Accurate mixed weibuII distribution fitting by differential evolution

Abstract

Mixed Weibull distribution is a probability distribution noted for its wide applicability in many diverse fields. The ability to accurately estimate mixed distribution parameters is essential for data-driven modeling, simulation, and analysis of the phenomena represented by mixed Weibull models. Nature-inspired metaheuristics for continuous parameter optimization have shown good potential for approximating parameters of complex statistical models. Differential evolution is a popular evolutionary real-parameter optimization method with good results in many areas. This work uses differential evolution to fit mixed Weibull distribution to data and analyzes the ability of different differential evolution variants to estimate mixture parameters.