THE APPLICATION OF PARAMETER OPTIMIZATION TECHNIQUES TO LINEAR OPTIMAL-CONTROL SYSTEM-DESIGN

The linear quadratic regulator (LQR) design technique produces controllers with well-known properties including good stability margins. A fundamental question with the LQR technique is the appropriate selection of the state and control weighing matrices to achieve given design specifications. A solution to this problem is presented whereby a parameter optimisation approach is employed to automate the LQR design procedure. Standard numerical algorithms are used to calculate the weighting matrices such that the specifications are satisfied. Decoupled command tracking specifications in the time domain are considered at present, and a multivariable proportional plus integral controller structure is employed. To the knowledge of the authors the design methodology is new and has the advantages of both optimal control and accommodation of additional design specifications. The application of the method is demonstrated in the design of a decoupled longitudinal control system for a vertical take-off-and-land (VTOL) aircraft.