Search

CN-121073230-B - Electricity behavior portrayal method and system based on edge AI

CN121073230BCN 121073230 BCN121073230 BCN 121073230BCN-121073230-B

Abstract

The application relates to the field of electricity behavior portrayal, in particular to an electricity behavior portrayal method and system based on an edge AI. The method comprises the steps of obtaining multi-dimensional data of operation of the electrical equipment, carrying out multi-state hierarchical analysis based on the multi-dimensional data of operation of the electrical equipment, judging the current life cycle state type of the electrical equipment, generating the current life cycle state type of the electrical equipment, establishing a health reference and deducting a health life cycle sequence of the equipment according to a time axis based on the current life cycle state type of the electrical equipment, generating a health state duration predicted according to a time rule, carrying out health comprehensive assessment based on the current life cycle state type of the electrical equipment and the health state duration of the equipment, and generating and outputting a user comprehensive portrait report for user decision. The application can extract high-value electrical equipment information in the process of electricity consumption behavior image drawing, and reduces unnecessary economic loss and safety risk.

Inventors

  • HUANG LI
  • LIN XIN
  • YUAN HAITAO
  • LIU JINHUI

Assignees

  • 福建百悦信息科技有限公司

Dates

Publication Date
20260512
Application Date
20251110

Claims (8)

