Research On The Optimal Problem For Promotion Combination Based On Genetic Algorithm

The genetic algorithm is a kind of probability searching method which simulates the natural evolution. As one probability searching method in Systems engineering, it is simple and easy to implement, especially doesn't need the special field knowledge and only uses the fitness function as the assessment index and, comparing with the traditional optimized algorithm, it is more practical. Therefore, since appearing in 1970's, it has arose many scholars' attention and favor. Now, it's widely applied in the fields of combination optimizing, system controlling, machine learning, and data warehouse and so on, and has got plentiful fruits.This paper mainly carries the analysis and researches on the genetic algorithm and its mixed strategy in the application of promotion combination's optimizing, expecting to enhance this topic's research to a quantification analysis level. The content is mainly composed by three major parts. The first part is literature summary, and in this part, it first outlines the basic theory of the promotion, promotion combination and domestic and foreign research situation, and then, narrates the genetic algorithm and its research development, including genetic algorithm's simple introduction, the basic technology as well as the appraisal and the application. The second part is the model construction and algorithm designing part, and also is this article's emphasis and innovation point. In this part, it constructs the promotion combination's optimization model by abstracting the prototypical question, based on which, combining the pattern research algorithm, it designs a hybrid GA to solve this model by choosing the appropriating encoding method and setting genetic operator. The third part is the example application part. Referred to a beer factory's data, it describes how to deduce the concrete model by obtaining its parameter and runs the model in matlab. Finally, through analysis, it gives some reasonable suggestions about using this model and hybrid GA and points out some insufficiency and limitation that needs further improvements.