A method and system for leak diagnosis of flexible pipes
By synchronously acquiring pressure and displacement information in flexible pipelines, constructing joint feature vectors, and using an isolated forest model for real-time leak detection, the problems of slow response and high false positives in traditional methods are solved, enabling rapid leak identification and safe response of flexible pipelines under large displacement conditions.
Patent Information
- Authority / Receiving Office
- CN · China
- Patent Type
- Applications(China)
- Current Assignee / Owner
- TIANJIN UNIV
- Filing Date
- 2026-01-22
- Publication Date
- 2026-06-02
AI Technical Summary
In urban gas, industrial media transportation, fire water supply and electromechanical pipelines of seismic isolation buildings, traditional pipeline leakage monitoring methods have long response cycles and low positioning accuracy, making it impossible to detect hidden leaks in a timely manner. Furthermore, diagnostic methods based on long sequence prediction are difficult to achieve real-time judgment at edge terminals, and cannot meet the safety requirements under large displacement conditions.
By synchronously acquiring pressure information from pipeline pressure measurement points and displacement information from the hinge area, a joint pressure-displacement feature vector is constructed. Real-time anomaly detection is performed using an isolated forest model, and combined with a failure probability model and reinforcement learning strategy, rapid leak determination and safety policy triggering are achieved.
It enables rapid identification and safe response to leaks in flexible pipelines under large displacement conditions, reduces misjudgments and delays, improves the timeliness and accuracy of leak detection, and shortens the accident handling time.
Smart Images

Figure CN122129654A_ABST