Hybird Genetic Algorithms On Solving Multi-Peak Optimization Problems And Application

Genetic Algorithm (GA) is a highly parallel, random and adaptive searchingprobabilistic method based on the mechanics of natural selection and genetic,whichdoesn't need special domain knowledge besides simple in design, strong robustness,etc. Itmade an amazing progress and has permeated to many fields. But the theory and methodof GA are not mature, algorithm needs for further improvement and consummation.Optimization problem is a common example of GA. So, this paper chooses multi-peakoptimization questions based on Hybrid Genetic Algorithms (HGAs) as all object of study.It analyzes and studies for combining GA with classical optimization methods andinspired optimization methods. Aim in gat the deficiency of general GA on solvingcomplex multi-peak optimization question, the paper proposes two HGAs. Finally,thepaper selects some typical complex multi-peak functions to have a test and compares. Thesimulation results demonstrate that the proposed algorithms gall not only preventpremature convergence effectively,and strengthen algorithms' local search.Firstly, it presents a HGA according to the short of GA's local search and prematureconvergence, which uses real code, and selects and retains the solutions from memoryoperator in improved trust region method to make HGA achieve the local solutions tomake up for the short above.Secondly, it proposes a HGA according to the short of GA's local search andpremature convergence, which starts GA firstly to format initial information pheromone todistribute for ant algorithm, then makes use of ant algorithm to search the local solutions,which could avoid the stagnation phenomenon of algorithm in search and enhancealgorithm's local search.Last, the HGA is applied to correct judgment matrix consistency in enterprise'sdecisions, it has high effective and high efficiency by examples and methods.