On the adaptive pattern recognition control

Abstract

The development of adaptive pattern recognition control was based on the analysis and description of cognition and thinking of a human control engineer, particularly its associative and qualitative nature. This analysis yielded a design procedure that can be divided into two basic tasks which can be treated separately: pattern description (Task 1) and assignment of control action to patterns. (Task 2). The adaptive pattern recognition controller (APRC) scheme uses an elementary pattern set approach for pattern description (Task 1) and an iterative adaptive controller adjustment strategy for a PI controller setting (Task 2). The parameter-adaptive PI controller based on this elementary pattern approach is described. It yielded good results in simulation studies for linear time-varying stable minimum/nonminimum phase processes with dead time. The application to two laboratory plants for the control of flow, pressure, and temperature was successful.<>