TITLE
RESEARCH
DWG NO.
AP-2026-003
REV
C
DRAWN BY
A. PANTER, EIT
DATE
2026-06-01
SCALE
1:1
SHEET
3 OF 5
1 OVERVIEW 2 EXPERIENCE 3 RESEARCH 4 ACADEMICS 5 PROJECTS
DETAIL A — GRADUATE RESEARCH · SMU LYLE · PRIMARY
GRADUATE RESEARCH ASSISTANT · SEP 2025 — PRESENT · SOUTHERN METHODIST UNIVERSITY

ML-Based Welding Defect Detection & Quality Assurance

SMU Lyle School of Engineering · Dept. Mechanical Engineering · Dallas, TX

Developing machine learning pipelines to automatically detect, classify, and localize welding defects from imaging data. Research targets the replacement of manual inspection bottlenecks in manufacturing environments with real-time, AI-driven quality assurance systems applicable to power generation fabrication and heavy equipment manufacturing.

Training deep learning models on weld imagery datasets to identify porosity, cracking, incomplete fusion, undercut, and spattering defects. Pipeline integrates computer vision preprocessing, feature extraction, and model inference with structured defect reporting outputs suitable for integration into existing quality management workflows.

RESEARCH DELIVERABLES
D-01Automated weld defect classification model — porosity, cracking, incomplete fusion, undercut
D-02Real-time inspection pipeline deployable on standard manufacturing camera feeds
D-03Structured defect reporting system for quality records and traceability
D-04Model training framework with labeled weld imagery dataset curation
D-05Integration pathway for existing manufacturing quality management systems and inspection workflows
INDUSTRY APPLICATIONS

ML-based weld inspection directly addresses manufacturing quality challenges in power generation, heavy equipment, and industrial fabrication. Automated defect detection reduces human error, enables real-time non-conformance tracking, generates traceable quality records, and scales to high-volume production environments where manual inspection creates throughput bottlenecks.

PROJECT STATUS
ACTIVE RESEARCH
MODEL TRAINING
PUBLICATION PENDING
FUNDING SECURED
DEFECT CLASSES
Porosity Cracking Incomplete Fusion Undercut Spattering Burn-Through
TOOLS & STACK
PythonPyTorchOpenCV TensorFlowYOLOAWS MATLABNumPy
INSTITUTION
Southern Methodist University
LYLE SCHOOL OF ENGINEERING
MECHANICAL ENGINEERING DEPT.
DETAIL B — APPLIED RESEARCH · NEXT LAB · ACU
ITEM 1 · ISOTOPE EXTRACTION ENGINEER · AUG 2023 — MAY 2025

High-Temperature Molten Salt Systems Research

Nuclear eXperimental Testing Laboratory · Abilene Christian University
COMPLETED

Applied engineering research in isotope extraction within high-temperature molten salt environments. Contributed to nuclear testing infrastructure through vessel design, failure analysis, and quality documentation — building foundational skills in manufacturing process evaluation under extreme operating conditions.

F-01Vessel geometry improvements for molten salt containment at temperatures exceeding 700°C — manufacturing process optimization under extreme thermal conditions.
F-02Systematic failure mode analysis protocols for high-temperature mechanical and electrical components — process gap identification and resolution.
F-03Authored comprehensive SOPs establishing reproducible safety and quality standards for laboratory high-temperature operations.
DOMAINNuclear Engineering · High-Temperature Fabrication · Quality Documentation
METHODSThermal Analysis · Failure Mode Analysis · SOP Development · Process Inspection