The Application Of Particle Swarm Optimization To Cigarette Blending Design

As a new research field, born in 1970, Swarm Intelligence is always the concern by many researchers. It has been a hot issue and front research field of Economic, Society, Biology, and Computer and so on. Swarm Intelligence provides a new method for people under the condition of no centralized control and provided no global model. Particle Swarm Optimization Algorithm is a kind of Swarm Intelligence Algorithm. It is created by Eberhart and Kennedy in 1995. This algorithm imitates birds and fish community looking for food in the migration and solves the real number problem through the coordination of global best value, personal best value and weight value. It is becoming a hot research issue in the past decade.The quality of cigarette is influenced by many factors, including fragrance and flavor, auxiliary materials, wrapping paper, filter tip and so on. Otherwise, the combination of tobaccos is the most important factor. Though the feeling quality of cigarette is decided by its chemical composition, examine the chemical composition is needed, the cigarette blending is based on the combination of tobaccos, what is need to optimize is the combination of tobaccos, not the chemical composition. This is also one of the complex roots of cigarette. The motivation of this article is to use PSO to solve the question like tobacco combination.PSO is used in this thesis to solve the problem of cigarette blending design. Personal best value is replaced by average value in PSO's velocity formula to improve particle's performance in looking for the best value. Then genetic strategy and chaos are introduced respectively to the PSO modified previously to improve PSO's performance and accuracy. The modified PSO in this thesis is better than genetic algorithm in solving cigarette blending design is found in the last part of this thesis through compare the results, which can provide better designs for the designers. The application research of PSO in this thesis can also provide a new method to solve problems which are similar to the cigarette blending design.