Control of systems subject to sudden change in character

Dynamic models in which the coefficient processes exhibit sudden changes at random time points can be used to make manifest the influences of subsystem failures or rapid changes in the environment in which a system must operate. Control and regulation of such systems is complicated by the uncertainty which surrounds the effect of future events on system behavior. A satisfactory control policy must quantitatively discount these events of uncertain occurrence to provide good performance for a variety of operating conditions. This paper describes controller synthesis procedures for specific classes of such systems and gives some of the ways in which these controllers differ from those derived under the assumption of perfect knowledge of future plant behavior.