Pointwise Elimination Method on the Abnormal Data and Robust LS Polynomial Fitting of Wind Power

Abstract

In this paper, we establish the 5th order polynomial robust LS curve-fitting model, research the pointwise elimination method on the abnormal data of wind power, and develop an improved stopping-rule combining with the method of T test. Under the nonlinear model, we prove that when the stopping-parameter is 2 ( 1) 1 ( 1) t m w m a − − = − , we can maximally eliminate the abnormal data and reserve the good data, where α is the confidence level and m is the number of observation data. Finally, numerical experiments show simulation results and illustrate the effectiveness of our method.