Data Modelling and Prediction of OBD-II Time Series using ARIMA

Current vehicles include networks of interconnected sensors that collect, exchange and analyze data related to the vehicle operations on the purpose of providing insights and services. Such measurements can be extracted using diagnostics systems, such as OBD-II (On Board Diagnostics) system. Monitoring engine operations gives insight into the behavior of the driver and the engine, the identification of faults and malfunctioning as well as the improvement of certain characteristics. In this work, we consider and model vehicle engine monitored data. Specifically, we employ the ARIMA (Autoregressive Integrated Moving Average) models to model key engine parameters based on OBD-II time series data and investigate their predictability. The ARIMA model is applied on six engine operation parameters. The predictability of each parameter is evaluated and verified according to the fit of the ARIMA model in additional values of these parameters.