ANN-BASED MODEL FOR DAILY SOLAR RADIATION PREDICTION WITH A LOW NUMBER OF HIDDEN NEURONS AND OPTIMAL INPUTS

Tóm tắt

Solar radiation has been the focus of many studies over the past years due to its usefulness in clean energy generation through photovoltaic
(PV) systems. For both standalone and grid-connected PV systems, it is necessary to have solar radiation information beforehand (accurate
prediction) as it is used for the PV systems design, for making power dispatching plans, for potential future PV system feasibility, etc.… In
this article, an Artificial Neural Network (ANN)-based model for daily global solar radiation prediction is proposed. This model is trained
with a backpropagation algorithm and make prediction using meteorological variables as inputs. While keeping the accuracy at high level,
the model is built using a low number of neurons in the ANN’s hidden layer and most effective input variables. As a result, the proposed
model is simpler as compared to many existing models. First, the minimum and the maximum numbers of hidden neurons are calculated.
Second, results of numerical simulations based on a trial-and-error method can show us the low number of hidden neurons and the optimal
inputs for the model. A simple model is preferred when the performance level is not affected. In addition, a simple model is easy to implement
and requires lower computer’s capacity; which is beneficial economically.