Application Of Computational Intelligence To Tobacco Formulating System

Since artificial intelligence based on symbolism advanced in 1956, it have been severed as a powerful research tool and applied in many fields successfully such as industry, agriculture, medicine and so on. But this method can do nothing about nonlinear and uncertain problems. Therefore researchers have to explore other approaches to solve these problems. Computational intelligence concluding artificial neural networks, evolutionary algorithms and fuzzy systems emerged as the times require. Its main characteristic is that can solve many hard problems without making complicated mathematical model. In fact computational intelligence provides artificial intelligence with a new way in many application fields.Cigarette Formulating is a typical uncertain and nonlinear problem because it involves complex data and technics. This thesis expatiates computational intelligence's basic principles and applications to cigarette formulating. The whole thesis consists of seven chapters.Chapter one introduces basic outline of computational intelligence, brings forward research content and significance. Chapter two points out the disadvantage of traditional tobacco classifying, puts forward a new method coupled experts' experience with Fuzzy Kohonen neural networks. Chapter three discusses the applications and researches on forecasting tobacco's sensory-quality and smoke using the integration of Kohonen and BP network, which can solve the hard problem of sensory-quality assessment and smoke measuring in tobacco product. Chapter four explains the method of tobacco formulating maintenance and design by genetic algorithm, and this method fill up vacancy in domestic tobacco industry. Chapter five describes the integration and relation of intelligent method used in the system. Chapter six analyzes the whole system structure, and illuminates designing process and data flow chart. In the end of this chapter, a part of running result and contrastive analyzing were listed. The last chapter concludes the whole thesis and explores the further developments of computational intelligence.