Methods of the controllable - aberrance, random - aberrance and an idiographic dynamic clonal selection algorithm are put forward. Sets of suspicions computer virus are established, which contribute to the reduction in the mistakes (FPR) of the immune system. A virus detection model based on the idiographic artificial immune system is put forward and realized. Emulation experiment shows that the model can identify not only old virus, but also new transmutation virus and is a self-adapted, and diverse model for virus detection.