CN115511016A - Incremental active learning-based electricity charge anomaly detection method and device - Google Patents
Incremental active learning-based electricity charge anomaly detection method and device Download PDFInfo
- Publication number
- CN115511016A CN115511016A CN202211479010.9A CN202211479010A CN115511016A CN 115511016 A CN115511016 A CN 115511016A CN 202211479010 A CN202211479010 A CN 202211479010A CN 115511016 A CN115511016 A CN 115511016A
- Authority
- CN
- China
- Prior art keywords
- abnormal
- user
- electricity
- charge data
- suspected
- 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
- 238000001514 detection method Methods 0.000 title claims abstract description 181
- 230000005611 electricity Effects 0.000 title claims abstract description 141
- 230000002159 abnormal effect Effects 0.000 claims abstract description 230
- 230000001960 triggered effect Effects 0.000 claims abstract description 7
- 238000000034 method Methods 0.000 claims description 27
- 239000013598 vector Substances 0.000 claims description 25
- 230000005856 abnormality Effects 0.000 claims description 23
- 238000004364 calculation method Methods 0.000 claims description 15
- 238000012549 training Methods 0.000 claims description 13
- 230000008859 change Effects 0.000 claims description 12
- 238000005259 measurement Methods 0.000 claims description 8
- 238000012986 modification Methods 0.000 claims description 7
- 230000004048 modification Effects 0.000 claims description 7
- 238000011084 recovery Methods 0.000 claims description 6
- 230000009467 reduction Effects 0.000 claims description 6
- 239000000725 suspension Substances 0.000 claims description 6
- 238000012795 verification Methods 0.000 claims description 5
- 238000012545 processing Methods 0.000 claims description 4
- 230000005540 biological transmission Effects 0.000 claims description 3
- 230000002547 anomalous effect Effects 0.000 abstract description 2
- 230000010365 information processing Effects 0.000 abstract description 2
- 230000008569 process Effects 0.000 description 12
- 238000000354 decomposition reaction Methods 0.000 description 4
- 230000007547 defect Effects 0.000 description 3
- 238000012423 maintenance Methods 0.000 description 3
- 239000011159 matrix material Substances 0.000 description 3
- 238000010276 construction Methods 0.000 description 2
- 238000005457 optimization Methods 0.000 description 2
- 230000005612 types of electricity Effects 0.000 description 2
- 238000009412 basement excavation Methods 0.000 description 1
- 238000012937 correction Methods 0.000 description 1
- 230000003247 decreasing effect Effects 0.000 description 1
- 230000001419 dependent effect Effects 0.000 description 1
- 238000013461 design Methods 0.000 description 1
- 238000005516 engineering process Methods 0.000 description 1
- 230000010354 integration Effects 0.000 description 1
- 230000007246 mechanism Effects 0.000 description 1
- 238000003062 neural network model Methods 0.000 description 1
- 230000001932 seasonal effect Effects 0.000 description 1
- 239000000126 substance Substances 0.000 description 1
Images
Classifications
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06F—ELECTRIC DIGITAL DATA PROCESSING
- G06F9/00—Arrangements for program control, e.g. control units
- G06F9/06—Arrangements for program control, e.g. control units using stored programs, i.e. using an internal store of processing equipment to receive or retain programs
- G06F9/44—Arrangements for executing specific programs
- G06F9/448—Execution paradigms, e.g. implementations of programming paradigms
- G06F9/4494—Execution paradigms, e.g. implementations of programming paradigms data driven
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06N—COMPUTING ARRANGEMENTS BASED ON SPECIFIC COMPUTATIONAL MODELS
- G06N20/00—Machine learning
- G06N20/10—Machine learning using kernel methods, e.g. support vector machines [SVM]
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06N—COMPUTING ARRANGEMENTS BASED ON SPECIFIC COMPUTATIONAL MODELS
- G06N3/00—Computing arrangements based on biological models
- G06N3/02—Neural networks
- G06N3/08—Learning methods
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06Q—INFORMATION AND COMMUNICATION TECHNOLOGY [ICT] SPECIALLY ADAPTED FOR ADMINISTRATIVE, COMMERCIAL, FINANCIAL, MANAGERIAL OR SUPERVISORY PURPOSES; SYSTEMS OR METHODS SPECIALLY ADAPTED FOR ADMINISTRATIVE, COMMERCIAL, FINANCIAL, MANAGERIAL OR SUPERVISORY PURPOSES, NOT OTHERWISE PROVIDED FOR
- G06Q50/00—Information and communication technology [ICT] specially adapted for implementation of business processes of specific business sectors, e.g. utilities or tourism
- G06Q50/06—Energy or water supply
-
- Y—GENERAL TAGGING OF NEW TECHNOLOGICAL DEVELOPMENTS; GENERAL TAGGING OF CROSS-SECTIONAL TECHNOLOGIES SPANNING OVER SEVERAL SECTIONS OF THE IPC; TECHNICAL SUBJECTS COVERED BY FORMER USPC CROSS-REFERENCE ART COLLECTIONS [XRACs] AND DIGESTS
- Y02—TECHNOLOGIES OR APPLICATIONS FOR MITIGATION OR ADAPTATION AGAINST CLIMATE CHANGE
- Y02D—CLIMATE CHANGE MITIGATION TECHNOLOGIES IN INFORMATION AND COMMUNICATION TECHNOLOGIES [ICT], I.E. INFORMATION AND COMMUNICATION TECHNOLOGIES AIMING AT THE REDUCTION OF THEIR OWN ENERGY USE
- Y02D10/00—Energy efficient computing, e.g. low power processors, power management or thermal management
Landscapes
- Engineering & Computer Science (AREA)
- Theoretical Computer Science (AREA)
- Software Systems (AREA)
- Physics & Mathematics (AREA)
- General Physics & Mathematics (AREA)
- Business, Economics & Management (AREA)
- General Engineering & Computer Science (AREA)
- Health & Medical Sciences (AREA)
- Economics (AREA)
- Computing Systems (AREA)
- Artificial Intelligence (AREA)
- Mathematical Physics (AREA)
- Data Mining & Analysis (AREA)
- Evolutionary Computation (AREA)
- General Health & Medical Sciences (AREA)
- Medical Informatics (AREA)
- Public Health (AREA)
- Computational Linguistics (AREA)
- Computer Vision & Pattern Recognition (AREA)
- Biomedical Technology (AREA)
- Biophysics (AREA)
- Life Sciences & Earth Sciences (AREA)
- Molecular Biology (AREA)
- Water Supply & Treatment (AREA)
- Human Resources & Organizations (AREA)
- Marketing (AREA)
- Primary Health Care (AREA)
- Strategic Management (AREA)
- Tourism & Hospitality (AREA)
- General Business, Economics & Management (AREA)
- Management, Administration, Business Operations System, And Electronic Commerce (AREA)
Abstract
