Real Time Brake Health Monitoring System A Machine Learning Approach

Abstract

Due to the recent technological advancements in the automobile industry, vehicle usage is rising day by day. Reliability must be guaranteed to survive in a competitive worldwide market. Many sensor fusion models have been implemented in order towards guaranteeingvehicle stability andan elevated degree of safety and security for drivers and passengers. To take the appropriate action, the sensor data is processed using various programming techniques. One such control element is the brake system, which requires close attention to ensure proper operation. It could result in major catastrophic repercussions including accidents, brake down, frequent brake down, and so on. if it is not carefully managed. As a result, the brake system should be regularly monitored. The primary objective of this research is to utilize vibration signatures to monitor the brake system s health condition. The feature-based analysis has been examined for use in light motor vehicle (LVM) hydraulic braking system failure diagnosis utilising a variety of computational approaches. The vibration transducer was used to gather the vibration signals from a real-time braking setup in both good and bad conditions. The captured vibration signals were used to extract the statistical, histogram, and wavelet features. To find the best features, an attribute evaluator and effect of number of feature study was done Various classifiers, including tree-based algorithms, rule-based algorithms, function-based algorithms, Rough Set, CHIRP, logitBoost, Kstar, Support Vector Machine, KNN classifier, Bayes Net, and Naive Bayes were used to classified the selected best characteristics. In order to figure out the optimal data framework for the accurate classification of braking states, the study focuses on the fault diagnosis procedure. A data model will be built and tested before actual implementation in vehicles. Based on the outcome, the tested model is being used for the online condition monitoring process.