Quantum-Inspired Load Forecasting Model Integrating STL Decomposition and Clustering Techniques for Intelligent Energy Management
Article Sidebar
Main Article Content
Abstract
This research presents a framework for analyzing and forecasting three-phase electrical load using Quantum-Inspired Machine Learning combined with STL Decomposition, K-Means Clustering, and Anomaly Detection techniques. Real-world data were collected from the Industrial Technician School building at RMUTSV over a period of 151 days (December 2024 – April 2025). The analysis revealed that Phase B exhibited the highest average current at 13.40A, compared to Phase A (6.02A) and Phase C (7.62A), indicating a significant phase unbalance, with an average A–B difference of 7.38A and a maximum differential of 17.9A. STL Decomposition indicated an upward trend in Phase B load, increasing from 7.0A in December 2024 to 21.5A in March 2025, along with multiple residual spikes exceeding ±5A, reflecting transient load fluctuations. Anomaly Detection (Z-Score + Isolation Forest) identified 17 abnormal points, with the highest anomaly recorded on March 12, 2025, reaching 36.0A. Using K-Means clustering (k = 3), the load was classified into three clusters Low ~7.1A Medium ~13.7A High ~19.3A These findings provide critical insights for developing effective energy management strategies within educational buildings.
Article Details
References
Ji X, Huang H, Chen D, Yin K, Zuo Y, Chen Z, Bai R. A hybrid residential short-term load forecasting method using attention mechanism and deep learning. Buildings. 2023;13(1):72. doi: 10.3390/buildings13010072
El Khantach A, Hamlich M, Belbounaguia N. Short-term load forecasting using machine learning and periodicity decomposition. AIMS Energy. 2019;7(3):382–394. doi: 10.3934/energy.2019.3.382
Farsi B, Amayri M, Bouguila N, Eicker U. On short-term load forecasting using machine learning techniques and a novel parallel deep LSTM-CNN approach. IEEE Access. 2021;9:31191–31202. doi: 10.1109/ACCESS.2021.3060705
Jiang Y, Liu Q, Li X. A data decomposition and attention mechanism-based hybrid approach for load forecasting. Complex & Intelligent Systems. 2024;10:1379–1394. doi: 10.1007/s40747-023-01995-1
Ahmed A, Awan A, Shah S, Hussain S. Load forecasting with machine learning and deep learning methods. Applied Sciences. 2023;13(13):7933. doi: 10.3390/app13137933
Lin Y, Zhang Z, Wang P. Forecasting very short-term power load with hybrid interpretable models. Engineering Applications of Artificial Intelligence. 2025;135:106746. doi: 10.1016/j.engappai.2025.106746
Patel H, Mehta P, Kumar V. Application of machine learning with gradient descent method for FF-DNN-based short-term load forecasting. J Electr Syst Inf Technol. 2025;12(1):46. doi: 10.1186/s43067-025-00246-5
Zhang J, Wang J, Li X. ES-dRNN: A hybrid exponential smoothing and dilated recurrent neural network model for short-term load forecasting. arXiv preprint arXiv:2112.02663. 2021. Available from: https://arxiv.org/abs/2112.02663.
Zhou H, Li T, Peng H. Non-intrusive load decomposition based on CNN-LSTM hybrid deep learning model. Energy Reports. 2021;7:5762–5771. doi: 10.1016/j.egyr.2021.09.044
Xiong T, Li C, Bao Y. Forecasting model combining STL decomposition and extreme learning machine for time series prediction. Neurocomputing. 2018;277:483–494. doi:10.1016/j.neucom.2017.11.053
