基於雙指標策略追蹤之SMC-PID控制演算法結合壓電感測應用於多軸機械手臂振動分析與精密定位之研究

Abstract

隨著21世紀邁入智慧高科技時代，工業4.0浪潮在人工智慧、物聯網與大數據等技術推動下，促使傳統人工作業逐漸朝向自動化生產轉型。機械手臂憑藉其高自由度、高能效與精確控制等優勢，已成為現代智慧工廠中關鍵性自動化設備，廣泛應用於物件抓取、搬運、焊接與高精密組裝等任務中。針對多軸機械手臂系統常見的非線性與動態不確定性特性，本研究提出一套結合滑模控制（Sliding Mode Control, SMC）與比例-積分-微分控制（PID）的複合式控制策略，藉以提升系統穩定性與軌跡追蹤精度。 本研究引入雙指標追蹤策略（Dual-Index Tracking Strategy），整合虛擬軌跡規劃（Pre-Pointer）與實體軌跡執行（Post-Pointer）兩階段控制機制，以進一步強化動態回應能力與控制穩健性。透過STM32F407微控制器與Arduino所構建之硬體平台，並結合MATLAB模擬系統進行實驗驗證，結果顯示：本控制架構相較於PID控制器，穩定時間從17.805秒顯著縮短至6.24秒，X軸與Y軸超越量分別降低81.25%與87.5%，而穩態誤差亦由0.9848mm明顯減少至0.1183mm，展現出優異的控制性能與系統響應特性。 此外，透過Modal 356A32壓電式加速度感測器所擷取之振動訊號，並結合FFT所進行之頻域分析，亦可作為後續振動抑制與智慧診斷系統開發之依據，進一步拓展本研究於智慧控制與穩健系統設計領域之應用潛力。 綜合上述成果，所提出之複合控制架構具備高度穩定性、即時性與系統彈性，具潛力應用於智慧製造、醫療手術機器人以及高精密自動化生產線等領域，對未來先進製程與智慧系統之發展具有高度應用價值與實務貢獻。

As the world enters the era of intelligent high technology, the wave of Industry 4.0-driven by artificial intelligence, the Internet of Things (IoT), and big data—has accelerated the transformation of traditional manual operations toward automated production. Robotic arms have emerged as critical components in smart factories due to their high degrees of freedom, energy efficiency, and precise controllability, and are widely applied in object manipulation, handling, welding, and precision assembly. To address the nonlinearities and dynamic uncertainties in multi-axis robotic arm systems, this study proposes a hybrid control strategy integrating Sliding Mode Control (SMC) with Proportional–Integral–Derivative (PID) control to enhance system stability and trajectory tracking. A Dual-Index Tracking Strategy combining virtual trajectory planning (Pre-Pointer) and physical trajectory execution (Post-Pointer) is also introduced to improve dynamic response and control robustness. The system was implemented on a hardware platform using an STM32F407 microcontroller and Arduino, and validated via MATLAB/Simulink simulation. Results show that compared to conventional PID control, the proposed scheme reduced settling time from 17.805 to 6.24 seconds, decreased X-axis and Y-axis overshoot by 81.25% and 87.5%, and lowered steady-state error from 0.9848 mm to 0.1183 mm, demonstrating superior control performance. Additionally, vibration signals collected from the end-effector using a Modal 356A32 piezoelectric accelerometer were analyzed via Fast Fourier Transform (FFT), providing a basis for future development in vibration suppression and intelligent diagnostics. In summary, the proposed control architecture offers high stability, responsiveness, and flexibility, with strong potential for applications in smart manufacturing, surgical robotics, and precision automation.