Air Pressure System Failure Prediction and Classification in Scania Trucks using Machine Learning

Abstract

Air Pressure System is a vital part of heavy vehicles like Scania trucks. The braking system in these vehicles is dependent on-air pressure and hence there is a need for the proper functioning of the air pressure system. Predictive maintenance in automobile industry reduces the maintenance cost and improve the performance of the vehicle. This can be achieved either manually or automatically. Manual predictive maintenance needs interference of the human task and may induce some errors. Automatic predictive maintenance through artificial intelligence techniques explores the hidden cause for failure of air pressure system in the Scania trucks.ln the proposed system, machine learning approaches are investigated for predictive maintenance of the trucks hised on condition of air pressure system. The dataset used in this work consist of 59,000 negative instances and 1,000 positive instances. Hence there is a need to address the class imbalance issue laefore applying machine learning algorithms. Many resampling techniques like under sampling, over sampling and SMOTE are analyzed for the efficiency of the classifier. After pre-processing the data, machine learning classification algorithms like Random Forest, Logistic Regression, Support Vector Machine, Naïve Bayes, Decision Tree, K-Nearest Neighbors and Stochastic Gradient Descent are implemented and accuracy of the classifiers are analyzed. Experimental results show that Stochastic Gradient Descent algorithm with under sampling outperforms other classifiers with accuracy of 99%.