本研究之目的在於設計一自調式類神經適應清洗濾波器,以期能在受限的動感平台運動空間內,針對不同的飛機操縱,能藉由類神經倒傳遞演算法則將清洗濾波器參數即時調整,使得位移與角度等穩態濾除,同時儘可能將具有加速感之動態行爲展現出來,藉以達到動感模擬之最高境界。於論文中,我們採用自調式類神經PID的概念,利用神經網路調整清洗濾波器之參數。 另外,我們對自調式最佳清洗濾波器提出一嶄新的架構,藉由最佳控制法則取得最佳參數後,再利用類神經網路作即時調整,藉此探討是否能產生更佳的效能。
The purpose of this research is to design an auto-tuning adaptive washout filter with neural network to demonstrate the dynamics of actual system. The parameters of the adaptive washout filter can be auto-tuned immediately with neural network by error back-propagation algorithm. It will be utilized to demonstrate the dynamic behavior of the acceleration motion cueing by washing out the steady state value, such as displacement and attitude angle, and passing al the transient signals as possible. In this study, we adopt the concept of auto-timing PID to tune the parameters of washout filter by using neural network. In addition, we propose a new architecture for the auto-tuning adaptive optimal washout filter. Using the optimal control methodology, we can obtain the optimal parameters and then use neural network to tune some parameters of optimal washout filter in real time. By this method, we can upgrade the performance of the optimal washout filter.