Recognition of Household Electrical Appliances Using Deep Neural Network
Zhibin Yu, Hong Chen, Chunxia Chen, Haina Rong and Xiantai Gou
Deep neural network has made a breakthrough in the applications of image and speech recognitions, and have been hot issues in the artificial intelligence in recent years. In this paper, the deep neural network model is applied to identify the household electrical appliances in nonintrusive load monitoring (NILM). The accuracy of traditional methods with the voltage current (V-I) trajectory feature is low to fulfill practical requirement in the complex environment with noise interference. Especially, when the number of the V-I trajectory feature sample data is very small, traditional methods are more difficult to recognize the household electrical appliances. Hence, based on the deep neural network and transfer learning theory, a deep transfer learning method is presented to solve the small-sample problem of data in status-monitor of electrical appliances. Meanwhile, aiming at the problem that a large number V-I trajectory images are not available due to noise, a preprocessing algorithm is proposed to improve the quality of the V-I trajectory image data. In lab test, the average recognition rate of the four kinds of household electrical appliances by using the proposed method reaches up to 99%. The experimental results show that the proposed method has higher recognition accuracy than traditional methods.
Keywords: Deep neural network, transfer learning, voltage current trajectory, household electrical appliance recognition