Hourly OEE Time Series Datasets for Short-Term Forecasting in Manufacturing Systems

This record contains three real-world industrial time-series datasets used for short-term Overall Equipment Efficiency (OEE) forecasting in manufacturing systems. The datasets are denoted GH2, H2, and GM2. GH2 and GM2 correspond to advanced production systems used in the manufacture of household appliance inner-body groups, while H2 corresponds to a hydraulic press system consisting of four presses used for stainless steel body component production. Each dataset contains hourly OEE measurements on a 1–60 scale. Unlike the conventional percentage-based OEE representation, this scale reflects the effective operating time within a 60-minute window. Lower values indicate limited or no operation, whereas a value of 60 indicates full effective operation during the hour. The datasets span approximately one month of operation and represent short-term industrial monitoring scenarios with high volatility, non-normal behaviour, and multiple seasonal patterns. The study identifies expected shift-level, daily, and weekly seasonalities corresponding to 8-hour, 24-hour, and 168-hour cycles. These characteristics make the datasets suitable for research on time-series forecasting, predictive maintenance, feature engineering, topological data analysis, and industrial analytics. Dataset summary: GH2: 648 hourly observations H2: 683 hourly observations GM2: 672 hourly observations Value range: 1 to 60 for all datasets This dataset collection supports the study of short-term OEE forecasting under dynamic manufacturing conditions and was used in a framework combining decomposition-based forecasting, statistical features, topological features, and SARIMAX modelling.