CN114970729A - Abnormal electricity consumption time interval detection method for smart power grid - Google Patents
Abnormal electricity consumption time interval detection method for smart power grid Download PDFInfo
- Publication number
- CN114970729A CN114970729A CN202210609391.1A CN202210609391A CN114970729A CN 114970729 A CN114970729 A CN 114970729A CN 202210609391 A CN202210609391 A CN 202210609391A CN 114970729 A CN114970729 A CN 114970729A
- Authority
- CN
- China
- Prior art keywords
- time interval
- abnormal
- abnormal electricity
- candidate
- real
- Prior art date
- Legal status (The legal status is an assumption and is not a legal conclusion. Google has not performed a legal analysis and makes no representation as to the accuracy of the status listed.)
- Granted
Links
- 230000002159 abnormal effect Effects 0.000 title claims abstract description 127
- 230000005611 electricity Effects 0.000 title claims abstract description 81
- 238000001514 detection method Methods 0.000 title claims abstract description 49
- 238000000034 method Methods 0.000 claims abstract description 18
- 238000000605 extraction Methods 0.000 claims abstract description 13
- 238000012216 screening Methods 0.000 claims abstract description 3
- 230000007246 mechanism Effects 0.000 claims description 4
- 238000010276 construction Methods 0.000 claims description 3
- 238000007689 inspection Methods 0.000 description 10
- 230000006870 function Effects 0.000 description 9
- 230000006399 behavior Effects 0.000 description 8
- 208000037170 Delayed Emergence from Anesthesia Diseases 0.000 description 6
- 238000010586 diagram Methods 0.000 description 6
- 238000011160 research Methods 0.000 description 5
- 230000008569 process Effects 0.000 description 3
- 238000005070 sampling Methods 0.000 description 3
- 238000012549 training Methods 0.000 description 3
- 238000003745 diagnosis Methods 0.000 description 2
- 230000006872 improvement Effects 0.000 description 2
- 238000011835 investigation Methods 0.000 description 2
- 238000010801 machine learning Methods 0.000 description 2
- 238000012360 testing method Methods 0.000 description 2
- ORILYTVJVMAKLC-UHFFFAOYSA-N Adamantane Natural products C1C(C2)CC3CC1CC2C3 ORILYTVJVMAKLC-UHFFFAOYSA-N 0.000 description 1
- 230000005856 abnormality Effects 0.000 description 1
- 230000009286 beneficial effect Effects 0.000 description 1
- 238000004422 calculation algorithm Methods 0.000 description 1
- 238000004364 calculation method Methods 0.000 description 1
- 238000004891 communication Methods 0.000 description 1
- 230000007547 defect Effects 0.000 description 1
- 238000009499 grossing Methods 0.000 description 1
- 238000012986 modification Methods 0.000 description 1
- 230000004048 modification Effects 0.000 description 1
- 230000009467 reduction Effects 0.000 description 1
- 238000007619 statistical method Methods 0.000 description 1
Images
Classifications
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06F—ELECTRIC DIGITAL DATA PROCESSING
- G06F18/00—Pattern recognition
- G06F18/20—Analysing
- G06F18/24—Classification techniques
- G06F18/241—Classification techniques relating to the classification model, e.g. parametric or non-parametric approaches
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06N—COMPUTING ARRANGEMENTS BASED ON SPECIFIC COMPUTATIONAL MODELS
- G06N3/00—Computing arrangements based on biological models
- G06N3/02—Neural networks
- G06N3/04—Architecture, e.g. interconnection topology
- G06N3/045—Combinations of networks
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06Q—INFORMATION AND COMMUNICATION TECHNOLOGY [ICT] SPECIALLY ADAPTED FOR ADMINISTRATIVE, COMMERCIAL, FINANCIAL, MANAGERIAL OR SUPERVISORY PURPOSES; SYSTEMS OR METHODS SPECIALLY ADAPTED FOR ADMINISTRATIVE, COMMERCIAL, FINANCIAL, MANAGERIAL OR SUPERVISORY PURPOSES, NOT OTHERWISE PROVIDED FOR
- G06Q50/00—Information and communication technology [ICT] specially adapted for implementation of business processes of specific business sectors, e.g. utilities or tourism
- G06Q50/06—Energy or water supply
-
- Y—GENERAL TAGGING OF NEW TECHNOLOGICAL DEVELOPMENTS; GENERAL TAGGING OF CROSS-SECTIONAL TECHNOLOGIES SPANNING OVER SEVERAL SECTIONS OF THE IPC; TECHNICAL SUBJECTS COVERED BY FORMER USPC CROSS-REFERENCE ART COLLECTIONS [XRACs] AND DIGESTS
