A Class Of Novel Evolutionary Algorithms For Multi-modal Functions

Problems in many fields can be summed up to optimization problems .There have been many classical methods with good results for some problems .But these traditional optimiztion algorithms have many defaults ,for example:they are restricted by continuity ,the results depend on the problem and the initial value, and they will stuck into local optima easily.In recent years,evolutionary algorithms have been applied in many fields.There are many results for functions optimization because algorithms'capacity of global optimization,high efficiency and parallelism.Compared to numerical optimization,Evolutinary algorithms have many strongpoints. They based on population and stochastic searching mechanism and attract many people.Various of evolutionary algorithms emerge in endlessly. The reprensentives are GT(Guotao algorithm)and PSO(Particle swarm optimization algorithm),which have high effeiciency and rapid speed.Most of researchers pay too many attention to singlet function ,but in our real life,many mathematical problems and engineering problems are multi-modal functions,for example:the problem of neural networks structure and weight optimization problem.For such problems,we can compute repeatly till servnal peaks are found,but it not only wastes time but also can not gunaratee to get all optima. In this paper,firstly we introduce the chaotic optimization into GT algorithm to improve the local searching capacity.Then improved PSO is introduced .Compared with PSO,the novel algorithm can get all optima neither one.Lastly, we combined GT and PSO and apply it into multi-modal function optimization to get all sohtions.The results show it perfomance well.Abstract:GuoTao algorithm,Particle swarm algorithm,chaotic opitimization,multi-modal functions...