Study On Risk Determination Method Of Oil And Gas Pipeline Based On Random Forest Algorithm

Is a big demand for oil and gas resources in our country,from the point of oil and gas transportation pipeline mileage,the domestic pipeline mileage increasing in our country,with the rapid growth of pipeline mileage,with oil and gas pipelines accidents,oil and gas pipeline risk assessment for the pipeline enterprise to prevent pipeline accidents,ensuring the safe operation of the best means,to achieve risk precontrol,guarantee the safe operation of pipeline.Although there are evaluation methods and models for oil and gas pipeline risk assessment,there are some problems such as risk obscuring and unreasonable evaluation.At present,the number of risk assessment experts in the domestic oil and gas pipeline industry is small and cannot cover all the oil and gas pipelines.Most of the personnel engaged in the oil and gas pipeline industry have insufficient experience in risk assessment and cannot independently complete the risk assessment work of the oil and gas pipeline.In this paper,the oil and gas pipelines at home and abroad and the cause of the accident statistics,combining with the characteristics of oil and gas pipelines,selection of control experience and the method of analytic hierarchy process combined with oil and gas pipeline risk factors identification,analysis of the characteristics and risk factors of properties of oil and gas pipeline risk factors can be divided into seven rule layer,and refine into the 40 index layer,set up index system of oil and gas pipeline risk factors.Analysis of the commonly used oil and gas pipeline risk assessment methods and procedures,is currently the most widely used method of index system to analyze its problems and disadvantages,points out the irrationality and a variety of problems,there is not prominent,risk of significant risk factors of mutual consideration insufficiency,the unreasonable weight division,the consequence factor proportion is bigger,the index factors such as a variety of problems.In this paper,the characteristics of common machine learning algorithms are analyzed and compared.Combined with the actual cases of oil and gas pipelines,the random forest algorithm is selected to construct the risk determination process of oil and gas pipelines,and the risk evaluation method based on the random forest algorithm is established.This article selects a X domestic pipeline,collect the basic information pipeline and pipeline integrity management,high consequence areas including pipe identification conditions,pipeline risk,pipeline inspection situation inside and outside the pipeline inspection situation,as the base case annotation,selecting typical 200 risk points,on the case marking after correlation analysis,The indicators with strong correlation were removed,the model was programmed with the Scikit-Learn machine learning module of Python programming language,and the model parameters were optimized by the combination of learning curve method and grid search method,and the evaluation results were compared.The risk determination method based on random forest algorithm can effectively solve the accuracy problem of existing pipeline risk assessment methods,and put forward targeted risk control measures,which has a great application prospect.And the risk determination software based on random forest algorithm is designed and developed for its GUI program.