RO-based self-organizing neuro-fuzzy approach for HDD positioning control

Chunshien Li, Kuo Hsiang Cheng, Jin Long Chen, Chih Ming Chen

研究成果: 雜誌貢獻期刊論文同行評審


The requirement for achieving the smoothness of mode transit between track seeking and track following has become a challenging issue for hard disk drive (HDD) motion control. In this paper, a random-optimization-based self-organizing neuro-fuzzy controller (RO-SNFC) for HDD servo system is presented. The proposed controller is composed of three designs. First, the concept of pseudo-errors is used to detect the potential dynamics of the unknown plant for rule extraction. Second, the propensity of the obtained pseudo-errors is specified by a cubic regression model, with which the cluster-based self-organization is implemented to generate clusters. The generated clusters are regarded as the antecedents of the T-S fuzzy "IF-THEN" rules. The initial knowledge base of the RO-SNFC is established. Third, the well-known random optimization (RO) algorithm is used to evolve the controller parameters for control efficiency and robustness. In this paper, a motion reference curve for HDD read/write head is employed. With the reference velocity curve, the RO-SNFC is used to achieve the optimal positioning control. From the illustrations, the feasibility of the proposed approach for HDD servo systems is demonstrated. Through the comparison to other approaches, the excellent performance by the proposed approach in access time and positioning smoothness is observed.

頁(從 - 到)2615-2625
期刊IEICE Transactions on Fundamentals of Electronics, Communications and Computer Sciences
出版狀態已出版 - 10月 2005


深入研究「RO-based self-organizing neuro-fuzzy approach for HDD positioning control」主題。共同形成了獨特的指紋。