A Data Aggregation Algorithm for Two-Time Scale Adaptive Control

We examine a two-time scale approach to adaptive control where the direct control function is implemented on the first-layer microprocessor controllers of a two-layer distributed control "system and the parameter identification and adaptation functions are carried out on a time-shared second-layer computer at a slower time scale. It is assumed that the parameters of the model used to describe the process change slowly (or infrequently) relative to the dominant transient response modes of the proces loops being controlled. A data aggregation algorithm is presented which reduces the data transmission rate between the local controller and the central computer by a factor of n/N (where n is the order of the process model and N is the level of aggregation). Algorithms are derived for identifying the process models based on the aggregated data. Computer simulation experiments on first and second-order dynamic models show that the approach works well for aggregation levels up to eight. However, as the aggregation level or the model order increases, the complexity of the computations increases significantly resulting in problems of time delay, bias and convergence in the parameter estimation. The results suggest, however, that further study might yield some modifications of the algorithms used that will extend the range of applicability of the approach.