油壓機械自從十九世紀進入工業生產以來[1]，已逐漸變成人類生活的一
部份。隨著時代變遷油壓機械也由早先只強調大出力、大扭力的雛形，演
變到強調定位與操控性的精密機械。舉凡交通工具、土木機械、建築機械
、航太機械、甚至講究精巧的醫療機械也處處可見它的蹤跡。但是由於油
壓伺服控制系統容易受到扭矩馬達磁滯現象的影響，加上零點漂移，流量
對負載壓力的非線性特性；甚至溫度變化引起壓力油黏性的改變，樣樣使
得系統具有複雜的非線性時變特性，造成模式取得與推導不易，徒然增加
控制時的困難。模糊控制利用人性化的語言變數(linguistic variable),
隸屬函數(membership function)及一群模糊控制法則巧妙的迴避了模式
取得障礙的問題也正因模糊控制這個不須取得數學模式的優點使得它在各
種控制領域中都有傑出的表現模糊控制雖然已解決許多複雜系統的控制問
題,然而它對經驗卻毫無累積的能力。本文中更進一步地結合了類神經網
路與模糊控制法則，利用逆傳遞學習法、針對一組控制資料在離線學習之
後，再進行線上參數選擇，以期找出最佳控制法則。並對類神經控制領域
的應用進行探討經由實驗結果顯示類神經模糊控制器具有很好的控制性
能.

Because of the nonlinear time-varying characteristics of
electro- hydraulic servo-cylinder caused by magnetic hysteresis
of torque- motor and change of the oil''s viscosity due to the
various temperatures, and so on, it is very difficult to get
accurate mathematical transfer function from modeling process,
and all this increases the difficulty in controlling.The fuzzy
control using linguistic variables, membership function, and
its own control rules wisely avoids the hardness of modeling.
Therefore, just because the fuzzy control doesn''t need to get
mathematical models , it can be executed in various control
fields, and it works excellently.Though the fuzzy control have
solved many control problems in the very complex systems, it
doesn''t have any ability to accumulate the experiences of
control processes. In this paper, we combine the on-line
searching method with the neuro- fuzzy network theory which is
based on backpropagation method to get optimal control rule.
After working on the application of the neuro network in the
control field, we have found that appliances based on the neuro-
fuzzy control theory and on-line searching method have great
control abilities.