Improving Approach Noise Predictions for Aircraft Environmental Impact Models Using Deep Neural Networks and Sound-level Monitor Data
Abstract/Contents
- Abstract
- In order to minimize the impacts of noise on overflown communities by re-designing flight procedures, predictive techniques such as the Federal Aviation Administration’s (FAA) Aviation Environmental Design Tool (AEDT) have been developed and are widely used in the aviation community. Such tools are usually based on the simulation of flight trajectories and noise generation as well as propagation models of various fidelity levels. To more accurately describe the impacts of noise and changes in noise metrics that result from flight procedure redesigns, errors in the predictions must be sufficiently small so that changes to air traffic patterns do not result in unforeseen consequences. At Stanford University, we have developed the Metroplex Overflight Noise Analysis (MONA) system to measure, predict, and analyze the noise generated by aircraft overflights in complex multi-airport settings and to provide accurate and actionable data to inform consensus-building and, possibly, procedure redesigns in the context of FAA’s NextGen system. The capabilities of the MONA system include generating vast amounts of data (approximately 3,000-5,000 flights/day around Bay Area airports alone). This data contains (a) detailed flight trajectories and equipment combinations extracted from ADS-B data feeds, (b) noise levels at several locations measured by a network of sound-level monitoring (SLM) stations, and (c) predictions of the noise produced by the aircraft/engine combinations computed by automated analyses using AEDT. Over an entire year, millions of predictions and measurements are collected and indexed resulting in curated datasets for further analysis. This paper explores an application of these datasets, using machine learning (ML) techniques to create models of the noise generated by aircraft on approach trajectories to San Francisco International airport (SFO). Using deep neural net ML architectures we show that the errors in the mean for Lmax noise metrics are reduced to within the margin of measurement error of the SLMs.
Description
| Type of resource | text |
|---|---|
| Date created | June 2022 |
| Publication date | May 15, 2023; April 16, 2023 |
Creators/Contributors
| Author | Shukla, Aditeya |
|---|---|
| Advisor | Alonso, Juan |
Subjects
| Subject | Recurrent Neural Network |
|---|---|
| Subject | Airplanes > Noise |
| Subject | Aviation Environmental Design Tool |
| Subject | San Francisco International Airport (Calif.) |
| Subject | Flight Trajectory |
| Subject | Federal Aviation Administration |
| Subject | Machine learning |
| Subject | Aeronautics, Commercial |
| Subject | Statistical Distributions |
| Subject | System Wide Information Management |
| Genre | Text |
| Genre | Thesis |
Bibliographic information
Access conditions
- Use and reproduction
- User agrees that, where applicable, content will not be used to identify or to otherwise infringe the privacy or confidentiality rights of individuals. Content distributed via the Stanford Digital Repository may be subject to additional license and use restrictions applied by the depositor.
- License
- This work is licensed under a Creative Commons Attribution Non Commercial 4.0 International license (CC BY-NC).
Preferred citation
- Preferred citation
- Shukla, A. (2023). Improving Approach Noise Predictions for Aircraft Environmental Impact Models Using Deep Neural Networks and Sound-level Monitor Data. Stanford Digital Repository. Available at https://purl.stanford.edu/xp881gf4013. https://doi.org/10.25740/xp881gf4013.
Collection
Undergraduate Theses, School of Engineering
View other items in this collection in SearchWorksContact information
Also listed in
Loading usage metrics...