應用線性化技術於撓性關節機械手臂之模態分析與參數鑑別

Abstract

本論文針對高速高精度機械手臂操作過程中，因結構撓性與摩擦行為所引發的振動問題，提出一套完整且系統化的建模、參數鑑別與控制方法。隨著智慧製造與高端自動化需求日益提升，傳統剛體動力學模型已無法全面描述機械手臂於高速運動下的動態行為。因此，本研究首先建立結合剛體與柔性結構的動力學模型，考慮手臂於不同姿態變化下的質量分佈與剛性變異。為確保模型參數的準確性，採用最小參數集法搭配全域優化策略進行動力學參數鑑別，降低識別過程中的耦合問題，並提高鑑別效率。 考量摩擦現象對於高精度控制的影響，引入 LuGre 摩擦模型，完整描述穩態摩擦與暫態摩擦特性，並透過實驗數據進行參數辨識，使模型具備更高的實務適用性。在撓性結構分析方面，建立雙質量撓性模型，深入探討不同運動姿態對自然頻率與振型的影響，進而掌握機械手臂在高動態作業中的模態變化行為。配合實驗模態分析（EMA），驗證理論模型的可靠性與精度。 最後，本研究運用模態更新技術與模型修正策略，將實驗數據與解析模型進行比對與校正，達成動態模型的即時調整與優化。實驗結果證明，所提出的方法能有效掌握並預測機械手臂於運動過程中的模態變異現象，望能提升結構振動的控制能力與作業精度。此研究成果將為未來智慧製造、高速加工與自動化生產系統提供理論依據與技術支援，具備高度的應用價值與推廣潛力。

This thesis addresses the vibration issues induced by structural flexibility and friction during high-speed and high-precision operations of robotic manipulators. A comprehensive and systematic methodology for modeling, parameter identification, and control is proposed. With the rapid development of smart manufacturing and advanced automation, conventional rigid-body dynamic models are no longer sufficient to accurately describe the dynamic behavior of robotic arms under high-speed motion. Therefore, this study establishes a dynamic model that integrates both rigid-body and flexible structural components, taking into account the variations in mass distribution and stiffness under different manipulator postures. To ensure the accuracy of the model parameters, a minimal parameter set combined with a global optimization strategy is adopted for dynamic parameter identification, which reduces coupling issues and enhances identification efficiency. Considering the significant impact of friction on precision control, the LuGre friction model is introduced to comprehensively describe both steady-state and transient friction characteristics. Experimental data are utilized to identify the friction parameters, improving the practical applicability of the model. In the structural flexibility analysis, a two-mass flexible model is constructed to investigate the effects of different operational postures on the natural frequencies and mode shapes, thereby capturing the modal variations during dynamic manipulator operations. Experimental Modal Analysis (EMA) is conducted to verify the reliability and accuracy of the theoretical model. Finally, this research employs modal updating techniques and model correction strategies by comparing experimental data with analytical models, achieving real-time adjustments and optimization of the dynamic model. Experimental results demonstrate that the proposed method effectively captures and predicts the modal variations occurring during manipulator motion, enhancing vibration control capability and operational precision. This study is expected to provide theoretical support and technical guidance for future smart manufacturing, high-speed machining, and automated production systems, offering substantial application value and broad potential for practical implementation.