Author(s) :

Jayden P. Chen.

Volume/Issue :

Abstract :

The rise of predictive maintenance models has revolutionized vehicle maintenance, promising significant improvements in performance and lifespan. This research aims to develop a robust predictive maintenance model for automotive engines using a hybrid approach that combines neural networks and logistic regression models. By analyzing patterns within a publicly available dataset of 19,503 engine cases, which includes features such as engine rotations per minute, temperatures, and pressures, the study trains hybrid models to predict when a vehicle requires maintenance. The methodology involves preprocessing the dataset, training individual models, and integrating them within a stacked ensemble framework. Neural networks are leveraged for their complex pattern recognition capabilities, and logistic regression models offer interpretability and simplicity. Metrics such as accuracy, precision, and recall evaluate the models’ performance. The ultimate goal is to enable vehicle owners and mechanics to address potential issues proactively, ensuring better vehicle performance and extending engine lifetimes. The hybrid models show enhanced success compared to traditional models, providing potential contributions to predictive analytics, and a new standard for various industries.