The Research Of Optimization Control Method On The Managed Pressure Drilling

Managed pressure drilling(MPD) technology is a deep stratigraphic oil and gas exploration drilling technology through the precise control of the bottom pressure. However, faced with increasingly complex drilling conditions, conventional pressure control methods have been difficult to meet site requirements. For this problem, the article designs model-free adaptive control method, neural network predictive control method and neural network PID control method according to the MPD experimental device for researching the performance and application significance of the MPD.(1) We create the MPD experimental system and design the automatic control system with the PC systems made up by the KingView and MATLAB. The experimental device and its automatic control system is the basis of the research of optimization control method in MPD.(2) We design the model-free adaptive control method for MPD. The method calculates the pseudo-gradient vector by the real-time bottom pressure data to get the current optimal throttle opening degree. This real-time optimization method can achieve a good adaptive control for bottom pressure. The experimental results show the effectiveness of model-free adaptive control method in the MPD applications.(3) We design the neural network predictive control method for MPD. The method predicts the future bottom pressures by the neural network model, correct the predictive value based on the error of the predicted value and the actual value, calculate the throttle opening degree according to the reference trajectory. The experimental results show the effectiveness of neural network predictive control method in the MPD applications, and it has better control effect than the model-free adaptive control method.(4) We design the neural network PID control method for MPD. The method could adjust the three control parameters of PID on time. So that it could follow the system changing to achieve the optimal throttle opening degree, achieve the best control for bottom pressure. The experimental results show the effectiveness of neural network PID control method in the MPD applications, and its performance is significantly better than the model-free adaptive control method and neural network predictive control method.Overall, the MPD pressure control research based on model-free adaptive control, neural network PID control and neural network predictive control shows that the three kinds of advanced control algorithms has good control effect in MPD, which has a certain reference significance for the research of optimization control in MPD.