| With the fast development of Internet, searching in Internet is becoming the important way for people to get useful information. If we leak the sensitive personal or country information, it will be dangerous and threat to individuals and the country. And hacks that come along with the network have become a major pollutant to the social. Their intrusion serious is harm to the normal social order, even threat to national security. Therefore, intrusion detection system has become an effective tool for network security. The study of the intrusion detection system is the main of the thesis.As an important branch of the artificial intelligence, artificial immune system has attracted many more scholars'attentions as much as Neural Network and Genetic Algorithms. And as for theories in AIS, such as the negative selection and etc, they have dynamical, self-adaptation and self-study characteristics, which is suitable to introduce into the classifier training process in Intrusion Detection System. So, it is one of innovations in this paper that using artificial immune system's relative theory into Intrusion Detection System model.Research works in this thesis focus on the excellent properties and algorithms of AIS and the main technology in intrusion detection system. The main research is concluded as follows:Firstly, summarize and analyze the special concepts, basic immune methods, immune algorithms and immune model for machine learning in AIS, expatiate RLAIS immune model and its critical algorithm emphatically, and analyze its advantage and disadvantage.Secondly, analyze the intrusion detection technology based on the anomaly and misuse intrusion detection, study their specific classification methods and the advantages and disadvantages of each method.Finally, through the research of negative selection algorithm, apply the anomaly detection technology based on it to the actual project. |