Research On The Particle Swarm Optimization For Solving The Problem Of Cotton-blending

Automatic Cotton-blending is a process that makes different kinds of cotton with different parameter of property together. Automatic Cotton-blending issue is a multi-constrained combination optimization, with a very high computational complexity. From a theoretical point of view for the calculation, it is an NP hard problem. An automatic Cotton-blending problem of difficulty lies in the constraints of complexity. How to meet the constraint condition of circumstances to find the optimal solution is what it differences form general combinatorial optimization problem of key points.Particle Swarm Optimization (PSO) is a new type of Swarm intelligence algorithm, forward by DR R.C Eberhart and DR Kennedy J in 1995. PSO comes from researching of bird's predator-prey behavior, is a recently developed evolutionary algorithm. System initials a group of random solutions, and searches the most optimal value through iterative. It is widely applied in function optimization, neural network training, data mining, and fuzzy system control and other fields.This article first studied the issue of Cotton-blending under multiple constraints, and used mathematical models to express them. Three commonly used intelligent algorithms in combinatorial optimization problems of were introduced briefly. They are genetic algorithm, ant colony algorithm and particle swarm optimization. It shows the basic principle and algorithm characteristics of the three algorithms. According to the characteristics of automatic cotton-blending, the article reposed the particle swarm algorithm as way for solving the automatic cotton-blending issue.At last, by studying the automatic cotton-blending achieved by the particle swarm algorithm, and by the experimental results, the article proposed two particle swarm optimization approaches. One was to integrate the genetic algorithm and particle swarm algorithm. And the other is the use of dynamic parameters and the simulated annealing in particle swarm algorithm. By comparison the experimental result of the two algorithms, the article got a conclusion:The convergence, the pros and cons of the result and the computational efficiency of the two algorithms has been significantly improved. But both of them have their own strengths in certain areas. Specifically, when the scale of the problem was not big, the former had a better performance. With the scale of the problem expanded, the advantage of the latter gradually revealed.