Motion Control of Elastic Joint Based on Kalman Optimization with Evolutionary Algorithm

Abstract

Actual industrial ambition is to remove a maximum of sensor to improve reliability and cost. Performances are then decreasing a lot, specially for a system with variable parameters and direct drives. Moreover, a two-mass system representing numerous class of industrial problem can become unstable. Keeping stability, a simple controller and observer tuning approach and a lower time consuming are main goals of this study. A previous calculated state feedback is used as base for two Kalman filters with special a noise matrix. An evolutionary algorithm optimizes observer's degrees of freedom to keep stability all over the stiffness variation. The results show that the stability and performances are kept on an experimental test bench.