Development of an IoT Based Process Plant Emission Controller

Climate change problems emanates largely from the process industries globally. A control scheme using Dynamic Resource Allocation (DRA) in an emissive process plant was developed in this thesis. A field survey was conducted at Coca Cola Nigeria Ltd. Plot 126 Trans Amadi Industrial Layout Road, Port Harcourt, Rivers State, Nigeria to ascertain better ways of managing Carbon dioxide (CO2 ) emissions in concurrent process plants. With the aim of overcoming the challenges of conventional process plants and inspired works by relevant authors, the work adopted an Unsupervised Machine Learning Algorithm based on Pattern Discovery and Reinforcement Learning (UML-PDRL). A DRA that is SMS activated for the automation aggregated process plants was implemented to address the issue of industrial plant concurrency for emission conservation. An SMS control signaling using UML-PDRL was used to give an on-demand SMS signal to a plant agent such that the running plant shuts down on its allocated time and triggers another plant and also shuts down on its own when it gets to a certain threshold to solve the problem of CO2 emission. C++ programming and Arduino Uno board was used to characterize an emulated process plants. A validation study on supervised, unsupervised and semi supervised algorithms were carried out through estimation inspection. The efficiency obtained were 4.24%, 4.94% and 90.82% respectively; the latency response was 33.33%, 41.67% and 25.00% respectively. The results showed better resource allocation in the clustered plants using unsupervised algorithm.