In the rapidly advancing world of automotive technology, ensuring safety and reliability is paramount. A vehicle’s braking systemis crucial for preventing accidents and mechanical failures. Monitoring the braking system performance is vital, as any degradationcan have severe consequences. This study focuses on identifying key features to evaluate braking system performance. Vibrationsignals were analyzed under various conditions including air in the brake fuid, mechanical fade, reservoir leak, and diferent brakepad wear states. The feature extraction process used statistical, histogram, and autoregressive moving average (ARMA) techniques.These features were then selected using the J48 decision tree to determine their signifcance. Each feature type was independentlyclassifed to assess its efcacy, achieving classifcation accuracies of 98.00%, 99.00%, and 97.00% for statistical, histogram, andARMA features, respectively, using the random forest (RF) classifer. To improve accuracy, combinations of selected features weretested with histogram and statistical, ARMA and histogram, ARMA and statistical and histogram, and ARMA and statistical. Thehistogram and statistical features and the combination of all three features produced a classifcation accuracy of 100.00% with theRF classifer. Histogram and statistical feature combination was deemed most reliable, balancing high accuracy with reducedinformation and time complexity.