TCN-BiLSTM Based Data-Driven Secure Control for Unmanned Marine Vehicles under FDI attacks

Due to the openness of information transmission during the unmanned marine vehicles’(UMVs) marine operations, it is highly susceptible to network attacks, which can compromise the system’s security. To address this issue, this paper proposes a data-driven secure control scheme by using Temporal Convolutional Network (TCN) and Bidirectional Long Short-Term Memory (BiLSTM) neural network. First, a deep learning predict network is designed by using TCN and BiLSTM, named T-BiLSTM, to predict the velocity and position information of UMVs. Based on it, a lumped information observer is designed by combining the T-BiLSTM predictor to estimate the external disturbances and FDI attacks. Compared with traditional extended state observer, it only rely on the input/output data and not depend on the model of UMVs. What’s more, a data-driven secure control method is designed, it ensures the UMVs can safely track the desired trajectory when subjected to external disturbances and FDI attacks. Finally, the stability of trajectory tracking error system for the UMVs is analyzed. The simulation example is used to demonstrate the effectiveness of the proposed control scheme in this paper.