Frequency Limited Model Reference Adaptive Control for Unmanned Underwater Vehicle

Controlling of unmanned underwater vehicles (UUV) is a complicated challenge due to the nonlinear dynamics, uncertainty in model parameters, and external disturbances that exist in underwater environments. Therefore, compared to classical control techniques, adaptive control is a promising solution to address these issues. However, to realize its complete potential in underwater vehicle applications, the trade-offs among rapid transient performance and smooth control signals must be carefully analysed. When using high learning rates to improve the transient performance of systems controlled using Model Reference Adaptive Control (MRAC), it has been observed that it results in undesirable high frequency oscillations in control signals. Frequency Limited Model Reference Adaptive Control (FLMRAC) is a technique that adds a mismatch term to the reference system of MRAC. Addition of this term provides better control compared to standard MRAC by filtering out high-frequency content of the tracking error. This results in filtering of high-frequency content from the control signal, which in turn helps to avoid high frequency oscillations. Simulation results also show that FLMRAC results in a lower tracking error as well as lower rise times compared to MRAC operating with the same learning rate. Furthermore, FLMRAC also shows robust behaviour in situations such as partial thruster failure.