Enhanced Detection of Pipeline Leaks Based on Generalized Likelihood Ratio with Ensemble Learning

Processes, Journal Year: 2025, Volume and Issue: 13(2), P. 558 - 558

Published: Feb. 16, 2025

To address the challenges of insufficient model generalization, high false alarm rates due to scarcity leakage data, and frequent minor alarms in traditional weak (the amount is less than 1%) detection methods for gas transmission pipelines, this paper proposes a real-time framework natural pipelines based on combination generalized likelihood ratio (GLR) ensemble learning. Compared methods, core innovations study include following: (1) For first time, GLR statistics are integrated with an learning strategy construct dynamic pipeline operating states through multi-sensor collaboration, significantly enhancing model’s robustness noisy environments by fusing pressure data from inlet outlet, as well outlet flow data. (2) An adaptive threshold selection mechanism that dynamically optimizes thresholds using distribution characteristics designed, overcoming sensitivity limitations fixed complex conditions. (3) decision module developed voting strategy, effectively reducing associated single models. The capability under normal conditions was validated self-built test platform. results show proposed method can achieve precise small 0.5% low-noise while rate zero. It also detect 1.5% strong noise interference. These findings validate its practical value industrial scenarios. This provides high-sensitivity, low-false-alarm, intelligent solution safety monitoring, which particularly suitable early warning leaks long-distance pipelines.

Language: Английский

Research on Leak Detection and Localization Algorithm for Oil and Gas Pipelines Using Wavelet Denoising Integrated with Long Short-Term Memory (LSTM)–Transformer Models

Sensors, Journal Year: 2025, Volume and Issue: 25(8), P. 2411 - 2411

Published: April 10, 2025

Traditional leakage prediction models for long-distance pipelines have limitations in effectively synchronizing spatial and temporal features of signals, leading to data processing that heavily relies on manual experience exhibits insufficient generalization capabilities. This paper introduces a novel detection localization algorithm oil gas pipelines, integrating wavelet denoising with Long Short-Term Memory (LSTM)-Transformer model. The proposed utilizes pressure sensors collect real-time pipeline applies eliminate noise from the signals. By combining LSTM’s feature extraction Transformer’s self-attention mechanism, we construct short-term average gradient-average instantaneous flow network model continuously predicts based gradient inputs, monitors deviations between actual predicted flow, employs curve distance accurately determine location. Experimental results Jilin-Changchun demonstrate possesses superior warning Specifically, accuracy reaches 99.995%, location error margin below 2.5%. Additionally, can detect leaks exceeding 0.6% main without generating false alarms during operation.

Language: Английский