摘要
考虑到常规BP神经网络算法容易陷入局部最优解,所建立的网络遗传流量检测模型检测效率低,准确率不高等问题,提出一种改进型GA优化BP神经网络算法,并使用其建立网络遗传流量检测模型。常规遗传算法在搜索过程中,往往会由于出现影响生产适应度高的个体而对遗传算法搜索过程产生影响的现象发生,因此需要对常规遗传算法进行改进。使用的方法是通过混合编码方式进行改进,同时对交叉算子、变异算子、交叉概率以及变异概率等参数进行优化修正。使用KDD CUP99数据库中的网络异常流量数据进行实验研究,研究结果表明,所提出方法的检测性能要明显优于常规算法,其对BP神经网络的结构、权值以及阈值进行同步优化,避免了盲目选择BP神经网络结构参数带来的问题,避免了常规BP神经网络容易陷入局部最优解的问题。
Since the conventional BP neural network algorithm is easy to fall into local optimal solution,and the established network abnormal flow detection model has low detection efficiency and poor accuracy,an improved BP neural network algorithm based on GA is proposed,by which the network abnormal flow detection model was established. The conventional genetic algorithm in the search process often influences on the search effect because of the high fitness individual,so it is necessary to improve the conventional genetic algorithm. The hybrid encoding mode is used to optimize and correct the parameters of crossover operator,mutation operator,crossover probability and mutation probability. The experiment study for network abnormal flow data in KDD CUP99 database is conducted,and the research results show that the detection performance of the proposed method is better than that of the conventional algorithm. The improved genetic algorithm is used to synchronous optimization the network,weight and threshold of BP neural,and can avoid the problem causing by the blind selection of BP neural network structure parameter,and avoid that the BP neural network is easy to fall into the local optimal solution.
出处
《现代电子技术》
北大核心
2016年第3期90-93,共4页
Modern Electronics Technique
关键词
网络异常检测
BP神经网络
遗传算法
异常流量
network anomaly detection
BP neural network
genetic algorithm
abnormal flow
