A Nonlinear Controller for Parallel DC-DC Converters with ZIP Load and Constrained Output Voltage

TL;DR

This paper introduces an adaptive nonlinear controller for parallel DC-DC converters with ZIP loads, ensuring voltage regulation, power sharing, and voltage constraints under load uncertainties using barrier functions and adaptive backstepping.

Contribution

It presents a novel barrier-function-based adaptive backstepping controller that guarantees voltage bounds and load adaptation in parallel DC-DC converters with ZIP loads.

Findings

Controller maintains output voltage within bounds during load variations.

On-line parameter identification improves control accuracy.

Simulation results validate effectiveness under uncertain conditions.

Abstract

In this paper, an adaptive nonlinear controller is designed for a parallel DC-DC converter system that feeds an unknown ZIP load, characterized by constant impedance (Z), constant current (I), and constant power (P), at the DC bus. The proposed controller ensures simultaneous voltage adjustment and power sharing in the large signal sense despite uncertainties in ZIP loads, DC input voltages, and other electrical parameters. To keep the output voltage within a desired range, we utilize a barrier function that is invertible, smoothly continuous, and strictly increasing. Its limits at infinity represent the upper and lower bounds for the output voltage. We apply the invertible transformation of the barrier function to the output voltage and then design the controller using the adaptive backstepping method. Using this barrier-function-based adaptive backstepping controller, uncertain parameters are identified on-line, and the voltage adjustment and power sharing objectives are established. Moreover, voltage constraint is not violated event in the presence of sudden and unknown large variations of load. The efficiency of the proposed nonlinear controller is evaluated through simulations of a parallel DC-DC converter system using the MATLAB/Simscape Electrical environment.