SHIP TRAJECTORY CLEANSING AND PREDICTION WITH HISTORICAL AIS DATA USING AN ENSEMBLE ANN FRAMEWORK

Ship trajectory provides crucial spatial-temporal maritime traffic information, which helps maritime traffic participants enhance the safety and efficiency for the traffic control and management. In that manner, significant focuses were paid to obtain accurate ship trajectory with the help of historical Automatic Identification System (AIS) data. To that aim, we proposed an ensemble Artificial Neural Network (ANN) model to cleanse and predict ship trajectory from the AIS data (i.e., latitude, longitude, speed). The proposed framework smoothed noises in the raw AIS data via the ensemble Hampel Filter (HF) and Butterworth Filter (BF). Then, the proposed framework normalized the smoothed AIS data to equalize the time interval between neighboring AIS samples. After that, we predicted the ship trajectory with the help of ANN model. The experimental results showed that our proposed model was effective and efficient in removing the AIS data outlier, and obtained satisfied ship trajectory prediction results.