本篇論文當中,主要探討回歸型小腦模型計算器分別於控制工程及訊號處理之應用。首先,於控制工程中本文提出一種具有穩定特性及良好控制性能的失效補償及調節系統的設計並應用於多輸入多輸出的人形機器人上。模擬結果顯示本系統可以達成良好的失效回復及性能回復能力。然後,於訊號處理的領域中,本文提出一種具有穩定性且能快速收斂的法則,並應用於通道等化器及雜訊消除上。從理論分析及模擬結果中,顯示本系統可達成系統干擾之強健特性及通道之近似特性。因此,本文所提出之系統能成功地應用於控制工程及訊號處理。
This thesis focuses on the design of recurrent cerebellar model articulation computer (RCMAC) and its application to control engineering and signal processing. In control engineering, the stability and performance properties of the proposed fault compensation and accommodation scheme in the presence of system failure are rigorously established. A design method of RCMAC for multi-input multi-output nonlinear systems is developed and is applied to a biped robot. Simulation results illustrate the effectiveness of the RCMAC-based fault compensation methodology in a nine-link robotic system. For signal processing, a stability and fast convergence algorithm is developed and is applied to a nonlinear communication channel equalization and an adaptive noise cancellation problems. The simulation results show the effectiveness of the proposed method for the channel equalization and noise cancellation problems. Thus, the proposed method can apparently applied to control engineering and signal processing.