P010

Smith Predictor-based Adaptive Generalized Predictive Control of AUV

A0019 Sun Jun College of Marine Engineering, Northwestern Polytechnical University

A0020 Xu Demin College of Marine Engineering, Northwestern Polytechnical University

In order to achieve fast response and improve robustness, a Smith

predictor-based adaptive generalized predictive controller for SISO

systems is presented and applied in course keeping and turning control

for AUV with time-delay in this paper. Instead of an optimal

predictor in generalized predictive control, this controller employs

Smith predictor which is made up of identifiable parameter using RVFF

(recursive variable forgetting factor method) on line. Then the

controller will simultaneously drive the actual process variable

towards the setpoint whether the course changes or a load disturbs the

AUV, even there is a mismatch between the process and the model. It

is believed that this controller has several advantages: improves the

mobility of AUV; overcomes predictive error aroused by the mismatch

between the process and the model; covers a wide range of linear

systems, namely non-minimum phase, time-delay and time-varying systems.