| Coal analysis is of great significance for an efficient use of coal resources as well as the optimization and management of power generation process in thermal power generation industry.Traditional coal analysis methods are usually offline and performed in laboratory,which are difficult to quickly and in-situ obtain the data of various analysis parameters of coal to monitor the running status of boiler in the thermal power plant.Laser-induced breakdown spectroscopy(LIBS)is a new elemental analysis technique with many unique advantages in the application of rapid coal quality detection.However,this technique has not yet been widely used in the field of online coal analysis due to the lack of researchers' attention to some engineering and technical issues.This thesis focuses on industrial coal quality online detection system using LIBS.And the design and development of the software as well as the study of quantitative algorithms in the system are the main research goal of this text.The detailed contents are as follows:Firstly,a LIBS setup which was used for quantitative analysis of coal quality was established.In order to ensure the long-term and reliable work of the setup,all optical components and hardware modules in the setup were encapsulated and integrated according to the actual demand of the industrial environment in the power plant.And the spectral data of40 standard coal samples were acquired though this setup.Four multivariate calibration methods: partial least squares regression,support vector machine regression,neural network regression,and principal component regression,were used and compared based on these spectra.As a result,the neural network regression had better prediction ability than the others,and the average absolute error for prediction of ash,volatile matter,calorific value and carbon content were 0.69%,0.87%,0.56MJ/kg,and 1.27% respectively.In terms of generalization ability and modeling efficiency of the calibration model,neural network regression can seek the best compromise in the prediction ability and modeling efficiency,and is demonstrated to be a more suitable calibration method for rapid and in-situ detection of coal quality usingLIBS in coal industry.Secondly,a full-featured and highly automated software developed by C++ and Python was customized for the online coal analysis system based on the LIBS hardware setup and specific requirements of the power plant.This software can realize many functions that required by the industrial coal analysis system,such as instrument control,spectral data acquisition and analysis,data transmission though the network,and working in an unattended manner.Finally,the self-developed software and quantitative analysis algorithm were integrated into the LIBS system,and the full control of coal sampling,drying,crushing,cake pressing and measurement were realized.Based on the neutral network algorithm,the calibration model of six common analysis parameters,namely moisture content,ash,volatile matter,hydrogen content,sulfur content and calorific value was established using coal samples collected from the power plant.The performance of the entire system was tested,and the result showed that the system meets the Chinese national standard of coal online analysis using the neutron activation analyzer. |
PDF Full Text Request