Cross-synthesis of Observer and controller for a two-mass uncertain system

Abstract

AbstractThe industrialists want a minimum of sensors to improve reliability and cost. Performances are then decreasing for systems with variable parameters or direct-drive. Moreover, such a flexible power train leads to a two-mass system which can become unstable. This paper proposes a solution to quickly and accurately tune an observer and a state feedback with a lower computer time consumption and lower conception time. A Linear Quadratic optimized state feedback and a Kalman observer both with special rating matrices are implemented. An evolutionary algorithm optimizes the observer and the controller degrees of freedom all over the variations underlining the benefits of a crossed synthesis. Experiments show that the stability and the performance are effectively maintained.