Research On Wellbore Trajectory Prediction And Optimization Design Method Of Trajectory To Be Drilled

With the increase of oil and gas exploration and development,precise well trajectory prediction and control are becoming more and more critical technology to restrict the success of drilling.Drilling borehole trajectory formed in the underground,drilling engineers only through monitoring sensor(MWD)measurement while drilling data to monitor the changes of borehole trajectory,measurement data cannot be real reaction and the actual situation of drilling drilling borehole trajectory and design track will happen deviation,trajectory adjusted frequently,these conditions restrict the drilling construction.Based on the previous research and related trajectory prediction methods,this thesis propose the method of well trajectory data cleaning and processing.Through the analysis of real-time engineering data and inclinometer data,the data cleaning,data fusion and missing value completion are carried out by using Python’s numpy and pandas toolkit.The well trajectory data processing flow and method are established.Four physical calculation models are summarized and applied by single well data.Combined with data processing and influencing factors of well trajectory,an intelligent prediction method of well trajectory is established by using two machine learning algorithms: Support vector machine(SVR)and recurrent neural network(RNN).The optimization design method of trajectory to be drilled was established.The optimization method was solved by elite reserved genetic algorithm(SEGA),which was quick and convenient in practice and could provide reference for field wellbore trajectory control.Finally,according to the well trajectory prediction method and the optimization design method of the trajectory to be drilled,the software for monitoring and adjusting the trajectory of horizontal wells while drilling is developed.