標題: | 比例閥開迴路學習控制系統之研究 Research of open-loop learning control of proportional valve |
作者: | 許洽道 Sheu, Shya-Daw 唐佩忠 Pei-Chong Tang 電控工程研究所 |
關鍵字: | 開迴路;控制;比例閥;open-loop;control;proportional valve |
公開日期: | 1997 |
摘要: | 比例閥在工業上是很普遍使用的電磁閥,由於比例閥的頻寬低以及時間延 遲的原因,大致上都是使用開迴路控制,因此無法做到比較精確的控制, 本論文的研究目的就是希望能夠在開迴路的控制基礎上,使用學習控制的 方式,提昇比例閥動作的精度。在我們的研究中,射出成形機的速度控制 皆使用比例閥,我們從射料系統的射出速度響應量測得比例閥的特性,並 利用系統判別方法判別出比例閥的一階線性系統,且依據量測得到的非線 性特性,建立比例閥的非線性模型。我們首先以一階線性系統作為受控系 統,來分析開迴路學習系統的特性,控制方法使用類似閉迴路控制中使用 的PID控制方式,並針對saturation、disturbance、雜訊、時間延遲以及 其他非線性現象進行各種模擬,模擬的結果顯示開迴路學習控制可以增進 系統響應的精度,對於閉迴路難以控制的時間延遲問題可以有效的解決, 對於以上談到非線性的影響也有抗拒能力。最後我們以比例閥的非線性模 型進行開迴路學習控制模擬,結果顯示可達到不錯的響應。未來,我們希 望能夠直接應用在實際系統上,改善開迴路控制的精度。 The use of hydraulic proportional valves in industry is very popular. Because ofthe low frequency response and the effect of time delay in valve dynamics, open-loopcontrol is usually used in proportional valves. So it can not get high- accuracyperformance. In our study, a open-loop learning control system is proposed to improvethe accuracy of proportional valves based on the hardware of original open-loopcontrol system. The system we faced is the inject system of an inject machine and the injecting speed is controlled by a proportional valve. First, we got the first-order linearmodel by the method of identification and then we built the nonlinear model of this system based on the nonlinear characteristics from the analysis of the data we measured. To analyze the property of open-loop learning control system, a similar PID controller like in closed-loop control system was used in a first-order linear system in our simulation of open-loop learning control system. Some nonlinear situations like saturation, disturbance, noise, delay, etc. were also included in our simulation. The results have shown that open-loop learning control can improve the accuracy of the system and could overcome the effect of some nonlinear conditions. In the end, we applied the nonlinear model of the proportional valve to the simulation, the results also showed that it could achieve good response. In the future, we hope to apply the open-loop learning control system to a real system to improve the accuracy. |
URI: | http://140.113.39.130/cdrfb3/record/nctu/#NT860591022 http://hdl.handle.net/11536/63198 |
Appears in Collections: | Thesis |