Load Feature Image Identification Method Based on Convolutional Neural Network

Inhalt:
Load identification technology which can identify types of electrical appliances through their power consumption information is the key of non-intrusive load disaggregation technology. At present, most of the commonly used load identification technologies are based on the combination of load steady-state or transient characteristics, and lack of deep information mining of feature combination. In this paper, a load feature image identification method based on convolutional neural network is proposed, which presents the combined information of load characteristics in the form of image, and we use deep convolutional neural network with residual mechanism to dig out the deep-level information of load feature image. Finally, a public dataset is used as an experimental sample to verify the method put forward in this paper. The experimental results show that the proposed method has good identification accuracy on most household appliances.