- Y04—INFORMATION OR COMMUNICATION TECHNOLOGIES HAVING AN IMPACT ON OTHER TECHNOLOGY AREAS
- Y04S—SYSTEMS INTEGRATING TECHNOLOGIES RELATED TO POWER NETWORK OPERATION, COMMUNICATION OR INFORMATION TECHNOLOGIES FOR IMPROVING THE ELECTRICAL POWER GENERATION, TRANSMISSION, DISTRIBUTION, MANAGEMENT OR USAGE, i.e. SMART GRIDS
- Y04S10/00—Systems supporting electrical power generation, transmission or distribution
- Y04S10/50—Systems or methods supporting the power network operation or management, involving a certain degree of interaction with the load-side end user applications
Landscapes
- Engineering & Computer Science (AREA)
- Theoretical Computer Science (AREA)
- Physics & Mathematics (AREA)
- Business, Economics & Management (AREA)
- Data Mining & Analysis (AREA)
- Health & Medical Sciences (AREA)
- General Physics & Mathematics (AREA)
- Artificial Intelligence (AREA)
- Economics (AREA)
- General Engineering & Computer Science (AREA)
- Life Sciences & Earth Sciences (AREA)
- Evolutionary Computation (AREA)
- General Health & Medical Sciences (AREA)
- Software Systems (AREA)
- Human Resources & Organizations (AREA)
- Mathematical Physics (AREA)
- Molecular Biology (AREA)
- Computational Linguistics (AREA)
- Biophysics (AREA)
- Biomedical Technology (AREA)
- Public Health (AREA)
- Water Supply & Treatment (AREA)
- Computing Systems (AREA)
- Marketing (AREA)
- Primary Health Care (AREA)
- Strategic Management (AREA)
- Tourism & Hospitality (AREA)
- General Business, Economics & Management (AREA)
- Bioinformatics & Cheminformatics (AREA)
- Bioinformatics & Computational Biology (AREA)
- Computer Vision & Pattern Recognition (AREA)
- Evolutionary Biology (AREA)
- Supply And Distribution Of Alternating Current (AREA)
Abstract
The invention discloses an abnormal electricity utilization time interval detection method for an intelligent power grid, which comprises the following steps: screening out a candidate interval based on expert knowledge, and judging whether the candidate interval is a real abnormal electricity utilization time interval or not according to whether the overlapping rate of the candidate interval and the real abnormal electricity utilization time interval exceeds a threshold value or not; inputting abnormal power consumption time sequence data into a feature extraction network to obtain a feature map, performing regional weighting on the feature map through different candidate intervals to obtain a corresponding weighted feature map, and inputting the weighted feature map into a classification network and a regression network to obtain predicted candidate interval labels and predicted abnormal power consumption time intervals; calculating loss by using a real candidate interval label and a real abnormal electricity utilization time interval as real labels until the model converges; and detecting the abnormal electricity utilization time interval by using the converged model. The method can effectively extract key information from the long-time data with more redundant information and noise, thereby improving the detection speed and the detection accuracy.
Description
Technical Field
The invention belongs to the technical field of abnormal power utilization detection, and particularly relates to a method for detecting an abnormal power utilization time interval for a smart grid.
Background
With the reform of the power industry and the further construction of the smart grid, the automation, informatization and intelligentization degrees of various power equipment and systems are improved to a greater extent, and the explosion-type growth of a large amount of electric energy metering data follows. In the process of collecting electric energy metering information, a large amount of data are abnormal due to reasons such as power grid fluctuation, equipment faults, communication faults and management modes. The normal electric energy metering data can ensure the accuracy of electric energy metering, and the abnormal electric energy metering data also contains a plurality of important information in the power grid, which influences the dynamics, effectiveness and accuracy of the electric energy metering information to a certain extent.
