Multiscale Self-Attention Architecture in Temporal Neural Network for Nonintrusive Load Monitoring
Zihan Shan, Gangquan Si, Kai Qu, Qianyue Wang, Xiangguang Kong, Yu Tang, Yang Chen
Abstract
Non-intrusive load monitoring constitutes a significant function of the smart grid in the future. The purpose is to ameliorate the consumption and supply of electricity by disaggregating the total load to appliance-level load without intrusive monitoring. Recently, energy disaggregation is improved with the emergence of deep learning, but the imbalanced datasets and long sequences bring multiple difficulties to model training. The distribution of on/off states and the limitation of the model lead to massive false positive samples and undetected events. To tackle these problems, we proposed a multiscale self-attention temporal neural network (MSANet) to utilize the global temporal correlation and local sequential features. Specifically, the dilated window self-attention mechanism is proposed to compute the local attention and the multi-branches structure is to exploit sequential features of different scales. Furthermore, the embedding of global temporal information is introduced to improve global contextual awareness, and subtask networks are designed for different tasks respectively to alleviate the effect of the imbalance. The proposed model is evaluated on REDD and UK-DALE datasets and shows outstanding performance on MAE and F1 score compared to baseline algorithms.