Pipeline integrity prediction method based on big data and neutral network

In order to comprehensively and objectively evaluate the integrity of oil and gas pipeline, predict the potential threats and intensifies the safety management of pipelines, this paper proposes the framework and method to learn and predict the potential threats and pipeline conditions by adopting artificial neutral network machine learning theory. According to the definition of potential threat types and pipeline conditions, this method simulates the human thinking and learning, with pipeline construction, operation, failure and detection as the potential influential factors. It can learn the pipeline threats and conditions objectively and effectively, and provide the order of importance of potential influential factors. With the information obtained, it makes objective prediction on the pipeline threats and conditions, and then applies the prediction results to evaluate pipeline risks and define pipeline detection period. Unlike previous analysis methods that rely on experts' opinions and feature strong subjectivity, this method can make full use of various pipeline data sets to predict the pipeline integrity objectively, which is of great importance to improving pipeline safety and efficiency of decision-making.