To address issues such as low automatic identification system(AIS)data frequency

trajectory nonlinearity

and high maneuverability

an interacting multiple model(IMM)Kalman filtering algorithm is proposed for ship trajectory prediction.By preprocessing AIS data to correct abnormal speeds and positions

the accuracy of the data is improved.The algorithm employs multiple state models to estimate the ship’s state in parallel

utilizes a Markov chain to achieve model probability transitions

and fuses the results to obtain an optimal estimation.Experimental results show that the mean fluctuation error of ship trajectory prediction is less than 5 meters

with an error standard deviation of approximately 20，and it is less affected by target maneuvers.The algorithm demonstrates good real-time performance and precision

effectively handling nonlinearity and maneuverability variations

thereby providing reliable support for intelligent shipping.