The Research Of The Population Migration Algorithm Based On Artificial Fish Swarm Algorithm

Population Migration Algorithm (PMA) is a new kind of Simulated Evolutionary Algorithm (SEA) proposed recently by Zhou Yonghua and Mao Zongyuan, which simulates the principle of Population Migration. PMA is different from other SEAs, because it set up the model through simulating the principle in society that people migrate along with economic center and disperse as population pressure increases instead of some optimal characteristic of groups in nature. But because of the randomness of population's movement affect the performance of Algorithm. Artificial Fish Swarm Algorithm (ASFA) is an animal's autonomous method that bases on the principle of artificial intelligent. It has some characteristics, such as has no restrict demand on the object functions, being insensitive to the initial values, tolerating wide range of values of parameters, intrinsic parallel processing nature and global search ability. Its swarm intelligence also contributes to improve the performance of the algorithm. How to use the swarm intelligence of AFSA to conquer problems caused by the randomness of population's movement in PMA is the key point of this thesis.The major contributions are as follows:A new search of mechanism is proposed for the visual effect to the convergence of the ASFA. From the analysis of experiments, it is obvious that when searching area contracts, the convergence is improved greatly. The numerical experiments show that the mean iteration generation and the least successfully iteration generation of the proposed algorithm is less than that of ASFA. And the improved Algorithm shows better local search ability and better convergence stability.Aiming at the low searching efficiency led by the movement of population, we use the behavior of prey in AFSA instead of the movement of population in PMA, and a PMA based on AFSA is proposed. Then we describe the evolution of the hybrid scheme as an abstract stochastic process, and conclude that hybrid scheme is essentially a kind of SEA.Finally ,we revise the original definition of the properties of selection operator, and compute the concrete characteristic numbers of all operators, Further more, by using the properties of operators, we prove that the hybrid scheme with "parent-offspring competition strategy" is convergent with probability, we also prove the convergence of the new hybrid scheme which is based on the ASFA and the PMA .At last, numerical experiments not only show the effectiveness of the hybrid scheme, but also make known that the setup of parameters provided in this thesis can well improve the convergence rate of the hybrid scheme.