CN-122026605-A - Electricity consumption monitoring index construction method and system based on user portrait

CN122026605ACN 122026605 ACN122026605 ACN 122026605ACN-122026605-A

Abstract

The invention discloses a user portrait-based electricity consumption monitoring index construction method and system, and relates to the technical field of electricity consumption monitoring, wherein the method comprises the steps of collecting electricity consumption power load data of a user and extracting power load characteristics of each window; the power load characteristics of each window are processed through multi-granularity behavior unit decomposition according to the power load characteristics of each window, multi-granularity electricity utilization behavior sequences are identified, sequence similarity calculation is conducted on all users to obtain multi-granularity behavior pattern clusters, multi-granularity electricity utilization monitoring index threshold construction is conducted, and the optimized electricity utilization monitoring index threshold is obtained. The method solves the technical problem of insufficient electricity consumption abnormality identification accuracy caused by lack of suitability of the electricity consumption monitoring index threshold in the prior art, achieves the aims of realizing accurate construction and optimization of the electricity consumption monitoring index threshold, and improves the accuracy of electricity consumption abnormality identification and the technical effect of monitoring index suitability.

Inventors

CHEN JIANHUA
ZHAO JIAN
ZHANG JIE
LU YAN
Kan tianyang
MA RUI

Assignees

国网冀北电力有限公司

Dates

Publication Date: 20260512
Application Date: 20251215

Claims (10)

