• 作者： Chang, Ming-Ping; Chang, Der-Shing; Yu, Cheng-Ching
• 中文摘要： 狀態估測是程序監測及控制重要的一部分。但由於計算及模式化的複雜，這是程序系統工程較少被實際應用的一環。本文以可變結構類神經網路來進行狀況估測，這減少了模式化的負擔。此類神經網路介於前饋式網路及迴遞式網路之間。若系統狀態是不可量測，部分網路結構會自我調整成迴遞式網路。換言之，網路結構是由系統可量測情況來決定。我們以一個反應系統來描述此類神經網路的建構及狀態估測過程，結果顯示這是一個相當有效的狀態估測方式，而且這可由標準的類神經網路軟體來執行。
• 英文摘要： State estimation is an integral part of process monitoring, diagnosis and control. Due to the mathematical complexity of nonlinear model, optimal state estimation is much less established in practice. In this work, the variable-structure neural network (VSNN) of Tung et al. is employed for nonlinear state estimation. Similar to the Kalman filter approach, the filter gain is adjusted according to the ratio of noise and error covariance. An algorithm is proposed to approximate the error covariance and, subsequently, results in a non-iterative procedure in the filter gain calculation. Moreover, when some of the states are not measurable, the VSNN naturally results in a RecN-like architecture for the unmeasured states. A chemical reactor example is used to demonstrate the effectiveness of the VSNN state estimation scheme. Results show that the variable-structure neural network provides effective state estimation for nonlinear chemical reaction systems. More importantly, all the on-line computing can be carried out with standard neural network software in a non-iterative manner.
• 中文關鍵字： 非線性狀態估測; 卡門濾波器; 類神經網路; 網路架構
• 英文關鍵字： Nonlinear State Estimation; Kalman Filter; Neural Network; Network Architecture