Intelligent Modeling And Coordinated Optimization Of Rate Of Penetration And Mud Pit Volume In Geological Drilling Process

Exploration and development of geological resources can provide sufficient resources and energy security for the country’s economic and social development,and it is also an important support for national security.As the pre-procedure engineering of resource and energy exploration and development,geological drilling is faced with complex geological environment and technological process,nonlinearity,coupling and data pollution,which makes it difficult to accurately judge and adjust the changes in rate of penetration and mud pit volume.The efficiency and safety of geological drilling process need to be further improved.In view of the characteristics of the drilling process of geological vertical wells,considering the process and technology of the actual drilling process,this thesis uses data-driven modeling and intelligent optimization methods,focuses on the research of the modeling of rate of penetration and mud pit volume,the evaluation and decision-making of drilling system operating states and coordinated optimization between systems.The main research work and achievements of the thesis are as follows:(1)Rate of penetration(ROP)prediction model based on corrected data and support vector regression(SVR)As an important index for drilling efficiency,ROP has complex nonlinearity with drilling parameters,and abnormal data will affect the model accuracy.Aiming at these problems,a ROP prediction model based on corrected data and SVR is established to achieve accurate prediction of ROP and provide model basis for subsequent drilling optimization.First,the abnormal data is identified by local outlier factor.Based on the identification result,the nearest normal data is determined by Euclidean distance to realize the correction of abnormal data.Then the SVR method is used to establish the ROP prediction model,and the nonlinear relationship between drilling parameters and ROP is established.Further,a modified bat algorithm(MBA)is developed to improve the global search ability,which can solve the non-convex problem of model hyperparameter design,improve the accuracy of the ROP model and achieve accurate prediction of ROP.Compared with the five algorithms on the IEEE CEC2005 benchmark function,the superiority of the MBA is verified.The experimental results based on actual drilling data show that the proposed ROP model can solve the nonlinear prediction problem and correct abnormal data.(2)Hybrid prediction model for mud pit volume(MPV)combined with time series fine-tuningThe change of the MPV has nonlinear and time series characteristics,and it is difficult for a single model to meet the modeling needs of the MPV.To solve the prediction problem of the MPV,a hybrid prediction model combined with time series fine-tuning is proposed.First,the SVR method and the back-propagation neural network method are used to establish the prediction sub-model for MPV respectively.For each sub-model,a hybrid particle swarm optimization(HPSO)algorithm is studied to design the best model parameters and improve the prediction accuracy of the sub-model.Then,based on the results of the model on different evaluation indicators,the weight of the combination of sub-models is determined to improve the nonlinear prediction accuracy.Finally,a long short term memory neural network(LSTMNN)is used to establish a time series model of the MPV.According to the difference in time series prediction between the combined model and LSTMNN model,the combined model is fine-tuned to improve its time series prediction accuracy.The experimental results based on the IEEE CEC2015 benchmark function verify the global search ability of the HPSO algorithm.The comparison results based on actual drilling data show that the proposed model can meet the nonlinear and time series requirements of MPV prediction.(3)Drilling state evaluation and decision-making method based on fuzzy comprehensive evaluationBefore optimizing the drilling process,it is necessary to evaluate and make decisions on the states of the drilling systems to determine whether the current drilling process needs to be optimized and the adjustment range of the operating parameters.The evaluation and decision-making of the operating state of the drilling system need to consider multiple working conditions,multiple indicators and the ambiguity of the evaluation indicators.Therefore,an evaluation and decision-making method based on fuzzy comprehensive evaluation for drilling states is proposed.First,an evaluation strategy of juxtaposed structure is designed.The fuzzy comprehensive evaluation method is used to evaluate the operating states of the drill string system and the circulation system respectively.Then,the fuzzy C-means method is used to divide the drilling conditions,and determine the condition to which the current state belongs.Finally,based on the evaluation results and the determined working conditions,a comprehensive decisionmaking method is designed to determine whether the drilling state needs to be adjusted and the adjustment range of the operating parameters.The simulation results based on actual drilling data show that the proposed method can effectively evaluate the operating state of the drilling system,and make comprehensive decisions to provide the adjustment range of operating parameters for subsequent optimization.(4)Coordinated optimization strategy between systems with vertical well trajectory constraintsTo improve the efficiency and safety of and solve the optimization problem of the geological drilling process,considering the constraints of the vertical well trajectory,the adjustment time interval of operating variables and the interaction between the systems,a coordinated optimization among systems with vertical well trajectory constraints is proposed.First,the SVR method is used to establish the prediction model of ROP and MPV respectively.In order to determine the best value of model parameters,a novel hybrid bat optimization algorithm(NHBA)is designed based on the MBA and HPSO algorithm,which improves the global search ability and stability of results.Then,the influence of between weight on bit(WOB)and vertical well trajectory is analyzed,and the constraints of WOB are determined.Considering that the drilling operating parameters have different adjustment time intervals,a long time scale and short time scale optimization strategy is designed,and the non-dominated sorting genetic algorithm II is used to improve the ROP and reduce the fluctuation of the MPV.The comparison results with the three algorithms in the IEEE CEC2005 benchmark function show that the NHBA has better global search ability and result stability.The comparison results based on the actual drilling data and the experimental system show that the proposed method can improve ROP and reduce the fluctuations of MPV.By studying the rate of penetration modeling,mud pit volume modeling,drilling state evaluation and decision-making method and coordination optimization strategy with vertical well trajectory constraints,it provides reliable solutions for drilling process modeling and optimization problems and also lays an important foundation for improving the efficiency and safety of the geological drilling process.