Wood Classification Recognition And Water Content Prediction Based On THz-TDS

Wood is an important living resource.The classification and identification of wood and the detection of moisture content are prerequisites for the rational utilization of wood.Research on advanced wood identification and physical property detection methods has important scientific significance and application value,and is also an important part of wood science development.THz wave has strong spectral resolution and security,and has broad application prospects in material detection.Studying the feasibility of THz technology in the field of wood detection can effectively complement and develop the traditional wood detection methods.Based on the background of wood classification recognition and moisture content detection,this paper studies the methods of THz spectral feature extraction,classification recognition and regression prediction modeling of wood.The main work includes the following aspects:1)A wood classification and recognition model based on terahertz time-domain spectroscopy(THz-TDS)was established.Common wood(Pinus densiflora?Cunninghamia?Pinus sylvestris?Douglas fir)with five different structural differences and five rare woods with similar appearance and structure(Dalbergia bariensis?Dalbergia oliveri?Dalbergia cochinchinensis?Bois de rose?Pterocarpus santalinus)extracting THz optical parameters,combined with principal component analysis(PCA)and continuous projection algorithm(SPA),feature extraction and selection of THz absorption coefficient spectrum and refractive index spectrum,using support vector machine(SVM),random forest and other modes The recognition algorithm establishes a wood classification recognition model,and compares the classification and recognition effects of the model.The results show that the PCA-SVM model based on the absorption coefficient spectrum has better classification and recognition effect on the four common woods,and the SPA-RF model based on the refractive index spectrum has better classification and recognition effect on the five kinds of rare woods,indicating THz.-TDS technology has great potential in wood classification identification.2)A wood moisture prediction model based on terahertz time-domain spectroscopy(THz-TDS)was established.The effects of different water content on the THz spectrum of the same wood(Douglas fir)were studied.The relationship between wood moisture content and wood THz spectral characterization was analyzed.The PLS prediction model of wood moisture content was established by combining SPA extraction characteristic bands.The correlation indicators such as correlation coefficient and mean square error are used to compare and analyze the prediction effect of the model.The results show that the moisture content of wood below the fiber saturation point is positively correlated with the THz absorption coefficient spectrum of wood.The fitting effect and prediction result of SPA-PLS model established by absorption coefficient spectrum and second derivative spectrum in training set and test set are obtained.Both are more accurate and can predict the moisture content of wood.3)The timber detection software of THZ-TDS was developed by using MATLAB.By analyzing the THz wood detection method,the design and use flow of the software was determined.The software set wood THz technology classification identification and moisture content detection as one,built-in model based on pattern recognition and regression prediction algorithm,integrated data reading,image display,model selection and other functions,to achieve THz for different situations of wood Spectral analysis and detection of wood type and moisture content are characterized by convenient operation and rich functions.The software can be applied to the analysis of frequency domain in the THz spectrum and the study of THz optical parameters.