Variation Law of Anatomical Shape Characteristic during Wood Across-compression Based on Neural Network Recognition

摘要

Feedforward BP algorithm is used to distinguish and analyze cell types and analyzed the variation law of anatomical shape characteristic during wood across-compression. Experimental analysis shows the neural network had strong robustness and good capacity of cell recognition. The researches of cell microcosmic variations of three typical coniferous woods, which are P.koraiensis, A.nephrolepis and Lgmelini, resulted in mathematical models reflecting the variation law of anatomical shape characteristic in the course of across-compression and realized the quantitative description to wood stress-strain about wood across-compression.