Cultural Algorithm Based On Artificial Fish For Global Optimization And Their Application

This paper focus on the research of Cultural Algorithm based on Artificial Fish and its application in the location of the distribution centers. Its engineering background is a complex continuous facility location problem. From the mathematical model of perspective, the distribution center location problem belongs to multi-source Weber problem, which has the characteristics with NP-hard; from the optimization point of view, it belongs to the scope of global optimization; from a practical point of view, it has a wide range of applications value and far-reaching practical significance.Since the last century facility location problem has been closely watched by academics and engineers.At present there have been a series of sophisticated location model and algorithm, such as the center of gravity method, Baumol-Wolfe model, mixed integer programming model and evolutionary computation, etc. Because this study is the continuous location problem, The Cooper proved that the mathematical model of the problem is neither convex nor concave function, there may be also a large number of local optimal solution, So it is very important to choose a better method for global optimization. This paper has research on the characteristics of artificial fish algorithm and culture algorithm, it is found that Artificial fish easily trapped into local optimization algorithm and also it has a slow convergence in the late search. In order to overcome this shortcoming, this thesis proposes A Cultural Algorithm based on Artificial Fish, and it is also used to solve the continuity of distribution center location problem.The main work of this paper is as follows:(1)Cultural Algorithm based on Artificial Fish for global optimization is proposed in this thesis, by inserting artificial fish and global explorations into a new population-based framework. In this algorithm, AFSA is used to achieve local optimization, combined with the artificial fish's global jumping strategy. This algorithm performs better global search ability. The search knowledge of the belief space which is updated by behaviors of individuals and genes is used to actively guide the search. When this algorithm's search process runs slower or it's search results at a standstill state, this thesis can make a greater search precision which employs influence strategies taken by Gaussian mutation.(2) Making numerical simulation experiments of the logistics distribution center location problem. First this thesis analyzes the current situation of distribution center location problem, then at the lowest distribution cost of criteria, establishing the simplified mathematical mode, and proposes a algorithm solving distribution center location problem, combined CA-AF and ALA. At last comparing the simulation results with AFSA and GA algorithm, the results show that the feasibility and effectiveness of this algorithm, and its performance of global search are better than the other two's. This algorithm can provide decision makers with an effective optimization tool.This subject and method provides a valuable reference value for In-depth study location problem.