Study of vessel-following behavior in port channels based on an improved GM model
To address the escalating congestion and navigational safety risks in port waterways driven by the rapid growth of global seaborne trade, this study investigates vessel-following behavior from a microscopic traffic-flow perspective. Using simulation data from ship-following experiments conducted in the main channel of Qingdao Port, we demonstrate that vessel-following dynamics exhibit variation patterns comparable to those observed in road traffic, thereby supporting the extension of classical car-following theory to maritime traffic systems. On this basis, the nonlinear General Motors (GM) stimulus-response model was introduced to characterize vessel-following behavior. The experimental results confirmed that the GM model can capture the speed and spacing evolution of vessels in port channels with high fidelity. However, because the original formulation does not account for the physical length of vessels, it may generate serious rear-end collision risk under extreme deceleration conditions. To overcome this limitation, a dynamic minimum safe-spacing mechanism and a piecewise control strategy were incorporated to develop an improved GM vessel-following model. The proposed model preserved the high-accuracy fitting capability of the original GM formulation for experimental navigation data while fundamentally eliminating collision risk in simulation. These findings provide a more physically consistent and safety-oriented framework for modeling vessel interactions and offer valuable theoretical support for intelligent ship collision avoidance, port traffic-flow management, and channel capacity assessment.