Short-Term Bus Load Forecasting Based On Intelligent Similar Day Selection And Deviation Self-Correction

Short-term bus load forecasting is of great significance for ensuring the stable operation of the power grid and the orderly conduct of electricity market transactions. In view of the problems such as complex bus load conditions and low prediction accuracy of short-term bus load forecasting, we firstly propose an intelligent strategy of the optimal similar day selection by considering the influence factors of short-term bus load in this paper, taking the bus load of similar day as the result of first forecasting step. The deviation caused by first forecasting step is analysed, on this basis, real-time feature vector is constructed by combining real-time meteorological status. Then, this paper presents a deviation self-correction model based on an efficient and accurate machine learning algorithm - XGBoost. The case study of a bus in Guangdong province of China shows that the model and algorithm proposed in this paper have high accuracy and stability in short-term bus load forecasting.