Research on oil rig automatic feed drilling system based on MOBP neural network

This paper focus on AC frequency conversion electric drill of automatic feed drilling system and an MOBP neural network is used to control the WOB (weight on bit) and realize the research of constant pressure automatic feed drilling. The same type of high quality wells´ drilling parameter is normalized as a network training set. A more effective optimization algorithm called momentum method is used to design a suitable improved BP neural network for automatic feed drilling system. Modular neural network is established by Matlab/Simulink and compared with conventional PID controller and Fuzzy controller. The simulation results show that in the condition of hysteresis, MOBP has better stability, better robustness and smaller steady-state error than conventional PID and Fuzzy control. The application of Neural Network in automatic feed drilling system has a significance of guidance to improve the performance.