Research On Grid Short-term Bus Load Data Preprocessing And Forecasting Model

The short-term bus load forecasting provides a basis for the grid security early-warning and economic analysis.The forecasting model's popularity, forecast accuracy and forecast speed are the key study points of short-term bus load forecasting. The paper studied the short-term bus load forecasting from three aspects:load characteristic classification, data preprocessing and forecast method. Firstly, a characteristic classification method based on gray incidence matrix was proposed, which set the stage for its adaptive prediction. Secondly, a new data preprocessing method was proposed to process the three typical abnormal data, which was based on the modified data across comparision and wavelet threshloding denoising method. At last, it put forward a hybrid method based on least squares support vector machines(LSSVM) and markov chain to forecast the bus load, in which a generalized grid-search algorithm was used to optimize the selection of model parameter. The case study proved that the proposed data preprocessing method and hybrid forecast method are available to satisfactory forecast accuracy.