Application of Machine Learning to Multiple Radar Missions and Operations

Simple item page
dc.contributor.advisorZhang, Yan
dc.contributor.authorShrestha, Yunish
dc.contributor.committeeMemberMcGovern, Amy
dc.contributor.committeeMemberMetcalf, Justin
dc.contributor.committeeMemberRyzhkov, Alexander
dc.contributor.committeeMemberYu, Tian-You
dc.date.accessioned2022-07-29T21:11:13Z
dc.date.available2022-07-29T21:11:13Z
dc.date.issued2022-08-04
dc.date.manuscript2022
dc.description.abstractThis dissertation investigated the application of Machine Learning (ML) in multiple radar missions. With the increasing computational power and data availability, machine learning is becoming a convenient tool in developing radar algorithms. The overall goal of the dissertation was to improve the transportation safety. Three specific applications were studied: improving safety in the airport operations, safer air travel and safer road travel. First, in the operations around airports, lightning prediction is necessary to enhance safety of the ground handling workers. Information about the future lightning can help the workers take necessary actions to avoid lightning related injuries. The mission was to investigate the use of ML algorithms with measurements produced by an S-band weather radar to predict the lightning flash rate. This study used radar variables, single pol and dual-pol, measured throughout a year to train the machine learning algorithm. The effectiveness of dual-pol radar variables for lighting flash rate prediction was validated, and Pearson's coefficient of about 0.88 was achieved in the selected ML scheme. Second, the detection of High Ice Water Content (HIWC),which impact the jet engine operations at high altitudes, is necessary to improve the safety of air transportation. The detection information help aircraft pilots avoid hazardous HIWC condition. The mission was to detect HIWC using ML and the X-band airborne weather radar. Due to the insufficiency of measured data, radar data was synthesized using an end-to-end airborne weather system simulator. The simulation employed the information about ice crystals' particle size distribution (PSDs), axial ratios, and orientation to generate the polarimetric radar variables. The simulated radar variables were used to train the machine learning to detect HIWC and estimate the IWC values. Pearson's coefficient of about 0.99 was achieved for this mission. The third mission included the improvement of angular resolution and explored the machine learning based target classification using an automotive radar. In an autonomous vehicle system, the classification of targets enhances the safety of ground transportation. The angular resolution was improved using Multiple Input Multiple Output (MIMO) techniques. The mission also involved classifying the targets (pedestrian vs. vehicle) using micro-Doppler features. The classification accuracy of about 94% was achieved.en_US
dc.identifier.urihttps://hdl.handle.net/11244/336447
dc.languageen_USen_US
dc.subjectRadaren_US
dc.subjectMachine Learningen_US
dc.subjectLightningen_US
dc.subjectHIWCen_US
dc.subjectMIMOen_US
dc.thesis.degreePh.D.en_US
dc.titleApplication of Machine Learning to Multiple Radar Missions and Operationsen_US
ou.groupGallogly College of Engineering::School of Electrical and Computer Engineeringen_US
shareok.nativefileaccessrestricteden_US

Files

Original bundle
Now showing 1 - 1 of 1
Thumbnail Image
Name:
2022_Shrestha_Yunish_Dissertation.pdf
Size:
10.34 MB
Format:
Adobe Portable Document Format
Description:
Download
License bundle
Now showing 1 - 1 of 1
No Thumbnail Available
Name:
license.txt
Size:
1.71 KB
Format:
Item-specific license agreed upon to submission
Description:
Download

Collections

OU - Dissertations