The invention relates to the technical field of information processing and electric power marketing, and particularly discloses an electric charge abnormity detection method based on incremental active learning, which comprises the following steps: performing primary anomaly detection on the acquired current electric charge data of the plurality of target users, and outputting suspected anomalous users in the plurality of target users if an anomaly rule is triggered; performing secondary abnormal detection on the electricity charge data of the suspected abnormal user to obtain a suspected abnormal user detection result, and if the uncertainty of the suspected abnormal user detection result is lower than a preset threshold value, directly outputting the suspected abnormal user detection result; and if the uncertainty is higher than the preset threshold, the electricity charge data of the suspected abnormal users with high uncertainty is finally researched and judged, and normal users in the suspected abnormal users with high uncertainty are output. The invention also discloses an electricity charge abnormity detection device based on the incremental active learning. The invention can solve the problems that the current accounting rule system has low hit rate and can not independently complete the iterative updating of the model by applying the service data.
Description
Technical Field
The invention relates to the technical field of information processing and electric power marketing, in particular to an electric charge abnormity detection method based on incremental active learning and an electric charge abnormity detection device based on incremental active learning.
Background
The existing electric charge abnormity detection method mainly comprises two types: one is an anomaly detection method based on simple rules, and the other is an anomaly detection algorithm based on data driving.
The anomaly detection algorithm based on the simple rules mainly depends on service experts to summarize common problems in the service to form formal language description, and the rules are realized through logic operation in the program language; the common anomaly detection algorithms include KNN, OCSVM and the like based on sample distance measurement, HBOS, MCD and the like based on sample statistics, algorithms based on an integration method such as IForest, and neural network models such as AutoEncoder, VAE and the like.
Although the existing anomaly detection algorithm is deeply applied to the business link of power marketing, the existing methods have certain defects. The method is not dependent on complex model design, realizes the excavation of abnormal problems through simple logic judgment, for example, users with suddenly increased or decreased electric quantity are screened out through designing a threshold value strategy, and has the characteristics of high efficiency judgment and simple realization, but the flexibility and the expandability of the method are not high, for example, when the calculation cost data of the users are influenced by seasonal or regional factors, the rules cannot be adjusted adaptively according to the monthly data, so that a large number of problems of false report or missing report occur; and the rule-based detection method is often directly embedded in the fee calculation code, so that the updating and maintenance cost is high, and the risk cost of iterative optimization is high. Although the data-driven anomaly detection algorithm can realize the self-adaptive adjustment of the model based on data, the computational complexity of the computational cost model is high, the requirement on the computational cost time efficiency cannot be met, and the method is more applied to problem verification and attribution afterwards. Moreover, the algorithm needs to face the influence of the data quality problem, and when the data quality used for training is not high, the detection performance of the model often cannot meet the requirements of practical application. In addition, both the calculation rule discovered based on the process problem and the algorithm model based on the after-the-fact check depend on a large amount of human cost for construction and maintenance, for example, iterative optimization of the calculation rule requires a service person to analyze and attribute monthly abnormal data, and summarize the optimized content of the refinement rule, while for the data-driven algorithm model, the acquisition and processing of data also occupy a large amount of human cost, and these conditions limit the performance of the existing detection model and are difficult to meet increasingly complex and variable service application scenarios.
Disclosure of Invention
Aiming at the defects and shortcomings in the prior art, the invention provides an incremental active learning-based electricity charge abnormity detection method, which is used for solving the problems of low flexibility, high construction and maintenance cost and limited data quality of the existing electricity charge abnormity detection algorithm in the prior art.
As a first aspect of the present invention, there is provided an electricity rate abnormality detection method based on incremental active learning, including:
step S1: acquiring current electric charge data of a plurality of target users, wherein the current electric charge data of each target user comprises archive data, service change data and price data;
step S2: performing primary anomaly detection on the current electricity charge data of the target user based on an accounting rule system, and if the current electricity charge data of the target user does not trigger an anomaly rule, directly outputting a detection result that the target user has no anomaly; if the current electricity charge data of the target users trigger the abnormal rule, outputting suspected abnormal users in the plurality of target users;
and step S3: performing secondary abnormal detection on the electricity charge data of the suspected abnormal user based on an SVDD abnormal detection model to obtain a suspected abnormal user detection result, and judging the uncertainty of the suspected abnormal user detection result to determine whether to output the suspected abnormal user detection result;
and step S4: if the uncertainty of the detection result of the suspected abnormal user is lower than a preset threshold value, the detection result of the suspected abnormal user is directly output; and if the uncertainty of the detection result of the suspected abnormal user is higher than the preset threshold, performing final abnormal study and judgment on the electricity charge data of the suspected abnormal user with high uncertainty, and outputting a normal user of the suspected abnormal users with high uncertainty.
