A Real-time Adaptive Forecasting Algorithm for Electric Power Load

Forecasting of electric power load is the basis of achieving the power system automatic operation, real-time and accurate forecasting of load determines the run-time of most power equipment in the system, and is the foundation of accomplishing the safety and economy of power system running, it influences the life period and reliability directly, and is of great importance in theory and practical use. Real-time adaptive forecasting algorithm for power load in this paper is time series analysis method, which is based on ARIMA (p, d, o) model structure and whose parameter estimation adopts the recursive forgetting factor least square method (RFFLS). The Astrom forecasting algorithm, which is based on linear minimum square error of prediction, is used for forecasting. Modeling procedure of the algorithm is relatively simple compared with traditional time series method, and the algorithm avoids the fussy process of model structure recognition and verification, less data is needed and accuracy of parameters estimation is fine. Experimental verification of the algorithm is made using power load data of certain place in Hebei district. The result shows that the forecasting algorithm can estimate the model parameters online, and meet the requirement of dynamically predicting power load, solving the problem of great effect of random disturbance and uncertainty on one step real-time forecasting of power load. Good forecasting performance is acquired while the algorithm is applied in the working days, weekends and different seasons. The results show that the algorithm has fine adaptability