Power consumption characteristic waveform extraction and analysis architecture based on edge calculation

Abstract

The invention discloses a power consumption characteristic waveform extraction and analysis architecture based on edge calculation, which comprises a terminal and a cloud. The terminal comprises a device real-time data collection module, a real-time data calculation and analysis module and a lightweight characteristic library; the cloud comprises a fault classification model module based on machine learning and a lightweight feature library extraction module; the terminal is deployed at a power utilization side, collects a time sequence of power utilization data of a user in real time and at high frequency, then calculates a characteristic value of time sequence data in real time, and then uploads the characteristic value to the cloud, and the cloud constructs a lightweight characteristic library through a fault classification model module and a lightweight characteristic library extraction module and feeds back the lightweight characteristic library to the terminal. And then the terminal performs feature extraction and fault analysis on the collected power consumption data according to the deployed lightweight feature library. According to the invention, the side cloud is utilized to cooperatively and efficiently realize the abnormal fault detection of the power utilization side.

Description

Technical field [0001] The present invention relates to the field of electricity data analysis, power data abnormal prediction, and edge sharing calculations, and specific to an electrical characteristic waveform extraction and analysis architecture based on an edge calculation. Background [0002] In recent years, there are two typical methods for abnormal fault detection around the electricity side. The first method sets a simple threshold at the terminal device to be diagnosed, and the data is stored in the storage device. After the electricity is abnormal, then the data is subjected to data from the terminal to the field by offline method, this method is more Focus on the post-analysis; the second method, the terminal transmits the data in real time to the cloud, and the model training and online fault diagnosis are performed by the cloud. The diagnostic fault method used in the first type of method is simple, and the effect is not satisfactory, and the second even if the clo...

AI Technical Summary

Problems solved by technology

The first type of method uses a simple fault diagnosis method, and the effect is not satisfactory, while the second type even if the cloud participates in the fault diagnosis to improve the accuracy of fault identification, but because it is not performed locally, the data transmission There will be a delay in the process, which cannot meet the requirements of timely response when a fault occurs

Therefore, a large number of computing tasks cannot be simply processed on the cloud, and the terminal equipment is only used to collect and upload data. It is necessary to "reduce the burden" on the cloud and appropriately "increase the burden" on the terminal equipment to achieve full utilization of resources and The goal of timely response to fault data

Images