Further, the current electricity charge data of the target user is subjected to primary anomaly detection based on an accounting rule system, and if the current electricity charge data of the target user does not trigger an anomaly rule, a detection result that the target user is not anomalous is directly output; if the current electricity charge data of the target user triggers the abnormal rule, outputting suspected abnormal users in the plurality of target users, and further comprising:
judging whether each abnormal rule is triggered by the current electricity charge data of the target user according to the definition of each abnormal rule in the accounting rule system;
and calculating the condition of triggering an abnormal rule for each current electric charge data of the target user, counting the target user triggering at least one abnormal rule into suspected abnormal users, and waiting for secondary abnormal detection.
Further, still include:
selecting a plurality of abnormal rules in the accounting rule system to carry out verification;
if it is not
If yes, the current electricity charge data x of the target user triggers a kth abnormal rule; if it is not
It means that the current electricity rate data x of the target subscriber is not triggered
A bar exception rule; the preliminary detection result of the current electricity rate data x of each of the target users is expressed as
Wherein p represents the number of abnormal rules in the accounting rule system.
Further, the performing, by the SVDD anomaly detection model, a secondary anomaly detection on the electricity charge data of the suspected abnormal user to obtain a suspected abnormal user detection result, and determining an uncertainty of the suspected abnormal user detection result to determine whether to output the suspected abnormal user detection result, further includes:
according to an abnormal rule triggered by the suspected abnormal user, adopting a corresponding SVDD abnormal detection model, calculating distance measurement between the electricity charge data of the suspected abnormal user and a support vector description model central point, and carrying out secondary judgment on the suspected abnormal user according to the measurement value to obtain a suspected abnormal user detection result;
and according to the abnormal distribution characteristics of the electricity charge data of the suspected abnormal users, determining the uncertainty of the detection result of the suspected abnormal users, and manually re-studying and judging the electricity charge data of the suspected abnormal users with high uncertainty.
Further, still include:
preliminary detection result of current electricity fee data x for each of the target users
If, if
If yes, indicating that suspected abnormal users exist in the target users, and then carrying out secondary abnormal detection on the electricity charge data of the suspected abnormal users; according to the abnormal category corresponding to each abnormal rule, an SVDD abnormal detection model is realized
Is shown as
The SVDD anomaly detection models corresponding to the anomaly rule have p SVDD anomaly detection models in total;
assuming that all normal electricity charge data are surrounded by a minimum boundary in a high-dimensional space, the electricity charge data positioned on the minimum boundary is called a support vector, and whether the electricity charge data of the suspected abnormal user is positioned in the minimum boundary is detected through an SVDD (singular value decomposition/direct decomposition) abnormality detection model to judge the abnormal user;
and calculating the uncertainty of the SVDD model detection result based on an information entropy formula.
Further, still include:
and performing incremental training on the SVDD anomaly detection model based on the electricity charge data of the normal user.
Further, the archive data comprises the electricity utilization type, the marketization attribute, the voltage grade, the metering mode, the operation capacity, the contract capacity, the pricing strategy type, the power factor assessment mode, the basic electricity charge calculation mode, the power quantity, the electric quantity calculation mode, the participation power factor calculation mode, the temporary electricity utilization mark, the industry type, the energy utilization type and the time-sharing electricity utilization mark of the user.
Further, the service change data comprises new capacity increase, suspension, capacity reduction, classification, metering equipment fault handling, voltage change, metering equipment replacement, suspension recovery, capacity reduction recovery and power receiving facility modification.
Further, the data of the measuring price fee comprises the type of electricity price, the electricity price of transmission and distribution, the electricity price of electricity degree, the electricity price of charging, the reading number of the last time, the reading number of the current time, the active electricity quantity, the demand reading number, the reactive electricity quantity, the basic electricity fee, the electricity price and the electricity adjusting fee.
As a second aspect of the present invention, there is provided an electricity rate abnormality detection device based on incremental active learning, including:
the system comprises an acquisition module, a processing module and a display module, wherein the acquisition module is used for acquiring current electric charge data of a plurality of target users, and the current electric charge data of each target user comprises archive data, service change data and pricing charge data;
the first anomaly detection module is used for carrying out primary anomaly detection on the current electric charge data of the target user based on an accounting rule system, and directly outputting a detection result that the target user is not abnormal if the current electric charge data of the target user does not trigger an anomaly rule; if the current electricity charge data of the target users trigger the abnormal rule, outputting suspected abnormal users in the plurality of target users;
the second abnormal detection module is used for carrying out secondary abnormal detection on the electricity charge data of the suspected abnormal user based on the SVDD abnormal detection model to obtain a suspected abnormal user detection result, and judging the uncertainty of the suspected abnormal user detection result to determine whether to output the suspected abnormal user detection result;
a third abnormal detection module, configured to directly output the detection result of the suspected abnormal user if the uncertainty of the detection result of the suspected abnormal user is lower than a preset threshold; and if the uncertainty of the detection result of the suspected abnormal user is higher than the preset threshold, performing final abnormal study and judgment on the electricity charge data of the suspected abnormal user with high uncertainty, and outputting a normal user of the suspected abnormal users with high uncertainty.
