Identification Of Cunninghamia Lanceolata And Cryptomeria Fortunei Based On Terahertz Time-Domain Spectroscopy

Wood is a kind of natural polymer composite material,which plays an important role in the national economy.Cunninghamia lanceolata and Cryptomeria fortunei are important timber species in China.The economic value and use value of Cunninghamia lanceolata are higher than that of Cryptomeria fortunei,but the shape of Cunninghamia lanceolata and Cryptomeria fortunei is similar,so it is difficult to identify them,which often causes huge economic losses.The traditional wood identification methods have the disadvantages of slow speed,damaged wood,and complicated procedures,and cannot realize fast and non-destructive wood identification.The emergence of terahertz time-domain spectroscopy provides a new way to solve this problem.THz radiation has the characteristics of fingerprint spectrum and high penetrability,and most of the rotation and vibration frequency of wood biomolecules are concentrated in THz frequency band.Using terahertz time-domain spectroscopy technology can achieve fast and non-destructive identification of wood.This article focuses on Cunninghamia lanceolata and Cryptomeria fortunei,and terahertz time-domain spectroscopy combined with radial variation of wood(heartwood,sapwood)and different sections(cross section,radial section and tangential section)are used to carry out wood identification research.The effect of radial variation and different cuts on wood identification is analyzed.The identification studies of wood are carried out based on different data frequency bands,spectroscopy preprocessing methods,modeling methods and optimization algorithms.The main research contents and conclusions are as follows:(1)The collection and characteristic analysis of wood terahertz spectroscopy.The terahertz time-domain spectroscopy of samples were collected using the terahertz time domain spectroscopysystem,and the THz frequency domain spectroscopy was calculated by Fourier transform.The refractive index and absorption coefficient are extracted according to the parameter model,and the terahertz spectroscopy of the sample is compared and analyzed.The results show that the terahertz spectroscopy of Cunninghamia lanceolata and Cryptomeria fortunei differs depending on the location of the heartwood and sapwood,and differs depending on the section.The results show that because the samples belong to the heartwood,sapwood parts and the different sections,the amplitude and absorption peak of the terahertz spectroscopy are different.(2)Modeling of wood identification model.Spectroscopy data of different terahertz frequency bands are selected and preprocessed by different spectroscopy preprocessing methods.These data are used to establish a wood identification model of Cunninghamia lanceolata and Cryptomeria fortunei based on BP neural network,learning vector quantization neural network and extreme learning machine method,and then the genetic algorithm is used to optimize the model.The results show that the BP neural network identification model based on THz spectroscopy data of 0.2?1.5THz band processed by S-G convolution is the best.When the model is optimized by genetic algorithm,the accuracy is improved obviously,which can realize the accurate identification of Cunninghamia lanceolata and Cryptomeria fortunei.(3)Research on the influencing factors of wood identification.The wood identification model was established based on the spectroscopy data of heartwood,sapwood and cross section,radial section and tangential section.The identification models of the same and different types of prediction samples and modeling samples were studied.The influence of the wood radial variation(heartwood and sapwood)and section variation on wood identification were analyzed.The results show that the radial variation factor has little effect on wood identification,and the section factor has greater effect on wood identification.