Stephen Nsikak Umurie
0%
Research

Dissertations & Research

Undergraduate research on machine learning for predictive maintenance in automobile engines — from an initial decision-tree classifier to a two-stage ensemble diagnostic pipeline.

Extended Paper

A Two-Stage Random Forest Ensemble for Condition-Based Fault Detection and Root-Cause Diagnosis in Automobile Engines, with a Deployed Diagnostic Dashboard

Stephen Nsikak Umurie · Extension of the original dissertation

Extends the original decision-tree work into an ensemble and gradient-boosting framework tested on two datasets. A two-stage Random Forest pipeline (detect, then diagnose) reaches 100% accuracy on binary fault detection and 64.8% on fault-subtype identification on EngineFaultDB, while a validation-tuned gradient boosting classifier nearly doubles fault-class recall on the original dataset. Both trained models are packaged into a public, browser-only diagnostic dashboard — no install, no internet connection required.

Original Dissertation

Development of a Decision Tree-Based Predictive Model for Condition-Based Fault Detection in Automobile Engines

B.Eng. (Hons.) Mechanical Engineering · Covenant University, Ota · June 2025

Final-year dissertation developing a decision-tree predictive model to classify automobile engine condition as GOOD or BAD from six sensor parameters — engine RPM, fuel pressure, lubricant oil pressure and temperature, coolant temperature and pressure — with a root-cause feature that identifies which parameter is responsible for a bad reading.