IoT-Based Driving Pattern Analysis and Engine Sensor Damage Prediction Using Onboard Diagnostics

Abstract

Car diagnostic tools use sophisticated software to quickly and accurately identify problem areas in a car’s engine or elsewhere. The adaptation from mechanical systems to electrical paved the way for much research in the area of vehicle monitoring through a server. Although many studies focus on the tracking of vehicles, fault detection has also gained considerable attention. From our background study, we noticed that most of the systems currently available display the data from OBD and do not perform any functions using that data. We propose a machine learning-based driving pattern analysis and sensor damage prediction system using the data collected from an Onboard Diagnostics (OBD) port. Our system consists of an ELM327 adapter to read data from an OBD port, a mobile application, and a cloud backend. Using the time series data collected, we perform driving pattern analysis and sensor damage detection from the backend server. The smartphone application makes the findings of these analyses visible. The findings can be accessed by the drivers directly via the smartphone app. Drivers are notified of alerts created in the backend as a result of unfavorable conditions.