Study On Rapid Moisture Measurement Of Biomass Based On NIRS

With the concept of "clear water and green mountains" put forward in recent years,and the basic national policy of resource conservation and environmental protection,energy conservation,environmental protection and sustainable development have become the themes of The Times.China’s energy structure is based on coal combustion of thermal power,but coal resources are not renewable,the current widely concerned renewable energy mainly includes wind energy,water,solar energy and biomass energy.Compared with other renewable energy,biomass energy has huge reserves in China,but its utilization rate is still very low,so there is still a lot of room for development and improvement.Biomass fuel needs to be processed before entering the boiler for combustion,because in the biomass boiler,the moisture content of the biomass fuel can not be too high.In addition,too high moisture content in the fuel will also reduce the available calorific value,affect the conversion rate of biomass fuel,and further affect the utilization rate.And collected from around the biomass moisture content is relatively high,can not be directly into the boiler for combustion.Therefore,the real-time,efficient and rapid detection of biomass moisture is of great significance to effectively ensure the safe and efficient operation of biomass power plants.In this paper,near infrared spectroscopy technology is adopted to measure the real moisture of biomass fuel by drying method through experiments,and infrared spectral data are collected.And collect the infrared spectrum data,through the near infrared moisture spectrum data acquisition,establish the relevant model and discuss the relevant principles of subsequent data processing,on this basis,a rapid moisture measurement model was established for near infrared spectroscopy.Different pretreatment schemes were used to eliminate noise and select characteristic wavelengths.Both linear and nonlinear methods were used to establish the moisture model.And through the corresponding error calculation and analysis,the aim is to improve the efficiency of the measurement of biomass moisture content.In the selection of noise removal and characteristic wavelength in data preprocessing,principal component analysis,Maharanobis distance and Laida’s rule are mainly used to remove abnormal data from the NIR original spectrum.Linear regression model was used to cross-validate biomass moisture to verify the elimination effect.Partial least squares method and correlation coefficient method are selected for the processing of characteristic wavelengths.These two methods have good processing results for the characteristic wavelengths of the prior data.The correction model is used to verify the result of characteristic wavelength selection.Finally,linear and nonlinear models of biomass moisture were established.Through the subject of biomass spectrum and water data processing and data modeling,using Laida’s law to remove abnormal data and noise wavelength.Data preprocessing was normalized by the standard normalization method,and the first derivative was used to highlight the peak and trough characteristics of the NIR spectra of biomass.The combination of multiple scattering correction and second derivative is the best method to select the characteristic wavelength.Using multiple linear regression to establish mathematical model is the best.