The incremental active learning-based electricity charge abnormity detection method provided by the invention has the following advantages:
(1) The anomaly detection model is combined with the existing accounting system, the existing accounting rule anomaly study and judgment mode is reserved, and the performance bottleneck that the existing accounting rule is difficult to accurately study and judge the abnormal electricity charge is broken through by introducing the data-driven anomaly detection model;
(2) The existing abnormal studying and judging mode is expanded, and a new method is provided to combine the two modes, which is different from the study and judgment of a simple dependence rule or a model, so that the model can be automatically updated iteratively through data, the flexibility of the model is expanded, the support of business expert knowledge can be obtained under the active learning technology of the model, and the stability of the model is expanded;
(3) By the incremental learning strategy, the defect that the detection model needs to be retrained in the iterative updating process is avoided, incremental data only need to be acquired from monthly business work in each model updating process, the data selection process is automatically completed by the model, and the workload of manually selecting the data is avoided.
Drawings
The accompanying drawings, which are included to provide a further understanding of the invention and are incorporated in and constitute a part of this specification, illustrate embodiments of the invention and together with the description serve to explain the principles of the invention and not to limit the invention.
Fig. 1 is a flowchart of an electricity rate abnormality detection method based on incremental active learning according to the present invention.
Fig. 2 is a flowchart of an embodiment of a power rate abnormality detection method based on incremental active learning according to the present invention.
Detailed Description
It should be noted that the embodiments and features of the embodiments may be combined with each other without conflict. The present invention will be described in detail below with reference to the embodiments with reference to the attached drawings.
In order to make those skilled in the art better understand the technical solution of the present invention, the technical solution in the embodiments of the present invention will be clearly and completely described below with reference to the drawings in the embodiments of the present invention, and it is obvious that the described embodiments are only a part of the embodiments of the present invention, and not all of the embodiments. All other embodiments, which can be derived by a person skilled in the art from the embodiments given herein without making any creative effort, shall fall within the protection scope of the present invention.
It should be noted that the terms "first," "second," and the like in the description and claims of the present invention and in the drawings described above are used for distinguishing between similar elements and not necessarily for describing a particular sequential or chronological order. It is to be understood that the data so used may be interchanged under appropriate circumstances in order to facilitate the description of the embodiments of the invention herein. Furthermore, the terms "comprises," "comprising," and "having," and any variations thereof, are intended to cover a non-exclusive inclusion, such that a process, method, system, article, or apparatus that comprises a list of steps or elements is not necessarily limited to those steps or elements expressly listed, but may include other steps or elements not expressly listed or inherent to such process, method, article, or apparatus.
In the present embodiment, a method for detecting an abnormal electricity rate based on incremental active learning is provided, and fig. 1 is a flowchart of the method for detecting an abnormal electricity rate based on incremental active learning provided by the present invention. As shown in fig. 1, the incremental active learning-based power rate abnormality detection method includes:
step S1: acquiring current electric charge data of a plurality of target users, wherein the current electric charge data of each target user comprises archive data, service change data and volume price data;
step S2: performing primary anomaly detection on the current electricity charge data of the target user based on an accounting rule system, and if the current electricity charge data of the target user does not trigger an anomaly rule, directly outputting a detection result that the target user has no anomaly; if the current electricity charge data of the target users trigger the abnormal rule, outputting suspected abnormal users in the plurality of target users;
it should be noted that not triggering an exception rule refers to not triggering any exception rule in the accounting rule system, and triggering the exception rule refers to triggering any one or more exception rules in the accounting rule system.
And step S3: performing secondary abnormal detection on the electricity charge data of the suspected abnormal user based on an SVDD abnormal detection model to obtain a suspected abnormal user detection result, and judging the uncertainty of the suspected abnormal user detection result to determine whether to output the suspected abnormal user detection result;
and step S4: if the uncertainty of the detection result of the suspected abnormal user is lower than a preset threshold value, directly outputting the detection result of the suspected abnormal user; and if the uncertainty of the detection result of the suspected abnormal user is higher than the preset threshold, performing final abnormal study and judgment on the electricity charge data of the suspected abnormal user with high uncertainty, and outputting a normal user of the suspected abnormal users with high uncertainty.
It should be noted that, based on the active learning strategy, manual study and judgment (a set of study and judgment mechanism operated according to expert experience) is further performed on the abnormal user with high uncertainty, so as to obtain the final abnormal user.
The following describes in detail a specific implementation process of the incremental active learning-based electricity rate anomaly detection method provided by the present invention with reference to fig. 2.
Preferably, the primary anomaly detection is performed on the current electricity charge data of the target user based on an accounting rule system, and if the current electricity charge data of the target user does not trigger an anomaly rule, a detection result that the target user is not abnormal is directly output; if the current electricity charge data of the target user triggers the abnormal rule, outputting suspected abnormal users in the plurality of target users, and further comprising:
judging whether each abnormal rule is triggered by the current electricity charge data of the target user according to the definition of each abnormal rule in the accounting rule system;
and calculating the condition of triggering an abnormal rule for each current electric charge data of the target user, counting the target user triggering at least one abnormal rule into suspected abnormal users, and waiting for secondary abnormal detection.
Preferably, the method further comprises the following steps:
selecting a plurality of abnormal rules in the accounting rule system to carry out verification;
it should be noted that 12 abnormal rules with the top-ranked monthly trigger times in the existing accounting rule system are selected to perform algorithm verification.
