Soft sensingforsamplehandling system

Abstract

This thesis aimstodevelop a soft-sensing model for sample handling system and demonstrate their usefulness in sensor validation (Monitors the performance of the sensor to maintain or improve performance of sensor) and sensor backup (Problem of unavailability of data from plants when sensors become unavailable or may have been damaged). Various machine learning algorithms (Linear Regression, Decision tree, Random forest, Support Vector regressor) were studied for the development of the soft sensing model and it was found Support Vector Regressor to be performing best. Therefore, a Support Vector regressor was used to estimate a particular sensor measurement based on other sensor values which are correlated.