The reasons for causing the abnormal state of the electric energy metering equipment are many, and the electric energy metering equipment can be classified according to different angles: according to the position where the abnormal information appears, the following conditions are generally adopted: the fault of the electric energy meter, the fault of the metering cabinet and a plurality of factors in an electric energy metering loop; according to the reasons of abnormal information, two conditions, namely human factors and non-human factors, are mainly adopted, the human factors are mainly caused by illegal electricity utilization behaviors such as human error or subjective electricity stealing and civil and commercial electricity utilization, and the non-human factors are caused by self faults of equipment.
Since abnormal electricity consumption behaviors such as electricity stealing and the like bring great harm to a power grid, a large amount of abnormal electricity consumption detection related work is emerging in recent years. The collection of mass power data also opens up a new solution and idea for the field of metering abnormity diagnosis, abnormal power utilization behavior clients and fault equipment are intelligently, efficiently and accurately identified by a machine learning method and are quickly pushed to detection personnel, and the detection personnel perform on-site investigation and implementation, so that the solution gradually becomes a research hotspot problem.
However, in the process of on-site inspection by the inspection personnel, whether the time interval of the occurrence of the abnormal electricity utilization behavior is overlapped with the time interval of the inspection by the inspection personnel is an important factor influencing the success rate of the inspection, because in the on-site inspection, the user does not perform the abnormal electricity utilization behavior during the inspection possibly, the inspection fails; if the key information of the user is mastered, the success probability of the field survey can be greatly improved. However, in the current research work aiming at abnormal electricity utilization detection, the related research on the time interval for detecting the abnormal electricity utilization behavior is insufficient, and the support strength for field inspection is insufficient.
In summary, the current research work related to power metering abnormality diagnosis is insufficient in research on detecting the time interval of the abnormal power consumption behavior, insufficient in support of field inspection, and short in detection method for the abnormal power consumption time interval of the smart grid.
Disclosure of Invention
Aiming at the defects and improvement requirements of the prior art, the invention provides a smart grid-oriented abnormal electricity utilization time interval detection method, which aims to detect the time interval of the abnormal electricity utilization behavior of a user and provide time-related information so as to improve the success probability of field investigation.
In order to achieve the above object, in a first aspect, the present invention provides a method for constructing an abnormal electricity consumption time interval detection model for a smart grid, including the following steps:
s1, acquiring multiple abnormal power consumption time sequence data, wherein each abnormal power consumption time sequence data corresponds to one or more real abnormal power consumption time intervals;
s2, screening out a candidate interval based on expert knowledge, and judging whether the candidate interval is a real abnormal electricity utilization time interval according to whether the overlapping rate of the candidate interval and the real abnormal electricity utilization time interval exceeds a threshold value, so as to label the candidate interval as a real candidate interval label;
s3, building an abnormal electricity utilization time interval detection model comprising a feature extraction network, a classification network and a regression network; inputting abnormal power consumption time sequence data into a feature extraction network to obtain a feature map, and then respectively carrying out region weighting through different candidate intervals to obtain a corresponding weighted feature map; inputting the weighted feature maps into a classification network and a regression network respectively to obtain predicted candidate interval labels and predicted abnormal electricity utilization time intervals;
and S4, calculating loss by using the real candidate interval label and the real abnormal electricity time interval as real labels until the abnormal electricity time interval detection model converges.
Further, in S2, the overlapping rate IoU of the candidate section and the real abnormal electricity consumption time section is calculated as follows:
wherein x is 1 、x 2 As coordinates of start and end points of the candidate interval on the time axis, y 1 、y 2 Coordinates of start and end points on time axis for real abnormal electricity consumption time interval, s inter Indicating the interval in which the two overlap.
