Abstract: The traditional genetic algorithm may be premature in the case of insufficient data, it has a strong dependence on the search range, and the genetic algorithm with large search range often has a better performance. In order to solve the above problems, we can use the Spark mass storage and parallel computing ability to solve the genetic algorithm, implementing a coarse-grained parallel model, executing genetic algorithm selection, crossover and mutation operations in parallel using Spark. Executing genetic algorithm in parallel with Spark can greatly improve the search scope and parallel running speed. The application of improved genetic algorithm to the logistics and distribution problems and the experimental results show that compared with the serial and traditional parallel program, running time of genetic algorithm based on Spark is significantly reduced, and premature phenomenon is eased also.

Key words: genetic algorithm, distributed parallel computing, Spark, logistics distribution algorithm