Engine Performance Optimization using Machine Learning Techniques

The purpose of this paper is to integrate the concept of Supervised Learning Algorithms in Engine tuning. These days Machine learning has become a very valuable tool for prediction. A given subset of this domain involves using supervised algorithms to intake data, analyze the data and 'learn' from it. The more the data that is processed by it (training stage), the better it learns (Fitting Parameters on Training Set) and the better it will be able to predict (Prediction Stage). By feeding data to the system we are teaching the system about how the input parameters (plenum volume, exhaust and intake runner length, Engine rpm) in the data are inter-related with one another and how the values of a set of variables can change by changing the value of any one variable. The efficiencies of various regression models were used and neural networks were also implemented.