Enhancing the Efficiency of Internal Combustion Engine Repair: A Machine Learning-Based Predictive Maintenance Approach

Abstract

In Kazakhstan, a large part of the vehicle fleet is characterized by a high degree of wear, harsh climatic conditions, and unstable fuel quality, which leads to premature engine failures. Traditional reactive repair methods, aimed at eliminating failures only after they occur, lead to longer downtime and higher operating costs. Preventive maintenance does not always ensure the required reliability because it relies mainly on fixed time intervals rather than the actual technical condition of engine components.

Therefore, the implementation of predictive maintenance methods based on machine learning technologies is becoming increasingly important in modern transport engineering. This study presents the results of experimental research conducted in Karaganda, where a machine learning model was developed using sensor data from internal combustion engines to detect potential failures at an early stage. The results show that the proposed approach reduces unplanned failures, extends engine service life, improves fuel efficiency, and decreases harmful emissions.