These are experimental data for the paper: Pawel Bielski, Aleksandr Eismont, Jakob Bach, Florian Leiser, Dustin Kottonau, and Klemens Böhm. 2024. Knowledge-Guided Learning of Temporal Dynamics and its Application to Gas Turbines, 15th ACM International Conference on Future Energy Systems (e-Energy '24), Singapore The data consist of:

1. experimental time series data collected from a micro gas turbine
2. results from the experiments and the corresponding code to create plots used in the paper The corresponding GitHub repository: https://github.com/Energy-Theory-Guided-Data-Science/Gas-Turbine

Micro Gas Turbine Data

Overview

These experimental data support the paper "Knowledge-Guided Learning of Temporal Dynamics and its Application to Gas Turbines", presented at 15th ACM International Conference on Future Energy Systems (e-Energy '24), Singapore. The data was collected from a commercial micro gas turbine designed for residential use, generating approximately 3 kW of electrical power. Its purpose was to model the turbine's behavior over time using machine learning techniques.

Folder Structure

• data: Contains 8 experimental time series data in CSV format, collected from the micro gas turbine.
• plots: Includes results from experiments and the code used to generate plots from the paper.
• plots/create_plots.ipynb: A Jupyter notebook containing code to create the plots.

Time Series Data

Each time series represents a separate experiment where the input control voltage was varied over time, and the resulting output electrical power of the micro gas turbine was measured. The data has a resolution of approximately 1 second and is structured in a CSV file with the following columns:

• time: Time in seconds, denoted as $t$.
• input_voltage: Input control voltage in volts, representing the control signal $x_t$.
• el_power: electrical power in watts, representing the output signal $y_t$.

Prediction Task

The data was used for a time-series prediction task, aiming to predict el_power based on input_voltage. In the paper, the objective was to forecast the output $y_t$ given the control inputs $x_t, x_{t-1}, \dots, x_{t-N+1}$.

Additional Information

Requirements for running create_plots.ipynb:

• Python 3.8.17
• Jupyter Notebook 6.5.4
• Pandas 1.2.2
• Matplotlib 3.5.2
• Seaborn 0.12.2 When using this dataset, please cite the following paper: Pawel Bielski, Aleksandr Eismont, Jakob Bach, Florian Leiser, Dustin Kottonau, and Klemens Böhm. 2024. Knowledge-Guided Learning of Temporal Dynamics and its Application to Gas Turbines, 15th ACM International Conference on Future Energy Systems (e-Energy '24), Singapore. For more details and the code used in the experiments, visit the GitHub repository.