Deep Learning Model for Prediction of Air Mass Deviation Faults

Abstract

Major Systems of an internal combustion Engine are Air System, Fuel system, and Exhaust system. Any malfunction in these systems increases emissions. OBD legislation mandates to monitor these systems for any faults and appropriate action should be taken in case of the any faults which increase vehicle emissions. The idea of the paper is to find the Air mass flow deviation faults using datamining and machine learning based approach. Detection of fault is classifying whether system is faulty or not. Objective is to create a deep learning model using the available vehicle data to classify the system for a fault. Three main inputs for the Air Mass flow in an internal combustion Engine are
1) Fresh Air which measure using Mass Air Flow sensor
2) Low Pressure EGR
3) High Pressure EGR
During vehicle lifetime, due to different real vehicle operating conditions and environmental conditions, deviation in the set point of air mass flow and actual mass flow are possible to an extent, which can affect vehicle emissions. Deviation in the Air Mass flow can be caused by intake Air mass, LP-EGR, HP-EGR. The Aim of the project is to create the deep learning model for Air Mass Flow Hi and Low faults using the available data, and associate the fault to the component in the Intake Air System.

Keywords: OBD, LPEGR, HPEGR, Machine Learning, Emissions, Deep Learning, Air Mass Deviation, Internal Combustion Engine, Air Mass Flow, Hi and Low faults, Intake Air System