PID-neuron controller design for rotary inverted pendulum system

Abstract

Rotary inverted pendulum is a nonlinear system and a popular model in testing the control algorithm. When using PID control for rotary inverted pendulum system, parameters Kp, Ki, Kd are selected by experimental methods. Nevertheless, those parameters make the system stable but with large swing. This paper applies artificial neural network to calibrate online PID parameters (PID-neuron). These parameters are changed progressively to ensure the stability of the system and smaller swing. The simulation performed in Matlab/Simulink environment shows that parameters Kp, Ki, Kd which are adjusted online make the system stable. The control algorithm is tested in the real rotary inverted pendulum in Matlab/Simulink environment with data acquisition card DSP-F2812. Experimental results show that the proposed PID-neuron controller gives less swing for the pendulum angle than the static PID controller does.