Degradation Assessment Through Prediction Approaches for Solar PV System in South India

An essential element that directly affects the economic viability and financial sustainability of solar photovoltaic (PV) systems throughout their operational lifetime is the rate of degradation of these systems. The performance of PV systems has traditionally been believed to degrade linearly over time, declining at a fixed rate. The degradation rate, however, is frequently nonlinear, as observed from the operational nature of the PV system. The considered system for the present objective includes a section of 380 kWp PV system evaluated for a period of 10 months. Three prediction-based approaches for evaluating degradation rate have been formulated which includes feed-forward neural network, radial bias (exact fit) and radial bias (fewer neurons). The RMSE found for these methods were 8.755 × 10–8, 0.00181, and 0.00169, respectively. Amongst the developed models, feedforward backpropagation based model approach predicts a close match with the experimental/actual degradation rate factor (Rd). PV system owners and investors may experience financial instability if this nonlinearity in the degradation rate is ignored.