AI-Driven Prediction of Electricity Production and Consumption in Micro-Hydropower Plant

Osman Safi; Gul Muhammad Khan; Gul Rukh Khattak

Authors

Osman Safi UET Peshawar
Gul Muhammad Khan National center of AI, UET Peshawar
Gul Rukh Khattak National Center of AI, University of Engineering and Technology Peshawar

Keywords:

Artificial Intelligence, Micro-Hydropower Plant, Time Series Forecasting, LSTM, CGP, Hourly Electricity Prediction

Abstract

Micro hydropower plants must effectively manage demand response to preserve operational firmness and prevent system breakdowns. This research focuses on accomplishing a fine balance while predicting consumption and production, which is significant for upholding system integrity. The study delves into predictive modeling methods to forecast patterns in the production and consumption of electricity over an array of time horizons. We adopted a custom sliding window mechanism, in which actual and predicted values are used to predict the next hour of electricity. We set a baseline to resolve this and examined various algorithms, focusing on RNN-LSTM and CGP-LSTM. The CGP-LSTM forecasting output sequences with different time horizons precisely outperform the RNN-LSTM. The dataset utilized is downloaded from the Kaggle website. 50% of the data is used to train the models, and the rest is used to test the models. This work deals with the complex fluctuations in the demand response system and provides electricity production and consumption predictions. CGP-LSTM model gave a training MAPE of 6.67 (Accuracy of 93.33%) and a testing MAPE of 6.68 (accuracy of 93.32%) for the next three hours; on the other hand, LSTM gave a training MAPE of 6.53 (accuracy of 93.47%) and testing MAPE of 7.46 (accuracy of 92.54%) for the next three hours. The results offer a base for further developments and improvements in the field, drawing attention to more effective and reliable energy management capabilities in micro hydropower plants.

CGP-LSTM model gave a training MAPE of 6.67 (Accuracy of 93.33%) and a testing MAPE of 6.68 (accuracy of 93.32%) for the next three hours; on the other hand, LSTM gave a training MAPE of 6.53 (accuracy of 93.47%) and testing MAPE of 7.46 (accuracy of 92.54%) for the next three hours.

The results offer a base for further developments and improvements in the field, drawing attention to more effective and reliable energy management capabilities in micro hydropower plants.

Author Biography

Gul Muhammad Khan, National center of AI, UET Peshawar

Post Graduate Advisor and the author of Electrical Engineering Department and the author of Evolution of Artificial Neural Development: In search of learning genes (Studies in Computational Intelligence Book 725).

References

A. K. Singh, Ibraheem, S. Khatoon, M. Muazzam, and D. K. Chaturvedi, “Load forecasting techniques and methodologies: A review,” ICPCES 2012 - 2012 2nd Int. Conf. Power, Control Embed. Syst., 2012, doi: 10.1109/ICPCES.2012.6508132.

R. Weron, “Electricity price forecasting: A review of the state-of-the-art with a look into the future,” Int. J. Forecast., vol. 30, no. 4, pp. 1030–1081, Oct. 2014, doi: 10.1016/J.IJFORECAST.2014.08.008.

M. Rafik, A. Fentis, T. Khalili, M. Youssfi, and O. Bouattane, “Learning and Predictive Energy Consumption Model based on LSTM recursive neural networks,” 4th Int. Conf. Intell. Comput. Data Sci. ICDS 2020, Oct. 2020, doi: 10.1109/ICDS50568.2020.9268733.

J. F. Chen, W. M. Wang, and C. M. Huang, “Analysis of an adaptive time-series autoregressive moving-average (ARMA) model for short-term load forecasting,” Electr. Power Syst. Res., vol. 34, no. 3, pp. 187–196, Sep. 1995, doi: 10.1016/0378-7796(95)00977-1.

G. Juberias, R. Yunta, J. Garcia Moreno, and C. Mendivil, “New arima model for hourly load forecasting,” Proc. IEEE Power Eng. Soc. Transm. Distrib. Conf., vol. 1, pp. 314–319, 1999, doi: 10.1109/TDC.1999.755371.

“Forecasting of a hydropower plant energy production with Fuzzy logic Case for Albania”, [Online]. Available: https://www.jmest.org/wp-content/uploads/JMESTN42352182.pdf

K. Kandananond, “Forecasting Electricity Demand in Thailand with an Artificial Neural Network Approach,” Energies 2011, Vol. 4, Pages 1246-1257, vol. 4, no. 8, pp. 1246–1257, Aug. 2011, doi: 10.3390/EN4081246.

N. Amjady and F. Keynia, “A New Neural Network Approach to Short Term Load Forecasting of Electrical Power Systems,” Energies 2011, Vol. 4, Pages 488-503, vol. 4, no. 3, pp. 488–503, Mar. 2011, doi: 10.3390/EN4030488.

P. P. et Al., “Systematic analysis and review of stock market prediction techniques,” Comput. Sci. Rev., vol. 34, p. 100190, Nov. 2019, doi: 10.1016/J.COSREV.2019.08.001.

E. Yuniarti, Nurmaini, B. Y. Suprapto, and M. Naufal Rachmatullah, “Short Term Electrical Energy Consumption Forecasting using RNN-LSTM,” ICECOS 2019 - 3rd Int. Conf. Electr. Eng. Comput. Sci. Proceeding, pp. 287–292, Oct. 2019, doi: 10.1109/ICECOS47637.2019.8984496.

F. U. M. Ullah, A. Ullah, I. U. Haq, S. Rho, and S. W. Baik, “Short-Term Prediction of Residential Power Energy Consumption via CNN and Multi-Layer Bi-Directional LSTM Networks,” IEEE Access, vol. 8, pp. 123369–123380, 2020, doi: 10.1109/ACCESS.2019.2963045.

E. Izgi, A. Öztopal, B. Yerli, M. K. Kaymak, and A. D. Şahin, “Determination of the Representative Time Horizons for Short-term Wind Power Prediction by Using Artificial Neural Networks,” Energy Sources, Part A Recover. Util. Environ. Eff., vol. 36, no. 16, pp. 1800–1809, Aug. 2014, doi: 10.1080/15567036.2011.561274.

F. A. Olivencia Polo, J. Ferrero Bermejo, J. F. Gómez Fernández, and A. Crespo Márquez, “Failure mode prediction and energy forecasting of PV plants to assist dynamic maintenance tasks by ANN based models,” Renew. Energy, vol. 81, pp. 227–238, Sep. 2015, doi: 10.1016/J.RENENE.2015.03.023.