In this paper, the Particle Swarm Optimization (PSO) is used to optimize the randomly generated detectors in the Negative Selection Algorithm (NSA). In our method, with a certain number of detectors, the coverage of the non-self space is maximized, while the coverage of the self samples is minimized. Simulations are performed using both synthetic and real-world data sets. Experimental results show that the proposed algorithm has remarkable advantages over the original NSA.