A sophisticated modeling approach for photovoltaic systems in load frequency control

Abstract

This paper proposes a new model to mimics accurately the real dynamic and steady-state power generation profiles of a photovoltaic (PV) power plant connected to the power system. The model is targeted at load frequency control (LFC) studies, which takes into account the effect of the dynamic behavior of the system frequency on the PV output power. The artificial neural network using the radial bias function (ANN-RBF) is adopted to model the nonlinear behavior of the PV power plant only considering the frequency deviation as one of the inputs at different operating conditions. Also, the ANN prevents going into complex mathematical analysis to get the dynamic behavior of the output power from the PV plant. The input layer of the ANN-RBF receives the temperature, humidity, irradiation level, and frequency deviation, while the output layer represents the output power from the PV power plant. The main idea to include the frequency deviation in the PV modeling is to represent the dynamic behavior of the PV power that is injected by the PV inverter that is synchronized with the power system using the utility grid angle, i.e. system frequency. The ANN-RBF has been trained using actual data collected from a PV power plant in Aswan, Egypt. The ANN dataset contains significant variations of operating parameters to generalize the output modeling of the PV array. Moreover, the shortages in the previous models for the PV system used in the LFC studies, which are the first-order model, second-order model, and noise-based model, are discussed in detail. The simulation results show that the proposed model of the PV system based on the ANN-RBF is more suitable for the LFC studies than the well-known models based on the first-order representation as it takes into account the dynamic behavior of the system frequency during the load disturbance.