If it is not
If yes, the current electricity charge data x of the target user triggers a kth abnormal rule; if it is used
Indicating that the current electricity charge data x of the target user does not trigger the kth abnormal rule; the preliminary detection result of the current electricity rate data x of each of the target users is expressed as
Wherein p represents the number of abnormal rules in the accounting rule system.
Preferably, the performing secondary abnormal detection on the electricity fee data of the suspected abnormal user based on the SVDD abnormal detection model to obtain a suspected abnormal user detection result, and determining an uncertainty of the suspected abnormal user detection result to determine whether to output the suspected abnormal user detection result further includes:
according to an abnormal rule triggered by the suspected abnormal user, calculating distance measurement between the electricity charge data of the suspected abnormal user and a support vector description model center point by adopting a corresponding SVDD abnormal detection model, and carrying out secondary judgment on the suspected abnormal user according to a measurement value to obtain a suspected abnormal user detection result;
meanwhile, in order to ensure that the abnormal user is not misjudged, even if the abnormal user is already researched and judged as abnormal by the model, the uncertainty of the detection result of the suspected abnormal user is still determined according to the abnormal distribution characteristics of the electricity charge data of the suspected abnormal user, and the electricity charge data of the suspected abnormal user with high uncertainty needs to be researched and judged manually.
Because the model studying and judging process is accompanied with actual business work, the manual re-studying and judging mode of the abnormal sample is consistent with the current business personnel re-studying and judging mode based on the accounting rule, the method has the advantages that the introduced uncertainty measurement further reduces the abnormal quantity needing manual studying and judging, and simultaneously, along with the complete expansion of the model, the unknown abnormal distributed sample is memorized by the model in a mode of being added into a support vector set, so the workload of manual studying and judging is continuously reduced.
Preferably, the method further comprises the following steps:
preliminary detection result of current electricity fee data x for each of the target users
If, if
If yes, indicating that suspected abnormal users exist in the multiple target users, and then carrying out secondary abnormal detection on the electricity charge data of the suspected abnormal users; according to the abnormal category corresponding to each abnormal rule, an SVDD abnormal detection model is realized by using F k The SVDD abnormality detection models corresponding to the k-th abnormality rule are represented, p SVDD abnormality detection models are shared,
the SVDD anomaly detection model comprises a set of support vectors
And model parameters
；
Assuming that all normal electricity charge data are surrounded by a minimum boundary in a high-dimensional space, the electricity charge data positioned on the minimum boundary is called a support vector, and whether the electricity charge data of the suspected abnormal user is positioned in the minimum boundary is detected through an SVDD (singular value decomposition/direct decomposition) abnormality detection model to judge the abnormal user; the above process is formally described as
Wherein z refers to electricity charge data samples for training if
If so, the electricity charge data of the suspected abnormal user
Is an anomaly in which
Electric charge data representing the suspected abnormal user in a high-dimensional space
The minimum boundary encloses the square sum distance measure between the centers,
radius metric, R, representing the space encompassed by the minimum boundary 2 Can be obtained by solving according to the support vector;
the abnormal user detection results correspondingly output by the p detection models can obtain a vector set
In which
Then calculating the detection result based on the information entropy formula
Uncertainty of
In which
；
If the uncertainty H (q) is greater than the set empirical threshold
Then, the final judgment is performed through the manual judgment link, and the judgment result is recorded as
Wherein
. If it is used
Then the electricity charge data of the final abnormal user does not have the abnormality under the k-th rule, and the electricity charge data of the normal user is obtained
Adding to set X k In (1). After completing a batch of new user detection each time, p sample sets are obtained
}。X k Refers to the electricity rate data of all normal users that do not trigger the k-th class exception rule.
Preferably, the method further comprises the following steps:
and performing incremental training on the SVDD anomaly detection model based on the electricity charge data of the normal user.
In particular, according to set X k Performing incremental training on p SVDD abnormal detection models, adopting a FIVDD algorithm in a training mode, and assuming a feature vector set of an original model for a single SVDD modelIs synthesized into S t k And the formed similarity matrix is marked as A t Where t is the reference of the training round, the parameter vector is recorded as
For each from the sample set X k The training process of the newly added sample z and SVDD anomaly detection model is as follows:
(1) According to
Judging whether the current sample z is judged as an abnormal sample under the current detection model parameters, if so, judging whether the current sample z is judged as the abnormal sample under the current detection model parameters
If the sample z is an abnormal sample, go to step (2), if it is
Directly continuing to investigate the next sample;
(2) Adding the current abnormal sample z into the support vector set, and updating the similarity matrix:
updating model parameters simultaneously
The updated formula is
Wherein
Representing a full 1 vector. In order to simplify the calculation process,
can pass through
And
and (3) realizing incremental calculation:
wherein
To do so
，
And
are intermediate variables used to simplify the model representation.
