非線性不確定系統模糊類神經網路控制器的硬體實作

Abstract

本篇論文中，以TS形式的模糊類神經網路模型為基礎的即時智慧型適應控制方法，用硬體實作出來，控制行星倒單擺系統。這個硬體平臺是dSPACE DS 1104 R&D control board，可以搭配MatLab在Windows 2000 下運作。在模擬與實作上都獲得很好的結果。並且經由模擬與實作，探討了運算的時間延遲對這個控制器效能的影響。從實作探討出來的運算時間延遲來看，我們可以使用更經濟的硬體平台來實踐這個控制方法，應用到工業上。

In the thesis, true hardware implementation of an on-line intelligent adaptive TSK FNN controller is performed to control the planetary inverted pendulum. The hardware platform is dSPACE DS 1104 R&D control board under Windows 2000 running with MatLab. Excellent agreements have been obtained between theoretical simulation and hardware implementation. The effects of computational time delay for controller is also investigated through both software simulation and hardware emulation and by building SimuLink blocks in MatLab. The estimated maximum computational time delay can be quite practical for the industrial applications to choose cheaper hardware platform with less cost.