  1. 1. An electricity behavior image-drawing method based on an edge AI is characterized by comprising the following steps: Disposing an edge computing node in an electrical equipment scene, acquiring electrical equipment operation multidimensional data, calling a multi-state hierarchy analysis model of an edge AI based on the electrical equipment operation multidimensional data, performing multi-state hierarchy analysis, judging the current life cycle state type of the electrical equipment, and generating the current life cycle state type of the electrical equipment, wherein the method comprises the following steps: The electrical equipment operation multidimensional data comprise a current harmonic frequency spectrum, a high-frequency current waveform, a steady-state power consumption track and a reference voltage phase signal; identifying the sustainable degradation drift of the harmonic structure of the current harmonic spectrum by tracking the component proportion change of the current harmonic spectrum on time sequence, and generating a sub-health state judgment result; based on the element degradation trend indicated by the sub-health state judgment result, identifying oscillation attenuation abnormality or rising edge flattening struggling of the waveform distortion characteristics of the high-frequency current waveform in a starting transient stage, and generating a pre-death state judgment result; after the pre-death state judgment result confirms that the equipment enters a functional failure critical point, analyzing the steep drop or surge phenomenon of the current power consumption track by taking the steady-state power consumption track as a standard and the step-out or jump phenomenon of the current voltage phase signal by taking the reference voltage phase signal as a standard, identifying the associated characteristics representing the termination of the equipment function, and generating a death state judgment result; Comprehensively generating the current life cycle state type of the electric appliance based on the sub-health state judgment result, the pre-death state judgment result and the death state judgment result; based on the current life cycle state type of the electric appliance, establishing a health reference according to a time axis and deducing a health life cycle sequence of the equipment to generate a health state duration of the equipment predicted according to a time rule; based on the current life cycle state type of the electric appliance and the equipment health state duration, carrying out health comprehensive evaluation, generating and outputting a user comprehensive portrait report for user decision, wherein the comprehensive portrait report comprises the following components: deducing and giving a user electricity behavior portrait tag based on the current life cycle state type of the electrical equipment and the corresponding equipment health state duration; If the device is in a sub-health state and the lifetime indicates that a maintenance window exists, defining a user behavior portrait as a risk avoidance type user, wherein the type represents that the user has tolerance to potential fault risks of the device and is prone to maintenance in a plan; if the device is in a pre-death state and the lifetime indicates that an emergency risk exists, defining a user behavior portrait as a cost-sensitive user, wherein the type characterizes that the user is highly sensitive to loss caused by sudden shutdown of the device and tends to take emergency damage stopping measures; Generating the user comprehensive portrait report matched with the electricity behavior portrait tag based on the electricity behavior portrait tag: Providing preventive maintenance planning and risk early warning suggestions based on the duration for the risk avoidance type user; for the cost sensitive users, reporting emphasis provides decision advice based on emergency loss prevention, spare part preparation or immediate replacement for a duration.
  2. 2. The method of claim 1, wherein the generating sub-health status decision results by tracking component proportion variations of the current harmonic spectrum over time, identifying persistent degradation drift of its harmonic structure, comprises: extracting harmonic component distribution in a typical normal working period from historical operation data of electrical equipment, and establishing a dynamic reference baseline reflecting the proportional relation of amplitude values of subharmonics; Slicing the current harmonic spectrum according to a time window, comparing the current harmonic spectrum with the dynamic reference base line one by one, capturing abnormal increase and decrease of specific subharmonic amplitude values, and generating a harmonic proportion offset event; Trend analysis is carried out on the harmonic proportion offset events in the continuous time window, and when the specific subharmonic proportion shows unidirectional continuous deviation and the amplitude is increased, the formation of the continuous degradation drift characteristic is judged; And mapping the harmonic frequency and the deviation direction corresponding to the persistent degradation drift characteristics to a preset electrical element degradation knowledge base, and generating a sub-health state judgment result containing specific degradation element types and stages.
  3. 3. The method of claim 2, wherein the generating a pre-death state decision by extracting waveform distortion characteristics of the high frequency current waveform at a start transient stage, identifying oscillation damping anomalies or rising edge plateau struggles thereof, comprises: when the sub-health state judging result indicates that the specific power element or the power mechanism is continuously degraded, at each starting moment of the electrical equipment, starting transient state information of a high-frequency current waveform is extracted, and the instant dynamic behavior of the waveform profile is analyzed: Analyzing the attenuation process of the high-frequency current waveform oscillation after reaching a peak value, identifying whether the oscillation amplitude does not recover to a threshold level within expected times or the oscillation period number is obviously shorter than that of a standard starting mode, and generating an attenuation abnormal mark representing the inertial loss of a mechanism; tracking the gradient continuity of the current in the current rising stage, and generating rising edge struggling marks representing blocked driving power when detecting that rising kinetic energy is interrupted to form a platform-shaped flat top or to present step-shaped delayed climbing; And carrying out space-time correlation on the attenuation abnormal mark and the rising edge struggling mark, and judging that the equipment core moving part or the driving circuit is in functional failure and generating a pre-death state judging result if the attenuation abnormal mark and the rising edge struggling mark are displayed in the same starting event.
  4. 4. A method according to claim 3, wherein said analyzing the steep drop or surge of the current power consumption trace and the out-of-step or jump of the current voltage phase signal based on the reference voltage phase signal based on the steady-state power consumption trace, identifying the associated features characterizing the termination of the device function, and generating the death-state decision result comprises: when the pre-death state judgment result simultaneously comprises the attenuation abnormal mark and the rising edge struggling mark, capturing cliff-type falling or sudden high-impact phenomenon which occurs after the pre-death state judgment result is separated from a normal working interval based on the steady-state power consumption track, and generating a power consumption extreme abnormal event; Based on the reference voltage phase signal, identifying continuous angle offset or instantaneous angle jump of the reference voltage phase signal relative to the standard phase of the power grid, and generating a phase unlocking event; time-axis alignment of the power consumption extreme anomaly event with the phase loss lock event is associated with causal logic: when the steady-state power consumption track is broken cliff-type and continuous angle offset occurs along with the reference voltage phase signal synchronously, judging that the associated characteristic of open circuit of a main circuit or complete disabling of a driving core is formed; when the steady-state power consumption track is suddenly fluctuated and the instantaneous angle jump occurs along with the reference voltage phase signal synchronously, judging that the associated characteristics of breakdown short circuit of an internal component or mechanical locking of a power mechanism are formed; based on the establishment of any associated feature, the device is directly judged to lose basic operation function, and a death state judgment result representing irreversible shutdown of the device is generated.
  5. 5. The method of claim 4, wherein the establishing a health benchmark and deriving a device health lifecycle sequence based on the appliance current lifecycle state type based on a time axis generates a time-regularly predicted device health state duration, comprising: Taking the moment when the electrical equipment is marked as the sub-health state judgment result for the first time as a life deduction sequence starting point, integrating all the electrical equipment operation multidimensional data recorded from the starting point to the current moment and the corresponding life cycle state types thereof, and constructing an equipment health life cycle sequence; Constructing a first life decay model by fitting a duration of an evolution from a sub-healthy state to a pre-dead state to a trend slope of the harmonic proportional offset event based on the device health lifecycle sequence; Constructing a second life decay model by fitting the duration of the evolution from the pre-death state to the frequency of occurrence of the decay abnormality signature and the rising edge struggle signature; the first life decay model is used for representing a duration prediction relation of evolution from a sub-health state to a pre-death state, and the second life decay model is used for representing a duration prediction relation of evolution from the pre-death state to the death state; and based on the first life attenuation model and the second life attenuation model, merging the duration time and the degradation rate of the current state, calculating the remaining time until the function is terminated, and generating the equipment health state duration time.
  6. 6. The method of claim 5, wherein the constructing a first life decay model by fitting a duration of time from a sub-health state to a pre-death state to a trend slope of the harmonic scaling event comprises: locating a first historical progression period from the sub-health state determination to the pre-death state determination in the device health lifecycle sequence; analyzing all harmonic proportion offset events in the first historical evolution duration interval, extracting key harmonic components which cause state judgment in each event, calculating the average change rate of the amplitude of the key harmonic components deviating from the dynamic reference baseline, and generating a harmonic degradation trend slope; When the absolute value of the harmonic degradation trend slope exceeds a preset accelerated degradation threshold, the speed of the harmonic degradation trend slope to a pre-death state is obviously accelerated; the first life decay model dynamically predicts a time required to evolve from a current sub-health state to a pre-death state based on the evolving correlation analysis and the accelerated degradation threshold.
  7. 7. The method of claim 6, wherein said constructing a second life decay model by fitting a duration of an evolution from a pre-death state to a frequency of occurrence of the decay abnormality signature and the rising edge struggle signature comprises: Locating a second historical duration from the first occurrence of the pre-death state decision to the generation of the death state decision in the device health lifecycle sequence; analyzing all equipment starting events in the time interval, and respectively counting the occurrence times and the occurrence frequencies of the attenuation abnormal mark and the rising edge struggling mark in a unit time window; Superposing the occurrence frequency of the attenuation abnormal mark and the occurrence frequency of the rising edge struggling mark to generate a joint frequency index representing the integral function instability degree; The destabilization association analysis is carried out on the second historical duration and the joint frequency index, namely, the duration time of the equipment from the pre-death state to the death state is recognized to be shortened sharply after the joint frequency index exceeds a critical destabilization threshold value; And dynamically predicting the second life attenuation model of the time required for the current pre-death state to evolve to the death state based on the critical destabilization threshold and the destabilization correlation analysis result.
  8. 8. An edge AI-based electricity consumption portrayal system, applied to the method according to any one of claims 1-7, comprising: the state analysis module is used for acquiring the operation multidimensional data of the electrical equipment, carrying out multi-state hierarchical analysis based on the operation multidimensional data of the electrical equipment, judging the current life cycle state type of the electrical equipment and generating the current life cycle state type of the electrical equipment; The state prediction module is used for establishing a health reference according to a time axis and deducing a device health life cycle sequence based on the current life cycle state type of the electric appliance, and generating a device health state duration predicted according to a time rule; and the comprehensive portrait module is used for carrying out comprehensive health assessment based on the current life cycle state type of the electric appliance and the equipment health state duration, and generating and outputting a user comprehensive portrait report for user decision.

