DEVELOPING A MATHEMETICAL MODEL USING MACHINE LEARNING IN FAULT DIAGNOSIS OF AUTOMOBILE SYSTEMS

Abstract

Online fault diagnosis system level for a mechatronic system makes use of a number of sensors which provide number of data from various components that they are linked with the help of the reliability of any system. It increases reducing the run time error by providing accurate results from the previous data measurement. The current technological development in automobile industry demand Accuracy and Precision more than anything to meet such high expectations. One must incorporate the knowledge of various learning techniques which provide a gateway to overcome this reverberating situation. This is where machine learning plays vital role because of number of options it provides to solve a particular problem. Decision tree and regression techniques being one among them are the ones incorporated here in order to perform fault diagnosis on any system. The interrelations between all the sensor
measurements must be timely monitored to collect data which can be put in any one of the machine learning technique to enhance the output.

This is challenging and often demands for better understanding of the techniques so as to select the best one for our problem which in the end generates the required output in this project. We try to enhance the diagnosis process with the help of basic level machine learning techniques which meet the preliminary requirements of the data collected from automobile that is the Go Kart which we made use of. A sufficient amount of data is collected with the help of sensors and then later used for enhancing the output of the system.