Open AccessA Non-intrusive Load monitoring Algorithm Based on Seq2point
Author(s) -
Zhen Zhang,
Jun Ma,
Menghou Li,
Haofang Li
Publication year - 2021
Publication title -
journal of physics. conference series
Language(s) - English
Resource type - Journals
SCImago Journal Rank - 0.21
H-Index - 85
eISSN - 1742-6596
pISSN - 1742-6588
DOI - 10.1088/1742-6596/2005/1/012062
Subject(s) - computer science , artificial neural network , decomposition , set (abstract data type) , field (mathematics) , filter (signal processing) , data set , sequence (biology) , training (meteorology) , training set , point (geometry) , artificial intelligence , time sequence , real time computing , algorithm , data mining , computer vision , mathematics , ecology , genetics , physics , geometry , meteorology , pure mathematics , biology , programming language
Aiming at the problems of long training time and poor decomposition performance of neural network in the application of Sequence-to-point deep learning method in the field of non-intrusive load monitoring, a non-intrusive load monitoring optimization model based on Seq2point was proposed. In this model, the training speed of the neural network is optimized by the forward gated recurrent unite, and the decomposition performance is improved by the median filter and the standardized data pre-treatment method. Experimental results on public data set AMPDS2 show that this algorithm can effectively improve the accuracy of load decomposition and shorten the training time of network.