Further, in S3, the formula of the area weighting mechanism is as follows:
A(X mn ，bbox)＝alpha·X mn ，alpha＝[B 1 ，O，B 2 ]
B 1 ＝[a 1 ，…，a i ]，a 1 ＝…＝a i ＝α
O＝[b 1 ，…，b j ]，b 1 ＝…＝b j ＝β
B 2 ＝[c 1 ，…，c k ]，c 1 ＝…＝c k ＝α
i＝bbox l ，j＝bbox r -bbox l +1，k＝n-bbox r +1
wherein, X mn Is an input feature map, m represents the number of channels of the feature map, n represents the length of the feature map, alpha represents a weight score vector, bbox is the coordinate of a candidate interval, bbox l As a start point coordinate value, bbox r For the endpoint coordinate values, α and β are hyper-parameters.
Further, the feature extraction network is a one-dimensional residual convolution network.
Further, the classification network is a three-layer full-connection layer network plus a Sigmoid function.
Further, the regression network is a three-layer full-connection layer network.
In a second aspect, the invention provides a smart grid-oriented abnormal electricity consumption time interval detection method, which includes: and inputting the abnormal power consumption time sequence data to be detected into the abnormal power consumption time interval detection model constructed by the construction method of the intelligent power grid-oriented abnormal power consumption time interval detection model in the first aspect, and outputting the abnormal power consumption time interval detection result.
In a third aspect, the present invention provides a machine-readable storage medium storing machine-executable instructions which, when invoked and executed by a processor, cause the processor to implement a method for constructing a smart grid-oriented abnormal electricity consumption time interval detection model according to the first aspect and/or a smart grid-oriented abnormal electricity consumption time interval detection method according to the second aspect.
Generally, by the above technical solution conceived by the present invention, the following beneficial effects can be obtained:
(1) the abnormal electricity consumption time interval detection model provided by the invention is suitable for one-dimensional long-time sequence data, such as electricity consumption time sequence data with longer sequence length; the model is based on knowledge embedding to generate a candidate frame, and is additionally provided with a region weighting mechanism, so that the model is more suitable for small target detection of long time sequence data, and key information can be effectively extracted from the long time sequence data with redundant information and more noise.
(2) Different from a statistical method and other machine learning models for detecting time sequence abnormal values, the method uses a one-dimensional data target detection method, takes the abnormal values as targets and other values as backgrounds, and generates a candidate frame based on knowledge embedding, so that the detection speed is increased, and the detection accuracy is also improved.
Drawings
Fig. 1 is a schematic flow chart of a method for constructing an abnormal electricity consumption time interval detection model for a smart grid according to an embodiment of the present invention;
FIG. 2 is a block diagram of a detection model for abnormal power consumption time intervals according to an embodiment of the present invention;
FIG. 3(a) is a diagram of a feature extraction network architecture provided by an embodiment of the present invention;
FIG. 3(b) is a residual block diagram of FIG. 3 (a);
FIG. 4 is a diagram of a classification network architecture provided by an embodiment of the present invention;
fig. 5 is a diagram of a regression network architecture according to an embodiment of the present invention.
Detailed Description
In order to make the objects, technical solutions and advantages of the present invention more apparent, the present invention is described in further detail below with reference to the accompanying drawings and embodiments. It should be understood that the specific embodiments described herein are merely illustrative of the invention and are not intended to limit the invention. In addition, the technical features involved in the embodiments of the present invention described below may be combined with each other as long as they do not conflict with each other.
Referring to fig. 1 and fig. 2 to 5, the invention provides a method for constructing an abnormal electricity consumption time interval detection model for a smart grid, which includes operations S1 to S4.
In operation S1, a plurality of abnormal power consumption time series data are obtained, where each abnormal power consumption time series data corresponds to one or more real abnormal power consumption time intervals.
In this embodiment, the abnormal power consumption time series data of the user is obtained, the time span of the data is one week, the sampling interval period is not higher than one hour, and the real abnormal power consumption time interval coordinate corresponding to each abnormal power consumption data sample is obtained again.
In operation S2, a candidate interval is screened out based on the expert knowledge, and whether the candidate interval is the real abnormal electricity consumption time interval is determined according to whether the overlapping rate of the candidate interval and the real abnormal electricity consumption time interval exceeds a threshold, so that the candidate interval is labeled as a real candidate interval label.
