Short-term and long-medium term power load forecasting method based on machine learning model |
| CN115034519A (en) * | 2022-07-29 | 2022-09-09 | 广东电网有限责任公司 | Method and device for predicting power load, electronic equipment and storage medium |
Non-Patent Citations (1)
| Title |
|---|
| 基于智能电表大数据的异常用电检测;潘磊;杨延;连浩;方龙泉;计算技术与自动化(第02期);全文 * |
Also Published As
| Publication number | Publication date |
|---|---|
| CN115758255A (en) | 2023-03-07 |
Similar Documents
| Publication | Publication Date | Title |
|---|---|---|
| CN115758255B (en) | Power consumption abnormal behavior analysis method and device under fusion model | |
| CN110347835B (en) | Text clustering method, electronic device and storage medium | |
| CN108737406B (en) | Method and system for detecting abnormal flow data | |
| CN112383052B (en) | Power grid fault repairing method and device based on power internet of things | |
| CN111199244B (en) | Data classification method and device, storage medium and electronic device | |
| CN111126820B (en) | Method and system for preventing electricity stealing | |
| CN115795329B (en) | Power utilization abnormal behavior analysis method and device based on big data grid | |
| CN111027629A (en) | Power distribution network fault outage rate prediction method and system based on improved random forest | |
| CN115905959B (en) | Defect factor-based power circuit breaker relevance fault analysis method and device | |
| CN112115965A (en) | SVM-based passive operating system identification method, storage medium and equipment | |
| CN115130538A (en) | Training method of text classification model, text processing method, equipment and medium | |
| CN117041017A (en) | Intelligent operation and maintenance management method and system for data center | |
| CN115659244A (en) | Fault prediction method, device and storage medium | |
| CN111882034A (en) | Neural network processing and face recognition method, device, equipment and storage medium | |
| CN117155771B (en) | Equipment cluster fault tracing method and device based on industrial Internet of things | |
| CN106484913A (en) | Method and server that a kind of Target Photo determines | |
| CN114238764A (en) | Course recommendation method, device and equipment based on recurrent neural network | |
| CN113282433A (en) | Cluster anomaly detection method and device and related equipment | |
| CN116865994A (en) | Network data security prediction method based on big data | |
| CN115809796B (en) | Project intelligent dispatching method and system based on user portrait | |
| CN113392868A (en) | Model training method, related device, equipment and storage medium | |
| CN110414229B (en) | Operation command detection method, device, computer equipment and storage medium | |
| CN113392867A (en) | Image identification method and device, computer equipment and storage medium | |
| CN116739408A (en) | Power grid dispatching safety monitoring method and system based on data tag and electronic equipment | |
| CN104200222B (en) | Object identifying method in a kind of picture based on factor graph model |
Legal Events
| Date | Code | Title | Description |
|---|---|---|---|
| PB01 | Publication | ||
| PB01 | Publication | ||
| SE01 | Entry into force of request for substantive examination | ||
| SE01 | Entry into force of request for substantive examination | ||
| GR01 | Patent grant | ||
| GR01 | Patent grant |