Robust multi-model controllers design without impulse in switching times for non-linear vibrations suppression of sandwich plate

Abstract

Generally multi-model controllers design has three stages: Decomposition of complicated non-linear systems into local linear models, designing the local controllers, and composition of them in sub-regions. In the present paper an algorithm is proposed in which all the three stages are simultaneously done to decrease the number of controllers. Determining a specific threshold based on non-linear model characteristics is a good criterion for this purpose. The threshold depends on the knowledge of the system features and designing method, and the improper selection of the threshold will increase the number of local models and the complexity of the global controller structure. This non-convex optimization problem is solved by the genetic algorithm whose cost function is defined by the complementary sensitivity function to guarantee the robust stability and optimal performance of the close loop the system, respectively. Another challenge in designing multi-model controllers is the transient performance degradation when switching from one local model to another. In the present research, the problem is solved by the system input generated by the online controller error signal and the feedback coefficient of the offline controller. The proposed robust controller in the presence of the uncertainties is evaluated for vibration Generally multi-model controllers design has three stages: Decomposition of complicated non-linear systems into local linear models, designing the local controllers, and composition of them in sub-regions. In the present paper an algorithm is proposed in which all the three stages are simultaneously done to decrease the number of controllers. Determining a specific threshold based on non-linear model characteristics is a good criterion for this purpose. The threshold depends on the knowledge of the system features and designing method, and the improper selection of the threshold will increase the number of local models and the complexity of the global controller structure. This non-convex optimization problem is solved by the genetic algorithm whose cost function is defined by the complementary sensitivity function to guarantee the robust stability and optimal performance of the close loop the system, respectively. Another challenge in designing multi-model controllers is the transient performance degradation when switching from one local model to another. In the present research, the problem is solved by the system input generated by the online controller error signal and the feedback coefficient of the offline controller. The proposed robust controller in the presence of the uncertainties is evaluated for vibration suppression of a sandwich plate.