| China’s economy is currently in a stage of rapid development,and in recent years,China’s energy demand is also constantly improving. By 2015,China also called for the implementation and emphatically put forward to improve the structure of energy utilization system.and how to make rational and effective use has been found and is developing new energy sources,make full use of renewable energy guidelines replace the original fossil energy,biomass in recent years,as the newly discovered energy and has full use of renewable energy an important part in the future could become the most commercially valuable raw material. How to combine our current situation,development and rational utilization of biomass energy to ease the energy shortage.By means of chemical,physical and biochemical,biomass could be converted into the bio fuels that can be used,and the economic benefits generated by the transformation process which will be directly affected by the conversion output rate.However it is restricted by limitations on the inspection process cumbersome and traditional chemical technology,could not meet the experiment or production personnel on the biomass composition ratio of rapid detection,so a lot of the production process lack clear ratio of guidance,the output rate is not very clear.Therefore, this experiment will take this as a start point.the soybean straws after removing the soybean as experimental samples,according to the content of soybean straw cellulose and hemicellulose content in the map are fitted by using near infrared spectroscopy,to explore the possibility of rapid detection,the next main research such as:(1) In all regions of Heilongjiang province,a collection of different varieties of soybean straw 173,and the cellulose and hemicellulose content it contains and the original calibration spectra were collected,the samples collected by normal use histogram statistical analysis to ensure that the selected sample has a certain representation.(2) The derivative and smoothing processing method were used, the original spectra did simple denoising processing,the use of the 3D view based on X-Y residual and leverage value analys is method to eliminate sample abnormal,after analysising, correction accuracy of the model also had the greatly improved.(3) For the cellulose and hemicellulose of denoised spectra of the characteristic bands of the optimal screening and interval selection of 35 and 50 Interval Partial Least Squares(IPLS), number of evaluations for 50 and 100 times of the genetic algorithm(GA),21 to 51 window size Moving Window Partial Least Squares method(mwpls) and random noise variables of 100 to 1000 Uninformative variables elimination(uve) were selected.And the parameters of the band feature selection method were used to make a variety of experiments,and then the calibration model of each characteristic wave band was established. Finally,it was concluded that the validation results of each method are quite different,but the results of the calibration model with respect to the whole spectrum were greatly improved.(4) Established the prediction model of hemicellulose and cellulose,established PLS regression model,and used BP neural network to complete the BP validation model, and comparing the prediction results.And cellulose and hemicellulose in two kinds of forecasting models were also see the pros and cons of in hemicellulose prediction,we could see the PLSR prediction ability far stronger than the BP neural network.However in the prediction of cellulose was diametrically opposite results.In summary,in this paper,the near infrared model of soybean straw cellulose and hemicellulose of rapid detection showed certain feasibility,but also to solve some essential problems encountered in the past to examine method,in order to use the increasing development of the technology and improve the rapid detection for future biofuels production process to provide a new way. |
PDF Full Text Request