Long-Distance Oil/Gas Pipeline Failure Rate Prediction Based on Fuzzy Neural Network Model

With an aging underground long-distance oil/gas pipeline, ever-encroaching population and increasing oil price, the burden on pipeline agencies to efficiently prioritize and maintain the rapidly deteriorating underground utilities is increasing. Failure rate prediction is the most important part of risk assessment, and the veracity of the failure rate impacts the rationality and applicability of the result of the risk assessment. This paper developed a fuzzy artificial neural network model, which is based on failure tree and fuzzy number computing model, for predicting the failure rates of the long-distance oil/gas pipeline. The neural network model was trained and tested with acquired Lanzhou-Chengdu-Chongqing product oil pipeline data, and the developed model was intended to aid in pipeline risk assessment to identify distressed pipeline segments. The gained result based on fuzzy artificial neural network model would be comparatively analyzed with fuzzy failure tree analysis to verify the accuracy of fuzzy artificial neural network model.