機械學習法應用於牛樟漂流木與非漂流木性質之研究

本研究目的為針對牛樟非漂流木（Non-driftwood）及漂流木（Driftwood）區分心材與邊材二類，探討兩者來源木材之機械性質。實驗試體由行政院農業委員會林務局屏東林區管理處所提供之牛樟木樣本，以漂流木與非漂流木進行比較，其中包括密度、抗壓、抗彎及抗剪測試。再利用微軟AZURE雲端運算服務進行機械學習分類兩類試材為漂流木或非漂流木，所學習之成果為抗剪決策樹0.980>抗壓決策樹0.941>抗彎類神經0.811，由此可得知，抗剪決策樹較其他兩項所學習之成效佳。藉由所衍生之判定系統能以較客觀方式判別受測試材是否屬於漂流木。

關鍵字

木材機械性質；機械學習；漂流木；邊心材

並列摘要

The aim of this study was to distinguish between heartwood and sapwood for non-driftwood and driftwood of Cinnamomum kanehirai Hayata, and to explore the mechanical properties of the woods. The experimental specimens were provided by the Pingtung Forest District Office of the forestry bureau of the executive yuan agricultural committee. The wood samples of the main section of the burdock logs were compared with the appearance and physical properties of the driftwood and the non-driftwood timber. Using the statistical calculation and AZURE mechanical learning system, in order to determine whether the test material belongs to the driftwood or non-driftwood. The results of mechanical learning 0.980 > compression 0.941 > bending resistance 0.811, which were shown that the shear decision tree was better than the others. The results of the project had been training well. Using mechanical learning, the training results of determination system could classify the driftwood or non-driftwood timber more objective.

並列關鍵字

Wood Mechanical Properties ； Machine Learning ； Driftwood ； Sapwood and Heartwood

主題瀏覽