Automatic Smith-predictor tuning using optimal parameter mismatch

The Smith-predictor is famous for its delay-free characteristic and suitable for regulating systems with an excessively long time delay. However, it is sensitive to plant parameter variations based on some frequency domain studies. In this paper, detailed analyzes of the system performance through parametric and temporal mismatch are investigated and accompanied with closed form solutions. Based on the derived solutions, the optimal performance and stability characteristic of a Smith-predictor compensated control system is examined. It is found that the system performance can be improved by introducing an intentional parametric or temporal mismatch and the corresponding optimal tunings are determined based on the value of a normalized plant parameter. The theoretical results are also summarized to form an automatic Smith-predictor tuning procedure and verified experimentally on a fluid temperature control system for a quiet hydraulic precision lathe.