In this embodiment, the candidate intervals are generated based on expert knowledge, for example, if the candidate intervals are set to two high load periods of each day of the week, the first high load period is 10 o 'clock to 14 o' clock, and the second high load period is 17 o 'clock to 21 o' clock, the sample data of the week corresponds to 14 candidate intervals.
The overlap ratio, IoU, is calculated as follows:
wherein x is 1 、x 2 As coordinates of start and end points of the candidate interval on the time axis, y 1 、y 2 Coordinates of start and end points on time axis for real abnormal electricity consumption time interval, s inter Indicating the interval in which the two overlap.
Preferably, the threshold value is taken to be 0.5.
It should be noted that each abnormal power consumption time sequence data corresponds to one or more real abnormal power consumption time intervals, and when each abnormal power consumption time sequence data corresponds to a plurality of real abnormal power consumption time intervals, the candidate interval is determined to be a real abnormal power consumption time interval only if the overlapping rate of the candidate interval and one of the real abnormal power consumption time intervals exceeds a threshold value; otherwise, judging that the candidate interval is not the real abnormal electricity utilization time interval.
Operation S3, building an abnormal electricity utilization time interval detection model comprising a feature extraction network, a classification network and a regression network; inputting abnormal power consumption time sequence data into a feature extraction network to obtain a feature map, and then respectively performing region weighting through different candidate intervals to obtain a corresponding weighted feature map; and respectively inputting the weighted feature maps into a classification network and a regression network to obtain predicted candidate interval labels and predicted abnormal electricity utilization time intervals.
As shown in fig. 2, in the abnormal electricity consumption time interval detection model provided in the embodiment of the present invention, the feature extraction network is preferably a one-dimensional residual convolution network, the sampling frequency is preferably half an hour, and the feature map is output after the abnormal electricity consumption data sample is input into the feature extraction network. Before the feature map is input into the classification network and the regression network, the feature map is subjected to region weighting according to coordinates of 14 different candidate intervals. Finally, classifying the weighted feature map by a classification network, outputting a predicted classification result, and calculating the loss between a predicted candidate interval label and a real candidate interval label, wherein the used loss function is a binary cross entropy loss function; the regression network regresses the coordinates of the candidate interval, outputs the predicted abnormal electricity utilization time interval, calculates the loss between the predicted abnormal electricity utilization time interval and the real abnormal electricity utilization time interval, uses the loss function of smoothing L1 loss, adds the two together to iterate the model parameters, and the optimizer is preferably an Adam optimizer.
The characteristic extraction network structure is ResNet-25 as shown in fig. 3(a), the input is an abnormal power consumption data sample, the data size is 1x336, the abnormal power consumption data sample is input into two residual blocks after being input into one-dimensional convolution layer of one layer, the residual block structure is shown in fig. 3(b), each residual block is provided with four Bottleneck modules, the stride of the convolution layer of the first Bottleneck module is 2, the strides of the other two layers are 1, and the strides of the three Bottleneck modules are 1. In a total of 25 one-dimensional convolution layers, the channel of the residual block 1 is 64, the channel of the residual block 2 is 128, and finally the characteristic diagram of 256 × 84 is output.
The formula of the region weighting mechanism is as follows:
A(X mn ，bbox)＝alpha·X mn ，alpha＝[B 1 ，O，B 2 ]
B 1 ＝[a 1 ，…，a i ]，a 1 ＝…＝a i ＝α
O＝[b 1 ，…，b j ]，b 1 ＝…＝b j ＝β
B 2 ＝[c 1 ，…，c k ]，c 1 ＝…＝c k ＝α
i＝bbox l ，j＝bbox r -bbox l +1，k＝n-bbox r +1
wherein, X mn Is an input feature map, m represents the number of channels of the feature map, n represents the length of the feature map, alpha represents a weight score vector, bbox is the coordinate of a candidate interval, bbox l As a starting point coordinate value, bbox r For the endpoint coordinate values, α and β are hyper-parameters. Preferably, α is 0.5 and β is 1.
