Hybrid Energy Storage Dataset

This repository contains the dataset and implementation code used in the study entitled “Explainable Multi-Task LSTM–GRU Framework for Power Loss Prediction and Energy Efficiency Assessment in Hybrid Renewable Energy Storage Systems.” The dataset contains time-ordered observations recorded at 5-minute intervals and represents the operational behaviour of a synthetic Hybrid Energy Storage System (HESS). It includes renewable energy generation, grid power, energy storage conditions, hydrogen production, load demand, supplied power, power loss, and efficiency-class information. The main variables include solar power, wind power, grid power, battery state of charge, supercapacitor charge, hydrogen production, load demand, power supplied, and power loss. The efficiency classes are defined using the following power-loss thresholds: High efficiency: power loss ≤ 2 kWMedium efficiency: power loss between 2 and 4 kWLow efficiency: power loss > 4 kW The accompanying Jupyter Notebook implements the preprocessing, time-series sliding-window construction, multi-task LSTM and GRU architectures, regression and classification analyses, statistical evaluation, and SHAP- and LIME-based explainability analyses reported in the associated manuscript. Original dataset source:Ziya (Kaggle username: ziya07), “Hybrid Energy Storage Dataset,” Kaggle.https://www.kaggle.com/datasets/ziya07/hybrid-energy-storage-datasetAccessed: 3 August 2026. The original dataset is distributed on Kaggle under the CC0: Public Domain license. The dataset is archived here together with the implementation code to preserve the exact version used in the study and to support transparency, accessibility, and reproducibility. The dataset is synthetic and does not contain personal, confidential, human-subject, or geographically identifiable information.