An Improved Permutation Entropy Algorithm for Non-intrusive Load State Change Detection

摘要

Non-intrusive load monitoring (NILM) was one of the developing direction of future load monitoring, which could provide technical support for the management of demand response. When the transient characteristics are used for NILM, the detection of the load state change is vital, because accurate detecting could provide solid basis for accurate analysis of transient characteristics. In this paper, an improved permutation entropy (PE) algorithm is proposed, aiming at promoting the performance of PE algorithm about its anti-noise capability and appropriate sampling window. The main points of the improved algorithm are as follows: first, the differential of multi-scale permutation entropy is analyzed to determine the interval in which the load encounters a state change, and second, the signal-to-noise ratio (SNR) of the interval is deployed to locate the load state change. Simulation results showed that the proposed algorithm could detect load state change more accurately, and it worked well with noisy signals of high-frequency sampling, thus the algorithm could greatly facilitate the load identification in subsequent.