Sanket Sudhakar Thul, Ujwal Pankaj Tiwari, Rani Singh | International Journal of Renewable Energy and its Commercialization | Vol 12, Issue 02 | ISSN: 2582-4120
Abstract
India is trying to use renewable energy such as solar and wind to produce extreme and unpredictable fluctuations within India's National Smart Grid. These extreme fluctuations create difficulties for traditional load forecasting methods, including Autoregressive Integrated Moving Average (ARIMA) models, to provide stable and efficient operation through the creation of accurate short-term load forecasts. To accomplish this, the research aims to develop high-quality predictive models by comparing three modelling methods: the statistical Autoregressive Integrated Moving Average (ARIMA), the Recurrent Long Short-Term Memory (LSTM) Network, and the Simulated Transformer Model, which is a new way of modelling based on transformers. ́The modelling methods were trained and validated using actual historical load and weather data from four renewable resource-rich states of India. The models were evaluated using standard performance metrics (Mean Absolute Error (MAE) and Root Mean Squared Error (RMSE)). The results demonstrate that the Simulated Transformer provides superior predictive capacity over ARIMA model predictions hence, the Simulated Transformer reduces the error generated by ARIMA by approximately 71.8% when developing predictions.The findings of this research support the conclusion that the Transformer model was able to understand complex, long-term trends in variable net load profiles through its self-attention mechanism. Therefore, based on this study’s results, it was determined that the Transformer architecture is currently the best-performing and most accurate method of short-term load forecasting (STLF) for this difficult forecasting environment. A prototype Grid Operator Dashboard is also created in this study that converts accurate forecasts into meaningful real-time information for grid reliability.
Keywords—
Load Forecasting, Transformer Model, Renewable Energy Integration, Smart Grid Stability, Performance Evaluation, Deep Learning, LSTM, ARIMA, Time-Series Forecasting, India.
References
[1] Salehimehr S., “Short-term load forecasting in smart grids using artificial intelligence,” IET Generation, Transmission & Distribution, 2022. [2] Upadhyay A., Garg D., Singh M., “Short Term Load Forecasting for Smart Grids Using Apache Spark and a Modified Transformer Model,” Computing and Informatics, vol. 42, no. 1, pp. 75-97, 2023. [3] Abumohsen M., et al., “Electrical Load Forecasting Using LSTM, GRU, and RNN,” Energies, vol. 16, no. 5, 2023. [4] Hasan M., “A state-of-the-art comparative review of load forecasting,” Elsevier, 2025. [5] Sarker M.A.A., “Enhancing smart grid load forecasting: An attention-based approach,” Energy and Buildings, 2024. [6] Jain A., “Evaluation of electrical load demand forecasting using computational intelligence methods,” Frontiers in Energy Research, 2024. [7] Elakkiya E., “Stacked hybrid model for load forecasting: integrating neural networks with Time Series Transformers and Fuzzy Logic Transform,” Scientific Reports, 2025. [8] Zairi S. & Freihat M., “Electric Load Forecasting using Machine Learning for Peak Demand Management in Smart Grids,” Eng. Technol. Appl. Sci. Res., vol. 15, no. 3, 2025. [9] Sen A., Mazumder A.R., Sen U., “Differential Evolution Algorithm based Hyper-Parameters Selection of Transformer Neural Network Model for Load Forecasting,” arXiv, 2023. [10] Emshagin S., Halim W.K., Kashef R., “Short-term Prediction of Household Electricity Consumption Using Customized LSTM and GRU Models,” arXiv, 2022. [11] Brauns K., Scholz C., Baier A., Jost D., “Vertical Power Flow Forecast with LSTMs Using Regular Training Update Strategies,” arXiv, 2020. [12] Panda S.K., “Electrical load and solar power forecasting using machine learning,” Complexity, 2025. [13] Upadhyay A., Garg D., Singh M., “Smart grid load forecasting with modified Transformer,” Computing & Informatics, vol. 42, no. 1, 2023. [14] J. Shi, S. Wang, P. Qu, X. Liu, and Y. Zhang, “Time series prediction model using LSTM-Transformer neural network for mine water inflow,” Scientific Reports, vol. 14, no. 18284, pp. 1–13, 2024. doi: 10.1038/s41598-024-69418-z. [15] Upadhyay A., Garg D., Singh M., “Short Term Load Forecasting for Smart Grids Using Apache Spark and a Modified Transformer Model,” Computing and Informatics, 2023. [16] Jain A., “Evaluation of electrical load demand forecasting using computational intelligence methods,” Frontiers in Energy Research, 2024. [17] Abumohsen M., et al., “Electrical Load Forecasting Using LSTM, GRU, and RNN,” Energies, 2023. [18] Sarker M.A.A., “Enhancing smart grid load forecasting: An attention-based approach,” Energy and Buildings, 2024. [19] Elakkiya E., “Stacked hybrid model for load forecasting,” Scientific Reports, 2025. [20] Hasan M., “A state-of-the-art comparative review of load forecasting,” Elsevier, 2025. [21] Zairi S., Freihat M., “Electric Load Forecasting using Machine Learning for Peak Demand Management in Smart Grids,” Eng. Technol. Appl. Sci. Res., 2025. [22] Salehimehr S., “Short-term load forecasting in smart grids using artificial intelligence,” IET Generation, Transmission & Distribution, 2022. [23] Sen A., Mazumder A.R., Sen U., “Differential Evolution Algorithm based Hyper-Parameters Selection of Transformer Neural Network Model for Load Forecasting,” arXiv, 2023. [24] Emshagin S., Halim W.K., Kashef R., “Short-term Prediction of Household Electricity Consumption Using Customized LSTM and GRU Models,” arXiv, 2022. [25] Brauns K., Scholz C., Baier A., Jost D., “Vertical Power Flow Forecast with LSTMs Using Regular Training Update Strategies,” arXiv, 2020.
