Machine Learning based Predictive Modeling of Stochastic Systems

Complex signals are ubiquitous in our daily lives, and interpreting and modeling them is vital for scientific advancement. Traditional methods for predictive modeling of complex signals include statistical signal processing and physics-based simulations. However, statistical signal processing methods often struggle to fully utilize complex and rich datasets, while physics-based simulations can be computationally demanding. As an alternative approach, machine learning (ML) offers a more effective method for the predictive modeling of complex signals. This research explores the applicability of ML-based predictive modeling to a biomedical and a mechanical system through two case studies. The first case study focuses on developing a machine learning-based model for early-stage glaucoma detection using electroretinogram signals, which has been a challenging problem in ophthalmology. By leveraging medically relevant information contained in ERG signals, the study aims to establish a novel and reliable predictive framework for the early detection of glaucoma using a machine-learning-based algorithm. The results demonstrate that machine-learning-based models, trained using advanced wavelet-based features, can effectively detect the early stage of glaucoma from ERG stochastic signals. The second case study centers on developing a machine learning-based model for stall delay correction in wind turbines. Existing stall delay correction models rely on 2D airfoil characteristics, which can lead to inaccuracies in predicting aerodynamic loads during design and, consequently, result in structural failure due to excessive load. To address this issue, the study proposes a novel stall delay correction model based on the soft computing technique of symbolic regression. The model offers high-level precise aerodynamic performance prediction through the blade element momentum process, making it a promising alternative for accurate and efficient stall delay correction in wind turbines.

Name thesis:degree_name

Ph.D. (Doctor of Philosophy)

Level thesis:degree_level

Doctoral

Discipline thesis:degree_discipline

Mechanical Engineering (UMKC)

Grantor

University of Missouri--Kansas City

Year dc:date.issued

2023

Author dc:creator

• Gajendran, Mohan Kumar

Advisor dc:contributor.advisor

• Mehdizadeh, Amirfarhang

Handle dc:identifier.uri

https://hdl.handle.net/10355/95128

OAI identifier oai:identifier

oai:mospace.umsystem.edu:10355/95128