Research On Design And Application Of Heuristic Genetic Algorithm

Genetic algorithm is simulation of natural biological evolution theory-survival of the fittest,a stochastic search algorithm fittest genetic mechanisms obtained,also belong to the heuristic algorithms.Since the 1990 s,genetic algorithm has been widely developed and applied,has been widely used in combinatorial optimization,many areas of signal processing,machine learning,adaptive control and artificial life.Algorithm itself also has a strong plasticity,in recent years been more with other algorithms such as simulated annealing algorithm,the smallest collision algorithm,greedy algorithm combined with the use.This paper describes a genetic algorithm in mining protein PPI network,application and algorithm N-queens problem and the shortest path problem of improvement.First,unlike other nodes based clustering method is proposed based on clustering links(edges clusters)heuristic algorithm-genetic algorithm GA(Genetic Algorithm).Because of the unique relationship between the links(edges)usually represent a node,using the link clustering will find links to groups with the same characteristics.Experiments using yeast protein data DIP,obtained through individual coding,re-evaluation of the individual,the middle after iteration select populations,and ultimately select the best individuals,and to obtain a protein complex prediction set by decoding.The resulting prediction set with two standard composite data set matching calculation,and the other two algorithms are representative of performance analysis and comparison.Experimental results show that the algorithm can recognize the standard composite data set more known proteins,in the calculation of the recall rate,aspect harmonic mean better performance.Secondly,the study N-queens problem,the search for the genetic algorithm efficiency will increase as the index of the state space and reduce the weaknesses,the greedy algorithm is proposed initial population,and after the genetic algorithm mutation operator to add the minimum cross-collision algorithm,thus ensuring good progeny effect,improving overall efficiency of the algorithm.Finally,the shortest path problem,a plurality of parallel traveling salesman problem for the test experiments proposed hybrid parallel genetic algorithm based on coarse-grained and fine-grained,and proved that the hybrid parallel algorithm in finding the right path,and the minimum time efficiency than the individual coarse-grained or fine-grained parallel genetic algorithm for parallel good effect.