DefinePK

Abstract

The scrutiny of combustion quality and its equivalent NOx emissions from flame images in thermal and gas turbine power plants is of immense significance in the realm of image processing. The principal goal is in detection, recognition and understanding of combustion conditions ensuring low NOx emissions. In this work, soft sensors using feed forward neural network trained with Back Propagation Algorithm (BPA) and Ant Colony Optimization (ACO) are used for flame image classification. The scheme uses the information from the color of the flame images as fundamental which is dependent on the combustion quality and NOx emissions. The initial gait is to describe a facet vector for each flame image including 10 feature elements. The distinctive attributes of the captured images is enhanced using curvelet transform. The perception of object (flame feature) recognition and classification of the flame image is conceded out to measure the flame temperature, combustion quality and NOx emissions from the flame color. The samples including 51 flame images, parts of which are used to train and test the model. Finally, the entire samples are recognized and classified. Experiments prove this method to be effective for classification of flame images. The solution includes the Internet of Things (IoT) where the intelligent sensors are connected to the embedded computing system to monitor the fluctuation of parameters relating to combustion quality. This form is flexible and dispensable for the infrastructural environment that needs continuous monitoring, controlling and behavior analysis in power plants. The working performance of the proposed model is evaluated using prototype implementation, consisting of Arduino UNO board, intelligent sensors and MATLAB with Arduino hardware support package. The implementation is tested for monitoring the combustion quality with respect to the normal operating conditions which provide a feed control for NOx emissions monitoring to make the environment smart.