(3) If it is used
If the current parameter does not meet the assumption of the minimum enclosure of the model, at least one support vector is in the current enclosure, the model needs to be adjusted, the support vector set which does not meet the requirements is removed, and the model parameter in the step (2) is recalculated
。
The detailed training process is as follows:
inputting: SVDD model parameter α t k， Support vector set S t k Incremental training sample set T k ；
And (3) outputting: updated model parameter α t+1 k And support vector set S t+1 k ；
Step 1, sequentially acquiring a single sample z in a training sample set according to a sample storage sequence;
step 2. Vector set S is supported t k Calculating whether the training sample z is positioned in a sample space formed by a support vector set;
and Step 3, generating an extended sample set according to the judgment result in the Step 2: if the sample z falls into the sample space, discarding the sample, otherwise, combining the sample and the original support vector set to form an extended sample set;
step 4, calculating model parameter alpha according to the similarity matrix formed by the extended sample set t+1 k Completing the correction of the extended sample set by the numerical value constraint of the model parameters, and dividing the extended sample set into a new support vector set and a reserved set;
and Step 5, carrying out secondary data check on the samples in the reserved set, adding the samples meeting the conditions into the support vector set again to form a final support vector set S t+1 k And updating the corresponding model parameter alpha according to the latest support vector set t+1 k 。
Preferably, the profile data includes a power utilization type, a marketable attribute, a voltage level, a metering mode, an operation capacity, a contract capacity, a pricing strategy type, a power factor assessment mode, a basic power charge calculation mode, a power quantity, an electric quantity calculation mode, a participation power factor calculation mode, a temporary power utilization sign, an industry category, an energy utilization category, a time-sharing power utilization sign and the like of the user.
Preferably, the service change data includes new capacity increase, suspension, capacity reduction, modification, metering equipment failure handling, voltage change, metering equipment replacement, suspension recovery, capacity reduction recovery, power receiving facility modification and the like.
Preferably, the data of the electricity rate fee includes the type of electricity rate, the electricity transmission and distribution rate, the electricity rate added, the reading number of the last time, the reading number of the current time, the active electricity amount (total), the active electricity amount (peak), the active electricity amount (flat), the active electricity amount (valley), the demand reading number, the reactive electricity amount (total), the basic electricity fee, the electricity rate fee, the power adjusting fee and the like.
As another embodiment of the present invention, there is provided an electricity rate abnormality detection device based on incremental active learning, including:
the system comprises an acquisition module, a processing module and a display module, wherein the acquisition module is used for acquiring current electric charge data of a plurality of target users, and the current electric charge data of each target user comprises archive data, service change data and pricing charge data;
the first anomaly detection module is used for carrying out primary anomaly detection on the current electric charge data of the target user based on an accounting rule system, and directly outputting a detection result that the target user is not abnormal if the current electric charge data of the target user does not trigger an anomaly rule; if the current electricity charge data of the target users trigger the abnormal rule, outputting suspected abnormal users in the plurality of target users;
the second abnormal detection module is used for carrying out secondary abnormal detection on the electricity charge data of the suspected abnormal user based on the SVDD abnormal detection model to obtain a suspected abnormal user detection result, and judging the uncertainty of the suspected abnormal user detection result to determine whether to output the suspected abnormal user detection result;
a third abnormal detection module, configured to directly output the detection result of the suspected abnormal user if the uncertainty of the detection result of the suspected abnormal user is lower than a preset threshold; and if the uncertainty of the detection result of the suspected abnormal user is higher than the preset threshold, performing final abnormal study and judgment on the electricity charge data of the suspected abnormal user with high uncertainty, and outputting a normal user of the suspected abnormal users with high uncertainty.
In summary, the incremental active learning-based electricity charge anomaly detection method provided by the invention combines the existing accounting rule system and the mainstream anomaly detection model SVDD, and solves the problems that the current accounting rule system has low hit rate and cannot independently complete model iterative updating by applying service data.
It will be understood that the above embodiments are merely exemplary embodiments adopted to illustrate the principles of the present invention, and the present invention is not limited thereto. It will be apparent to those skilled in the art that various modifications and improvements can be made without departing from the spirit and substance of the invention, and these modifications and improvements are also considered to be within the scope of the invention.
Claims (10)
1. The method for detecting the abnormal electricity charge based on the incremental active learning is characterized by comprising the following steps of:
step S1: acquiring current electric charge data of a plurality of target users, wherein the current electric charge data of each target user comprises archive data, service change data and volume price data;
step S2: performing primary anomaly detection on the current electric charge data of the target user based on an accounting rule system, and if the current electric charge data of the target user does not trigger an anomaly rule, directly outputting a detection result that the target user has no anomaly; if the current electricity charge data of the target users trigger the abnormal rule, outputting suspected abnormal users in the plurality of target users;
and step S3: performing secondary abnormal detection on the electricity charge data of the suspected abnormal user based on a SVDD abnormal detection model to obtain a suspected abnormal user detection result, and judging the uncertainty of the suspected abnormal user detection result to determine whether to output the suspected abnormal user detection result;
and step S4: if the uncertainty of the detection result of the suspected abnormal user is lower than a preset threshold value, directly outputting the detection result of the suspected abnormal user; and if the uncertainty of the detection result of the suspected abnormal user is higher than the preset threshold, performing final abnormal study and judgment on the electricity charge data of the suspected abnormal user with high uncertainty, and outputting a normal user of the suspected abnormal users with high uncertainty.
2. The incremental active learning-based electricity charge abnormity detection method according to claim 1, wherein the accounting-rule-based system performs initial abnormity detection on the current electricity charge data of the target user, and if the current electricity charge data of the target user does not trigger an abnormity rule, directly outputs a detection result that the target user has no abnormity; if the current electricity charge data of the target user triggers the abnormal rule, outputting suspected abnormal users in the plurality of target users, and further comprising:
judging whether each abnormal rule is triggered by the current electricity charge data of the target user according to the definition of each abnormal rule in the accounting rule system;
and calculating the condition of triggering an abnormal rule for each current electric charge data of the target user, counting the target user triggering at least one abnormal rule into suspected abnormal users, and waiting for secondary abnormal detection.
