Abstract
The cell operating temperature of a photovoltaic (PV) panel is an important information required to calculate its electrical power output. Two computing schemes proposed in this study to estimate the PV cell temperatures employ neural networks to learn and perform nonlinear functional mapping. Output of each computing scheme is the PV cell temperature estimated at 5 minutes intervals corresponding to a given profile of weather inputs, namely the solar radiation intensity and the ambient temperature. A feed forward neural network trained using the back propagation algorithm realizes the nonlinear relationship between weather inputs and the corresponding PV cell temperature. Results of the study show that the cell operating temperature of a PV panel can be very accurately estimated using this approach. The method can be applied in the PV systems simulators and controllers. The same technique may be used for modeling in the other areas where it is hard to obtain deterministic models but easy to obtain input-output data.
