Abstract: The study analyzes the characteristics of various types of Web application system security vulnerabilities. Based on the deformation of Web attack, according to the principle of Web application vulnerabilities occur, attack methods and targets, the attack characteristics is extended, Proposed a structural model of intrusion prevention based on the feature recognition. The experiments showed that the feature recognition-based Intrusion Prevention System can ensure higher performance in high-speed attack traffic network environment.