Full
Outline
Wood density, as a key parameter for evaluating wood quality, is a focal point in both wood trading and processing. The development of efficient, non-destructive, and cross-species detection technologies for wood density is of great importance for the exploitation of forestry resources and the efficient utilization of wood.
This study selected two typical hardwood species from northeast China, specifically Manchurian walnut (Juglans mandshurica) and Manchurian ash (Fraxinus mandshurica), as research subjects to establish a near-infrared (NIR) spectral transfer model for cross-species wood density prediction. A stable predictive feature selection (SPFS) algorithm was designed to score and rank spectral bands. The elbow method was applied to determine the optimal number of bands and to select common feature bands. A stacked autoencoder (SAE) was introduced as a feature transformation mechanism. Combined with an artificial backpropagation (BP) neural network prediction model, a new cross-species spectral model transfer method was constructed.
Following the transfer of the Juglans mandshurica model based on SAE, the coefficient of determination (R2) for the Fraxinus mandshurica prediction set rose from 0.172 7 to 0.827 1, and its root mean square error of prediction (RMSEP) fell from 0.180 1 to 0.049 3 g/cm3. In the opposite transfer direction, applying SAE to the Fraxinus mandshurica model improved the R2 for the Juglans mandshurica prediction set from 0.000 9 to 0.884 9, and reduced the RMSEP from 0.187 8 to 0.032 9 g/cm3. These results indicate that the SAE-based method achieves substantial improvements in both R2 and RMSEP metrics when compared to traditional techniques such as piecewise direct standardization and slope/bias correction algorithms.
The findings of this study demonstrate that the SAE-based NIR transfer model is effective for cross-species applications, allowing for the rapid and efficient prediction of wood density in different tree species. This approach promotes model sharing across species and offers both a theoretical basis and technical support for improving the universality of NIR models.
18
1
| 科 Family | 属数 Number of genus | 种数 Number of species | 占总种数比例 Percentage of total species (%) | 属 Genus | 种数 Number of species | 占总种数比例 Percentage of total species (%) |
|---|---|---|---|---|---|---|
| 鹅膏菌科Amanitaceae | 2 | 11 | 5.26 | 鹅膏菌属 Amanita | 10 | 4.78 |
| 小菇科 Mycenaceae | 2 | 12 | 5.74 | 丝盖伞属 Inocybe | 5 | 2.39 |
| 多孔菌科 Polyporaceae | 8 | 14 | 6.70 | 蜡蘑属 Laccaria | 5 | 2.39 |
| 红菇科 Russulaceae | 3 | 23 | 11.00 | 小皮伞属 Marasmius | 6 | 2.87 |
| 小菇属 Mycena | 11 | 5.26 | ||||
| 光柄菇属 Pluteus | 5 | 2.39 | ||||
| 红菇属 Russula | 17 | 8.13 | ||||
| 栓菌属 Trametes | 5 | 2.39 |
关闭全屏
No. 86 Xueyuan South Road, Haidian District, Beijing
010-62662699(Operation and Technical Support)
qkjq@cast.org.cn
Copyright © 2025 China Association for Science and Technology. All rights reserved. For all open access content, the relevant licensing terms apply.
Sponsored by the Office of the Leading Group for Cybersecurity and Informatization of CAST, and supported by Science and Technology Review Publishing House
