An Efficient Binary Differential Evolution with Parameter Adaptation

Abstract

Differential Evolution (DE) has been applied to many scientific and engineering problems for its simplicity and efficiency. However, the standard DE cannot be used in a binary search space directly. This paper proposes an adaptive binary Differential Evolution algorithm, or ABDE, that has a similar framework as the standard DE but with an improved binary mutation strategy in which the best individual participates. To further enhance the search ability, the parameters of the ABDE are slightly disturbed in an adaptive manner. Experiments have been carried out by comparing ABDE with two binary DE variants, normDE and BDE, and the most used binary search technique, GA, on a set of 13 selected benchmark functions and the classical 0-1 knapsack problem. Results show that the ABDE performs better than, or at least comparable to, the other algorithms in terms of search ability, convergence speed, and solution accuracy.

Empirical Comparison of Differential Evolution Variants for Industrial Controller Design

“…The parameters to be adapted can be encoded directly in each individual and undergo mutation and crossover operations. 19 Good control parameter values will bring forth better individuals and propagate them to future generations. Another adaptive approach is to obtain feedback information from the search and update control parameters accordingly.…”