Self-adaptive non-intrusive load identification method based on random forest

Abstract

The invention discloses a random forest-based self-adaptive non-intrusive load identification method. The method comprises the steps of establishing an electrical load characteristic database; extracting required load characteristics from each switching event; normalizing the obtained load characteristics to obtain required sample points; processing the sample points by an unknown pattern recognition module, and distributing known labels or unknown labels to the sample points; wherein all labels are known sample points, and obtaining an identification result by using a random forest algorithm;wherein all the labels are unknown sample points and are processed by an online clustering module, and if new clustering is generated, enabling a user to select whether to distribute the class labelsto the cluster or not; performing new clustering with labels, updating the random forest through an online updating module, and updating the existing knowledge through an unknown mode recognition module; and enabling the unknown points to obtain identification results through a new random forest. The load which is easy to wrongly classify can be identified as unknown. Correct identification is completed after new knowledge is obtained, and effective identification of an unknown load mode is facilitated.

Description

technical field [0001] The invention relates to the technical field of non-invasive load monitoring, in particular to an adaptive non-invasive load identification method based on random forest. Background technique [0002] Non-intrusive load monitoring technology is an effective means to support energy management and demand side management, and it is also the basis for realizing friendly interaction between power grids and residents. The load identification is one of the key technologies of the non-invasive load detection technology. Although the non-intrusive load identification technology has been continuously improved, with the development of the home appliance industry, the types, brands and models of home appliances are increasing, which greatly increases the difficulty of load identification and is prone to misidentification and misidentification. Therefore, how to acquire new knowledge from actual recognition samples and form a new learning scheme to correctly recog...

AI Technical Summary

Problems solved by technology

Although the non-intrusive load identification technology has been continuously improved, with the development of the home appliance industry, the types, brands and models of home appliances are increasing, which greatly increases the difficulty of load identification and is prone to misidentification and misidentification.

Images