Study On The Biomass Fuel Properties Based On Laser-induced Breakdown Spectroscopy

The inhomogeneity and diversity of the raw materials for biomass pellet fuel make it difficult to monitor the quality of the production process and combustion process.At present,the quality supervision of biomass pellet fuel and the design and operation of biomass combustion system basically depend on several main characteristics of biomass fuel,namely the gross calorific value,volatile matter and ash content of biomass fuel.The traditional biomass fuel property analysis test mainly relies on the laboratory chemical analysis method,which is time-consuming and tedious,and it is difficult to meet the demand of grasping fuel information in real time.In addition,the existing rapid measurement technology have some limitations.Therefore,this paper applied laser-induced breakdown spectroscopy(LIBS)to analyze the characteristics of biomass fuel.Since the pelletization parameters of biomass pellet fuel will vary with the manufacturer,it may affect the performance of the obtained laser-induced plasma spectral characteristics and the LIBS measurement model of biomass fuel properties.Therefore,the effect of four main pelletization parameters(pelletizing pressure,pelletizing temperature,moisture content and particle size)on LIBS spectra of biomass pellet fuel were studied before LIBS technology is applied to determine the properties of biomass pellet fuel.The results show that in the range of commonly used pelletization parameters,the influence of pelletizing pressure and moisture content on the LIBS spectra can be ignored,while the pelletizing temperature and particle size had a little fluctuation on the LIBS spectra measured by experiment,but the influence is not large,and the fluctuation can be reduced by data processing method,and thus achieving a relatively stable condition.The properties of biomass fuels made from different types of raw materials vary greatly.Therefore,the rapid identification of the types of biomass fuels before entering the furnace can provide guidance for the further precise combustion of biomass boilers.Firstly,the qualitative analysis of LIBS spectra was carried out by Principal Component Analysis,(PCA).Then,four commonly used classification algorithms(Support Vector Machine,K-Nearest Neighbor,Random Forest,Naive Bayes)were selected to construct the classification model ofbiomass fuels respectively in combination with LIBS.By comparing the results of test set samples of these four classification models,it was found that support vector machine has the best classification performance on biomass fuels,with the highest recognition rate of 99.83%,thus the feasibility of using LIBS to quickly identify the types of biomass fuels were demonstrated.Finally,Partial Least squares(PLS)combined with LIBS was used to quantitatively analyze the three main properties of solid woody biomass pellet fuel without sample pretreatment,namely gross calorific value,volatile matter and ash content.The quantitative analysis model with the best predictive performance was selected by comparing the models obtained by different input variables and different data preprocessing methods,and the feasibility of using LIBS and PLS to quickly analyze the properties of biomass fuel was verified.