Broad band active vibration isolation of automobile engine using linear and nonlinear feedback control is considered. The objective is to reduce the forces transmitted to the chassis and body, and, thus, reducing vibrations and structure borne noise inside the vehicle compartment when the engine is subjected to diﬀerent excitations. Moreover, the ability of the original passive suspension system to deal with high load transient excitation, e.g. due to a dropped clutch operation, has to be preserved. Engine excitations corresponding to idle and driving engine operating conditions as well as internal and external transient excitations have been investigated.
Solutions based on classical control and LQG (Linear Quadratic Gaussian) control methodologies have, to some extent, been treated. However, it turns out that the desired loop gain requires a control design method more suitable for shaping the loop gain and, at the same time, obtaining closed-loop stability. Using H2 control theory and Gain Scheduling, a MIMO (multi-input multi-output) control algorithm dealing with the above mentioned excitations when taking system nonlinearities into account, is developed.
The active engine suspension system design has been performed making use of a virtual simulation, analysis, and veriﬁcation environment providing powerful opportunities to deal with time varying system characteristics.
Except for some restrictions originating from non-linearities, feedback loop shaping technique is found to be a suitable way to achieve desired closed-loop characteristics when dealing with such MIMO system. Where all engine excitations except those corresponding to high ramping speed or extremely high nominal engine torque, are successfully dealt with. However, to guarantee closed-loop stability, two kinds of non-linearities, reﬂecting the time varying system characteristics, have to be taken into account. Those are non-linear material characteristics of the engine mounts and large angular engine displacements. This requires the linear H2 control theory to be extended using a non-linear Gain Scheduling control scheme.
The eﬀects of input saturation have been investigated using describing function analysis for two diﬀerent controller implementations, using computed and applied control force for state observation. It has, unexpectedly, been found that, avoiding closed-loop self-oscillations due to input saturation requires computed control force to be used for state observation.
Linköping: Linköping University , 2002. , 138 p.