Algorithm for Tuning the CMAC Regulator Scalin Gains via Automatic System Step Response Anlysis

Abstract

The paper presents the tuning of a CMAC (Cerebellar Model Articulation Controller) regulator, with B-spline basis functions of order 3, implemented in an automatic daylight control system. After the tuning process of the regulator scaling gains, a set of ranges for scaling gains resulted. For these ranges of scaling gains the automatic lighting control system (ALCS) satisfied the desired performances. After the tuning process, the behavior of the ALCS in the presence of the illuminance disturbances was studied. At the final of the paper, an experimental tuning algorithm of the regulator scaling gains via ALCS step responses analysis is proposed. Even if the used experimental model of the process is a gross approximation of the real model of the process, using this experimental tuning algorithm, the designer can find a set of ranges of the regulator scaling gains for which the ALCS is stable and satisfy the desired performances.

Keywords

• daylight
• scaling gains
• CMAC network
• learning rate
• visual comfort