A hybrid discrete differential evolution algorithm for stochastic resource allocation

It is widely admitted that the differential evolution (DE) algorithm is a very powerful optimization method for continuous-valued numerical optimization. However, DE has seldom been used to solve combinatorial optimization problems. In this paper, a new hybrid discrete DE (HDDE) is proposed to solve stochastic resource allocation problems, e.g., the classical weapon-target allocation (WTA) problem which arises from military operations research. In HDDE, solutions are represented by the permutation of all resources to be assigned. In design of DE mutation operators, modulo arithmetic is adopted to revise the solution vectors which go beyond the value range, converting these solutions into feasible regions. As for the crossover operator, the same pairing in target vector and trial vector will be preserved to well inherit historical assignment results. In the process of removing repetitions and inverting omitted integers, stochastic permutations are generated. Besides, local search operator is added into the discrete DE so as to improve exploitation ability. Computational results show that the proposed HDDE performs better than the other DE variants.