時刻歴データを誤差因子とした過渡時の基本機能の研究

－フィードバック制御の最適化－

抄録

Today's automobiles are full of controllers, and many man-hours are spent in optimizing their control parameters. This study was aimed at reducing the man-hour requirement. When a controller is embedded in a product, engineers have been taught to optimize the product exclusive of the controller. Although more and more modern products cannot work without feedback control, quality engineering has never been applied to optimize feedback control itself. In this study, feedback colltrol was optimized by applying quality engineering with dynamic characteristics, using time history data as a noise factor and variation of the target value as a signal. Since optimizing a simple PID control of the type often found in textbooks failed to provide adequate stability and response, more complex logic was employed and the number of control factors was increased to 16. Despite large interaction effects, the main factor effects were identified and conditions providing good stability and response were found.