Estimation of Weather Sensitive Load Power Based on Dual-Branch Neural Network

The proportion of weather sensitive load in the power system is increasing. Estimating weather sensitive load accurately can help in evaluating demand response capabilities and ensuring the safety and stability of the power system. Due to the difficulty in obtaining accurate weather sensitive load measurements, this paper proposes a method for estimating weatherly sensitive load using a dual-branch neural network model. The two branches are the baseline load estimation branch and the weatherly sensitive load estimation branch which can process the independent features respectively and output the estimated baseline load and weather sensitive load simultaneously. The weather sensitive load branch applies the convolutional long short-term memory(convLSTM) network to process the accumulative effect of temperature and the temperature-humidity effect. The effectiveness of the proposed method is verified. The effectiveness of the proposed method is verified by comparing other methods for estimating weather sensitive load in a real case.