Research For Multi-species Predator-prey Cellular Genetic Algorithm

With the development of science and technology, the mathematical optimizationproblem in engineering applications becomes increasingly complex. Such problemsusually have these characteristics: very difficult in optimization, wide search space,complex objective function, which is a technical problem that the traditionalmathematical methods can not effectively solve. Intelligent optimization algorithmopened up a new way for the typical complex problem solving. Intelligent optimizationalgorithms inspired by nature and biosphere, designing algorithms for problemaccording to its principle. The intelligent optimization algorithm have many notablefeatures, such as distributed, self-organization, collaboration, robustness, goodscalability and simple to implement, which makes it can solve a variety of complexissues quickly and reliably in the case of the absence of global information. As aprobabilistic search algorithm in intelligent optimization algorithm, it has beenresearched by a large number of scholars and achieved significant results. The cellulargenetic algorithm combines genetic algorithm with cellular automata, the spatialstructure of the environment and local complex interaction of population are introducedinto genetic algorithm which increase the ability of retain population diversity. The ruleof evolution in cellular genetic algorithm is random with great blindness. Predator-preycellular genetic algorithm made improvement that predator-prey mechanism isintroduced into cellular genetic algorithm. Predator-prey mechanism is inspired by thetheory which maintains prey and predator ecological balances in nature. Withpredator-prey mechanism replaces the original evolution rules, making the retention ofindividual genes to be more intelligent and better performance in population diversity.Predator-prey cellular genetic algorithm is a co-evolution algorithm that achievesoptimization goals by the interaction and restraint of two species. Existing co-evolutionalgorithm failed to distinguish different species in genotype and can not simulate thenatural genetic characteristics and behavioral features.This paper mainly from the point of species concept to study predator-prey cellulargenetic algorithm, defining different species from genotype and using appropriategenetic manipulation for each population, the main contents are as follows:(1) The relationship between predation theory and optimization algorithm is researched in this paper, predation mechanism is used to replace the evolutionary rulesin CEGA which simulates the natural predator-prey relationship. During theevolutionary process, the population size of the two populations is controlled in adynamic balance state in order to coordinating the two global exploration and localoptimization. The experimental results show that predation cellular genetic algorithmoptimization performance.(2) A novel multi-species predator-prey cellular genetic algorithm with linearmapping is proposed. A mapping matrix is applied into the calculating process ofpopulations’ fitness, which changes the mapping relationship between genotype andphenotype and makes different species carry different genetic information. In the courseof evolution, species uses different crossover method and an adaptive mechanism whichused to adjust the mapping matrix coefficients based on the dispersion degree ofpopulations is applied to enhance the ability of escaping from local optimum.(3) The principle of new algorithm is analyzed in this paper, the contents of thestudy include: how the linear mapping changes the solving difficulty of targetproblem、the process of escaping from local optimal solution、the way of speciescrossover、control of the population size. The result of analysis shows that when theoptimal solution of the objective function in the mapping area, the mapping operationcan reduce the difficulty of algorithm optimizes for the objective function; the mappingoperation designed in this paper is conducive to algorithm to escape from local optimalarea; the two genetic ways designed for preys have their own characteristic, they arecomplementary to each other; population size control of two species can effectivelymaintain the dynamic balance between prey and predator group.