Video-Sensing Characterization for Hydrodynamic Features: Particle Tracking-Based Algorithm Supported by a Machine...

Area of Science:

• Environmental science
• Fluid dynamics
• Machine learning

Background:

• Open channel flow monitoring is crucial for flood prevention and infrastructure protection.
• Traditional methods may lack accuracy or real-time capabilities for dynamic environmental conditions.
• Video imaging offers a promising approach for monitoring flow and geo-environmental constraints.

Purpose of the Study:

• To improve the accuracy of video-based water flow monitoring in open channels.
• To develop an advanced particle tracking algorithm for hydrodynamic analysis.
• To enhance environmental and hydrodynamic safety through reliable monitoring.

Main Methods:

• An improved particle tracking algorithm for flow hydrodynamics was developed.
• A machine learning approach using a convolutional neural network-evolutionary fuzzy integral (CNN-EFI) was employed.
• A sub-comparison was performed using a multi-layer perceptron (MLP) algorithm.
• Video signals from a CMOS camera monitoring a rainwater channel were processed.

Main Results:

• The CNN-EFI approach demonstrated reliable results in processing video signals for water flow analysis.
• The developed algorithm enhanced the accuracy of video imaging for hydrodynamic monitoring.
• The system effectively monitored water flow conditions and potential geo-environmental constraints.

Conclusions:

• The combined machine learning and particle tracking approach offers a reliable solution for environmental and hydrodynamic safety.
• Accurate monitoring of open channel flow is vital for flood mitigation and public safety.
• This technology can significantly contribute to safeguarding properties and lives from flood risks.