Optimized Battery Controlling for Smart Homes with Dynamic Insensitive-Zone Adjustment

Lo Pang-Yun Ting, Yu-Hao Chiang, Yu-Hsiang Chang, Michihiro Konagai, Kun-Ta Chuang

Abstract: Renewable energy sources have become increasingly important, highlighting the significance of battery energy storage systems. However, existing methods for controlling battery storage systems typically require a substantial amount of historical data to develop accurate charging and discharging strategies. This becomes challenging when there is insufficient past charging information for a household. In our paper, we introduce the DIA framework, which aims to address data scarcity in early-stage household operations. The core of DIA is to augment household electrical data by dynamically adjusting the insensitivity of the model, thereby reducing bias and accommodating variations in data quality and quantity. Our experiments on a real-world dataset reveal that our method can effectively reduce household costs compared to baseline approaches.