Estimation of Leak Time Parameter Based on Fusion of Leak Behavior and Data Distribution Characteristic from Gas Flow Sensor Data

The accurately estimation of leak parameter is an essential part of pipeline leakage risk assessment and guarantee of the quality of leak samples, while today’s estimation methods leverage only coarse time estimation and rely on expert experience. This paper presents a leak time parameter estimation framework fusing leak behavior and data distribution characteristic from gas flow sensor data. Under the proposed framework, a leak behavior and a data distribution characteristic extraction modules are established for guaranteeing automatic estimation of leakage time parameters in a fine-grained time range. The estimation of leak starting time for 69 different leak events are implemented based on gas flow sensors data. The estimation accuracy of 94.2% demonstrates the effectiveness of the proposed method.