Non-intrusive load monitoring method and system based on motif mining and semi-supervision method

Abstract

The invention discloses a non-intrusive load monitoring method and system based on motif mining and a semi-supervision method, and the method comprises the steps: S1, obtaining the load information of a user from a user total ammeter interface, and detecting all equipment switching events according to the load information; s2, dividing equipment operation windows by using all detected equipment switching events to obtain a plurality of determined equipment operation windows and a plurality of undetermined equipment operation windows; s3, extracting an equipment feature vector of each piece of equipment from the plurality of determined equipment operation windows and the plurality of to-be-determined equipment operation windows to obtain an equipment feature vector set; and S4, identifying the equipment type of each piece of equipment by adopting a semi-supervised learning method according to the equipment characteristic quantity set. According to the invention, load identification is carried out by using a harmonic function-based semi-supervised learning algorithm of a small number of marked samples, the accuracy and the practicability of a non-intrusive load monitoring technology can be improved, and normal production, life and privacy of a user are not affected.

Description

technical field [0001] The invention relates to the technical field of load monitoring research, in particular to a non-invasive load monitoring method and system based on phantom excavation and a semi-supervised method. Background technique [0002] Non-intrusive load monitoring technology has very important practical significance to realize the two-way interactive service mode between users and the power grid, and to respond to energy conservation and emission reduction. The running time information of various equipment analyzed by this technology can help users reduce electricity bills by replacing energy-saving appliances, adjusting electrical parameters, and shifting peak power consumption. Users can also share information with power companies to obtain more value-added services. Through non-intrusive load monitoring, electric power companies can grasp the energy consumption information of users in more detail, and scientifically formulate the development planning and ...

AI Technical Summary

Problems solved by technology

However, the feature database is collected in an intrusive manner, which affects the normal production, life and privacy of users

Moreover, with the increase in the type and quantity of user equipment, equipment with similar operating characteristics is gradually increasing. When such equipment is switched on and off, it is difficult to accurately analyze the equipment running time.

[0005] Second, in the process of load feature extraction, waveform parameters and harmonic analysis require a high sampling frequency, which requires high configuration of measuring equipment and increases the difficulty of storage and analysis.

[0006] Third, in the process of load identification, the supervision method adopted has a high demand for the number of marked samples, and the analysis results need to be manually marked, which makes the corresponding non-invasive load monitoring face high cost, cumbersome implementation and low comfort of residents in the application.

In addition, unsupervised methods have problems such as low classification accuracy and manual identification of analysis results.

Images