Research On Particle Swarm Optimization Algorithm And Its Application

There are lots of optimization problems in economic planning, engineering design, manufacturing, transportation, information processing, and many other fields, while optimization problem is the problem of finding the best one from numerous feasible solutions. To solve these problems effectively will not only have very significant meaning for entire society, but also bring great economic benefits. As an independent branch of mathematics, the objective of optimization is to solve the optimization problem and the specific method for which is called optimization method. Traditional optimization methods, such as Newton method, iterative method are based on mathematics and normally strict in the description of the problem. For example, the objective function and constraints of the problems must be continuously differentiable. With the development of advanced manufacturing technology, the practical optimization problems are becoming more and more complex, which makes the traditional optimization methods being unable to cope with. Therefore, it is an urgent need to explore some new optimization methods.Evolutionary computation originated from observation and simulation the natural or biological phenomena. Compared with traditional optimization methods, the bionic intelligent optimization methods represented by evolutionary computation are widely applied in scientific research and industrial production in virtue of their higher adaptability, robustness and parallel processing capability for various kinds of complex optimization problems. As one of the evolutionary algorithms, Particle Swarm Optimization (PSO) algorithm has been used successfully in many fields because of its features of being easily implementation and having fewer parameters. The single objective programming and production scheduling problems based on PSO are researched and some new algorithms are presented for solving these problems in this thesis. The performances of these algorithms are validated through a large number of experiments. The results demonstrate that the proposed algorithms can overcome premature convergence effectively, while the solution quality is also improved. The main contents of this thesis are as follows:1. Focusing on the single objective nonlinear programming problem, a hybrid method PSO-EM is proposed. This algorithm combines two heuristic optimization techniques: one is self-improvement ideas (each particle obtains knowledge through continuous information exchange) of PSO, the other is attraction-repulsion mechanism of Electromagnetism-like Mechanism (EM) algorithm. PSO and EM algorithm executes alternately, when PSO is completed, EM algorithm starts to run and exerts attraction-repulsion mechanism on the particle's best previous position (individual best), which will force the individual best move to the better position again. As a result, updating individual best will depend on not only previous global best position (global best), but also other particles'individual best. Thus, the convergence speed of this algorithm will be increased in this way. The experimental results show that the PSO-EM algorithm has an obvious improvement in whatever the convergence speed, convergence precision as well as success rates.2. An improved particle swarm optimization method (IMPSO) based on multi-population is presented for solving single objective nonlinear programming problem. The algorithm adopts uneven swarm division manner through which the whole swarm is divided into three sub-populations with different population size based on the talent hierarchy. Different speed updating strategies are adopted for different sub-population. The better-population aims to accelerate the convergence speed while the worse-population intents to explore the search space which has not been visited previously, so as to reduce the possibility of getting into local optimum. The objective of middle-population is to achieve a balance between exploration and exploitation in the search space. Moreover, the mutation strategy is introduced to execute refined search in local area, and the crossover strategy among different sub-populations maintains the diversity of the swarm to avoid premature convergence. The experimental results demonstrate that the proposed algorithm is both effective and efficient by comparing with some classic and new representative versions of PSO.3. Permutation flow shop scheduling problem is a typical production scheduling problem: a number of jobs will be processing on different machines, and each machine must process the jobs in the same order. A hybrid metaheuristic particle swarm algorithm HDCPSO which combines PSO and Iterative Greedy (IG) is proposed. The Destruction and Construction (DC) mechanism of IG algorithm is utilized for the mutating the particle, and the concept of individual hovering is also introduced to control when the particle mutate, which can prevent the swarm from stagnation and slow down the probability of premature convergence. In addition, an efficient population re-initialization strategy is adopted in this algorithm. Parts of the poor particles (small diversity or poor fitness) are re-initialized to maintain diversity of swarm and avoid premature convergence further. Moreover, an insert (shift) neighborhood search is introduced to improve the particle's searching ability and increase the convergence speed. Finally, the proposed algorithm is tested on different scale benchmarks and compared with some existing algorithms. The results have shown that HDCPSO is superior to other algorithms both in the solution quality and the stability.4. For the two-stage assembly scheduling problem, a discrete PSO algorithm (DPSO) is proposed. The two-stage assembly scheduling problem is an extension to flow shop scheduling problem. In this problem, the whole machine processing falls into two stages and each job must be processed on all the machines. The final operation must be conducted at the second stage and the last operation of each job can not start until all the operations at the first stage are completed. Firstly, this thesis redefined the particle velocity representation and modified particle movement accordingly. Moreover, individual strength is defined, which is used to control the particle's adaptive mutation, and mutation mode is decided by individual fitness. Furthermore, an exchange neighborhood search is introduced to enhance the local search ability of particle and increase the convergence speed. Finally, the effectiveness of DPSO is validated by some contrastive experiments on different scales. This algorithm has an advantage over other compared algorithms in solution quality and remains quite competitive in runtime.In recent years, research on PSO algorithm as well as its application has been paid more attention by many scholars, and numerous improved versions of PSO and new applications have emerged correspondently. This thesis mainly focuses on researching PSO for programming problem and production scheduling problem, and some more effective optimization methods are proposed. It is of both theoretical and practical significance of the research on the improvement of PSO algorithm, MAs and more application fields.