Effectiveness of neural network and fuzzy approaches in the control of crewless ships

Purpose of research of this work is the development and evaluation of control models for autonomous underwater and surface vessel using fuzzy logic and neural network technologies. The influence of different approaches on the control accuracy and motion stability of uncrewed, autonomous vessels is investigated. Methods. In this work, we used the fifth-order Rung-Kutta method for numerical modeling of the dynamics of an autonomous vehicle. This method allows to accurately calculate the state of the AV in time, taking into account the various parameters of its motion. The method used was fuzzy modeling, which includes the development of fuzzy controllers. These controllers take into account the peculiarities of AV dynamics and provide robustness under changing environmental parameters. Fuzzy modeling allows the use of linguistic variables to describe the different states of the system and takes into account the uncertainties that may arise in the control of the AV. Method of neural network technology in AV control. The use of neural networks provides the possibility of automatic training and adjustment of control parameters based on the results. Results. The simulation results showed that the use of fuzzy models significantly improves the control performance of AV compared to mathematical models. The implementation of neural networks achieved the best RMS error rate (REM) compared to both other models, which confirms the effectiveness of this approach. In particular, for the X direction, the RMSE for the neural network was 6.4321, which is the best among all models. Conclusion. Research has shown that integrating fuzzy logic and neural network technology into AV control results in significant improvements in control accuracy and stability in complex environments. Neural networks provide additional adaptability, allowing the system to respond effectively to changes in the external environment and improving the overall performance of the BEC.