Description

Electricity behavior portrayal method and system based on edge AI Technical Field The application relates to the field of electricity behavior portrayal, in particular to an electricity behavior portrayal method and system based on an edge AI. Background In the field of intelligent power consumption management and equipment health diagnosis, the refined monitoring and state evaluation of the user side electrical equipment are core links for realizing energy conservation and consumption reduction, guaranteeing power consumption safety and realizing predictive maintenance, the accuracy and the prospective of an analysis result are directly related to the user energy cost, the equipment service life and the timely elimination of potential safety hazards, and the method is a key technical foundation for constructing a novel power system and digital energy ecology. However, the existing electrical equipment state monitoring method lacks a mechanism for deep fusion and collaborative portrayal of equipment health state and user electricity behavior patterns when facing high-dimensional, asynchronous and reflecting electricity behavior data of different physical processes, so that visual and high-value information for guiding user decision is difficult to extract from mass data, and maintenance decision delay or misjudgment is more likely to be caused, thereby causing unnecessary economic loss and safety risk. Disclosure of Invention The application provides an electricity behavior portrayal method and system based on an edge AI, which are used for solving the technical problems. The application provides an electricity consumption behavior image method based on an edge AI, which comprises the steps of obtaining operation multidimensional data of electrical equipment, carrying out multi-state hierarchical analysis based on the operation multidimensional data of the electrical equipment, judging the current life cycle state type of the electrical equipment, generating the current life cycle state type of the electrical equipment, establishing a health reference according to the current life cycle state type of the electrical equipment according to a time axis, deducing a health life cycle sequence of the equipment, generating a health state duration predicted according to a time rule, carrying out health comprehensive evaluation based on the current life cycle state type of the electrical equipment and the health state duration of the equipment, and generating and outputting a user comprehensive image report for user decision. According to the technical scheme, edge computing nodes are deployed in an electrical equipment scene, electrical parameters are collected through an intelligent ammeter, an operation log is read by combining equipment with an interface, the operation log is integrated into multi-dimensional data of the electrical equipment, a multi-state hierarchical analysis model of an edge AI is called, life cycle state types such as sub-health state of the equipment are judged, health standards are established according to the types and combining histories and equipment data of the same type, a time sequence prediction model is used for deducing a health life cycle sequence and a health state duration of the equipment, finally multi-dimensional information is integrated for health comprehensive evaluation, and a comprehensive user portrait report containing user types, equipment states, predictions and suggestions is output to a user. The method has the advantages of early warning faults in advance by monitoring the equipment state in real time, shortening maintenance response time, reducing inconvenience of family life and production stopping loss of enterprises, improving operation and maintenance efficiency, helping users to adopt low-cost maintenance at the early stage of equipment performance attenuation, prolonging equipment life, identifying high-energy-consumption equipment, guiding energy-saving reformation, reducing use and energy cost, meeting low-carbon requirements, reporting, visually presenting and adapting to different user requirements, reducing decision thresholds, locally processing data by edge nodes, guaranteeing privacy and adapting to multiple scenes. The method comprises the steps of carrying out multi-state hierarchical analysis on the basis of multi-dimensional data of electric equipment operation, judging the current life cycle state type of the electric equipment, and generating the current life cycle state type of the electric equipment, wherein the multi-dimensional data of the electric equipment operation comprise a current harmonic frequency spectrum, a high-frequency current waveform, a steady-state power consumption track and a reference voltage phase signal, identifying continuous degradation drift of a harmonic structure of the current harmonic frequency spectrum through tracking component proportion change of the current harmonic frequency spectrum in time sequence, gen