The classification network structure is shown in fig. 4, the number of output channels of three layers of full-connection layer networks is 2048, 512, and 2, and finally a layer of Sigmoid function is added, preferably 0.5 is used as a threshold, and a label of 0 or 1 is output finally, where 0 represents that the candidate interval is detected to be not a real abnormal electricity consumption time interval, and 1 represents that the candidate interval is detected to be a real abnormal electricity consumption time interval.
The regression network structure is shown in fig. 5, the number of output channels of the three-layer full-connection layer network is 2048, 512 and 2, and the output predicted abnormal power utilization time interval comprises an initial coordinate and a termination coordinate.
In operation S4, a loss is calculated using the true candidate interval label and the true abnormal electricity usage time interval as the true labels until the abnormal electricity usage time interval detection model converges.
In this embodiment, the loss is divided into two parts, one part is the classification loss of the classification network, the other part is the regression loss of the regression network, and the two parts are added to form the overall model loss.
The loss calculation of the classification network is the loss between the predicted candidate interval label and the real candidate interval label, the loss function is a common cross entropy loss function, the loss function calculates the cross entropy of the predicted category probability distribution and the real label distribution probability, and the formula is as follows:
where N represents a total of N samples, i represents the ith sample, p ic A probability score representing that the sample belongs to the class c; y is ic Indicating that the sample isIf not, the category is 1, and not 0; then all classes are summed, and then all samples are summed and averaged.
The loss of the regression network is calculated by the loss between the predicted abnormal electricity utilization time interval and the real abnormal electricity utilization time interval, and the loss function uses the smooth L1 loss, and the formula is as follows:
and x is the difference value between the predicted abnormal electricity utilization time interval and the real abnormal electricity utilization time interval.
The specific training process is shown as algorithm 1:
and in the model application stage, inputting the abnormal electricity consumption time sequence data to be detected into the collected abnormal electricity consumption time interval detection model, and outputting the detection result of the abnormal electricity consumption time interval.
It should be noted that, during output, only the candidate interval with the classification network output label of 1 is determined as the real abnormal electricity consumption time interval, and the abnormal electricity consumption time interval output by the corresponding regression network is output.
Application case
Because a large amount of abnormal electric energy metering data of low-voltage users are lacked, abnormal data modeling is carried out on a CER normal low-voltage user electric power data set from ISSDA, and normal user electric power consumption data are changed into abnormal electric power consumption data through a mathematical modeling method, so that a large amount of abnormal electric energy metering data of the low-voltage users which can be used for training are obtained. The abnormal electricity utilization data set comprises 10 ten thousand samples, wherein 8 ten thousand samples are training sets, 2 ten thousand samples are testing sets, the sampling interval period is half an hour, and the time span of each sample is one week.
The criterion for determining whether or not the time interval generated by the abnormal electricity consumption time interval detection model is the real abnormal electricity consumption time interval is whether or not IoU, which is the overlap ratio between the generated time interval and the real abnormal electricity consumption time interval, exceeds a threshold value.
The accuracy of the time interval of model generation on the test set under different IoU score thresholds is shown by experimental results, which are shown in table 1 below.
TABLE 1 accuracy of anomaly time information in end-to-end generative model generated diagnostic information
It can be seen that the accuracy of the generated time information is reduced to some extent with the increase of the IoU score threshold, but the reduction is not much, and the overlapping rate of the time interval generated by the visible model and the real abnormal electricity utilization time interval is higher, thus proving the effectiveness of the invention.
It will be understood by those skilled in the art that the foregoing is only a preferred embodiment of the present invention, and is not intended to limit the invention, and that any modification, equivalent replacement, or improvement made within the spirit and principle of the present invention should be included in the scope of the present invention.
Claims (8)
1. A construction method of an abnormal electricity utilization time interval detection model for a smart grid is characterized by comprising the following steps:
s1, acquiring multiple abnormal power consumption time sequence data, wherein each abnormal power consumption time sequence data corresponds to one or more real abnormal power consumption time intervals;
s2, screening out a candidate interval based on expert knowledge, and judging whether the candidate interval is a real abnormal electricity utilization time interval according to whether the overlapping rate of the candidate interval and the real abnormal electricity utilization time interval exceeds a threshold value, so as to label the candidate interval as a real candidate interval label;
s3, building an abnormal electricity utilization time interval detection model comprising a feature extraction network, a classification network and a regression network; inputting abnormal power consumption time sequence data into a feature extraction network to obtain a feature map, and then respectively performing region weighting through different candidate intervals to obtain a corresponding weighted feature map; inputting the weighted feature maps into a classification network and a regression network respectively to obtain predicted candidate interval labels and predicted abnormal electricity utilization time intervals;
and S4, calculating loss by using the real candidate interval label and the real abnormal electricity time interval as real labels until the abnormal electricity time interval detection model converges.
