The Improvement Of Genetic Algorithm In Multiple-tables Query Of Database

In a relational database system.the database performance seems more and more important in the database application, especially the query efficiency.Modern database contains more information which usually leads to the data are too large. As a result, complex queries of the large data including join multi-table,GROUP BY,etc,are frequent operation. Meanwhile,the joins of multi-table increase the number of the relationship between the tables.With the increase of the number of the relationship,the number of query execution plans increases exponentially,which leads to the extreme expansion of search space and the great computation complexity.Therefore, the optimization of multi-table queries becomes a very important research topic in the database optimization areas.Though traditional database has optimized the data manipulation and query processing, in most cases,its optimization algorithm is mainly exhaustivity.Such search algorithm efficiency is very low,When the database queries have a large number of the relationships.Multi-database query optimization problem is a NP problem,it's a better method to use the heuristic or random algorithm. When face to database multi-table query problem,the genetic algorithm is more excellent.Compared with traditional methods,the optimization speed and effect of genetic algorithm all had very big enhancement,but genetic algorithm will appear "premature" phenomenon and plans to generate longer defects when optimize problem.In order to make multi-table query more optimization,this paper apply immune genetic algorithm for improved genetic algorithm.In the study of immune genetic algorithm, some researcher has lead memory cell of immunology into this algorithm.Because immune genetic algorithm memory cells can store the better individuals when algorithm iteration and in addition to the next iteration of new generation of populations, thus improving the algorithm convergence speed. But because of the excellent individual to join, the diversity of population will be destroyed easily, so as to achieve local convergence rather than global convergence, so this paper improved the immune genetic algorithm. Training with a vaccine inoculation process, namely parallel genetic algorithm replace memory cells, the purpose is the antibodies to the original memory cells for crossover and mutation genetic operation, thus enhanced the vaccine to the new generation after the diversity of population, effectively prevent "premature" phenomenon and local convergence. Although this will increase the algorithm executive price in some extent, but its convergence rate is greatly improved, thus the plan to generate time diminishes quickly, so this algorithm can achieve the purposes of database multi-table query optimization.