Submarine pitch and depth control using FCMAC neural networks

A non-model based nonlinear adaptive control scheme was developed to control the depth and the pitch of submarines operating in shallow water. This control scheme utilizes a special type of neural networks called the fuzzy cerebellar model arithmetic computer (FCMAC) to compensate the nonlinear dynamics of submarines. An online tuning scheme with no off-line training phase was used to update the weights in the FCMAC. Simulation results show that this control scheme has superior performance and robustness under various scenarios. Preliminary results also show that FCMAC control can enhance the failure recovery capability.