Load impedance estimation and iterative-learning control for a single-phase three-wire inverter

This paper presents a load impedance estimation and iterative-learning control for a single-phase three-wire inverter acting as an uninterruptible power supply (UPS), which can supply unbalanced load, linear load and rectified load. The proposed load impedance estimation scheme is based on an RLC equivalent circuit concept to determine load parameters, RLC, insuring that the inverter can generate sinusoidal voltage. With an iterative learning control, the steady-state error of the output voltage can be reduced significantly. Additionally, the proposed control can accommodate wide inductance variation of the inverter, reducing core size significantly. In the design and implementation, the inverter inductances corresponding to various inductor currents are measured at the start-up and stored in the memory of the controller for scheduling loop gain every switching cycle, and the RLC load parameters are estimated online to determine control (duty ratio). Experimental results measured from a 5 kVA inverter have verified the analysis and discussion.