Analysis About The Genetic Algorithm For Portfolio And Flow Shop Scheduling

Genetic Algorithm is a stochastic search optimization algorithm which is more used and effective. From 1960, a stochastic optimization technology called"Evolutionary Algorithms", which simulates the nature evolutionary process, shows its good qualities in solving the complex optimization problems which are hard done by the traditional optimization algorithms. Now the Evolutionary Algorithms has three research locations: Genetic Algorithm, Evolutionary Programming and Evolution Strategies. And Genetic Algorithm is most useful, it's not only used for single objective optimization problems, but also can better solve lots of multi-objective optimization problems.The paper studies and analyzes the theories and general applications of the Genetic Algorithms and it includes last contents.1. The paper introduces related biology knowledge and gives a brief review of the development history of EAs. Then the paper concludes the main characters of EAs and sums up the present situation of the theories and applications of EAs.2. Based on the classical Markowitz portfolio model, an improved portfolio is proposed for portfolio selection with minimum transaction lots, transaction costs and upper limit on the maximum amount of invested capital in any security. The portfolio selection modeling, as a nonlinear integer programming problem, is difficult with the traditional optimization methods. A genetic algorithm base on integer coding genetic operation is designed to propose the model. It is illustrated via a numerical example that the genetic algorithm can be used to solve the portfolio optimization problem efficiently.3. For the bi-objective Flow Shop scheduling problems where the objectives are taken to be the minimization of makespan and total tardiness time, a multi-objective genetic algorithm is proposed to obtain the Pareto optimal solutions efficiently. This method builds on the selection strategy of NSGA-Ⅱ, Computational experiment is performed on the test problems generated randomly, and the results demonstrate the efficiency and robustness of the suggested algorithm.