Control of Single Degree of Freedom Magnetic Pendulum with NonLinear Feedback Linearization Method

Yıldız H.

4th International Conference on Scientific and Academic Research, 19 - 20 Temmuz 2024, ss.201, (Özet Bildiri)

  • Yayın Türü: Bildiri / Özet Bildiri
  • Sayfa Sayıları: ss.201
  • İstanbul Üniversitesi-Cerrahpaşa Adresli: Evet

Özet

Torque-based motion actuator (motor) control is a widely used in the control of robotic arms and transport systems. However, producing small-sized motion generators that can be used in the design of micro- and macro-scale systems poses great challenges in applications. Robotic systems designed at this scale can be controlled using sound waves, light, and magnetic fields. Sound waves and light are affected by noise in the environment. The magnetic field is the most frequently used source of motion in the literature since it has not affected by the body environment. In micro and macro-sized robotic systems, the motion mechanism is controlled by external control mechanisms. Although robots are designed as simple systems, they are difficult to control due to the non-linear behavior of the magnetic field. In this study, the control of a single-degree-of-freedom magnetic pendulum system using the feedback linearization (FL) control method, one of the non-linear control methods, is studied. The controller was examined for step function, triangular form, and sinusoidal form input signals. The controller performance was compared with the PID (Proportional-integral-derivative) and PD (Proportional-derivative) control methods. For the step function, RMSE error values for PD, PID, and FL methods were obtained as 3.909364252°, 3.909365135°, and 4.048496914°, respectively; for the triangular function, they were 0.279721839°, 0.279722571°, and 0.206830204°; and for the sinusoidal function, they were 0.300646945°, 0.300646186°, and 0.217901812°. While the PD and PID controller results were very close to each other, they caused permanent errors. In the FL controller, the error was eliminated within 0.1 seconds. Although the PD and PID RMSE errors were lower for the constant reference signal, it was observed that the performance of the FL method was better for time-dependent reference signals, although it required more current compared to linear control methods.