• 作者： Chen, Cheng-Liang; Huang, Hsiao-Ping; Chao, Yung-Cheng
• 中文摘要： 本文針對多變數動態系統, 以參數型方式識別程序之轉變函數動態模式。目前為止, 識別多變數動態模式之最大困難在於其中各個時延項通常需單獨估測, 且必須為取樣時間之整數倍。本文以一轉換方式處理識別過程所遭遇的連續時間模式及其參數靈敏度函數, 因此, 模式之時延可以不必限制為取樣時間之整數倍, 時延也可以與其它模式參數同時估計而不必另作處理。本文以一個模擬的例子及一個小型蒸餾塔之動態模式識別的實例, 說明所提多變數動態模式識別方法之適用性。
• 英文摘要： This article proposes a parametric method for identifying an MIMO transfer-function matrix (TFM) as the dynamic model for a multivariable dynamic system. One of the reported difficulties in the identification of MIMO systems is the determination of dead times for each element in the TFM. The proposed method uses modified z-transforms to formulate the parametric optimization problem so that dead times in the TFM can be estimated together with other parameters by the gradient methods. The proposed method removes the limitation that estimated dead times should be integer multiples of the sampling interval. Results of identification of a simulated process with noisy observed data and those of a pilot-scale distillation column demonstrate that the method is useful in obtaining a low- order MIMO parametric model for the purpose of chemical process control.
• 中文關鍵字： 識別; 轉變函數矩陣; 時延; 參數型方法
• 英文關鍵字： Identification; Transfer-Function Matrix; Dead-Time; Parametric Method