SPXY样本划分法及蒙特卡罗交叉验证结合近红外光谱用于橘叶中橙皮苷的含量测定

在近红外光谱PLS定量模型的建立过程中训练集样本的选取和潜变量数的确定是十分重要的。因此,该研究以橘叶中橙皮苷的含量检测为例,分别比较了random sampling(RS),Kennard-Stone(KS),duplex,sample set partitioning based on joint x-y distance(SPXY)四种训练集样本的选取方法对模型的影响,以及留一交互验证法和蒙特卡罗法对潜变量数确定的影响。结果表明,SPXY法选取的训练集建立的模型优于其他三种方法,蒙特卡罗法能够较好地确定模型的潜变量数并有效地减少过拟合风险,所建模型的交互验证均方根,预测均方根及预测集相关系数分别为0.7681,0.7369,0.9752。

A new method to calculate lateral force acting on stabilizing piles based on multi-wedge translation mechanism

A new method based on the multi-wedge translation mechanism is presented to calculate the lateral force acting on the stabilizing piles. At first, there is no assumption for the shape of potential sliding surface, it is just considered that the potential sliding surface is a composite of a number of straight lines. And then, the potential sliding mass is divided into a number of triangular wedges take with these straight lines as its base. The kinematic theorem of limit analysis is adopted to calculate the rate of external work and the rate of energy dissipation for each triangular wedge, respectively. Furthermore, the multivariate functions are established to calculate the lateral force acting on the stabilizing piles. The lateral force and the corresponding potential sliding surfaces can be obtained by an optimizational technique. At last, an example is taken to illustrate the method. The effect of soil strength parameters, slope angle and pile roughness on the lateral force and the corresponding potential sliding surface are analyzed.The result are compared with those obtained using other methods.

Limiting stand density and basal area projection models for even-aged Tecomella undulata plantations in a hot arid region of India

This paper presents equations for estimating limiting stand density for Z undulata plantations grown in hot desert areas of Raj asthan State in India. Five different stand level basal area projection models, belonging to the path invariant algebraic difference form of a non-linear growth function, were also tested and compared. These models can be used to predict future basal area as a function of stand variables like dominant height and stem number per hectare and are necessary for reviewing different silvicultural treatment options. Data from 22 sample plots were used for modelling. An all possible growth intervals data structure was used. Both, qualitative and quantitative criteria were used to compare alternative models. The Akaike＇s information criteria differ- ence statistic was used to analyze the predictive ability of the models. Results show that the model proposed by Hui and Gadow performed best and hence this model is recommended for use in predicting basal area development in 12 undulata plantations in the study area. The data used were not from thinned stands, and hence the models may be less accurate when used for predictions when natural mortality is very significant.