Research For Power System Bus Load Forecasting

In the operational mechanism of power system, short-term load forecasting (STLF) is not only a compass of power system network to make profit, but also a very important reference to middle-term and long-term load forecasting, so short-term load forecasting has being an important task for years. However, a good load management strategy can be achieved by staggering of demand in different zones in a decentralized manner, which will require the bus load forecasting at distribution substations.At first, this paper analyses the historical hourly substation bus load data collected from the supervisory control and data acquisition system (SCADA), describes the structure, characteristic and the difficulty of bus load forecasting, and discusses the direction of bus load forecasting.Secondly, this paper utilizes ARIMA (Auto-Regressive Iterated Moving-Average) to predict the Bus Load, puts foreword a new method to reduce the order of stochasticmodel by normalizing history bus load data sequence, which will prompt the precision of parameter estimation and speed up the calculation procedure.Thirdly, in order to handle the effect of whether factors such as temperature to bus load, this paper also provide a back propagation neural network to deal with bus load prediction. In order to find out the most effective variables which will be inputs of artificial neural network, this paper presents a new method, which involves correlation coefficient method to quantities the relation between many factors and bus load. So we can make sure which one will affect the forecasting most, from which we can ensure the structure of neural network more reasonable. The forecasting result shows that this method is efficient.Further more, in order to enhance the precision of forecasting, this paper also provides a new method which combine the ARIMA and BP Neural Network together. It is based on the principle of optimal credit coefficient of sequence forecasting result. The forecasting result is better than any singular model used in this paper.The algorithmic models of bus load forecasting presented in this paper are simple with innovation, the results are accurate enough and of great value for practical operation and dispatching.