一种新型的多目标优化混合量子进化算法

Multi-objective optimization based on novel hybrid quantum-inspired evolutionary algorithm

针对复杂多目标优化问题,提出一种混合量子进化算法,并利用它求解多目标函数优化问题。该算法根据多目标优化的特点,创建外部集合保存历代搜索到的非支配解,利用其中的精英个体设计了一种旋转角自适应调整的量子门更新策略,并对量子比特表示的概率幅设置最大和最小阈值,以防止量子群体早熟收敛。借鉴量子门引入了专门针对量子个体的旋转交叉算子,同时小概率地对量子比特进行取反变异操作。对所提算法的计算复杂度进行了理论分析。与另一种已有的多目标量子进化算法的比较结果表明,所提算法具有更好的收敛性能、分布特性及求解效率。

This paper proposed a hybrid quantum-inspired evolutionary algorithm for complex multi-objective optimization problems,and it used to solve multi-objective function optimization problems.In consideration of the characteristics of multi-objective optimization,it used an external set to reserve the non-dominated solutions found so far.Making use of the eltism solutions in it,it gave a self-adaptive method for tuning the rotation angle in the quantum gate.Meanwhile,it set the maximum and minimum threshold values for the possibility amplitude represented by the quantum bits,which was to prevent premature convergence.It introduced a rotation crossover operator specially designed for the quantum individual according to the quantum gate.And it adopted a not-operation to mutate the quantum individual with a small possibility.It analyzed the computational complexity of the algorithm theoretically.Comparisons with another existed multi-objective quantum evolutionary algorithm indicate that the proposed algorithm has better performance of convergence,distribution and efficiency.