论文部分内容阅读
用误差反向传播(BP)的人工神经网络模型及分子结构描述码作为输入特征参数预测非芳香族多硝基化合物的生成焓,研究了网络参数及分子结构描述码的影响,同时按分子结构描述码进行了多元线性回归,取得了满意的结果,其回归方程相关系数达到了0.9977,精度高于文献值。绝大多数相对误差在10%以内。
The BP neural network model and molecular structure descriptor were used as input parameters to predict the enthalpy of formation of non-aromatic polynitro compounds. The effects of network parameters and molecular structure descriptors were studied. At the same time, The descriptors were multivariate linear regression with satisfactory results. The correlation coefficient of the regression equation reached 0.9977, which was higher than the literature value. The vast majority of relative errors within 10%.