- 學位論文
應用粒子群最佳化演算法於雙軸壓電致動撓性平台之建模與控制
Modeling and control of a biaxial PZT actuated flexible-base-stage using PSO
摘要
本論文介紹了一種撓性架構的雙軸壓電致動精密定位平台。由於壓電平台移動時會產生遲滯現象，進而使得壓電致動平台的定位精度下降，為了補償遲滯現象所產生的定位誤差，設計了一個以遲滯模型為基礎的前饋控制器。在本文中使用三種遲滯模型來建立壓電平台的系統模型，為了獲得模型的最佳參數，使用了基因演算法、粒子群演算法及免疫克隆演算法三種搜尋最佳化參數的演算法來計算模型參數，並比較這三種演算法的計算速度及收斂性能;經由計算結果證明，粒子群演算法有較快的計算速度及較好的收斂性能。最後，我們將利用粒子群演算法所計算出來的最佳參數應用於前饋控制器，並完成一個及時定位控制實驗，實驗結果證明利用該參數所設計的前饋控制器能有效的改善遲滯現象造成的誤差。更進一步的，我們將前饋控制架構結合PI形式之回饋控制器，實現了高精度的定位控制。
並列摘要
This paper presents a micropositioner which utilizes a monolithic flexure-based mechanism and is actuated by the embedded piezoelectric actuators to achieve the translations in the X- and Y-axes. The PEA-actuated FBM is used to compensate the residual tracking error of the coarse stage. However, hysteretic nonlinearity limits the positioning accuracy of the PEA; To compensate nonlinear hysteresis problem, a feedforward controller based on the identification method is proposed. The dynamics of the hysteresis is formulated by three hysteresis models, and to identify the optimal parameters of the hysteresis model, the real-coded genetic algorithm (RCGA), the particle swarm optimization (PSO) and the clonal selection algorithm (CSA) are studied and discussed. We compared the performance of their convergence and computing time and the comparisons of numerical simulations and experimental results prove that the PSO has better performance and uses this identification method for the FBM is feasible. Finally, to verify the consistency, two micro-contouring tasks are implemented by the proposed feedforward controller with the feedback of linear optical scales in DSP based real-time control architecture. For further improvements, we compare the feedforward controller and the PI type feedback controller to realize the ultra-precision positioning control.