Adaptive hysteresis compensation control of a macro-fiber composite bimorph by improved reinforcement learning

Xingqiu Li,Kaiming Hu,Hua Li,Ban Wang,Suan Xu,Yuchen He
DOI: https://doi.org/10.1177/1045389x241273047
IF: 2.7
2024-10-01
Journal of Intelligent Material Systems and Structures
Abstract:Journal of Intelligent Material Systems and Structures, Ahead of Print. The hysteresis characteristics related to the frequency and amplitude of the control signal seriously affect the precision of the displacement tracking control of the macro-fiber composite (MFC) bimorph. The traditional feedforward compensator tends to exhibit low precision in controlling displacement. Although it enhances the control accuracy to a certain extent by incorporating a feedback controller, the existing feedback controller has a weak adaptive ability. Thus, the Q-learning (QL) algorithm is combined with the Bouc-Wen (BW) feedforward compensator in this study. The output voltage from the BW compensator has a more significant impact on the control accuracy and convergence speed of the QL algorithm. Therefore, this paper proposes an improved QL (IQL) algorithm that leverages the control error. Moreover, a BW-IQL adaptive control method is proposed to realize precise adaptive control of the MFC bimorph. Experimental comparisons are performed with the proposed method and the traditional BW, BW-PID, and BW-fuzzy PID controllers. The BW-IQL controller reduces the average relative errors of the three controllers by 86.9%, 59.9%, and 41.8%, respectively. Meanwhile, the Q̄ table plays a major role in error suppression in the IQL controller. These results verify that the BW-IQL control method has higher adaptability and accuracy.
materials science, multidisciplinary
What problem does this paper attempt to address?