A novel detector generation algorithm for Real-Valued Negative Selection Algorithms, i.e. the PTS-RNSA, is proposed in this paper, which is based on the iterative Partition-Test-Spread process. Different from traditional detector generation algorithms that are randomized algorithms, the PTS-RNSA is a deterministic algorithm. When the number of the detectors is large enough, the PTS-RNSA can ensure to cover the whole non-self space except the boundary area between the self space and the non-self space. Experiments are done to compare the PTS-RNSA with the state-of-the-art algorithm, i.e. the V-detector algorithm. Experimental results demonstrate that the performance of the PTS-RNSA is very competitive. Especially, the time cost of the PTS-RNSA is much better than the V-detector algorithm.