2. The method for constructing the abnormal electricity consumption time interval detection model for the smart grid according to claim 1, wherein in the step S2, the overlapping rate IoU of the candidate interval and the real abnormal electricity consumption time interval is calculated as follows:
wherein x is 1 、x 2 As coordinates of start and end points of the candidate interval on the time axis, y 1 、y 2 Coordinates of start and end points on time axis for real abnormal electricity consumption time interval, s inter Indicating the interval in which the two overlap.
3. The method for constructing the abnormal electricity consumption time interval detection model for the smart grid according to claim 1, wherein in S3, the formula of the area weighting mechanism is as follows:
A(X mn ,bbox)＝alpha·X mn ,alpha＝[B 1 ,O,B 2 ]
B 1 ＝[a 1 ,…,a i ],a 1 ＝…＝a i ＝α
O＝[b 1 ,…,b j ],b 1 ＝…＝b j ＝β
B 2 ＝[c 1 ,…,c k ],c 1 ＝…＝c k ＝α
i＝bbox l ,j＝bbox r -bbox l +1,k＝n-bbox r +1
wherein, X mn Is an input feature map, m represents the number of channels of the feature map, n represents the length of the feature map, alpha represents a weight score vector, bbox is the coordinate of a candidate interval, bbox l As a starting point coordinate value, bbox r For the endpoint coordinate values, α and β are hyper-parameters.
4. The method for constructing the abnormal electricity consumption time interval detection model for the smart grid as claimed in claim 1, wherein the feature extraction network is a one-dimensional residual convolution network.
5. The method for constructing the abnormal electricity consumption time interval detection model for the smart grid according to claim 1, wherein the classification network is a three-layer full-connection layer network plus a Sigmoid function.
6. The method for constructing the abnormal electricity utilization time interval detection model for the smart grid according to claim 1, wherein the regression network is a three-layer full-connection layer network.
7. The abnormal electricity utilization time interval detection method for the smart grid is characterized by comprising the following steps of: inputting the abnormal power consumption time series data to be detected into the abnormal power consumption time interval detection model constructed by the method for constructing the intelligent power grid-oriented abnormal power consumption time interval detection model according to any one of claims 1 to 6, and outputting the abnormal power consumption time interval detection result.
8. A machine-readable storage medium storing machine-executable instructions which, when invoked and executed by a processor, cause the processor to implement a method of constructing a smart grid-oriented abnormal electricity usage time interval detection model according to any one of claims 1 to 6 and/or a smart grid-oriented abnormal electricity usage time interval detection method according to claim 7.
