Research On Oil Pipeline Leak Detection Method Based On Adaptive Belief Rule Base

The safe transportation of oil pipelines is of great significance to environmental protection and pipeline management companies.In the long-term oil pipeline transportation process,due to factors such as the aging of the pipeline's own material,corrosion,and third-party damage,the safety of pipeline transportation is reduced to a certain extent,resulting in frequent safety accidents.Therefore,it is necessary to perform leak detection on oil pipelines and estimate the size of the leakage,so as to take corresponding repair measures in the future.Thus the detection of oil pipeline leakage has theoretical and practical application value.In oil pipeline transportation,leakage often evolves from a small amount of leakage to a large amount of leakage.When the initial leakage occurs,it is difficult to detect it manually,and the manual detection method has the drawback of its own scope of application.However,if the leak is detected by modeling method,neither the environmental protection department nor the pipeline management company will allow the continuous occurrence of the leak to cause losses to the environment and the economy.In other words,the low historical sample size for leak detection is also a thorny problem.Therefore,in the pipeline leakage detection,it is necessary to comprehensively consider the indicators and influencing factors of multiple information sources,so as to enhance the input information of the model,so that the domain knowledge of experts can be used in the modeling,and the few historical data can be used to optimize the model.To solve the above problems,this paper uses the belief rule base(BRB)modeling method to detect oil pipeline leakage,which can integrate multi-source information in the modeling.This paper focuses on the oil pipeline leak detection method based on the adaptive belief rule base,and the main contributions are reflected in the following aspects:(1)To solve the problem of incomplete expert knowledge representation,a belief rule base expert system with interval-valued references(BRB-IR)modeling method is proposed.In the proposed method,the referential value of the antecedent attribute is expanded from the original single-valued form to the interval form,so as to introduce the expert knowledge into the model more completely,then the optimal referential value can be obtained through the nonlinear optimization algorithm.The projection covariance matrix adaptation evolutionary strategies(P-CMA-ES)is utilized to optimize the remaining parameters of the model,and finally achieves the purpose of detecting pipeline leakage.(2)To simplify the model and improve the interpretability of the model,a rule adaptive belief rule base(BRB-RA)modeling method is proposed.Firstly,a new matching degree calculation method is proposed,which considers all the references belonging to the attribute rather than only the adjacent references in the calculation.Then the belief rules are reduced round by round,so as to reduce the number of rules and ensure the modeling accuracy.Finally,the model is simplified,the interpretability is enhanced,and the effectiveness of the method is verified in pipeline leakage detection.(3)To apply the research of interval referential BRB and rule adaptive BRB to practical production,an adaptive BRB oil pipeline leak detection system is developed.The core algorithms of BRB-IR and BRB-RA are embedded in the system to realize the interval adaptive optimization and rule adaptive optimization of the BRB system.Finally,it forms a complete process from theoretical innovation to case verification to application implementation.To sum up,aiming at the problems of incomplete expert knowledge representation and high model complexity in the traditional BRB system,this paper proposes the corresponding adaptive optimization schemes: BRB-IR and BRB-RA.The rationality of the proposed method is deeply analyzed from the perspective of theory,and the effectiveness of the method is verified in pipeline leak detection.Finally,through system development,the proposed method is applied to practice.