Smart Vehicle: Online Prognosis for Vehicle Health Monitoring

The automation industry is continu-ously increasing at a very large pace, thus like any other industry predictive maintenance has become more important. Vehicles in the modern world al-ready have built in ECUs and different sensors that make them smarter. In Addition, combination of ECUs with OBD2 (On-Board Diagnostics) port leads to much detailed data, leading to the generation of special codes using which the mechanics identifies the specific issues that might have occurred in the vehicular system but all this information is not suffi-cient to diagnose the failures in advance, therefore, use of additional sensors in required. With the use of Machine Learning analyzing sensor data and per-forming failure predictions is feasible. This paper deals on how to integrate OBD2 port with different additional sensors into a single system using a three-layered IOT Architecture and a proposal is put for-ward for fault prediction using all the data from the system and show up the end results in an Android application. The entire system incorporates the four main sub-assembly of the vehicle - fuel, ignition, ex-haust and cooling system. Interesting Patterns are learned using the Random Forest algorithm, these patterns are then used to predict the failures that might occur in the vehicle in the future. Also, the data set for the same has been developed by creating a sheet as per the maximum and minimum possible value for the different parameters. The major mo-tive behind this is to let the driver know about the issue before time and to provide the details on a portable Android mobile application.