The trailing suction hopper dredger (TSHD) is a ship that excavates sediments from the sea bottom while sailing. Soil properties have a strong effect on the dredging process. The parameter with the greatest importance is the real-time soil grain diameter dm. This, however, cannot be directly measured by available sensors. In this paper, an alternative method is proposed to solve this problem. A new estimation method is developed on the basis of an existing sedimentation model and measuring system data. The loading process with several grain diameters is simulated to perform a sensitivity analysis of soil type. Simulation results show that the dredging efficiency is strongly affected by fine soil. This soil-related estimation problem is solved with a continuous–discrete feedback particle filter (CD-FPF), which is a recently developed filter for a CD time system. For comparison, a bootstrap particle filter (BPF) is also used to simulate the steplike changes in dm in both the no-overflow and constant-volume loading phases. The results show that the CD-FPF outperforms the BPF in terms of accuracy and applicability. Thus, it is recommended to be applied in the estimator of artificial intelligence (AI) dredging systems.