Short-Term Load Forecasting in Power Systems Using a Hybrid CNN-LSTM Deep Learning Model 
Huang Xuanyi

Raman University

Abstract: Short-term load forecasting (STLF) is essential for for the reliable and economic operation of power systems. Traditional forecasting methods often struggle with nonlinearity and volatility in electricity demand data. This paper proposes a hybrid deep learning model combining Convolutional Neural Networks (CNN) and Long Short-Term Memory (LSTM) networks to improve STLF accuracy. The CNN layers extract spatial features from multi-dimensional input data (e.g., historical load, temperature, day type), while the LSTM layers capture temporal dependencies. The model was trained and tested using actual load data from a regional grid. Experimental results show that the proposed CNN-LSTM model has lower MAPE and RMSE than benchmark models such as LSTM, SVR, and ANN on the test set. The results show that the hybrid deep learning architecture has application potential in short-term load forecasting of smart grids. For example, similar hybrid frameworks have been validated in other studies, confirming their effectiveness in handling complex load and renewable variability [4].

Keywords: Short-term load forecasting; deep learning; CNN; LSTM; power system automation; smart grid

References

[1] Hochreiter, S., & Schmidhuber, J. (1997). "Long short-term memory". Neural computation, 9(8),1735-1780.

[2] Wang, Y., et al. (2019). "Deep learning for load forecasting: A hybrid CNN-LSTM approach". IEEET ransactions on Smart Grid, 10(3), 3163-3173.

[3] ISO New England. (2021). "Load data archives. Retrieved from [URL]".

[4] K. Ullah et al. (2024). "Short-Term Load Forecasting: A Comprehensive Review and Simulation Study With CNN-LSTM Hybrids Approach". IEEE, vol. 12, pp. 111858-111881.

[5] K. Tae-Geun et al. (2025). "Very Short-Term Load Forecasting for Large Power Systems with Kalman Filter-Based Pseudo-Trend Information Using LSTM". Energies 18, no. 18: 4890.

[6] Ladjal, B., Nadour, M., Bechouat, M. et al. (2025). "Hybrid deep learning CNN-LSTM model for forecasting direct normal irradiance: a study on solar potential in Ghardaia, Algeria". Sci Rep 15, 15404.

[7] Tian, C., Ma, J., Zhang, C., & Zhan, P. (2018). "A Deep Neural Network Model for Short-Term Load Forecast Based on Long Short-Term Memory Network and Convolutional Neural Network". Energies, 11(12), 3493.

[8] Diederik P, K., & Jimmy, B. (2017). "Adam: A Method for Stochastic Optimization". arXiv:1412.6980.

[9] H. Drucker et al. (1997). "Support vector regression machines". Advances in Neural Information Processing Systems 28(7):779-784

[10] L. Su-Chang et al. (2022). "Solar Power Forecasting Using CNN-LSTM Hybrid Model". Energies 15, no. 21: 8233.