1. The electricity consumption monitoring index construction method based on the user portrait is characterized by comprising the following steps of: collecting power load data of a user, carrying out sliding window scanning on the power load data, and extracting power load characteristics of each window; Decomposing the multi-granularity behavior units according to the power load characteristics of each window, and identifying a multi-granularity power consumption behavior sequence, wherein the multi-granularity power consumption behavior sequence comprises a power consumption behavior sequence formed by instantaneous behavior units, a power consumption behavior sequence formed by long-time behavior units, a power consumption behavior sequence formed by periodic behavior units and a power consumption behavior sequence formed by frontal loss behavior units; Performing sequence similarity calculation on all users based on the instantaneous-power consumption behavior sequence, the long-time-power consumption behavior sequence, the periodic-power consumption behavior sequence and the frontal loss-power consumption behavior sequence to obtain a multi-granularity behavior pattern cluster; And constructing a multi-granularity electricity consumption monitoring index threshold according to the multi-granularity behavior pattern cluster, and carrying out fusion verification optimization on the multi-granularity electricity consumption monitoring index threshold to obtain an optimized electricity consumption monitoring index threshold.
2. The method of claim 1, wherein the multi-granularity behavioral unit decomposition is performed according to the power load characteristics of each window to identify a multi-granularity power consumption behavioral sequence, the method comprising: recording the positive power load characteristics of all electric equipment; Identifying the positive power load characteristics of all the electric equipment and the power load characteristics of each window to obtain an electric behavior unit of each electric equipment; And decomposing the electricity behavior units of each electric equipment according to a plurality of time sequence granularity factors, and identifying a multi-granularity electricity behavior sequence, wherein the time sequence granularity factors comprise time sequence step factors, time sequence steady-state factors, time sequence periodic factors and time sequence loss factors.
3. The method of claim 2, wherein the decomposing the power usage units of each powered device according to the plurality of timing granularity factors, the identifying the power usage sequence of instantaneous power usage units comprises: extracting power load characteristics corresponding to the power consumption behavior units according to the power consumption behavior units of each piece of electric equipment; Calculating a first-order difference of the power load characteristics corresponding to the power consumption behavior units, and comparing a first-order difference calculation result with a step detection threshold value as the time sequence step factor to identify an instantaneous behavior unit set, wherein the step detection threshold value is obtained through a first-order difference mean value and a first-order difference standard deviation configuration; And forming an instantaneous-electricity behavior sequence according to the instantaneous behavior unit set in the electricity behavior units.
4. The method of claim 2, wherein decomposing the power usage behavior units of each powered device according to a plurality of timing granularity factors, the method of identifying a power usage behavior sequence of long-term behavior units comprises: Calculating steady-state fluctuation of the power load characteristics corresponding to the power consumption behavior units, comparing a steady-state fluctuation calculation result as a time sequence steady-state factor with a steady-state fluctuation threshold value, and identifying a steady-state behavior unit set; and according to the steady-state behavior unit set in the electricity utilization behavior units, a long-time electricity utilization behavior sequence is formed.
5. The method of claim 2, wherein the decomposing the power usage behavior units of each powered device according to the plurality of timing granularity factors, the identifying the power usage behavior sequence of periodic behavior units comprises: Calculating a window power standard deviation of the power load characteristic corresponding to the power consumption behavior unit; The window power standard deviation calculation result is used as a time sequence periodic factor to identify a candidate periodic behavior unit set; Performing power load characteristic autocorrelation analysis on the candidate periodic behavior unit set to obtain an autocorrelation analysis result, and obtaining a periodic behavior unit set by using the autocorrelation analysis result; and forming a periodic-electricity behavior sequence according to the periodic behavior unit set in the electricity behavior units.
6. The method of claim 2, wherein the decomposing the power usage units of each powered device according to the plurality of timing granularity factors, the identifying the power usage sequence of loss units comprises: Calculating the corresponding power load characteristic of the electricity consumption behavior unit, wherein the corresponding power load characteristic comprises reactive power, a power factor, a current total harmonic distortion rate and three-phase unbalance; identifying a frontal loss behavior unit set by taking the frontal loss related features as time sequence loss factors; and forming a frontal loss-electricity consumption behavior sequence according to the frontal loss behavior unit set in the electricity consumption behavior units.
7. The method of claim 1, wherein the sequence similarity calculation is performed for all users based on the instantaneous-power-use behavior sequence, the long-term-power-use behavior sequence, the periodic-power-use behavior sequence, and the frontal-loss-power-use behavior sequence to obtain the multi-granularity behavior pattern cluster, the method comprising: respectively carrying out characterization processing on the instantaneous electricity consumption behavior sequence, the long-time electricity consumption behavior sequence, the periodic electricity consumption behavior sequence and the frontal loss electricity consumption behavior sequence to obtain multi-granularity electricity consumption behavior sequence characteristics; constructing a multi-granularity characteristic distance matrix according to the multi-granularity electricity utilization behavior sequence characteristics; Hierarchical clustering is carried out on the multi-granularity characteristic distance matrix to obtain a multi-granularity behavior mode set; carrying out consensus fusion on the multi-granularity behavior mode set, and outputting four-dimensional comprehensive behavior feature vectors of each user; calculating the fusion similarity of the four-dimensional comprehensive behavior feature vectors to obtain a fusion similarity matrix; And carrying out hierarchical clustering on the fusion similarity matrix to obtain a multi-granularity behavior pattern cluster.
8. The method of claim 1, wherein constructing the multi-granularity power usage monitoring indicator threshold according to the multi-granularity behavior pattern cluster, the method comprising: Matching the multi-granularity behavior pattern clusters, and defining structural electricity utilization monitoring indexes; calculating the individual deviation degree of each mode cluster in the multi-granularity behavior mode clusters; And carrying out self-adaptive threshold configuration on the structural electricity consumption monitoring index of each mode cluster according to the individual deviation degree to obtain a multi-granularity electricity consumption monitoring index threshold.
9. The method of claim 8, wherein the multi-granularity power usage monitoring indicator threshold is fusion validated for optimization, the method comprising: Performing candidate threshold configuration on the structural electricity utilization monitoring index of each mode cluster according to the individual deviation degree to obtain a candidate threshold set; predicting a false alarm rate set of the candidate threshold set, and carrying out self-adaptive configuration according to the false alarm rate set and the expected false alarm rate to obtain an initial solution of the power consumption monitoring index threshold; And verifying the fusion monitoring conflict of the primary solution of the electricity consumption monitoring index threshold value, and carrying out feedback optimization according to a conflict verification result to obtain an optimized electricity consumption monitoring index threshold value.
10. A user portrayal-based electricity usage monitoring index construction system for implementing the user portrayal-based electricity usage monitoring index construction method according to any one of claims 1-9, the system comprising: the sliding window scanning module is used for collecting the power load data of a user, carrying out sliding window scanning on the power load data and extracting the power load characteristics of each window; The electricity consumption behavior sequence identification module is used for carrying out multi-granularity behavior unit decomposition according to the power load characteristics of each window and identifying multi-granularity electricity consumption behavior sequences, and comprises an electricity consumption behavior sequence formed by instantaneous behavior units, an electricity consumption behavior sequence formed by long-time behavior units, an electricity consumption behavior sequence formed by periodic behavior units and an electricity consumption behavior sequence formed by forehead loss behavior units; The behavior pattern cluster acquisition module is used for carrying out sequence similarity calculation on all users based on the instantaneous-power consumption behavior sequence, the long-time-power consumption behavior sequence, the periodic-power consumption behavior sequence and the frontal loss-power consumption behavior sequence to acquire a multi-granularity behavior pattern cluster; And the fusion verification optimization module is used for constructing the multi-granularity electricity consumption monitoring index threshold according to the multi-granularity behavior pattern cluster, and carrying out fusion verification optimization on the multi-granularity electricity consumption monitoring index threshold to obtain the optimized electricity consumption monitoring index threshold.

