Parameter estimation of weibull mixtures based on improved cuckoo search algorithm

Abstract

As to the problem that it is difficult to estimate the parameters of Weibull mixtures accurately, a new parameter estimation method for Weibull mixtures based on the improved cuckoo search algorithm (CSA) is proposed. The optimization model, which is solved by the improved CSA, is established based on the idea of minimizing the residual sum of squares (RSS). The algorithmic improvement has three schemes including the improvement of discovering probability, the improvement of step-size scale and their integrated scheme. In the case study, craft windshields are regarded as the objects, and two-fold Weibull mixture is considered as the fitting function. Four parameter estimation methods based on CSA and three improved CSAs are respectively used to estimate the parameters for 2000 times. The results are compared and the result proves that the accuracy of the method based on the integrated scheme of CSA is improved by the integrated scheme is better than other three algorithms.