Fault Diagnosis of Subsea Blowout Preventer Based on Artificial Neural Networks

Subsea blowout preventer is an important tool for ensuring safety of drilling activities and rig personal. In case of faults, it might cause severe damages to the environment and oil companies. This paper presents the method to perform fault diagnosis of subsea blowout preventer (BOP) based on artificial neural network (ANN).BP ANN of the BOP are proposed on the basis of the typical faults of the BOP in the process of opening and closing. In order to obtain higher training speed and precision, BP ANN are improved with gradient descent with momentum and adaptive LR gradient descent methods. Besides, RBF network is also presented for getting a better model for diagnosis. Compared with BP network, RBF network has better performance concerning training speed and precision in this case. However, BP network will show stronger flexibility in the complex model with plenty of fault types.