Genetic algorithm for optimization of tobacco-group formulas design

Aiming at the shortcoming of subjectivity, high cost and long period in traditional cigarette formula design, the new method based on the genetic algorithm (GA) and industry expert knowledge was presented in this paper to search the rational tobacco-group´s formula schemes in complex and huge solution space. In the meanwhile, intelligent evaluation models were established according to the relativity among tobacco-group´s physical-chemical ingredients, sensory-quality indexes and smoke indexes. Through this model, the recommended formula schemes were evaluated and relatively optimum formula schemes can be obtained. Through this process, a man-computer cooperated intelligence-aided formula design can be realized, which overcomes the bottleneck of traditional expert system in knowledge obtaining, quicken the searching speed and enhance the system performance