Akademik Veri Yönetim Sistemi
5th International Eurasian Conference on Science, Engineering and Technology (EurasianSciEnTech 2024), Ankara, Türkiye, 26 - 28 Haziran 2024, ss.909-915
Pneumatic Artificial Muscles (PAM) are a type of soft actuator with properties of high force-to-weight ratio, low cost, and inherent compliance which facilitates their use for exoskeletons and rehabilitation robots. On the other hand, the most important factor restricting the use of PAM is its inherent nonlinear characteristics which create complications in modelling and control actions. To overcome the restriction issue, there are many modelling studies in literature, such as geometrical, empirical, and phenomenological models. Those approaches are limited to implementation or approximate ones, such as varying stiffness models, utilizing nonlinear direct input-output relationships. Based on the experiments to characterize the PAM in our test bed, the dynamic response of PAM to variation in pressure input results in a simultaneous force generation and muscle length change. Therefore, the existing direct input-output identification based modelling methods are not suitable enough for modelling complex
dynamic behaviour. Hence, the inverse dynamic modelling approach is proposed to combine simultaneous force generation and muscle length change to utilize PAM in control applications. As an addendum to our previous inverse modelling approaches, in this work, a black box inverse dynamic model is developed as the Hammerstein-Wiener estimation, using physical data obtained from the PAM test bed. According to validation results, the Hammerstein-Wiener-based inverse model has demonstrated a satisfactory performance that concludes it could be used as a simple and effective solution for PAM modelling and control.