Aero-Engine Multi-Station Thermodynamic Temperature Regression Dataset

Description

This dataset is curated for researching multi-objective thermal state regression in aerospace engineering. It primarily focuses on the simultaneous prediction of thermodynamic temperatures at three critical engine stations (T25, T3, and T5). Key Technical Specifications: Data Source: Public aero-engine operational time-series records. Task Type: Multi-objective continuous temperature trajectory regression subject to external disturbances. Data Extraction Method: All samples are extracted using a fixed-size temporal sliding window approach to preserve dynamic feature dependencies and transient thermal boundaries. Intended Applications: Virtual sensing, thermal dynamics regression, cross-system transferability verification, and uncertainty quantification (UQ).

Steps to reproduce

1. Data Extraction: All continuous temperature trajectory samples were extracted from raw public aero-engine operational time-series records using a fixed-size temporal sliding window approach to preserve dynamic feature dependencies and transient thermal boundaries. 2. Dataset Structure: The dataset is packaged into compressed files containing 4 distinct NumPy array (.npy) files, divided into training and testing subsets: - Training inputs and target thermodynamic temperatures (T25, T3, T5). - Testing inputs and target thermodynamic temperatures (T25, T3, T5). 3. How to Use/Reproduce: - Step 1: Download and decompress the provided archive files. - Step 2: Use Python's NumPy library (`numpy.load()`) to directly load the 4 `.npy` files into your workspace. - Step 3: Feed the array data directly into regression or deep learning sequences (e.g., Virtual Sensing or Uncertainty Quantification models) for training and evaluation.

Institutions

• Harbin Institute of Technology

  Heilongjiang, Harbin

Categories

Licence