Priority Applications (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN202210609391.1A CN114970729B (en) | 2022-05-31 | 2022-05-31 | Abnormal electricity utilization time interval detection method for smart power grid |
Applications Claiming Priority (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN202210609391.1A CN114970729B (en) | 2022-05-31 | 2022-05-31 | Abnormal electricity utilization time interval detection method for smart power grid |
Publications (2)
| Publication Number | Publication Date |
|---|---|
| CN114970729A true CN114970729A (en) | 2022-08-30 |
| CN114970729B CN114970729B (en) | 2024-09-06 |
Family
ID=82958159
Family Applications (1)
| Application Number | Title | Priority Date | Filing Date |
|---|---|---|---|
| CN202210609391.1A Active CN114970729B (en) | 2022-05-31 | 2022-05-31 | Abnormal electricity utilization time interval detection method for smart power grid |
Country Status (1)
| Country | Link |
|---|---|
| CN (1) | CN114970729B (en) |
Citations (4)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN108985330A (en) * | 2018-06-13 | 2018-12-11 | 华中科技大学 | A kind of autoencoder network and its training method abnormal use electric detection method and system |
| WO2020196954A1 (en) * | 2019-03-26 | 2020-10-01 | (주)누리텔레콤 | Apparatus for determining whether heater is abnormal by using power consumption and method therefor |
| CN113505935A (en) * | 2021-07-26 | 2021-10-15 | 上海东方低碳科技产业股份有限公司 | Electric power abnormal fluctuation detection and prediction calculation method based on integrated algorithm |
| CN113534276A (en) * | 2021-07-12 | 2021-10-22 | 西安电子科技大学 | Railway intrusion behavior detection method based on fast R-CNN |
-
2022
- 2022-05-31 CN CN202210609391.1A patent/CN114970729B/en active Active
Patent Citations (4)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN108985330A (en) * | 2018-06-13 | 2018-12-11 | 华中科技大学 | A kind of autoencoder network and its training method abnormal use electric detection method and system |
| WO2020196954A1 (en) * | 2019-03-26 | 2020-10-01 | (주)누리텔레콤 | Apparatus for determining whether heater is abnormal by using power consumption and method therefor |
| CN113534276A (en) * | 2021-07-12 | 2021-10-22 | 西安电子科技大学 | Railway intrusion behavior detection method based on fast R-CNN |
| CN113505935A (en) * | 2021-07-26 | 2021-10-15 | 上海东方低碳科技产业股份有限公司 | Electric power abnormal fluctuation detection and prediction calculation method based on integrated algorithm |
Also Published As
| Publication number | Publication date |
|---|---|
| CN114970729B (en) | 2024-09-06 |
Similar Documents
| Publication | Publication Date | Title |
|---|---|---|
| CN111459700B (en) | Equipment fault diagnosis method, diagnosis device, diagnosis equipment and storage medium | |
| CN106780121B (en) | Power consumption abnormity identification method based on power consumption load mode analysis | |
| CN110703057B (en) | Power equipment partial discharge diagnosis method based on data enhancement and neural network | |
| Mohammadi et al. | Machine learning assisted stochastic unit commitment during hurricanes with predictable line outages | |
| CN109800995A (en) | A kind of grid equipment fault recognition method and system | |
| Zheng et al. | Real-time transient stability assessment based on deep recurrent neural network | |
| CN105471647A (en) | Power communication network fault positioning method | |
| Wu et al. | Gridtopo-GAN for distribution system topology identification | |
| Sun et al. | Small sample reliability assessment with online time-series data based on a worm wasserstein generative adversarial network learning method | |
| CN117557127A (en) | Power grid dispatching system supporting platform reliability assessment method, system and storage medium | |
| CN117674128A (en) | Automatic fault removal method based on power dispatching system | |
| CN111586728A (en) | Small sample characteristic-oriented heterogeneous wireless network fault detection and diagnosis method | |
| Bao et al. | Remaining useful life assessment of slewing bearing based on spatial-temporal sequence | |
| CN117791856B (en) | Power grid fault early warning method and device based on inspection robot | |
| CN110956281A (en) | Power equipment abnormity detection alarm system based on Log analysis | |
| CN118193954A (en) | Power distribution network abnormal data detection method and system based on edge calculation | |
| CN114707789A (en) | Low-voltage distribution network high-frequency measurement data processing and topology identification method and system | |
| CN106655181A (en) | Priority setting method and system for power grid nodes | |
| CN115936926A (en) | SMOTE-GBDT-based unbalanced electricity stealing data classification method and device, computer equipment and storage medium | |
| CN116225752A (en) | Fault root cause analysis method and system for micro-service system based on fault mode library | |
| CN114970729A (en) | Abnormal electricity consumption time interval detection method for smart power grid | |
| CN115964951A (en) | Power system operation reliability assessment method and device and electronic equipment | |
| CN115618286A (en) | Transformer partial discharge type identification method, system, equipment, terminal and application | |
| CN114118759A (en) | Distribution transformer area load overload state assessment method and device | |
| Šipoš et al. | A method for determining the location and type of fault in transmission network using neural networks and power quality monitors |
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 |