Innovative technique to estimate the performance of a photovoltaic generator using the parameters of the single diode model equivalent circuit

The present work has been performed at Politecnico di Torino and at the Univeristy of Jaen. The rated power of a PhotoVoltaic (PV) module is provided by the manufacturer in Standard Test Conditions (STC), corresponding to cell temperature Tc=25 °C and irradiance G=1000 W/m2. However, PV generators generally work in very different conditions and, in order to assess their performance, several models have been developed in literature. The Osterwald Model (OM) is the most used, in which, generated power is proportional to irradiance and dependent on temperature by the γ coefficient, provided by the manufacturer. Even if this model is one of the most accurate, it presents a not negligible error regarding the estimation of the maximum power and the productivity of a PV module (3%-6%). However, other models are required for a more accurate evaluation. The performance of a PV module can be described by several electrical equivalent circuits. In this work, the Single Diode Model (SDM) is considered for completely irradiated modules. The SDM is described by five parameters: the photovoltaic current Iph, the saturation current I0, the non ideality factor n, the series resistance Rs, and the shunt resistance Rsh. In the first part of the thesis, at Politecnico di Torino, the parameters of SDM for two PV modules (monocrystalline and polycrystalline silicon) are extracted starting from experimental I-V curves by a curve fitting procedure based on Levenberg-Marquardt (LM) algorithm. Then, correlations are identified, which describe the dependence of the parameters with respect to irradiance and cell temperature. These equations are obtained modifying the coefficients of correlations already present in literature by a nonlinear regression of the extracted parameters. Correlations are used to estimate the maximum power in measurement conditions and the obtained results are compared with experimental data and the results from OM. In the second part of the thesis, at the University of Jaen, the above described analysis is performed on another monocrystalline silicon PV module. The results are compared with the ones obtained from a different extraction procedure based on two numerical methods (Simulated-Annealing e Nelder-Mead, SA-NM). Finally, the energy produced by the module during 20 days of measurements is estimated with the two models and with OM. By comparing these results with the experimental measurements, both the analysed models (LM and SA-NM) estimate the maximum power and the generated energy with higher accuracy with respect to OM. In the first case, the root mean square error is equal to 3.40 %, 3.58 %, and 6.14 %, respectively for, LM, SA-NM e OM. Regarding the energy, the obtained errors are equal to 1.55 %, 2.23 %, and 6,61 % for, respectively, LM, SA-NM e OM. In conclusion, the proposed extraction procedure in Politecnico di Torino (LM) is the most accurate. In future, this technique will be implemented to performed the real time diagnostics of a PV generator. In particular, the state of health of the PV modules will be checked by comparing reference parameters, evaluated by the proposed correlations in real operating conditions, with experimental parameters.