Application and Improvement of Smoothing Model with Winston Linearity and Seasonal Index in Prediction of Power Load

摘要

Owing to periodicity rises and falls with year, month or day for unit, and tendency that moves upward year by year, power load fit for in with smoothing model with Winston linearity and seasonal index predicting. On the other hand, this model is the most sensitive as for the unusual value in power load data, and frequently leads to calculating that the result is contrary with the development trend of actual power load. In the light of power load appearing the complicated combined characteristic with period and growth, it is led in this paper that smoothing model with Winston linearity and seasonal index is used power load prediction. For enhancing stability of this method, the improved model is set up to get stronger unusual data digestion ability, and prediction result fits the overall development trend of power load further. This improved model could adapt to few unusual date, avoid contrary conclusion, and advance prediction precision. Furthermore, the power load data of Shanghai electric network is applied to carry on the real predicting process, and to prove the stability and practicality of the improved model.