This paper presents a practical on-line auto-tuning methodology for fast and precise positioning controllers using Genetic Algorithms (GA). In the system design for the proposed GA-based tuning, guidelines of the fitness evaluation and the high efficiency of the optimization are especially studied to establish the practical tuning interfaces. This GA-based optimization evaluates two fitness indexes to ensure the desired positioning performance: one is the absolute required positioning specifications, i.e., the settling accuracy and time, and the other is the residual vibration in response. In order to provide the practical tuning time and computation cost, on the other hand, the searching space of each parameter to be optimized is appropriately determined using the system stability analysis, which enables the tuning time to be reduced. The proposed auto-tuning algorithm has been verified as a practical controller design tool, paying attention to the tuning time for positioning devices in commercial machine tools.
J-STAGEがリニューアルされました！https://www.jstage.jst.go.jp/browse/-char/ja/
