Application of Negative Selection Algorithm in the Fault Detection and Diagnosis of LRE

Anbo MING, Wei ZHANG, Ganyang TIAN, Wei ZHENG

The High-Tech Institute of Shang Xi, Xi’an, CHINA

Abstract: To overcome the obstacles existed in the fault detection and diagnosis of liquid rocket engine(LRE) such as the lack of real-time, in-time, and veracity, the negative selection algorithm of artificial immune system was introduced. By constructing the artificial recognition ball (ARB) and adopting the maximum similarity rule, the fault detection and diagnosis were carried out. The experiment was performed with steady state data collected in ground test then. In the multi-dimensional modal space, the fault property was discussed further. Faults surround the normal state in different distance. If the parameters deflect more from the values which were supposed, the fault will be more obvious and it will move away from the normal state in the ARB. Some faults are separable, but some are also lapped over partly. Results show that this method has the function of quickly detection, high rate of correction diagnosis, and powerful ability of discovering the unknown fault. It can be widely applied in the state monitoring and fault diagnosis of LRE.

Keywords: Liquid Rocket Engine; Artificial Immune System; Fault Detection and Diagnosis; Negative Selection Algorithm; Artificial Recognition Ball