关键词: 木材密度, 微钻阻力法, 微损测量, 线性模型, 对数模型

Abstract:

Objective: At present, scholars mainly use linear model to study the relationship between drill resistance and wood density. There are great differences among the models, and the generality of the models is poor. This paper discusses the model form between drill resistance and wood density, and whether different tree species or similar tree species can share a mathematical model, in order to provide a basis for indirect measurement of wood density by micro drill resistance method. Method: 1) The resistance of 8 tree species including softwood and hardwood was measured by Resistograph 650-S. Two thirds of the measured data of each tree species was used as the modeling data set, and the remaining one third of the measured data was used as the test data set. 2)Using the total modeling data set, the linear model, logarithmic model and the mixed model of linear model and logarithmic model between the average drill resistance and wood absolute dry density were established respectively. The model with the highest determination coefficient was selected as the total model between the average drill resistance and wood absolute dry density, and the model form was selected to represent the relationship between drill needle resistance and wood absolute dry density. 3)The 2 class models of softwood and hardwood between drill resistance and absolute dry density were established by using the selected model form. 4)The sub models of each tree species between the average resistance of drill needles and absolute dry density were established respectively using the selected model form. 5)The test data set was used to test the estimation standard error and average estimation accuracy of the total model, class model and sub model. Result: 1) The determination coefficients of linear model, logarithmic model and mixed model of linear model and logarithmic model established using the total modeling data set are 0.942, 0.952 and 0.961 respectively. 2)The determination coefficients of softwood class model and hardwood class model are 0.780 and 0.864 respectively, and the determination coefficients of the sub model of each tree species are 0.397-0.943. 3)The total estimation standard errors of total model, class model and sub model are 32.222, 31.635 and 27.121 kg ·m^-3 respectively, and the average estimation accuracy is 95.554%, 95.636% and 96.292% respectively. 4)There was no significant difference in the estimation standard error and average estimation accuracy among the total model, the class models and the sub models at the significance level of 0.1. Conclusion: 1) The mixed model of linear model and logarithmic model has the highest determination coefficient. Therefore, it is more appropriate to use the mixed model of linear model and logarithmic model to study the relationship between drill resistance and wood absolute dry density. 2)The estimation standard error of sub model is the smallest and the average estimation accuracy is the highest. Therefore, it is appropriate to use the sub model to predict the absolute dry density of wood. 3)When the measurement accuracy is not high, the total model can be used to estimate the absolute dry density of wood to reduce the modeling workload.

Key words: wood density, micro drill resistance method, micro destructive measurement, linear model, logarithmic model