RESEARCH ON COMBINING K-MEDOIDS AND FEEDFORWARD NEURAL NETWORKS FOR ULTRA SHORT TERM SOLAR POWER FORECASTING
Keywords
DOI:
https://doi.org/10.34238/tnu-jst.15500Abstract
Feedforward neural networks are commonly employed for ultra-short-term photovoltaic power forecasting due to their fast computational speed and acceptable forecasting accuracy. However, because photovoltaic power output fluctuates significantly with weather conditions, forecasting errors can increase considerably under certain conditions. Therefore, combining data clustering techniques with feedforward neural networks can help improve forecasting performance. This paper proposes the use of the K-medoids algorithm to cluster the training dataset and assign each testing sample to the corresponding cluster. A separate feedforward neural networks is then developed for each cluster to perform training and forecasting. Finally, the forecasting results from all clusters are combined for evaluation. Experimental data were collected from a 500 kWp photovoltaic system installed in Soc Trang with a forecasting horizon of 5 minutes. The experimental results on the testing dataset show that the Mean Absolute Percentage Error decreased from 26.857% using a standalone feedforward neural networks to 20.763% when the K-medoids algorithm was integrated with feedforward neural networks. Furthermore, the study also proposes a method for determining the optimal number of clusters to achieve the lowest forecasting error.Downloads
References
W.-C. Tsai, C.-S. Tu, C.-M. Hong, and W.-M. Lin, "A Review of State-of-the-Art and Short-Term Forecasting Models for Solar PV Power Generation," Energies: Solar Cells and Photoelectrochemical Cells: Current Status, Challenges and Future, vol.16, pp. 1-30, 2023, doi: 10.3390/en16145436.
Z. Si, M. Yang, Y. Yu, T. Ding, and M. Li, "A Hybrid Photovoltaic Power Prediction Model Based on Multi-source Data Fusion and Deep Learning," IEEE 3rd Student Conference on Electrical Machines and Systems (SCEMS), 2020, pp. 608-613.
F. Wang, Z. Zhang, H. Chai, Y. Yu, X. Lu, and T. Wang, "Deep Learning Based Irradiance Mapping Model for Solar PV Power Forecasting Using Sky Image," IEEE Industry Applications Society Annual Meeting, 2019, pp. 1-9.
P. K. Nguyen, K. L. Tran, P. C. Ha, M. H. Nguyen, and C. H. Hoang, "Short-term forecasting of rooftop solar power generation using artificial neural networks," Can Tho University Journal of Science, Special Issue: Engineering and Technology, vol. 59, pp. 251-260, 2023.
A. Matteri, E. Ogliari, and A. Nespoli, "Enhanced Day-Ahead PV Power Forecast: Dataset Clustering for an Effective Artificial Neural Network Training," The 7th International Conference on Time Series and Forecasting, Gran Canaria, Spain, 2021, pp. 1-9.
G. Li, Y. Cheng, C. Wang, J. Shi, and L. Li, "Ultra Short-term Output Forecasting of Distributed Photovoltaic Power Station Based on Feature Extraction," IEEE/IAS Industrial and Commercial Power System Asia (I&CPS Asia), 2020, pp. 1114-1119.
M. Shi, R. Yin, P. Zhang, J. Tian, and Y. Wang, "Ultra short-term forecast of photovoltaic generation based on time-division LSTM neural networks," 12th IEEE PES Asia-Pacific Power and Energy Engineering Conference (APPEEC), 2020, pp. 1-5.
Z. Zhen, J. Liu, Z. Zhang, F. Wang, H. Chai, and Y. Yu, "Deep Learning Based Surface Irradiance Mapping Model for Solar PV Power Forecasting Using Sky Image," IEEE Transactions on Industry Applications, vol. 56, no. 4, pp. 99-110, 2019.
V. V. Huynh and N. T. V. Nguyen, "Forecasting power generation of solar panels using machine learning," HUIT Journal of Science, vol. 25, no. 1, pp. 156-164, 2025.
P. Marpaung, "Analysis of K-Means and K-Medoids Algorithms for Grouping Academic Values of Prospective New Students in Selection of Campus Scholarship Recipients," Formosa Journal of Science and Technology (FJST), vol. 4, no. 9, pp. 2975-2988, 2025.
L. Kaufman and P. J. Rousseeuw, "Finding Groups in Data: An Introduction to Cluster Analysis," Wiley Series in Probability and Statistics, John Wiley & Sons Publication, 1990.
M. Yang and X. Huang, "Ultra-short-term prediction of photovoltaic power based on periodic extraction of PV energy and LSH algorithm," IEEE Access, vol. 6, pp. 51200-51205, 2018.
