Research On Intelligent Recognition And Early Warning Method Of Coking Image Of Thermal Power Boiler Based On Machine Vision

Coking is a common problem in thermal power boilers,if it is light,it will reduce the efficiency of the boiler,if it is serious,it will lead to forced shutdown and serious economic losses,therefore,on-line coking monitoring of boiler heating surface has always been an important research direction of thermal power boiler,among them,the intelligent recognition and early warning of coking based on visual imaging monitoring is also a problem to be solved.This paper presents a boiler coking image recognition and early warning system based on infrared visual imaging,it mainly through the combination of infrared imaging,image processing,image classification,quantification and software development,the acquisition of coking image,image preprocessing and evaluation,coking image recognition and classification,quantification of coking area and the development of visualization software are realized.Firstly,starting from the principle of through flame imaging,select the appropriate camera and accessories,carry out the test to verify the feasibility of through flame imaging,design the relevant structure and layout scheme,and collect the coking image in the boiler.Based on the collected coking image in the furnace,a set of preprocessing methods for the image are studied,so that the image can be used as the original data for subsequent image recognition,classification and coking quantification.The pre trained network in MATLAB is used to study the coking recognition and classification,including the production of data set,the selection of network and optimizer,and the network with the best effect is selected for coking classification.The quantitative method of coking area is studied.In order to verify the accuracy of the quantitative algorithm,the simulated coking image is taken and the area quantitative experiment is carried out,which makes the coking data more intuitive.Based on the analytic hierarchy process,the coking degree scoring system is studied,and finally the visual coking identification and early warning software is developed by QT,it concentrates the functions of real-time monitoring,coking classification,coking area quantification and early warning in the software,and completes the design of coking identification and early warning system,which provides a reference value for the development of boiler intelligent monitoring.