Oil and gas pipeline leakage recognition based on distributed vibration and temperature information fusion

Abstract It has great significance to monitor the oil and gas pipeline leakage to reduce economic loss and environmental pollution. In this paper, we propose a method to recognize the leakage of oil and gas pipeline based on both vibration and temperature information according to the distributed optical fiber sensor’s measurement ability. After comparing various feature values and different classifier models, we choose six temperature feature values, five vibration feature values, and the random forest model as the optimum combination for the pipeline leakage recognition. The method can accurately recognize the states of leakage, interference, and normal operation. The average recognition accuracy is 98.57%, which is higher than the traditional single-parameter judgment method, and the recognition time is only 6.79 ms.