Abstract
This paper presents a comparative study of theoretical methods for estimating ship speed through water (STW), a critical parameter for optimizing vessel operations, fuel efficiency, and performance monitoring. Traditional sensors, such as Doppler Velocity Logs (DVL), are commonly used but suffer from measurement uncertainties due to environmental disturbances, sensor drift, and calibration difficulties. To address these limitations, three theoretical approaches are reviewed, and a novel Adaptive Robust Multi Sensor Fusion (ARMS) algorithm is proposed. The proposed ARMS algorithm adopts a graph-based fusion structure that uses sparsity to suppress noise and systematic deviations, along with smoothness constraints introduced to stabilize the estimated speed signal. Each method is mathematically formulated, with implementation details and limitations discussed. Using data from a bulk carrier equipped with both wave radar and DVL sensors, the study demonstrates the proposed ARMS algorithm achieves the highest accuracy.