Research on Underground Pipeline Fault Monitoring Method Based on Big Data

Abstract

Distributed optical fiber sensors can sensitively sense local temperature changes caused by leakage at any position of the pipeline, but massive monitoring data has significant spatio-temporal non-stationary characteristics, and it is difficult to directly diagnose leakage based on the monitoring data. Under the framework of statistical pattern recognition, a spatio-temporal big data analysis method based on sliding window outlier analysis is proposed. Only the internal characteristics of distributed temperature monitoring data can be used to realize the intelligent identification of pipeline leakage, and the sliding is determined. The value method of window length and abnormal state diagnosis window length, and the physical simulation of prototype insulation steel pipe leakage monitoring was carried out. The results show that when the pipeline is intact, this method will not cause false alarms. Once the pipeline leaks, the method can quickly identify the pipeline leakage event and accurately locate the leak location.