Machine Learning-Based Predictive Maintenance for Car Air Filters

¹ Department of Computer Applications, IFTM University, Rajput, Uttar Pradesh 244102, India
² Department of Computer Science and Engineering, University of Illinois, Springfield, United States
³ Department of Mechanical Engineering, R.M.K. Engineering College, Kavaraipettai, Tamil Nadu 601206, India
⁴ Independent Researcher, IEEE Member, CA, United States
⁵ Department of Mechanical Engineering, MIT Academy of Engineering, Pune, Maharashtra 412105, India
⁶ Research and Development Cell, Lovely Professional University, Phagwara, Punjab 144411, India

^* Corresponding author: This email address is being protected from spambots. You need JavaScript enabled to view it.

Published online: 2 March 2026

Abstract

Predictive maintenance is recognized as a more economical approach to the typical methods of prevention in the industry because it allows early fault detection. This article introduces the MOMENT system, an application of machine learning in the field of predictive maintenance. The proposed system is intended to evaluate the condition of the car engine air filter. The data was collected from the OBD-II system of real cars. In view of the above analysis, predictive maintenance can provide a better and cheaper approach compared to the conventional preventive approach. This paper aims to present MOMENT, which is a machine learning-based prediction maintenance system for automotive engine air filter condition assessment for automobiles using OBD II devices. A data set of over 32,000 samples was collected using a 2014 Suzuki Grand Vitara AWD vehicle. After data preprocessing and analysis, different machine learning algorithms such as Support Vector Machines, Random Forest classifier, and k-Nearest Neighbors classifiers were tested using grid-based hyper parameter optimization. From the experimental results, it can be seen that k-NN performed best in prediction, whereby the model showed an F1-score of 0.983 on the validation data. Additionally, a functional system was developed concerning the data preprocessing and recommendation aspects for filter replacements. This experiment establishes the possibility of applying practical machine learning systems for automotive vehicle maintenance purposes.