The gas regulator fault diagnosis based on PCA-RBF neural network

Abstract

In order to further improve the gas pressure regulator intelligent fault diagnosis system, RBF neural network (abbreviated RBF) algorithm [3] was employed to diagnose four kinds of fault status in the gas regulator system, then combined the actual gas regulator case was studied. The results show that fault recognition rate can reach 62.5%. For further enhancing the recognition rate of fault diagnosis, the introduction of the principal component analysis (PCA) and combination of PCA and RBF neural network technology based on PCA and RBF neural network were put forward in the gas regulator improved method of fault diagnosis, as provided a new method of fault diagnosis for gas regulator. The results of actual sampling data analysis and algorithm simulation have shown that PCA-RBF neural network fault recognition rate than RBF neural network fault recognition rate increased by 25%, thus proving the effectiveness method of fault diagnosis in gas regulator.