Machine Learning-Driven AI System for Automated Flow Control: Simple LR and Decision Table ML Algorithms

Abstract

This chapter proposes a new concept of Designing an automatic flow controller that is based on AI and using selected machine learning algorithms. The optimum smart controller uses machine learning concept by learning the knowledge often convectional flow controller. The acquired data set is used to train the built model using ML algorithm using a software tool called Weka 3.8.5 is an open-source software suite developed at the University of Waikato in New Zealand. A model with optimum performance is built, the performance criteria analysis like mean square error, root means square error. The Weka software is used to build a model with 60% of dataset and evaluated with 40% of data set. The evaluated result is Root mean square error ranges from 0.96 to 0.0028. The correlation coefficient is equal to 1 and the percentage Relative absolute error is 0.0467% for simple linear regression and 15.869% for Decision table ML algorithms. Hence the built model functions as smart flow controller and the output of this controller will be imitating the conventional PID controller.