Event-triggered Online MPC Model Calibration for Photovoltaic Power Optimizers

Model predictive control (MPC) is a discrete-time control method for power converters in which control efficacy depends on the model values. A critical issue with MPC is parameter mismatch. In power electronics, parameters of the converter model equations are subject to mismatch including but not limited to capacitor values, equivalent series resistances, and load values. Solar photovoltaic (PV) dc optimizers and microinverters placed on the back sheet of the solar panel are exposed to excessive heat compared to an inverter in a less austere environment. Thus, exacerbating the effects of parameter mismatch. The contribution of this paper is a calibration technique to identify system parameters, augmented within the MPC framework; hence, mitigating parameter mismatch and improving tracking efficacy. The significance of this method is utilizing an error signal within the MPC control loop as an event-based trigger for the calibration process without the need for any sophisticated detection methods.