摘要
采用量子化学MOPAC AM1法计算了 4 2种取代芳烃的生成热Hf、分子最高占有轨道能EHOMO、分子量MW、分子总表面积TSA及偶极矩 μ .分别采用线性回归分析法和人工神经网络法对所研究化合物的生物降解性参数BOD进行QSBR研究 .对训练组而言 ,线性方法和神经网络法的平均预测误差分别为 1 5 9%和 1 1 4 % ;而测试组化合物的平均百分误差分别为 1 4 5 %和 1 3 0 % .无论对于测试组还是训练组 ,神经网络法的预测都更精确 .
The heat of formation(H f),energy of the highest occupied molecular orbital(E HOMO),molecular weight(MW),total surface area(TSA),and dipole moment(μ)of 42 organic chemicals were calculated by quantum chemical method MOPAC6.0-AM1.The quantitative structure-biodegradability relationship study was performed with biodegradability parameter BOD by the linear regression analysis method and artificial neural network approach,respectively.For the training set,the percentage errors between the experimental and predicted values of BOD are 15.9% by the linear regression analysis method and 11.4% by the neural network approach;while for the testing set,they are 14.5% and 13.0%,respectively.It has been shown that the neural network method is able to provide a superior fit whether for the training set or for the testing set and produce a lower prediction error than the linear regression method.
出处
《环境化学》
CAS
CSCD
北大核心
2002年第1期62-67,共6页
Environmental Chemistry
基金
国家自然科学基金资助项目 :No 2 98770 0 4 .
