A hybrid genetic algorithm for multimodal function

There are some limitations that using generic algorithms to dispose multimodal function, so that this paper brings forward an improved hybrid genetic algorithm. The block crossover, hierarchical mutation and multimodal function searching are adopted, which based on the analysis of variance ratio. The improvement can not only expand the range of searching the individual with high fitness and accelerate the convergence rate, but also avoid the local convergence. Owing to analysis of variance ratio, optimal value and the tolerance of every parameter in problem are solved at the same time, which is very practical for actual engineering. Terminations based on the analysis of variance ratio can not only speed up the calculation but also avoid the slow convergence at the late stage of the traditional method. The hybrid coding of decimal and floating can fit in with the needs of the continuous variables and the dispersed variables in the actual engineering better. These above improved methods have passed the test of GA test functions successfully, which has better search precision, convergent speed and capacity of global search. Numerical result shows that this hybrid generic algorithm is high efficiency, less genetic generation, and high accuracy for multimodal function.