关键词: Spark计算框架,粒子群算法,并行PSO优化神经网络,电力通信网络,安全态势预测

Abstract: With the expansion of the scale of electric power communication network,the electric power communication network continuously produce huge amounts of data communication.At the same time,the communication network attack means is in constant evolution,which brings threats for the safety of the electric power communication network.To solve above problems,combining with the Spark big data computing framework and the advantages of PSO,the Spark memory computing framework of parallel PSO optimization neural network algorithm is put forward to predict the security situation of electric power communication network.This study first introduced the Spark computing framework,the Spark frame has the characteristics of memory computing and quasi real-time processing,accord with the requirement of electric power communication big data processing.Then PSO optimization algorithm was proposed to modify the weights of neural network,in order to increase the study efficiency and accuracy of neural network.Then with the combination of RDD parallel characteristic,this paper proposed a parallel PSO optimization neural network algorithm.Through experiment and comparison,you can see that Spark framework based PSO optimization neural network algorithm has high accuracy,and compared with prediction method based on Hadoop,its processing speed has improved significantly.

Key words: Spark computing framework,Particle swarm optimization,Parallel optimization PSO neural network,Power communication network,Security situation prediction