Application of Meta family Classifiers for monitoring hydraulic brake system using vibration based statistical learning approach

Abstract

In the modern days, use of vehicles is increasing rapidly. It is very essential that the vehicle must have a good control mechanism which ensures the safety of the vehicle. The brake system in automobile is one of the important control element which needs to be monitored. The unconditional brake leads to catastrophic failures. Hence, the brake system should be monitored regularly. An experimental study is proposed for the brake system monitoring using vibration signals. The vibration signals are captured under all possible brake conditions. The hidden information in the vibration are extracted as statistical features. We carry out the feature selection. Classification using the selected features is the final step in machine learning (ML). Meta family classifiers are used for the study. Among the considered classifiers, Bagging algorithm produced 80.8 % accuracy for monitoring the brake condition.