Using the Artificial Neural Network Approach to Develop an Illegal, Unregulated, Unreported Transshipment System Under Consideration of the Effects of Wind in Banda Water

Indonesia faces significant challenges from illegal transshipment and criminal operations in its marine sector. This study focuses on detecting of illegal, unregulated, unreported (IUU) transshipment by identifying trajectory anomalies in vessel movements, which may result from intentional deviations or environmental factors such as weather disturbances. The proposed system comprises three main components: anomaly detection, anomaly selection, and decision-making. These components incorporate key features including vessel distance from expected routes, wind speed, heading changes, and speed variations. Two distinct models were developed: the first employs neural networks with carefully selected input and output variables based derived from validated data, while the second utilizes type-2 fuzzy logic. Evaluation result show that the neural network-based model achieves an accuracy of 98.726%, whereas the type-2 fuzzy logic model attains 91.67%. The system was further using real-world vessel trajectory scenarios originating from Philippine waters and transiting through the Makassar Strait, yielding an average detection accuracy of 80.92%. The primary objective of this research is to develop a robust and reliable system capable of identifying potential IUU transshipment violations to support maritime surveillance and enforcement efforts.