Datasets
Standard Dataset
Simulation Results of Data-Driven LightGBM Controller for Spacecraft Attitude Control
- Citation Author(s):
- Submitted by:
- Dimitri Mahayana
- Last updated:
- Mon, 08/07/2023 - 00:28
- DOI:
- 10.21227/63dh-en16
- Data Format:
- License:
51 Views- Categories:
- Keywords:
Abstract
This dataset contains simulation data of the LightGBM controller for spacecraft attitude control. The data were generated using a closed-loop system of spacecraft attitude dynamics under an exact feedback linearization-based controller. The LightGBM controller was designed using supervised machine learning methodologies, and the training and testing datasets were generated from the input-output data of the closed-loop system. This dataset contains the results of simulations conducted to train a LightGBM controller for spacecraft attitude control using different data sizes ranging from 1,000 to 32,800,768 data points. The simulations were performed to demonstrate the importance of data size in training the LightGBM controller. The dataset can be used to analyze the performance of the LightGBM controller under various data sizes and to explore the relationship between data size and controller performance.
Instructions :
- This dataset contains simulation results of spacecraft attitude control using the LightGBM controller
- The dataset is in the form of a single Parquet file, with columns representing sample size, iteration, spacecraft attitude, spacecraft attitude derivative, and torque.
- The simulation was performed under various data sizes to show the importance of data size in LightGBM training, ranging from 10,000 to 32,800,768 in data size.
- The dataset is intended for researchers and practitioners interested in spacecraft attitude control, data-driven control, and machine learning applications in control systems. The dataset can be used to evaluate the performance of the LightGBM controller under different data sizes and to compare it with other control methods. The dataset can also be used as a benchmark for developing and testing new control algorithms.
