Performance improvement of a photovoltaic system using a controller redesign based on numerical modeling

• A method to model PV systems based on numerical simulation data is presented.
• A common PV system with DC output is modeled using the technique introduced.
• Empirical PI-based controllers are redesigned with model-based hybrid PI/LQG methods.
• Redesigned system outperforms the standard one under all atmospheric/load conditions.
• Redesigned system's robustness to uncertainty in system parameters is also shown.

This paper focuses on the utilization of numerical modeling and simulation to improve the performance of a theoretically designed stand-alone photovoltaic (PV) system with constant DC voltage. A theoretically designed system based on standard methods found in the literature is modeled and simulated numerically. This system reveals some unexpected behavior when it is subjected to certain irradiation, temperature and load changes. This behavior is due to the parameters and dynamics of real circuit elements not taken into account by the design. It is also problematic that one cannot use conventional linearization methods to model the system and design the controller because the system contains large nonlinearities caused by elements such as DC/DC switching converters driven by pulse width modulation (PWM). This requires the use of an alternate technique based on the use of simulated input/output data to determine an operating point around which a linear system model is derivable. The controllers for the PV system are redesigned using these models, and the closed-loop system is simulated with variable temperature, irradiation and load levels. Upon evaluating the system performance reveals that the redesigned control system is capable of operating the PV panel at its maximum power point under different atmospheric and load conditions and can provide a constant DC voltage to the critical load while charging the battery with the extra power from the panel.