A water supply pipe leakage localization method based on pressure-flow fused signal analysis and adaptive dynamic...

Area of Science:

• Engineering
• Hydraulics
• Signal Processing

Background:

• Accurate leak localization is critical for water supply pipeline stability.
• Traditional methods using single pressure signals and ideal parameters have limited accuracy and generalization.
• Pipeline leaks cause complex signal perturbations that challenge existing localization techniques.

Purpose of the Study:

• To develop a novel pipeline leak localization approach integrating pressure-flow fused signal analysis and adaptive dynamic programming (ADP).
• To enhance the accuracy and adaptability of leak localization in water supply systems.
• To address the limitations of traditional methods in real-world, variable conditions.

Main Methods:

• Established a linear propagation model using the method of characteristics (MOC) and finite difference approximation for fused signal attenuation.
• Reformulated leak localization as a nonlinear, constrained, multi-step dynamic optimization problem.
• Applied an adaptive dynamic programming (ADP) framework with a critic-actor neural network architecture to solve the optimization problem.

Main Results:

• Validated the proposed method through simulations in industry-grade software and experiments on an actual water supply system.
• Demonstrated effectiveness even for small leaks and high system variability.
• Outperformed traditional time delay and negative pressure wave attenuation methods in estimation accuracy and adaptability.

Conclusions:

• The proposed pressure-flow fused signal analysis with ADP offers superior leak localization performance for water distribution networks.
• The method shows significant potential for practical application due to its accuracy and adaptability under diverse conditions.
• This integrated approach overcomes the limitations of conventional leak detection techniques.