The Negative Selection Algorithm has been applied in the field of fault diagnosis successfully. To generate more excellent detectors and improve the algorithm’s performance, in this paper, a new method based on the dualoriented variation of the immune memory set for generating fault detectors is proposed. To reduce the redundancy information, this article defines an immune memory set and puts the self matching detectors or mature detectors into the immune memory set, and then performs dualoriented variation to the detectors. The proposed method utilizes real number coding. The simulations demonstrate that the proposed method provides better results of fault diagnosis and lowers the  omission rate. Moreover, the average running time is significantly shorter. From this perspective, the theme of this academic study has a particular theoretical value and is pragmatically significant.