Study Of Hybrid Genetic Algorithm And Compound Form Method Solving Engineering Optimization Problems With Continuous And Discrete Variables

Researchers have applied Genetic Algorithms in optimization with discrete variables for global search in different fields. Little work has been done about the combination of Genetic Algorithm and the compound form method. This thesis puts forward a new effective hybrid genetic algorithm with compound form method dealing with hybrid engineering optimizations with continuous and discrete variables. The algorithm tries to search the global optimum more effectively. It avoids the inherent difficulty for genetic algorithms to carry out the local search and the problem of generating initial compound form and mapping failure for the compound form method.The combination of Genetic Algorithms and compound form method leads to complementarities. Genetic Algorithm can search the solution globally including many local optimums. But if the search space is too large, Genetic Algorithms have inherent difficulties to converge in the global optimum with an adequate precision. On the other hand, the compound form method follows a different strategy, being able to find local optimum with great precision. The interaction of both algorithms improves the solutions gained by Genetic Algorithms and makes the searching converge to the global solution with fewer iterations. Several numerical examples including one engineering project are carried out to verify the effectiveness of the algorithm.