An international group of researchers has used feature selection-based artificial neural networks (ANN) with various machine-learning algorithms to predict the OTA of PV systems. They ran their experiments at 37 locations in India, and depending on the evaluated ANN, accuracy improvements ranged from 38.59% to 90.72%.
“The OTA of the solar panels is one of the crucial variables that determine the effective installation and operation of PV systems,” said the group. “An OTA allows the sun’s rays to be absorbed by the material to the maximum extent possible. When collecting solar energy, PV panels are impacted by the angle at which light from various directions reaches them. Consequently, the yearly energy production of the system is directly affected by the selection of the suitable tilt angle.”
The experiment was based on information taken from the NASA Power Data Access Viewer website and included the following parameters: global solar radiation (SR), diffuse SR, extraterrestrial SR, global SR on a tilted surface, and the clearness index. They extracted data from 37 Indian cities, including New Delhi, Mumbai, Bangalore, and Kolkata.
“In machine learning and data analysis, feature selection is a crucial step that involves determining which variables and predictors in a dataset are most significant and add to a model’s predictive potential,” explained the academic team. “Relevant feature selection minimizes overfitting, increases interpretability, and boosts model accuracy. A list of feature selection methods is provided in the following subsections.”
The team used feature selection techniques, including the Pearson correlation coefficient, to assess the strength of data associations and the signal-to-noise ratio to simplify the process in high noise conditions. After applying these methods, they chose not to include extraterrestrial SR in the ANN prediction models.
“The inputs of global SR, diffused SR and monthly average clearness index showed a stronger negative relationship with the output OTA as compared to the extraterrestrial radiation which showed a negligible relationship with the output of OTA,” they said. “The global SR on tilted surfaces showed a moderate positive relationship. A negative correlation means as global SR, diffused SR, and monthly average clearness index increase, OTA decreases. The input of extraterrestrial radiation has a lower signal-to-noise value, implying that its correlation with OTA is low.”
Following that, six ANN algorithms were tested, and their predicted tilt angle was compared to real OTA target values. Then, the improvement in prediction accuracy (IPA) was calculated to show how the feature selection had improved the mean squared error (MSE) in compression to the MSE of the calculation with all of the parameters.
The lowest IPA was recorded in the case of the scaled conjugate gradient (SCG), showing an improvement of 38.59%. That was followed by an IPA of 53.33% with the Levenberg -Marquardt (LM) case and 66.93% for the Polak-Ribiere conjugate gradient (PRCG). One-step secant (OSS) had an IPA of 86.88%, while the Broyden-Fletcher-Goldfarb-Shanno (BFGS) recorded 89.53%. Elman neural network (ELM) has provided the best improvement, of 90.72%.
“The developed models in this study are used to optimize energy production, increase efficiency, and make well-informed judgments, solar panel tilt angles,” said the academics. “The model assists industry participants in attaining improved results and promoting the use of OTA by predicting at different sites. Future research work can be focused on OTA prediction using highly accurate measured values of solar radiation and incorporating other factors such as dust, pollution, and aerosols. Other region-specific OTA models will be developed and validated across different climatic conditions.”
The presented their results in “Novel feature selection based ANN for optimal solar panels tilt angles prediction in micro grid,” which was recently published in Case Studies in Thermal Engineering. The study was conducted by scientists from India’s SR University, IIMT University, Government College Hamirpur, COER University, South Korea’s Gachon University, and Hungary’s Eötvös Loránd University.
This content is protected by copyright and may not be reused. If you want to cooperate with us and would like to reuse some of our content, please contact: editors@pv-magazine.com.
re OTA using complex software to decide. The trend is to make systems that are increasing in complexity. I find that somewhat troubling in that user’s will be unable to diagnose problems and will be leaning on developers more.