A Design of Neural-fuzzy Control System and Its Application in Servo Tracking Control

類神經模糊控制系統設計及其於伺服追蹤控制的應用

本文提出一類神經模糊控制結構，使伺服馬達系統去追蹤步進馬達的位移或預先定義的軌跡。由追蹤定值、隨機、正弦波及方波等四個參考軌跡的結果來評估控制器的性能。控制法則是以LabVIEW程式語言來撰寫，輸入及輸出信號是利用資料擷取卡NI PCI6221M來實現。由實驗結果知：利用本控制系統來追蹤一定值軌跡或緩慢變動的隨機軌跡是可行的。由於伺服馬達的本質特性及LabVIEW程式的限制，去追蹤一正弦波或方波軌跡的均方根誤差會比追蹤前面二個參考軌跡來得相對大。

In this paper, a neural-fuzzy control scheme is proposed such that the output of servomotor system tracks the displacement of a stepping motor or a pre-defined trajectory. The results of tracking four reference trajectories, constant, random, sinusoidal and square-wave reference inputs, are employed to evaluate the performance of the proposed controller. The control laws are implemented in LabVIEW programming, and the input and output signals are performed by data acquisition card NI PCI6221M. The experiment results indicate that the proposed control system tracking either a constant or a slowly varied random trajectory is available. Due to the intrinsic characteristic of servomotor and the limitation of LabVIEW programming, the tracking of sinusoidal or square-wave reference trajectory has a relatively big root mean square error compared with the tracking of the former reference trajectories.