混合整数优化问题的差分进化算法研究

Research on Mixed-integer Optimization Problem Based on Differential Evolution Algorithm

为求解混合整数优化问题,提出了混合整数差分进化算法(Mixed Integer Differential Evolution,MIDE)。该算法结合整数变量的特点,为整数类型变量设计了专用的变异算子,使整数变量可以在差分进化算法中直接进化;为了维持种群多样性,采用了灾变策略;采用双编码方式,使整数变量与连续变量并行进化,进而提出了混合整数差分进化算法。通过与其它混合整数优化算法的比较,证明该算法具有较好的收敛速度、全局收敛性及算法稳定性等优点。

In order to solve the mixed integer optimization problem,a mixed integer differential evolution(MIDE)algorithm is proposed.Combined with the characteristics of integer variables,the algorithm designs a special mutation operator for integer variables,so that the integer variables can evolve directly in the differential evolution algorithm.In order to maintain population diversity,catastrophic strategies are employed.The double coding method is used to make the integer variable and the continuous variable evolve in parallel,and then a hybrid integer differential evolution algorithm is proposed.By comparing with other mixed integer optimization algorithms,it is proved that the proposed algorithm has the advantages of good convergence speed,global convergence and algorithm stability.