Multivariable Control of a Submersible using the LQG/LTR Design Methodology

Abstract

A multivariable feedback control system is designed for a submersible. The control variables are the bow, rudder, and differential stern control surfaces; these are dynamically coordinated so as to cause the vehicle to follow independent and simultaneous commanded changes in yaw rate, depth rate, pitch attitude, and roll angle. Two designs were evaluated using a nonlinear submersible simulation. One used all four control variables so that active roll control was possible. The other used only three control variables, and active roll control was not employed. Both feedback systems were designed using the Linear Quadratic Gaussian (LQG) with Loop Transfer Recovery (LTR) design methodology so as to meet similar design specifications in the frequency domain. Both the linearized models, and the non-linear simulation have shown that active roll control yields a very significant improvement in submersible performance. Active roll control minimized unwanted depth changes in difficult commanded trajectory scenarios.