基于微分进化算子的量子粒子群优化算法及应用

Improved Quantum-behaved Particle Swarm Optimization Algorithm Based on Differential Evolution Operator and Its Application

分析了量子粒子群优化算法(Quantum-behaved Particle Swarm Optimization,QPSO)的进化方程,指出其存在的局部收敛问题,通过将微分进化(Differential Evolution,DE)的基本操作思想引入到QPSO中,提出了改进的QPSO算法(QPSO-DE);算法改进的方法是在粒子搜索过程中,以一定的概率对粒子的每一维执行微分进化操作,以增加粒子的随机性,从而减少了粒子群体因多样性缺失而易于陷入局部最优或停滞的情况,增强了粒子群体的搜索能力,提高了算法的优化性能;对多个标准测试函数及在IIR数字滤波器优化设计中的仿真实验结果表明,与PSO算法和QPSO算法相比,QPSO-DE算法能够取得更好的优化结果。

The evolution equation of Quantum-behaved Particle Swarm Optimization （QPSO） algorithm was analyzed and then the premature convergence problem in QPSO algorithm was pointed out. The idea of Differential Evolution （DE） was introduced into QPSO algorithm and the improved algorithm was proposed, which was called QPSO-DE. During the search procedure of particle swarm, every dimension of a particle executes the DE operator according to a certain probability. The DE operator in QPSO can increase the randomicity and enhance the particles＇ search abili（y and the ability of obtaining the optimal solutions. At the same time, the cases of trapping into the local minima and stagnancy in QPSO for the reason of diversity loss were decreased. The superior performance of the proposed algorithm was shown by comparing with PSO and QPSO on the benchmark functions and the design of fiR digital filter.