期刊文献+

基于光谱预处理的低温水曲柳原木含水率检测 被引量:12

Moisture content detection of Fraxinus mandshurica logs at low temperatures based on different spectrum pretreatments
下载PDF
导出
摘要 【目的】原木含水率检测关系木材的质量与高效利用,低温对含水率的无损检测带来困难。应用近红外光谱技术实现低温状态原木含水率的无损检测,可提高预测精度。【方法】在冬季-20℃室外温度时,从林场获取了不同含水率的水曲柳原木木块220块,分别采取了常温状态(20℃)与低温状态(-20℃)的近红外光谱。利用主成分分析比较了相同含水率情况下两种状态的差别;使用基线校正、散射校正、平滑处理和尺度缩放四大类算法的9种光谱预处理算法分别对光谱进行单一方法优化和组合优化。【结果】低温与常温状态的平均光谱在多个吸收峰的强度和位置都有不同,前三个主成分得分图上两种状态的样本有明显的分离,低温对水曲柳试样近红外光谱分析的影响不可忽略;单一预处理算法优化下,移动平均平滑法和SG卷积平滑法对光谱均有较好的去噪效果,RMSEP分别为0.456 2和0.445 7。3种尺度缩放算法中中心化与标准化要明显差于归一化的优化效果,RMSEP分别为0.532 9、0.520 2和0.399 4。一阶导数和二阶导数对光谱的基线校正的效果较为明显,但是同时也放大了光谱噪声,RMSEP分别为0.412 6和0.489 5。MSC与SNV处理了大部分光谱散射,其中SNV在单一预处理算法中表现最好,R_(p)为0.804 1,RMSEP为0.384 1。将筛选出来的单一预处理算法进行组合后,组合算法均普遍优于单一算法。SG平滑与一阶导数的组合解决了噪声放大的问题,RMSEP减小到0.233 1。组合预处理算法中,SG平滑、SNV和一阶导数的组合表现最好,验证集相关系数R_(p)为0.912 8,RMSEP降为0.177 4,验证集预测精度提高了69.85%。【结论】近红外光谱法可以实现低温木材含水率的无损检测,通过对不同预处理进行比较筛选组合进行光谱优化可以显著提高低温状态含水率检测模型的精度。 【Objective】The detection of log moisture content is related to the quality and efficient utilization of wood.Low temperature brings difficulties to the non-destructive detection of moisture content.This study aimed to apply near-infrared spectroscopy technology to achieve non-destructive testing of the moisture content of logs at low temperatures,and to improve prediction accuracy.【Method】220 pieces of Fraxinus mandshurica logs with different moisture contents were obtained from the forest farm at-20℃outdoor temperature in winter,and the near-infrared spectra were taken at normal temperatures(20℃)and low temperatures(-20℃).Principal component analysis was used to compare the difference between the two states under the condition of the same water content.9 spectral preprocessing algorithms using four types of algorithms(baseline correction,scattering correction,smoothing,and scaling)were used to optimize the spectrum by a single method and a combination of optimization.【Result】The average spectra of low temperature and normal temperature were different in the intensity and position of multiple absorption peaks.The samples of the two states on the first three principal component score maps had obvious separation,and the result of low temperature was similar to that of Fraxinus mandshurica.The influence of infrared spectrum analysis cannot be ignored.Under the optimization of a single preprocessing algorithm,the moving average smoothing method and the SG convolution smoothing method had good denoising effects on the spectrum,and the RMSEP was 0.4562 and 0.4457,respectively.Among the three scale scaling algorithms,the centralization and standardization were significantly worse than the optimization effect of normalization,and the RMSEP was 0.5329,0.5202,and 0.3994,respectively.The first-order derivative and the second-order derivative had more obvious effects on the baseline correction of the spectrum,but they also amplified the spectral noise.The RMSEP was 0.4126 and 0.4895,respectively.MSC and SNV dealt with most of the spectral scattering,and SNV performed the best in a single preprocessing algorithm,with a R_(p) of 0.8041 and RMSEP of 0.3841.After combining the selected single pre-processing algorithms,the combined algorithms were generally better than the single algorithm.The combination of SG smoothing and the first derivative solved the problem of noise amplification,and the RMSEP was reduced to 0.2331.Among the combined preprocessing algorithms,the combination of SG smoothing,SNV and first derivative performed the best.The verification set correlation coefficient R_(p) was 0.9128,the RMSEP would be 0.1774,and the verification set prediction accuracy was improved by 69.85%.【Conclusion】Near-infrared spectroscopy can realize the non-destructive detection of low-temperature wood moisture content.By comparing and screening different pretreatments,performing spectrum optimization can significantly improve the accuracy of the moisture content detection model under low temperatures.
作者 阚相成 李耀翔 王立海 解光强 孟永斌 李春旭 谢军明 李怡娜 KAN Xiangcheng;LI Yaoxiang;WANG Lihai;XIE Guangqiang;MENG Yongbin;LI Chunxu;XIE Junming;LI Yina(College of Engineering and Technology,Northeast Forestry University,Harbin 150040,Heilongjiang,China)
出处 《中南林业科技大学学报》 CAS CSCD 北大核心 2022年第11期154-163,共10页 Journal of Central South University of Forestry & Technology
基金 国家自然科学基金项目(31570547) 黑龙江省应用技术研究与开发计划项目(GA19C006)。
关键词 近红外光谱 光谱预处理 低温 木材 含水率 NIR spectral pretreatment low temperature wood moisture content
  • 相关文献

参考文献14

二级参考文献104

共引文献282

同被引文献219

引证文献12

二级引证文献12

相关作者

内容加载中请稍等...

相关机构

内容加载中请稍等...

相关主题

内容加载中请稍等...

浏览历史

内容加载中请稍等...
;
使用帮助 返回顶部