Game-theoretic DBN–CFD–GA optimization of laser scan paths for LNG leak detection

This study presents a comprehensive optimization approach to improve gas leakage monitoring in LNG stations. Conventional scanning path planning typically relies on experience-based placement of scan points, lacking theoretical support and struggling to adapt to complex station structures and dynamic risk changes. The process begins with the creation of a 3D model, which captures the spatial details of the equipment and its environment, providing the foundation for subsequent risk assessment. Dynamic Bayesian Networks (DBN) integrate historical data, real-time monitoring information, and expert insights to assess equipment status and identify high-risk zones. Computational Fluid Dynamics (CFD) simulations then model gas dispersion within these zones, generating detailed gas cloud distributions from which critical scanning points are extracted. Finally, a non-cooperative game-theoretic framework combined with Genetic Algorithms (GA) optimizes the laser platform’s scanning trajectory based on the extracted scanning points and identified high-risk areas. The proposed method demonstrates a 37.8 % reduction in scanning path length, a 22 % improvement in detection accuracy, and significant reductions in energy consumption and detection time. These improvements offer a more efficient and cost-effective solution for leakage monitoring, outperforming traditional methods in real-world LNG stations in terms of scanning efficiency, safety, and operational costs.