摘要
用正则化方法结合最小角回归算法(Lars)对影响多硝基芳香族化合物(PNACs)密度的主因子进行研究,通过以多硝基芳香族化合物中的分子结构描述码为参数,构造L1正则化模型,考虑不同的最大迭代次数,选取影响分子描述码,依据相关的平均影响程度给出相应主要影响分子描述码,预测密度值与参考值相对误差在7.7%以内。
The main factors affecting the densities of polynitroaromatic compounds（PNACs） are studied by using regularization method which is solved by least angle regression algorithm. The molecular structure describers（MSDs） are used as the input variable for constructing L1 regularization model and different maximum iteration times are considered for determining the number of MSDs. The most MSDs affecting the densities of PNACs are chosen and the corresponding MSDs are given according to their relative average degree of influencing. The relative error between the predicted values and literature ones of the densities of PNACs is within 7.7%.
出处
《价值工程》
2017年第28期213-215,共3页
Value Engineering
基金
西京学院科研项目(XJ160141)
