Research On The Optimization System Of Drilling Fluid And Completion Fluid System Based On Machine Learning

Drilling is an important link in the exploration and exploitation of petroleum and natural gas resources,and it is also an important means of exploration and development of petroleum and natural gas.Different geological environments may require different drilling fluid completion fluids.The so-called drilling fluid system in this paper refers to the basic components or major special functions of drilling fluids that are similar and form a series,which contains individual differences.There are many systems in the fluid category.In engineering practice,due to the complexity,ambiguity and non-linearity of oil and gas reservoir data,the manual selection of drilling fluid systems alone has problems such as low efficiency,large reliability errors,and large subjective factors.To overcome these problems,this paper conducts research on drilling fluid completion fluid system selection based on machine learning.The main work of this paper includes:First,analyze the geological parameters that affect the choice of drilling fluid completion system and the nonlinear relationship between them,and combine the influencing factors and selection criteria of the selection system to determine the characteristic parameters that affect the selection of the system.Required characteristic parameters.Secondly,the training data set is subjected to dimensionality reduction processing.Here,the principal component analysis method,singular value decomposition method,and gray correlation analysis method are used respectively.Finally,the data sets processed by the above methods are respectively learned and trained by using BP neural network to obtain their respective preferred results,and the three preferred results are compared and analyzed with the system results of the unprocessed data set selection,so that Given the system's preferred alternatives.The innovation point of this paper is to apply machine learning to the optimization of drilling fluid completion fluid system,to achieve a more comprehensive and accurate selection of drilling fluid completion fluid system.This system is preferred to manual drilling fluid completion fluid completion The selection of the system is more convenient,and it also reduces the complicated workload for petroleum engineers to select drilling fluids.This is anew technical attempt,and it also allows historical data to be effectively used.The difficulty of this research is how to use machine learning technology to build a combined optimization model to achieve the purpose of drilling fluid system optimization.