Research And Implementation Of Intrusion Detection System Based On Neural Network

After entering the 21 st century,the development speed of the mobile Internet is beyond the expectations of people,the Internet improves people's living conditions at the same time,the problem of network security puzzles people's lives,and network intrusion occurs frequently.How to detect network intrusion effectively has become a hot topic in the field of network security.At present,the network security protection technology and the network security protection method are difficult to meet the strong demand of the modern people on the network security performance.The most popular in recent years is the intrusion detection technology,which can help the traditional network security protection technology.Although the intrusion detection technology is gradually improved,the detection rate is low,and the error rate is incorrect.A high level of disadvantage.In this paper,the genetic algorithm,Kohonen network and pole-speed learning algorithm are applied to the intrusion detection,so as to improve the detection performance of the intrusion detection system.The main contents are as follows:(1)In this paper,some KDD99 data are selected as the main data set,which is composed of 41 dimensional feature attributes.In order to facilitate the processing of neural network,the non-numerical data are digitalized and the data are normalized.Based on genetic algorithm,the reciprocal of error square sum of test set data is selected as fitness function,the dimension reduction of data is carried out,the redundant features are screened out,and the data before and after dimension reduction are input into the classifier constructed by BP.algorithm to verify the effect of data dimension reduction.(2)By adding the output layer after the competitive layer of Kohonen to make it a supervised learning network,the problem of slow convergence speed of Kohonen network is effectively alleviated.The extreme speed learning algorithm is used to optimize the network weights between the hidden layer and the output layer of S-Kohonen algorithm,which reduces the training time of the model and makes the algorithm more efficient.Some KDD99 data after dimension reduction are used to simulate the intrusion detection method in this paper.The experimental results show that the proposed method can improve the time and false alarm rate to a certain extent.(3)Based on the intrusion detection model of this paper,an intrusion detection system based on neural network algorithm is implemented.The system uses Wireshark network data acquisition tool to capture the data packets in the network traffic,extract the effective features by analyzing the transport layer and protocol type in the packet,and predict whether there is malicious intrusion behavior through the designed intrusion detection model,and reflect it to the user in time.