本文中旨在設計一類神經網路控制器，用以線上控制垂直負 載電液
伺服缸的位移。採用倒傳遞網路學習方法將網路訓練成受 控系統的互補
模型，藉正弦波響應、步階響應和定位軌跡規劃響 應實驗來驗證控制方
法的可行性。並分別將供油壓力下降、油溫 升高和負載增加以探討類神
經控制器對此等系統參數變動的適應 能力。

A neural network controller was designed in this paper, which
was implemeted to control the displacement of a vertical
servohydrualic cylinder.The network was trained on line to
approach the inverse-dynamic model of the plant by theback
propagation learning method.The feasibility of proposed control
scheme was verified via the system response to a sinusoidal
input,a step input anda position scheduling input.In addition,
the supply pressure,oil temperature andload were changed,
respectively in the experimental study see the
controllerperformance under changes of these system parameters.