SMART-GTPP Dataset

Description

This dataset contains operational and environmental measurements collected from a utility-scale gas turbine operating in a combined-cycle gas power plant in the Kurdistan Region of Iraq. The data cover the period from 2019 to 2023 and were prepared for gas turbine power-output prediction, performance analysis, machine-learning research, and data-driven digital-twin development. The dataset contains 7,893 observations, eight input features, and one target variable. All variables are numerical, and the file contains no missing values. The prediction task is a supervised regression problem. The input features are: AT: Ambient temperature, representing the temperature of the air entering the gas turbine compressor. Higher ambient temperature usually reduces air density and gas turbine output. AP: Atmospheric pressure, which affects inlet-air density, compressor mass flow, and turbine performance. DF: Electrical or grid frequency, reflecting generator and power-system operating conditions. CH: Humidity-related parameter describing the moisture content of the ambient air. Humidity can influence air density, combustion, and turbine efficiency. GP: Fuel-gas pressure supplied to the gas turbine combustion system. Stable gas pressure is important for combustion stability and power generation. CPR: Compressor pressure ratio, defined as the ratio between compressor discharge and inlet pressure. It is an important indicator of compressor and thermodynamic performance. CPD: Compressor discharge pressure, measured at the compressor outlet before the compressed air enters the combustion chambers. TTXM: Average turbine exhaust temperature, which is related to fuel input, combustion conditions, turbine loading, and overall efficiency. The target variable is EP, representing the active electrical power generated by the gas turbine in megawatts. The dataset can be used for electrical power-output forecasting, gas turbine performance modelling, feature-importance analysis, anomaly detection, digital-twin development, and comparison of regression algorithms such as Support Vector Regression, Random Forest, XGBoost, K-Nearest Neighbours, and Artificial Neural Networks. The dataset can be used for: Gas turbine electrical power-output prediction Energy forecasting in combined-cycle gas power plants Development of data-driven digital twins Gas turbine performance assessment Operational deviation and anomaly detection Feature-importance and sensitivity analysis Comparison of machine-learning regression algorithms Investigation of environmental effects on gas turbine output Predictive monitoring and decision-support system development Academic teaching and research in energy systems and artificial intelligence

Institutions

• Salahaddin University-Erbil

  Erbil, Erbil

Categories

Licence