查詢結果分析
相關文獻
- 以支持向量迴歸技術為基礎之適應性智能控制
- 支持向量機應用於飛彈飛行控制之研究
- 適應性內模式船舶自航器之設計
- 以適應性方式控制磁力式新生兒搖籃擺盪的響應探討
- Gain Scheduling Adaptive Control of Superconducting Magnetic Energy Storage Unit for Power System Stabilization
- 人工神經網路在紡織品手感評估中的應用研究
- 未知延遲時間的線性控制系統之類神經適應性控制
- 自動化鑄件毛邊磨修機
- 考慮致動器動態效應於機械臂之適應性軌跡控制
- On the Adaptive Tracking Control for Manipulator Arms Considering Electrical Actuator Dynamics
頁籤選單縮合
題名 | 以支持向量迴歸技術為基礎之適應性智能控制=A Support Vector Regression Technique Based Adaptive Intelligent Control |
---|---|
作者姓名(中文) | 鄧世剛; 林聖義; 劉達明; | 書刊名 | 中正嶺學報 |
卷期 | 41:1(A) 2012.05[民101.05] |
頁次 | 頁177-186 |
分類號 | 448.94 |
關鍵詞 | 支持向量迴歸; 智能控制; 適應性控制; Support vector regression; Intelligent control; Adaptive control; |
語文 | 中文(Chinese) |
中文摘要 | 本研究以支持向量迴歸(Support Vector Regression, SVR)技術,設計一個能即時線上學習並進行適應性控制的智能控制器─支持向量迴歸控制器(SVR Controller, SVRC),並將之實現在直流馬達轉速控制。研究中以閉迴路系統回授機制來得到誤差信號,將此誤差信號轉換成比例、積分與微分等三種誤差信號,用來在每個時間步階裡不斷地重複訓練控制器,使其能精準掌握控制命令變化趨勢,進而達成適應性控制的目的。模擬實驗中以步階命令進行測試,逐步探討SVR參數選定、即時訓練資料點數量及時間步階大小對控制效能的影響,找出最適合直流馬達轉速控制的SVRC設計,最後並將模擬結果與傳統PID控制結果比較,證實SVRC具有極優異的控制效能。 |
英文摘要 | This research applied the Support Vector Regression (SVR) technique to design an on-line learning adaptive intelligent controller, called SVR controller (SVRC), to perform a DC motor speed control. In this research, we used the error signals feedback from the closed loop DC motor control system, transformed them to the proportional, integral and differential type of signals, to train the SVRC on-line in every time step repeatedly to handle the trend of command precisely. Then, we used the trained SVRC to perform the DC motor control on-line adaptively. Within the simulation, a unit step function was designed to perform the test works for selecting the SVR parameters, the quantity of the training data for on-line training, and the size of the time step for the efficiency of SVRC in motor speed control system. Finally, the SVRC simulation results were compare with the traditional PID controller results to conform that SVRC has excellent adaptive control ability. |
本系統之摘要資訊系依該期刊論文摘要之資訊為主。