本論文根據燃料電池之電化學反應動態模型,將一小功率的燃料電池簡化為一多輸入、多輸出之藕和系統。在不同的操作點,以疑似白色二值信號為輸入,使用線性迴歸模型對系統進行識別,嚐試使用固定參數模型,以及運用最小平方法所得的時變參數模型,以期找出能描述燃料電池系統動態的模型。 識別結果發現,使用固定參數並無法合宜描述系統動態,而以最小平方法所計算出的時變參數模型可以精準地描述系統動態。並且證實了系統動態在35至45度間變化非常小。同時發現當氫氣化學劑量超過1.2時,快速的改變氫氣流量,對燃料電池系統的電壓及電流並不造成影響。
Based on electro-chemical reactions, the small-power fuel cell system is simplified as a coupled multi-input and multi-output system. Utilizing the pseudo random binary sequence as the input and ARMAX as the model, system identification is performed at different operating points. Both fixed-valued parameters and time-varying parameters are used in ARMAX in order to find a suitable model. According to the identification result, ARMAX models with fixed-valued parameters can not describe the system dynamics suitably, while models with time-varying parameters can. System dynamics are also proven to be insensitive to temperature changes between 35 and 45°C. It is also shown that when the hydrogen stoichiometry is over 1.2, fast changes of the hydrogen flow rate does no influence on the fuel cell system performance.