3. The incremental active learning-based electricity rate abnormality detection method according to claim 2, further comprising:
selecting a plurality of abnormal rules in the accounting rule system to carry out verification;
if it is not
If yes, the current electricity charge data x of the target user triggers a kth abnormal rule; if it is not
Indicating that the current electricity charge data x of the target user does not trigger the kth abnormal rule; the preliminary detection result of the current electricity rate data x of each of the target users is expressed as
Wherein p represents the number of abnormal rules in the accounting rule system.
4. The incremental active learning-based electricity fee abnormality detection method according to claim 3, wherein the SVDD abnormality detection model performs secondary abnormality detection on the electricity fee data of the suspected abnormal user to obtain a suspected abnormal user detection result, and determines an uncertainty of the suspected abnormal user detection result to determine whether to output the suspected abnormal user detection result, further comprising:
according to an abnormal rule triggered by the suspected abnormal user, adopting a corresponding SVDD abnormal detection model, calculating distance measurement between the electricity charge data of the suspected abnormal user and a support vector description model central point, and carrying out secondary judgment on the suspected abnormal user according to the measurement value to obtain a suspected abnormal user detection result;
and according to the abnormal distribution characteristics of the electricity charge data of the suspected abnormal users, determining the uncertainty of the detection result of the suspected abnormal users, and manually re-studying and judging the electricity charge data of the suspected abnormal users with high uncertainty.
5. The incremental active learning-based electricity rate abnormality detection method according to claim 4, further comprising:
for the preliminary detection result r of the current electric charge data x of each target user, if
If yes, indicating that suspected abnormal users exist in the target users, and then carrying out secondary abnormal detection on the electricity charge data of the suspected abnormal users; according to the abnormal category corresponding to each abnormal rule, an SVDD abnormal detection model is realized, and F is used k Representing SVDD anomaly detection models corresponding to the k-th anomaly rule, wherein p SVDD anomaly detection models are provided in total;
assuming that all normal electric charge data are surrounded by a minimum boundary in a high-dimensional space, the electric charge data positioned on the minimum boundary are called support vectors, and whether the electric charge data of the suspected abnormal user are positioned in the minimum boundary is detected through an SVDD abnormality detection model to judge the abnormal user;
and calculating the uncertainty of the electricity charge data of the suspected abnormal user based on an information entropy formula.
6. The incremental active learning-based electricity rate abnormality detection method according to claim 1, further comprising:
and performing incremental training on the SVDD anomaly detection model based on the electricity charge data of the normal user.
7. The incremental active learning-based electricity charge abnormality detection method according to claim 1, wherein the profile data includes electricity usage type, marketable property, voltage class, metering mode, operation capacity, contract capacity, pricing strategy type, power factor assessment mode, basic electricity charge calculation mode, power supply quantity, electricity quantity calculation mode, participation power factor calculation mode, temporary electricity usage flag, industry category, energy usage category and time-sharing electricity usage flag of a user.
8. The incremental active learning-based electricity charge abnormality detection method according to claim 1, wherein the service change data includes new capacity increase, suspension, capacity reduction, modification, metering device failure handling, voltage change, metering device replacement, suspension recovery, capacity reduction recovery, and powered utility modification.
9. The incremental active learning-based electricity fee abnormality detection method according to claim 1, wherein the electricity fee data includes electricity fee type, transmission and distribution electricity fee, electricity degree electricity fee, additional collection electricity fee, last meter reading number, present meter reading number, active electricity quantity, demand quantity reading number, reactive electricity quantity, basic electricity fee, electricity degree electricity fee and power regulation electricity fee.
10. An electric charge abnormality detection method device based on incremental active learning is characterized by comprising the following steps:
the system comprises an acquisition module, a processing module and a display module, wherein the acquisition module is used for acquiring current electric charge data of a plurality of target users, and the current electric charge data of each target user comprises archive data, service change data and pricing charge data;
the first anomaly detection module is used for carrying out primary anomaly detection on the current electric charge data of the target user based on an accounting rule system, and directly outputting a detection result that the target user is not abnormal if the current electric charge data of the target user does not trigger an anomaly rule; if the current electricity charge data of the target users trigger the abnormal rule, outputting suspected abnormal users in the plurality of target users;
the second abnormal detection module is used for carrying out secondary abnormal detection on the electricity charge data of the suspected abnormal user based on the SVDD abnormal detection model to obtain a suspected abnormal user detection result, and judging the uncertainty of the suspected abnormal user detection result to determine whether to output the suspected abnormal user detection result;
a third abnormal detection module, configured to directly output the detection result of the suspected abnormal user if the uncertainty of the detection result of the suspected abnormal user is lower than a preset threshold; and if the uncertainty of the detection result of the suspected abnormal user is higher than the preset threshold, performing final abnormal study and judgment on the electricity charge data of the suspected abnormal user with high uncertainty, and outputting a normal user of the suspected abnormal users with high uncertainty.
