ABSTRACT
The energy crisis and global warming are the two significant factors that have led to the accelerated development of PV power plants. Accurate prediction of PV plant power output is crucial to enhance its competitiveness in the renewable energy market, maintain grid stability, and reduce reliance on conventional energy sources. The present study uses nine machine-learning methods to estimate PV plant energy output. The usefulness of the examined models for real-time energy output prediction is assessed to guarantee optimum security and management in the renewable energy sector. The datasets used in this study represent data from 01 Jan 2019 to 30 Nov 2022 and pertain to the solar PV plant located in Bhopal city. The specific application of these methods to the actual plant dataset adds practical relevance, providing insight into their feasibility in a real-world situation. The results were compared using five statistical metrics (MAE, MSE, RMSE, R2, and MAPE) to demonstrate the study’s quality and reproducibility. The results show that MAE, MSE, RMSE, R2, and MAPE values of all algorithms range from 19.7 kWh to 76.6 kWh, from 5408.0 kWh to 14,958.0 kWh, from 73.5 kWh to 122.3 kWh, from 86.5% to 98.4%, from 3.7% to 12.9% respectively. The ANN and the RF models demonstrated superior accuracy compared to other models, with ANN being the top-performing model.
Abbreviations
AI | = | Artificial intelligence |
ANN | = | Artificial neural network |
AOD | = | Aerosol optical depth |
CNN | = | Convolution neural network |
DL | = | Deep learning |
DT | = | Decision Tree |
GBT | = | Gradient boosting tree |
GRU | = | Gated Recurrent Unit |
k-NN | = | Kernel nearest neighbor |
KPI | = | Key performance indicators |
LR | = | Linear regression |
LSTM | = | Long short-term memory |
MAE | = | Mean absolute error |
MAPE | = | Mean absolute percentage error |
ML | = | Machine learning |
MLP | = | Multi-layer perceptron |
MODIS | = | Moderate resolution imaging spectroradiometer |
MSE | = | Mean Squared error |
PDC | = | Probability density curve |
PSO-ELM | = | Particle swarm optimization extreme learning machine |
PV | = | Photo Voltaic |
RF | = | Random forest |
RMSE | = | Root mean squared error |
RNN | = | Recurrent neural network |
SVM | = | Support vector machine |
Acknowledgements
The researchers are thankful to the E&M department of NIT Bhopal for providing us with the data to carry out this research work and to the faculties of NIT Bhopal for supporting this study. The researchers also thank all the authors and co-authors mentioned in the references.
Authors’ contributions
Bharat Girdhani: Conceptualization, Investigation, Formal analysis, Resources, Writing – original draft, Writing – review & editing. Meena Agrawal: review & supervision.
Disclosure statement
No potential conflict of interest was reported by the author(s).
Availability of data and materials
The datasets used and/or analyzed during the current study are available from the corresponding author upon reasonable request after prior permission from NIT Bhopal.
Additional information
Notes on contributors
Bharat Girdhani
Bharat Girdhani received his master’s in Renewable Energy, focusing on modeling and optimizing hybrid energy systems from the Energy Centre, Maulana Azad National Institute of Technology, Bhopal. He is a Ph.D. research scholar at the Energy Centre, MANIT Bhopal. His fields of interest are Artificial intelligence for Renewable energy and sustainability.
Meena Agrawal
Meena Agrawal is the Associate Professor at Energy Centre, MANIT Bhopal, having profound teaching experience of over 30 years, including international teaching experience in Rwanda, Africa. Her areas of interest are Renewable Energy Systems, Smart Grid & Microgrid, Smart Green Cities, Energy Audit & Conservation, Artificial Intelligence - Multi-Agent Systems, and Human Resource Energy management. She did her Ph.D. in Energy from MANIT, Bhopal, Bhopal, M.E. (Power Electronics) from SGSITS, Indore, and B.E. (Hons) in Electrical Engineering from MANIT, Bhopal.