An Improved Particle Swarm Optimization Algorithm And Its Application To Integer Programming And Reliability Problems

Optimization exists in many industrial engineerings, and it is very important to our everyday life. Optimization aims at finding the best plan for a problem so that the problem under consideration can be handled with the smallest cost and the highest efficiency. Optimization is ubiquitous in our everyday life, such as integer programming and reliability problems. Over the past decades a number of optimization algorithms have been used extensively on solving these problems.Many engineering optimization problems can be represented as integer programming problems, such as general assignment problems, production scheduling and resource allocation. An improved particle swarm optimization (IPSO) algorithm is proposed to solve integer programming problems in this paper. The IPSO makes use of three updating strategies to improve the velocity updating of particle swarm optimization (PSO) algorithm, and these three strategies are stable updating, conservative updating and radical updating, respectively. In addition, a new inertia weight is introduced into the velocity updating, and it is used to balance the global search and local search. We devise a normalized penalty function method to handle constraints. Experimental results show that the IPSO algorithm has stronger convergence and stability than the other two PSO algorithms on solving integer programming problems. The IPSO can be an efficient alternative for solving integer programming problems.A design engineer often tries to achieve the highest reliability for a system, for a reliability design of high quality enables a system to work more safely and efficiently. Generally speaking, the system reliability is limited by several constraints such as cost, weight, and volume. When designing a system, an important issue is how to obtain the highest reliability without violating any constraint. That is to say, there is a difficulty in finding a balance between reliability and other resource constraints. The IPSO algorithm is also used to solve reliability problems in this paper. We adopt a usual penalty function method to trade off the objective function value and constraint violations. Experimental results show that the solutions obtained by the IPSO are better than the previously reported best-known solutions in the recent literature.