Parameters Analysis And Assessment Of Pulverized Coal's Ignition Sensitivity Based On Terahertz Time-domain Spectroscopy

Pulverized coal,used as fuel or gasification materials,has the advantages of energy saving and high-efficiency,which ensures the clean utilization of coal resource.The ignition sensitivity of pulverized coal has a direct relation to failure-free operation and security of facilities using powdered coal.The traditional analytical techniques for analyzing ignition sensitivity of pulverized coal can't meet the demands.The spectrum technology is considered as a new detection method of measuring the properties of pulverized coal.However,because of the indirection method of this technology,the accuracy of measurement depends on the selected spectrum data and the modeling methods.The spectrum of pulverized coal is characterized by high dimensions and distinctive features,which can be affected by electromagnetic scattering.It is essential to study the scattering correction method,dimensional reduction,feature extraction and quantitative modeling method.THz-TDS is a newly development spectrum technology.The frequency range of THz contains the information of the vibrations and rotation of the molecular,which is concerned with the physical,chemical and structural features of the analyzed substance.Relaying on this advantage,THz-TDS has been applied into region of molecular,chemical component and macrostructure detection of the substance.Therefore,the study was focus on the qualitative and quantitative method of detecting of the fineness,properties and ignition sensitivity of the pulverized coal adopting THz-TDS.The main work and achievements are as follows:(1)Because of the electromagnetic wave scattering,spectral distortions are common in quantitative terahertz spectra analysis of the granulated coal mixed with HDPE powder.A scattering correction method for extracting the intrinsic characteristic of the pulverized coal was deduced.To investigate the contribution of scattering,then the feasibility of two effective field approximation(EFA)models and quasi-crystalline approximation(QCA)model were discussed.The experimental result indicated that the QCA model combined with Waterman-Truell EFA has a better performance of scattering correction.The investigation provided a reasonable scattering correction method of THz spectral quantitative analysis for the coal particles in different sizes.(2)The traditional method of sampling pulverized coal is complicated.During sampling process,a secondary crushing of coal dust is inevitable,and the concentration of coal particles should be strictly controlled.To deal with the problems above,a mean diameter estimation method of coal particles based on the chaos features of THz timedomain signal was proposed.The conception of chaotic nonlinear dynamical system was introduced into the analysis of the THz time-domain signal.The THz signals could be regarded as a time series generated by a complex nonlinear dynamical system.The chaos features of the coal samples were extracted from the measured THz time-domain signals,and then a qualitative relationship between the chaos features and the mean diameter of the coal particles were turned up.A mathematical model was established using SVM algorithm to quantify the relationship.Gravitational search algorithm(GSA)was employed to optimize the parameters of SVM,as a comparison,PSO algorithm was also introduced.The experimental results showed that the GSA-SVM combined with chaos features provided a higher accuracy and a better convergence.In the further experiment,a coal sample with unknown mean diameter was test by the model,and a reasonable result was obtained.(3)In the domain of the high dimensions and redundancy of the THz spectra,a quantitative analytical method of the ash,volatiles,carbon matter content and heat of coal using Least squares support vector machine(LSSVM)combined with manifold learning algorithms was proposed.Principal component analysis(PCA),Gaussian process latent variable model(GPLVM),Stochastic neighbor embedding(SNE)and Maximum variance unfolding(MVU)were adopted to extract the characteristic features of the extinction spectra as the input of LSSVM,and an adaptive chaotic particle swarm algorithm was used to optimize the parameters of LSSVM.The experimental result showed that the prediction of ACPSO-LSSVM combined with SNE algorithm has a higher correlation coefficient and lower root mean square error.Therefore,THz-TDS is a feasible technology for the quantitative analysis of coal property.(4)By using THz-TDS,an ignition sensitivity of pulverized coal estimation method was developed based on the manifold learning algorithm and chaos features.During the study,the ignition sensitivity of the coal samples was obtained by thermogravimetric analysis,and the experimental results proved that the ignition sensitivity of pulverized coal was obviously affected by the coal properties and the diameter of the coal particle.Then the THz time-domain signals and extinction spectra were measured by THz-TDS.According to the findings mentioned above,the chaos features of the THz time-domain signals were calculated,and the characteristic features of the extinction spectra were extracted by using SNE algorithm.The two kinds of features were joined together as the input of ACPSO-LSSVM and GSA-SVM algorithms.The experimental result demonstrated that a better performance was achieved by the GSA-SVM algorithm.Therefore,the method of joining the chaos features and characteristic feature of the extinction spectra together presented a more precise way to estimate the effect of the coal properties and particle size on the ignition sensitivity.In this paper,a quantitative analysis model of the ignition sensitivity of pulverized coal was proposed.During the study,with the knowledge of spectroscopy,coal chemistry and machine learning,the THz-TDS was adopted to analyze the related parameters of the ignition sensitivity of pulverized coal.The research findings will enrich the theory in both the THz spectra data handling,regression modeling,and enhance the accuracy of the relevant parameter analysis of the ignition sensitivity of pulverized coal.It also will cover the inadequacy of the traditional quantitative analysis method of the ignition sensitivity of pulverized coal.Besides,for the application of THz-TDS to the analysis of powdered coal,the achievements in this paper are of good theoretical and practical meaning.