Non-intrusive load event global optimization matching method and system

Abstract

The invention discloses a non-intrusive load event global optimization matching method, which comprises the following steps of: firstly, dividing a load operation full time period into balance windows with different levels and different complexities in combination with actual engineering conditions, and progressively processing one by one according to the principle that the levels are from high to low and the complexities are from low to high, so that the load event matching efficiency is improved, and interference possibly brought by other load events which do not belong to the current balance area is eliminated; then, matching, result marking and feature extraction are performed on the load events in each balance window one by one, and for the load events to be matched in the windows, a candidate load event sequence set possibly corresponding to the load events to be matched in the windows is generated by using a depth-first search algorithm according to the power change characteristics of the complete working cycle of the electric appliance, then, selecting a load event sequence subset meeting optimal fitting of the total load power from the candidate load event sequence set, so that the accuracy and efficiency of load event matching are improved.

Description

technical field [0001] The invention belongs to the field of advanced measurement systems and relates to a load event matching method for non-invasive electric load monitoring. Background technique [0002] Non-intrusive load monitoring (Non-Intrusive Load Monitoring, NILM) is a cost-effective and detailed monitoring technology for load power consumption. [1] , it only needs to install a sensor at the power supply entrance of the power load, and can obtain the power consumption information of each (main) electrical appliance inside the user by analyzing the total data. At present, a mainstream NILM method is based on the load event detection method, which uses the load characteristics related to the electrical working state transition process as the load imprint [2] . According to whether the implementation process needs to manually set the labeling information of the load imprint samples, it can be subdivided into supervised and unsupervised two schemes. Compared with th...

AI Technical Summary

Problems solved by technology

Existing research mainly focuses on two-state load event sequence matching [6-7] , due to the principle of time proximity, the existing method has the limitation of local search, the matching effect and efficiency of the multi-state load event sequence are not good, and it is easy to mis-match in the multi-appliance aliasing scene [8-9]

Images