Study On Oil Pipeline Vulnerability Evaluation Of CO₂ Flooding Reservoir Based On SVM

Industrial applications of carbon dioxide flooding have an extensive future, so that no one can afford to neglect the following safety problems, in which the close attention ofenormous threat caused by this application is particularly important. Oil pipeline is the lifeblood of the oilfiled system, whose failure can lead to paralysis of the entire system, and bring disastrous consequences to life, property and the environment.Therefor, focus on the safety of CO₂ flooding pipeline and ensure it apprating safely is of great importance mathematically.Implementing safety assessment to precisly recognizeweaknesses of system and adobt targeted measures will provide a scientific basis for the development of preventive measures and management decisions.Vulnerability has become a hot issue and important analysis tool in many areas, to which academia and even many international organizations pay close attention. As a scientific research method of intrinsically safe systems, vulnerabilityevaluation is more valuable to safety management compared to traditional risk assessment. This paper evaluated the synthesissafeeondition of CO₂ floodingpipeline by means of this.（1）Synthesizing the existing vulnerability research, according to the characterastics of CO₂ flooding oil pipeline and the practical investigative result, binding targets of this paper to be achieved, defines the vulnerabilityof CO₂ flooding pipeline from 3 respects of the hazards, hazard bearing body and hazard response.（2） According to the given concept of vulnerability, considering principles and methods to selece index, buildsa five-level vulnerability evaluation index system covering 37 indexis by analyzing vulnerability factors in depth. Divided vulnerability Levels and determined each grade ranges. For each index, the quantitative methodwas determined, and the range was graded in accordance with the scope of each vulnerability level.（3）After study of the pros and cons of common vulnerability research methods, selects the SVM regression method to quantify vulnerability and establishs a SVM vulnerability assessment model on base of Matlab R2013 a. Theoutput of the model and the expected output fits well. The trained SVM model has stronggeneralization abilityand high accuracy, themaximum relative errorbetweenthe model-evaluated value and the expected output in the confirmatory experiment is only 0.0268. The model can be easily used for the evaluation of new samples and simplify the revaluate process, which is more scientific, objective, and suitable for promotioncompared to the conventional method.（4）Give examples of application of the model. Assessment Results is consistent with the pipeline‘s actual operating conditions, which proves the applicability of the model. By using fuzzy comprehensive evaluation method to evaluate vulnerability of this paper’s subject, the resultis consistent with the SVM model, which verifies the model’s reliability.