Power System Load Forecasting Using Partial Least Square Method

Load models have a wide range of application in the electric power industry including monitoring load management policies, helping with the generator commitment problem, and aiding system planning. This work investigates partial least square (PLS) technique for power system load forecasting. A comprehensive load model is proposed based upon analyzing hourly data using PLS method. The proposed algorithm simplifies the calculating process greatly and gives better results in terms of higher accuracy of short term forecasting as compared with regular LS. The periodic occurrence based on seasonal weather from peak to peak for a short term forecast is shown successfully. Both theoretical analysis and computer simulation on ISO New England data verify its effectiveness.