Optimal Residential Battery Scheduling using Short-term Forecasts

Optimal day-ahead scheduling of batteries relies on reliable day-ahead forecast of residential load profile, solar insolation, and time-of-use tariff. However, actual values may not be the same as the forecasted values due to various factors like weather, cloud cover, consumer behaviour, etc. This work aims at optimally scheduling the batteries installed in a residence to minimize the cost of power consumption from the grid and increasing the battery lifetime, while accounting for the variations in the forecasted load profile, solar insolation and time-of-use tariff. The proposed method uses short-term forecasts and employs a sliding window technique to generate optimal schedule on an hourly basis. The sliding window technique helps in adjusting for the variations in the forecasts used. Artificial bee colony (ABC) algorithm is used as the optimization technique during each window. Effect of allowable maximum depth-of-discharge, and sliding window size on the optimized cost are also studied. The results are compared with manual scheduling.