Description

Electricity consumption monitoring index construction method and system based on user portrait Technical Field The invention relates to the technical field of electricity consumption monitoring, in particular to an electricity consumption monitoring index construction method and system based on user portraits. Background In the field of power monitoring, a unified and standardized configuration logic is mostly adopted in a traditional power consumption monitoring index and threshold construction mode, namely, fixed monitoring indexes and thresholds are set for all power users, and differentiation of different users in aspects of electric equipment types, power consumption habits, power consumption time periods, power consumption characteristics and the like is not fully considered. On one hand, the monitoring mode is difficult to accurately identify personalized electricity utilization anomalies of different user groups, for example, short-time power fluctuation of common resident users is mostly caused by starting and stopping of normal household appliances, but is easy to judge as anomalies by unified thresholds, hidden electric energy loss risks in periodic high-load electricity utilization of commercial users are too loose to be timely perceived, on the other hand, the traditional method lacks deep disassembly and cluster analysis on multi-dimensional electricity utilization behaviors of users, cannot form a monitoring system fitting the electricity utilization images of the users, and causes the problems of high false alarm rate, poor suitability, difficult anomaly tracing and the like of monitoring results, so that a large number of invalid investigation works are brought to electric operation and maintenance personnel, personalized requirements of different types of users on electricity utilization safety and energy efficiency management are difficult to be met, and the refinement and intelligent development of electric power monitoring services are restricted. In the prior art, the power consumption monitoring index threshold lacks suitability, so that the technical problem of insufficient power consumption abnormality identification accuracy is caused. Disclosure of Invention The application provides a user portrait-based electricity consumption monitoring index construction method and system, which are used for solving the technical problem of insufficient electricity consumption abnormality identification accuracy caused by lack of suitability of an electricity consumption monitoring index threshold in the prior art. In view of the above problems, the application provides a method and a system for constructing electricity consumption monitoring indexes based on user portraits. In a first aspect of the present application, there is provided a user portrait-based electricity usage monitoring index construction method, the method comprising: The method comprises the steps of collecting power consumption load data of a user, carrying out sliding window scanning on the power consumption load data, extracting power load characteristics of each window, carrying out multi-granularity behavior unit decomposition according to the power load characteristics of each window, identifying multi-granularity power consumption behavior sequences, including power consumption behavior sequences formed by instantaneous behavior units, power consumption behavior sequences formed by long-time behavior units, power consumption behavior sequences formed by periodic behavior units and power consumption behavior sequences formed by loss behavior units, carrying out sequence similarity calculation on all users based on the instantaneous-power consumption behavior sequences, the long-time-power consumption behavior sequences, the periodic-power consumption behavior sequences and the loss-power consumption behavior sequences, obtaining multi-granularity behavior pattern clusters, carrying out multi-granularity power consumption monitoring index threshold construction according to the multi-granularity behavior pattern clusters, carrying out fusion verification optimization on the multi-granularity power consumption monitoring index threshold, and obtaining the optimized power consumption monitoring index threshold. In a second aspect of the present application, there is provided a user profile-based electricity usage monitoring index construction system, the system comprising: The system comprises a power consumption monitoring index threshold value acquisition module, a sliding window scanning module, a power consumption behavior sequence identification module and a fusion verification optimization module, wherein the power consumption monitoring index threshold value acquisition module is used for acquiring power consumption load data of a user, carrying out sliding window scanning on the power consumption load data, extracting power load characteristics of each window, carrying out multi-granularity behavior unit decomposi