Prismatic Series Elastic Actuator: Modelling and Control by ICA and PSO-Tuned Fractional Order PID

This paper presents dynamic modelling and control of a linear prismatic series elastic actuator. Since this actuator has the capability of generating large torques, it is increasingly used in the human-assistive robotic systems. Due to having the human in the loop, the actuator requires precise control. A fractional PID controller is used for the control to improve performance because this controller has more additional degrees of freedom than the classical PID. The actuator has one servo driver and five controller gains to be tuned. The gains are optimized using both Particle Swarm Optimization (PSO) and Imperialist Competitive Algorithms (ICA). Comparison of the results from the two optimization methods illustrates that the PSO tuned FOPID controller has a slightly better performance, faster convergence and better settling time. Next, the PSO tuned controller is compared with a Genetic Algorithm (GA) tuned PID controller. It is shown that the PSO tuned FOPID controller continues to offer better performance, especially in terms of rise time and settling time.