Research on the Fitting Effect of General Extreme Value Distribution Based on Wind Speed Data Measured Actually at the Hang-zhou Bay Bridge Site

摘要

The general extreme value distribution (GED) is applied to fitting wind speed data, and the failure problem of parameter estimation for GED is put forward. In order to resolve the problem, probability plot correlation coefficient and maximum likelihood estimation are combined to establish the highly-efficient and stable parameter estimation method for GED. Comparison is further made by correlation coefficient for the fitting effect of GED and Type Ⅰ, Type Ⅱ and Type Ⅲ Distribution. Results indicate that the effective peak value of correlation coefficient for GED is almost equivalent to the maximum of those for Type Ⅰ, Type Ⅱ and Type Ⅲ Distribution, namely the fitting effect of GED is the same with that of the optimal among Type Ⅰ, Type Ⅱ and Type Ⅲ Distribution. So GED has obvious advantage in fitting wind speed data.