The main issue of Intrusion detection systems is large computation, feature selection was introduced to solve the problem. According to the shortcomings of existing methods, this paper uses improved Particle Swarm Optimization to search optimal feature subset, proposes a feature selection method based on hybrid CatfishPSO and Least Square Support Vector Machine, uses combined CatfishBPSO and CatfishPSO to select feature subset and optimize the parameters of LSSVM simultaneously, and build a Intrusion detection model based on the feature selection method above. Experiments on KDD Cup 99 show that the model has a good detection performance.