Priority Applications (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN202211479010.9A CN115511016B (en) | 2022-11-24 | 2022-11-24 | Incremental active learning-based electric charge anomaly detection method and device |
Applications Claiming Priority (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN202211479010.9A CN115511016B (en) | 2022-11-24 | 2022-11-24 | Incremental active learning-based electric charge anomaly detection method and device |
Publications (2)
| Publication Number | Publication Date |
|---|---|
| CN115511016A true CN115511016A (en) | 2022-12-23 |
| CN115511016B CN115511016B (en) | 2023-04-28 |
Family
ID=84513773
Family Applications (1)
| Application Number | Title | Priority Date | Filing Date |
|---|---|---|---|
| CN202211479010.9A Active CN115511016B (en) | 2022-11-24 | 2022-11-24 | Incremental active learning-based electric charge anomaly detection method and device |
Country Status (1)
| Country | Link |
|---|---|
| CN (1) | CN115511016B (en) |
Cited By (1)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN117390419A (en) * | 2023-08-25 | 2024-01-12 | 国网四川省电力公司营销服务中心 | Abnormal electricity utilization behavior detection method suitable for new energy distribution network |
Citations (6)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| US5566084A (en) * | 1993-03-02 | 1996-10-15 | Cmar; Gregory | Process for identifying patterns of electric energy effects of proposed changes, and implementing such changes in the facility to conserve energy |
| CN110866035A (en) * | 2019-10-31 | 2020-03-06 | 国网福建省电力有限公司厦门供电公司 | Method and system for detecting abnormal electricity quantity and electricity charge |
| CN113643113A (en) * | 2021-08-06 | 2021-11-12 | 国网山东省电力公司营销服务中心(计量中心) | Automatic electric charge auditing method, system, medium and electronic equipment |
| CN113743894A (en) * | 2021-08-30 | 2021-12-03 | 深圳供电局有限公司 | Method and system for establishing rechecking rule model for rechecking electric bill |
| CN114328625A (en) * | 2021-12-29 | 2022-04-12 | 深圳供电局有限公司 | Electric charge data rescreening method and system |
| CN114897064A (en) * | 2022-05-05 | 2022-08-12 | 华北电力大学（保定） | Abnormal electricity consumption behavior detection device based on BiGAN and SVDD |
-
2022
- 2022-11-24 CN CN202211479010.9A patent/CN115511016B/en active Active
Patent Citations (6)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| US5566084A (en) * | 1993-03-02 | 1996-10-15 | Cmar; Gregory | Process for identifying patterns of electric energy effects of proposed changes, and implementing such changes in the facility to conserve energy |
| CN110866035A (en) * | 2019-10-31 | 2020-03-06 | 国网福建省电力有限公司厦门供电公司 | Method and system for detecting abnormal electricity quantity and electricity charge |
| CN113643113A (en) * | 2021-08-06 | 2021-11-12 | 国网山东省电力公司营销服务中心(计量中心) | Automatic electric charge auditing method, system, medium and electronic equipment |
| CN113743894A (en) * | 2021-08-30 | 2021-12-03 | 深圳供电局有限公司 | Method and system for establishing rechecking rule model for rechecking electric bill |
| CN114328625A (en) * | 2021-12-29 | 2022-04-12 | 深圳供电局有限公司 | Electric charge data rescreening method and system |
| CN114897064A (en) * | 2022-05-05 | 2022-08-12 | 华北电力大学（保定） | Abnormal electricity consumption behavior detection device based on BiGAN and SVDD |
Cited By (1)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN117390419A (en) * | 2023-08-25 | 2024-01-12 | 国网四川省电力公司营销服务中心 | Abnormal electricity utilization behavior detection method suitable for new energy distribution network |
Also Published As
| Publication number | Publication date |
|---|---|
| CN115511016B (en) | 2023-04-28 |
Similar Documents
| Publication | Publication Date | Title |
|---|---|---|
| CN108520357B (en) | Method and device for judging line loss abnormality reason and server | |
| CN109308571B (en) | Distribution line variable relation detection method | |
| CN111738462B (en) | Fault first-aid repair active service early warning method for electric power metering device | |
| CN107329088A (en) | The health status diagnostic device and method of battery | |
| CN106569030B (en) | Alarm threshold optimization method and device in a kind of electrical energy measurement abnormity diagnosis | |
| CN110232405A (en) | Method and device for personal credit file | |
| WO2021100673A1 (en) | Evaluation device, computer program, and evaluation method | |
| CN115422788B (en) | Power distribution network line loss analysis management method, device, storage medium and system | |
| CN109613440A (en) | Stage division, device, equipment and the storage medium of battery | |
| JP6569985B2 (en) | Diagnostic method for power conversion device | |
| CN115511016A (en) | Incremental active learning-based electricity charge anomaly detection method and device | |
| CN112180279A (en) | Retired battery health state diagnostic expert system | |
| CN111179576B (en) | Power utilization information acquisition fault diagnosis method and system with inductive learning function | |
| CN110263945A (en) | Based on markovian Line Loss of Distribution Network System processing system | |
| CN110880055A (en) | Building intelligent ammeter system | |
| CN110443481B (en) | Power distribution automation terminal state evaluation system and method based on hybrid K-nearest neighbor algorithm | |
| Du et al. | Applying deep convolutional neural network for fast security assessment with N-1 contingency | |
| CN112288172A (en) | Prediction method and device for line loss rate of transformer area | |
| CN111612149A (en) | Main network line state detection method, system and medium based on decision tree | |
| CN114862229A (en) | Power quality evaluation method and device, computer equipment and storage medium | |
| Lyu et al. | A semiparametric clustering method for the screening of retired Li-ion batteries from electric vehicles | |
| CN107145986A (en) | A kind of charge capacity Forecasting Methodology and device | |
| CN114331001A (en) | Self-adaptive matching method and system for allocating abnormal work orders of power terminal | |
| CN114202174A (en) | Electricity price risk grade early warning method and device and storage medium | |
| CN109784777B (en) | Power grid equipment state evaluation method based on time sequence information fragment cloud similarity measurement |
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 |