Long-term planning of integrated local energy systems using deep learning algorithms

• Published in: International journal of electrical power & energy systems 2021-07, Vol.129, p.106855, Article 106855
• Main Authors: Taheri, Saman, Jooshaki, Mohammad, Moeini-Aghtaie, Moein
• Format: Article
• Language: English
• Subjects: Deep learning; Deep recurrent neural network; Energy consumption prediction; Gaussian process; Integrated local energy systems
• ISSN: 0142-0615; 1879-3517
• DOI: 10.1016/j.ijepes.2021.106855

•A bespoke deep RNN configuration is explored on the task of medium to long-term electricity and heat demand.•An energy hub planning optimization framework is developed that incorporates DRNN forecasts and GP regression.•The proposed DRNN configuration outperforms other ML techniques such as SVM and GB regression in terms of accuracy. Optimal investment and operations of integrated local energy systems (ILESs) require medium to long-term prediction of energy consumption. To forecast load profiles, deep recurrent neural networks (DRNNs) are becoming increasingly useful due to their capability of learning uncertainty and high variability of load profiles. However, to explore and choose a DRNN model, out of conceivably numerous configurations, depends entirely on the performing task. In this regard, we tune and compare seven DRNN variants on the task of medium and long-term predictions for heating and electricity consumption. The ultimate DRNN model outperforms two state-of-the-art machine learning techniques, namely gradient boosting (GB) regression and support vector machine (SVM) in terms of accuracy. Also, developing optimization frameworks that can employ deep learning algorithms to alleviate the significant computational burden associated with EH planning is neglected in previous literature. Thus, we aim at improving the efficiency of the optimization process through a strong coupling of deep learning algorithms and conventional optimization methods. In this regard, a step-by-step algorithm is formulated to facilitate the use of Gaussian Process (GP) regression within the optimization problem and obtain optimal results without explicitly performing the optimization problem. The overall computational time is decreased by a factor of 0.47 when the proposed optimization framework is implemented for a practical EH system.