Unsupervised non-intrusive load monitoring result autonomous labeling method

Abstract

The invention discloses an unsupervised non-intrusive load monitoring result autonomous labeling method. The method comprises the following steps: reading load monitoring result data; judging whether the to-be-labeled electric appliance is subjected to result autonomous labeling or not; judging the control rule characteristic and the use rule characteristic of the to-be-labeled electric appliance; determining the name of the to-be-labeled electric appliance by adopting two layers of decisions, wherein the first layer of decision determines the candidate rough subset of the to-be-labeled electric appliance, and the second layer of decision determines the final name of the to-be-labeled electric appliance according to a parameter characteristic statistical result of the to-be-labeled electric appliance. After the names of all the to-be-labeled electric appliances generated in all the target scenes are labeled, a user can obtain the electric appliance names of all the to-be-labeled electric appliances in the target scenes in the visual interface, and whether the labeling result is accurate or not can be fed back through interpersonal interaction. According to the method, the NILM result can be automatically endowed with a physical name, the method can be used as a subsequent step of any unsupervised NILM method to be integrated with the NILM method, and then a completely unsupervised NILM scheme is formed and used for supporting various advanced applications expected by the NILM.

Description

technical field [0001] The invention belongs to the field of electric load information monitoring, and in particular relates to an unsupervised non-invasive load monitoring result autonomous labeling system and method. Background technique [0002] Non-intrusive Load Monitoring (NILM) is a cost-effective and detailed monitoring technology for load consumption [1] . It can obtain real-time working status (including start and stop time) and cumulative power consumption information of each electrical appliance in it only by analyzing the total load power consumption data. The obtained information can enable advanced applications such as fine-grained analysis of user behavior, advanced power marketization, and demand-side management (including response potential assessment, precise load regulation and response behavior verification, etc.), and promote smart grid [2] develop. [0003] According to the degree of "non-intrusive" 0 Different, NILM can be divided into supervised ...

AI Technical Summary

Problems solved by technology

Due to the variety of brands and models of similar appliances, the behaviors of different users vary greatly, and the scene generalization ability of parameter characteristics is limited.

Images