Research On Application Of Laser-induced Breakdown Spectroscopy In On-line Detection Of Coal And Petroleum Coke In Thermal Power Plant

Coal is still China’s main source of primary energy.For thermal power plants,both coal and petroleum coke can be used as fuel for electricity production.The complexity and diversity of coal quality and the high pollution of petroleum coke directly affect the safety and economy of boiler operation in the power plant.The lack of real-time detection technology in the production process of coal and petroleum coke seriously restricts the development of safety and intelligence of related industries.Due to the lag of traditional laboratory detection methods,new technologies are urgently needed to realize the rapid detection of coal and petroleum coke.Laser-induced breakdown spectroscopy(LIBS)is a new analytical tool in the field of spectral analysis.LIBS uses a pulsed laser to excite the plasma on the surface of the sample and then uses a spectroscopic analysis device to extract spectral information of atoms,ions,and molecules from the plasma for qualitative and quantitative analysis.The advantages of LIBS include the analysis of all elements,fast real-time detection,low detection limit,simple sample pretreatment,etc.Therefore,LIBS has great advantages and prospects in the field of industrial online analysis.Based on LIBS technology,this paper adopts a combination of theoretical analysis,experimental research,and engineering application to research coal and petroleum coke.The important quantitative detection parameters,such as coal industry analysis,coal melting characteristics,and element contents of coal and petroleum coke,were analyzed in depth.It provides an experimental basis and theoretical basis for LIBS technology to be used in real-time measurement of related industries.Specific research contents are as follows:1.Set up an online detection system for coal and petroleum coke in power plants.National standard coal samples and coal samples from different thermal power plants provided by China Huadian Corporation LTD were selected as the research object of coal quality testing.Standard petroleum coke samples from Shandong Zhigu Carbon Research Institute were selected as the research object of petroleum coke detection.The experimental device of LIBS on-line detection was designed,and the experimental parameters,such as sample pressure,laser energy,focusing position,and acquisition delay time,were optimized by spectral analysis and plasma photography analysis.Background spectrum removal,spectral normalization,baseline removal,abnormal spectrum screening,and characteristic wavelength Lorentz fitting were used to pretreat LIBS spectra.These works can reduce the influence of LIBS spectral instability in practical engineering applications and improve the quantitative detection accuracy.2.The matrix matching method is used to model and predict the coal samples,which improves the quantitative analysis accuracy of coal ash content,volatile matter and calorific value.Determination of ash content,volatile matter,and calorific value in coal with LIBS is significantly affected by the matrix effect,due to the physical and chemical properties of different types of coals.The support vector machine(SVM)classification method optimized by genetic algorithm(GA)was used to match the matrix of different types of coal:101 groups of coal samples were divided into three categories according to ash content,and then the partial least squares regression(PLSR)method was used to establish different models for all each type of coals.With this detailed classification scheme,the coefficients of determination(R2)of the training set and test set of volatile matter were improved from 0.9269 and 0.9310 to 0.9959 and 0.9888,respectively.The root-mean-square error of cross-validation(RMSECV)and the root-mean-square error of prediction(RMSEP)of volatile matter were also reduced from 1.9940%and 1.8320%to 0.4989%and 0.7719%.respectively.The prediction performances of ash content and calorific value are also improved by this LIBS detection method based on coal ash matrix matching.3.The prediction of coal fusion characteristics and determination of slagging behavior are of great significance to the operation of thermal power plants.The traditional method needs a long time to measure coal ash fusion temperatures(AFTs)by heating the coal ash cone.Based on the correlation between AFTs and the content of various metal elements,the coal AFTs were predicted directly from coal rather than coal ash by LIBS technology.On the premise that ash content,volatile matter,fixed carbon,and moisture are known on the air-dry base state,these four parameters are added as the generalized spectrum to the AFTs prediction model.Four methods(support vector machine,neural network,random forest,partial least squares discrimination analysis)were used to classify coal samples with AFTs critical point 1500℃,and then support vector regression(SVR)was used to model the coal samples with AFTs less than 1500℃.By this method,the R2 of softening temperature(ST)and hemisphere temperature(HT)reached 0.9958 and 0.9856,the RMSECV reached 4.88℃ and 9.11℃,the RMSEP reached 8.15℃ and 11.3℃,respectively.The trend relationships between the AFTs and the intensities(or ratio)of the spectral lines corresponding to individual elements of the LIBS spectrum were analyzed.Then the variation trend of coal AFTs with specific element content(or ratio)is deduced qualitatively.Using LIBS to directly predict the AFTs of coal samples improves the working efficiency,which can be used for real-time guidance in the actual production process of thermal power plants.4.LIBS was used for the first time to detect petroleum coke,and the contents ofvanadium(V),iron(Fe),nickel(Ni)in petroleum coke were analyzed.Although the element quantitative detection accuracy of LIBS petroleum coke has a certain gap with atomic absorption spectrometry(AAS)and inductively coupled plasma atomic emission spectrometry(ICP-AES),it has the advantages of simple sample pre-treatment,faster measurement speed,less pollution of the measuring environment,and no reagents consumption.Combined with the NIST database and LIBS spectra of petroleum coke,after feature selection of the wavelengths corresponding to V,Fe,and Ni in the LIBS spectrum of petroleum coke,the element contents of the test set were predicted after SVR modeling for the three elements in the training set.The RMSEP of V,Fe,and Ni reached 36.4 mg/kg,36.04 mg/kg,and 14.94 mg/kg,respectively.The feature selection process is as follows:cyclic selection independent variable was used for SVR modeling and determined the weight of characteristic wavelength based on the root-mean-square error(RMES)calculated by the fitting result and true value,then the feature selection is carried out by sorting the weight results.The RMSEP of V and Ni were close to the reproducibility(accuracy)of element measurements(32 mg/kg and 14 mg/kg)using the wavelength dispersive X-ray fluorescence spectroscopy(WD-XRF)method in ASTM D6376-10 standard.This work has verified the feasibility of LIBS for the detection of trace elements in the petroleum coke industry,which provides a theoretical and application basis for the detection in the production process of petroleum coke as fuel and carbon products.To sum up,the quantitative detection of coal and petroleum coke in a thermal power plant was systematically studied based on LIBS technology.The industry analysis and calorific value analysis of coal with matrix matching were studied.Four ash fusion temperatures of coal are predicted based on the relationship between coal melting characteristics and element content.Full-spectrum modeling and feature selection methods were used to optimize the quantitative detection of major elements in coal and trace elements in petroleum coke to improve the prediction accuracy.The above research has been applied to the online analysis of coal quality in thermal power plants,and will strongly promote the development and application of LIBS technology in industrial fields.