Residential Load Identification Based on Load Profile using Artificial Neural Network (ANN)

摘要

Residential load identification is important for homeowners who wish to optimize their electricity usage. If residential loads are identified correctly, the homeowners may use this information to manage their electricity usage and thus can result in annual cost savings. This paper outlines a method to identify residential load types using an artificial neural network (ANN) with eight selective input features. ANN models with different combinations of layers and neurons were tested with three activation functions - Sigmoid, Exponential Linear Unit (ELU), and Rectified Linear Unit (ReLU). The trained ANN models were then used to identify four residential load types - 1) Dishwasher, 2) Refrigerator, 3) Furnace, and 4) Stove. The performance of these models were evaluated using metrics such as Mean Absolute Percentage Error (MAPE) and Root Mean Squared Error (RMSE). Initial findings indicate that the ANN model was able to identify two loads correctly such as stove and dishwasher, but unable to differentiate between the refrigerator and furnace loads.