Wireless Sensor Networks for Real-Time Health Monitoring of Electric Powertrains

This study presents a Wireless Sensor Network (WSN) based continuous monitoring system of which key parameters such as temperature, vibration and electrical current are continuously monitored over powertrain components. During pre-processing to improve data quality, a Kalman filter is applied to get rid of noise and preserve the real-time signals. Principal Component Analysis (PCA) is applied to the data for feature selection in order to eliminate dimensionality and keep only important fault-related features. Finally, a random forest classifier is used for multi-class classification of the component health state in to Normal, Warning, or Critical from the processed data. The proposed framework can effectively support intelligent fault diagnosis and predictive maintenance to reduce the downtime of electric vehicles and improve the lifecycle performance.