Profiling Of Single Particle Combustion And Flame Monitoring Through Digital Imaging

Solid fuels(e.g.,coal and biomass)are the major energy resource worldwide due to their abundant reserves and low prices.This will continue for a long period because of the high cost of future renewable energies.In recent years,environmental issues have attracted the global attention.Improving the combustion efficiency and emissions control are effective ways to deal with environmental problems.Therefore,the in-depth understanding of the combustion behaviors of solid fuel particles and effective monitoring of flames in furnaces are highly demanded in scientific research and industrial applications.Study of the combustion behaviors of single solid fuel particles is an essential step for the in-depth understanding of the combustion processes of the fuel.It can help for the optimization of the burner design and combustion control.Therefore,the combustion behaviors of single solid fuel particles are investigated in this dissertation.The main activities of research in this area are:(1)An investigation into the single solid fuel particle combustion behaviors based on high-speed imaging and digital image processing is carried out.Images of burning single particles are collected using a high-speed camera.The high-speed video images are then processed through image processing.Consequently,key physical parameters(i.e.,size,shape,rotation and brightness)of the single particles are defined and calculated,and finally,the characteristic parameters are used for profiling single particle combustion behaviors.(2)Experimental studies on the combustion behavioral characteristics of single pulverized coal particles of different sizes(particle sizes of 106-150?m and 150-212?m)and palm kernel expeller(PKE)particles(particle sizes of 75-125?m and125-150?m),especially the volatile combustion and the carbon/fixed carbon combustion.The experimental results are analyzed and compared.In practical combustion processes,on the other hand,fuels are conveyed into the furnace with gas flow and then burned in the furnace.Although research of the combustion behaviors of single particles is critical to an in-depth understanding of the combustion,it cannot replace studies of the actual combustion processes.In order to achieve effective control of combustion processes and reduce emissions,it is necessary to study practical combustion processes and monitor flames in furnaces.Therefore,the second part of this dissertation is aimed at studying the characteristics of the flame and monitoring combustion processes.The main activities of research with this aspect are:(1)A new flame image segmentation method is proposed based on multiscale color and wavelet-based textures(MCWT).The technical challenges of establishing such a method is to maintain the segmentation accuracy and a fast processing time particularly where the flame images suffer strong background noise and the variable operation conditions.In the proposed method,the color and texture features are combined as the feature matrix and the flame images are segmented in multiscale spaces to improve the quality of the image segmentation.Flame images captured from an industrial-scale coalfired test rig under different operation conditions are used to evaluate the performance of the proposed method.The accuracy and efficiency of the image segmentation by the proposed method are compared with that of other image segmentation methods.The results demonstrate that the performance of the MCWT-based flame image segmentation method is superior to other traditional methods.(2)The fractal dimensions of the thermal mixing layers of turbulent flames are calculated and studied through digital image processing techniques.According to the characteristics of turbulent flames,the images of a turbulent flame were acquired using a digital imaging system on an industry-scale pulverized coal-fired test rig and the fractal dimensions of the flame thermal mixing layers are calculated through morphology-based flame image processing techniques.The impact of the primary air flow and secondary/tertiary air ratio on the fractal dimensions of the thermal mixing layers is studied.(3)The existing imaging-based combustion process monitoring methods are designed for a constant individual process.In order to meet the requirement of multimode combustion process monitoring,a monitoring method for variable combustion conditions is proposed by combining digital imaging,principal component analysis and random weight network(PCA-RWN)techniques.The color and texture features of the flame images are extracted,and the dimensions of the features are reduced by a global PCA model.The combustion condition is then recognized using a RWN model.Finally,the Hotelling's T2 and squared prediction error(SPE)statistics are used to identify the state(normal or abnormal)of the combustion process.The proposed approach is examined by using flame image datasets obtained at an industry-scale combustion test facility.Operation conditions were varied by changing the primary air flow and the secondary to territory air(SA/TA)ratios.The experimental results demonstrate that,for both variable primary air flow and SA/TA ratio operation conditions,the operation conditions and the abnormalities are successfully identified by using the proposed PCARWN based multimode process monitoring method.