The uncertainty threshold principle - Fundamental limitations of optimal decision making under dynamic uncertaintyThe fundamental limitations of the optimal control of dynamic systems with random parameters are analyzed by studying a scalar linear-quadratic optimal control example. It is demonstrated that optimum long-range decision making is possible only if the dynamic uncertainty (quantified by the means and covariances of the random parameters) is below a certain threshold. If this threshold is exceeded, there do not exist optimum decision rules. This phenomenon is called the 'uncertainty threshold principle'. The implications of this phenomenon to the field of modelling, identification, and adaptive control are discussed.
Document ID
19770046002
Acquisition Source
Legacy CDMS
Document Type
Conference Proceedings
Authors
Athans, M. (Massachusetts Inst. of Tech. Cambridge, MA, United States)
Ku, R. (Massachusetts Inst. of Tech. Cambridge, MA, United States)
Gershwin, S. B. (MIT Cambridge, Mass., United States)