Statistics and Machine Learning in Aviation Environmental Impact Analysis: A Survey of Recent Progress

• Published in: Aerospace Vol. 9; no. 12; p. 750
• Main Authors: Gao, Zhenyu, Mavris, Dimitri N.
• Format: Journal Article
• Language: English
• Published: Basel MDPI AG 01-12-2022
• Subjects: Aeronautics; Air pollution; air transportation; Air transportation industry; Aircraft; Aircraft exhaust emissions; Airports; Algorithms; Aviation; aviation environmental impact; Aviation fuel; Big Data; Climate change; Data reduction; data-driven methods; Emissions; Environmental impact; Environmental impact analysis; Environmental impact assessment; Flying-machines; Gas turbine engines; Impact analysis; Machine learning; Neural networks; Noise; Prediction models; Statistics; Surveillance; Surveys; sustainable aviation; Trends; Workflow
• ISSN: 2226-4310; 2226-4310
• DOI: 10.3390/aerospace9120750

The rapid growth of global aviation operations has made its negative environmental impact an international concern. Accurate modeling of aircraft fuel burn, emissions, and noise is the prerequisite for informing new operational procedures, technologies, and policies towards a more sustainable future of aviation. In the past decade, due to the advances in big data technologies and effective algorithms, the transformative data-driven analysis has begun to play a substantial role in aviation environmental impact analysis. The integration of statistical and machine learning methods in the workflow has made such analysis more efficient and accurate. Through summarizing and classifying the representative works in this intersection area, this survey paper aims to extract prevailing research trends and suggest research opportunities for the future. The methodology overview section presents a comprehensive development process and landscape of statistical and machine learning methods for applied researchers. In the main section, relevant works in the literature are organized into seven application themes: data reduction, efficient computation, predictive modeling, uncertainty quantification, pattern discovery, verification and validation, and infrastructure and tools. Each theme contains background information, in-depth discussion, and a summary of representative works. The paper concludes with the proposal of five future opportunities for this research area.