查詢結果分析
來源資料
頁籤選單縮合
題名 | Online Positional Control Using the CMAC Methods= |
---|---|
作者 | 陶逖; 范恆康; |
期刊 | 光武學報 |
出版日期 | 19981100 |
卷期 | 23 1998.11[民87.11] |
頁次 | 頁15-32 |
分類號 | 448.942 |
語文 | eng |
關鍵詞 | 類小腦類神經網; |
英文摘要 | In online positional control process, they are usually three problems have to be solved. The first is choose a good controller that can havdle the uncertainties and nonlinearty of a plant. The PID controller has the good ability to solve above problems, so it is sometimes used to trace a reference position. But the parameters (Kp, Ki and Kd) of PID controller must be tuned in the control process. So the second problem is find a good method to tune control parameters adaptively. The neural network [3] methodology is always presented to learn the model of an unknown plant. Unfortunately, the learning process may require many interations to converge or even cannot converge due to various reasons. Therefore, such kinds of neural networks are inappropriate for online controlling. The third problem is how to online learn and control a system. The method of CMAC (Cerebellar Model Articulation Controller [1,3]) is proposed to overcome some of the above problems. CMAC can be said to be a learning mechanism that imitates the cerebellum of human being and has demonstrated good learning speed and convergence property in various researches [6,7]. In this paper, a based on CMAC PD controller is employed to online learn and trace a reference position, and the simulation results show the superiority of using the proposed method. |
本系統之摘要資訊系依該期刊論文摘要之資訊為主。