The Application Of Diploid Adaptive Genetic Algorithm In Function Optimization

The genetic algorithm is a group intelligent optimization algorithm,which learnsfrom Darwin’s natural selection theory of species evolution and Mendel’s geneticmutation theory. It is specially situable for handling the complicated and nonlinearproblems.However, the genetic algorithm itself still has some defects and needs furtherresearch and improvement.Based on the diploid genetic algorithm, this paper not only keeps genetic groups ofbiological diversity betterly,but also joins the adaptive operators,then formed diploidadaptive genetic algorithm. The advantage of this algorithm is that it can adapt to changesbetter in the environment and have stronger local search capability. It has both the global searchcapability of diploid genetic algorithm and rapid convergence ability of the diploid adaptivegenetic algorithm. To improve the design process of dominant and hidden code in theprocess of implementation of the algorithm, this paper improved the design process ofdominant and hidden code and the process of the show recessive operation of father generationgenotype chromosomes and put forward the binary dominant mapping on unit point which makesthe operation process of father generation to genotype simpler and more feasible, and avoids thediscontinuity of unnecessary code. In the selection process,to ensure the former generated dynastieswhich has high order, long distance, high average fitness not be destroyed,to ensure the combinationprosess of fine mode,we introduced the best retention policy on the basis of the method of choice inproportion,to aviod damage the best one in the historical records,which appered in the formerdynasties accumulation process accidently. To ensure that the genetic information exchange fully,andto exchange the information of father generation genotype chromosomes sufficiently, we introducethe concept of exchange in the father generation genotype chromosomes,then exchange the geneticinformation at a certain probability,so as to match a good individual betterly,and find out the optimalsolution of the problem. Through testing the optimization capability of standard geneticalgorithm, adaptive genetic algorithm, diploid genetic algorithm and diploid adaptivegenetic algorithm effect to multi-peak function which one-dimensional andmulti-dimensional variable, in the Visual C++6.0platform, we found that theimproved algorithm achieve the expected purpose from simulation and experimentalresults.