Adaptive Parameter Identification of an Accurate Nonlinear Dynamical Model for Marine Thrusters

Abstract

Recently reported finite-dimensional nonlinear thruster models employing empirically determined lift–drag curves have been shown to accurately model both the transient and steady state response of marine thrusters. These reports have employed the standard off-line paradigm for model parameter identification: First, real-time sensor data (force, torque, fluid velocity) is logged in laboratory experiments. Second, the experimental data is analyzed with a least-square regression technique to complete a best fit for the model parameters. This paper reports an on-line technique for adaptive identification of model parameters for marine thrusters. The stability of the proposed technique is shown analytically. The performance of the on-line adaptive identification technique, evaluated with respect to an experimentally validated plant model, is shown to compare favorably to its off-line counterpart, but does not require the thrust and torque instrumentation required by conventional off-line least-squares parameter identification techniques.