An Improved Adaptive Niche Differential Evolution Algorithm

Abstract

Differential evolution (DE) algorithm is a random search algorithm by referring to the natural genetic and natural selection mechanism of the biological world and it is used to process the complicated non-linear problems which are difficult to be solved by traditional computational methods. However, subject to its own mechanism and single structure, the basic DE algorithm is easy to get trapped into local optimum and it is difficult to handle high-dimensional and complicated optimization problems. In order to enhance the search performance of the DE algorithm, this paper uses the idea of niche, decomposes the entire population into several niches according to the fitness, perform population selection by integrating the optimum reservation strategy to realize the optimal selection of niche, adjusts the fitness of the individual of the population, designs the adaptive crossover and mutation operators to make the crossover and mutation probabilities change with the individual fitness and enhances the ability of DE algorithm to jump out of the local optimal solution. The experiment result of benchmark function shows that the method of this paper can maintain solution diversity, effectively avoid premature convergence and enhance the global search ability of DE algorithm.