Introduction: Introducing our fellow scholar Juwon Ajayi, a recent McNair scholar of the REM 28 cohort. Juwon is currently a Junior, majoring in computer engineering. This summer, he has been given the opportunity to participate at the University of North Carolina in Charlotte Summer Program to Increase Diversity in Undergraduate Research (SPIDUR).
Juwon is currently working on developing an analytical program which will analyze an image taken from a Particle Image Velocimetry(PIV). This will also be in correlation with utilizing machine learning to predict velocities, using data collected from fluid dynamics. We are excited to hear more about your research, Juwon! Always remember we are here to support you in any way we can. Keep striving and learning. We promise you, big things will happen.
Abstract: The purpose of this research is to develop a MATLAB code that will analyze the images taken from a Particle image velocimetry (PIV) set up. The images will be analyzed using crosscorrelation to generate velocity vector field. After completing this first process then develop a Machine learning (ML) algorithm that will be able to predict the flow of the particle from images taken from Particle image velocimetry set up. I am interested in implementing ML with PIV to analyze how accurate the ML model determines the flow of the fluid when relying on the input images from either computational or experimental results. Two research questions guide my research: First, how accurate the developed MATLAB code obtains the velocity vector fields compared to those from PIV results? Second, how accurate the develop ML model predict the flow field velocity values using the experimental/computational input data? For MATLAB, online courses and available resources in the research group will enable developing a simple, yet robust code. For ML, several courses are taken, and regression models on some sample data are being developed. The results of this research will be beneficiary to other researchers in the team to (1) educate them on how the cross-correlation in a commercial software (i.e., Dantec Dynamics for PIV) works and (2) how a data-based approach (i.e., ML) can use the physics-based data to obtain a faster and accurate prediction of flow fields.