Skip to Main Content
A multilayer control strategy for the control of a class of static nonlinear multivariable systems is described, and the economic tradeoff between the costs of implementing the approach and the performance level achieved is investigated. The definition of the control function at each layer in the control hierarchy is based on a partitioning of the manipulated variables according to a measure of performance sensitivity. The sensitivity of a variable is defined with respect to a measure of the performance degradation produced by variations of the variable from its optimal value. The more sensitive manipulated variables are updated by the lower layers at a higher frequency, while the less sensitive variables are updated by the higher layers at a lower frequency. The design variables of importance to the economic tradeoff are the number of layers in the hierarchy, the variables to be updated by each layer, and the corresponding periods of control action. To ease the computational requirements necessary to carry out the tradeoff analysis, an approximate method of analysis is described in which 1) a simplified approximate tradeoff measure is used to determine the number of layers in the hierarchy, the number of variables to be updated by each layer and the corresponding periods of control action, and 2) the sensitivity functions are used to determine which particular variables are to be updated by each layer. The use of the approximate method of analysis is demonstrated with reference to a simple example.