Data Ensemble Model for Prediction of Oxygen Content in Gas fired Boiler for Efficient Combustion

Measurement of the oxygen concentration in a boiler is crucial for efficient combustion in a gas-fired boiler, which results in the reduction of the amount of release of toxic gases in the environment. This research work describes a method to improve the effectiveness and performance of the Boiler combustion efficiency by the prediction of oxygen content using a Data Ensemble model. More useful data about the flame's physical characteristics can be viewed using CCD flame pictures. Images of boilers at various temperatures, air pressures, and gas conditions have been acquired through on-site tests in a genuine combustion system and classification models have been used. By applying a deep learning prediction model to these images, and comparing the results with the test data set, the accurate value of oxygen content is determined. An accuracy of 98% and a loss of 0.02 have been obtained. The unsupervised and semi-supervised, multilevel based deep ensemble models are applied and oxygen content is predicted.