QSAR结合人工神经网络预测磺酰脲类除草活性

摘要

采用误差反传前向人工神经网络模型研究了18种磺酰脲类化合物的结构与除草活性之间的关系.以18种磺酰脲类化合物的量子化学参数作为输入,除草活性作为输出,构建网络模型,取得了较好的预测结果.网络的自相容能力和交叉检验结果良好.该方法还可作为QSAR研究及对有机化合物其他性质进行预测的一种有效手段.%The relationship between the structures of 18 sulfonylurea compounds and their herbicidal activi- ties was studied by using the neural network based on the back propagation algorithm. The prediction results were examined by a self-consistency test and a cross-validation test. The self-consistency test and the cross-validation test obtained good results when using the quantum chemical parameters about structure as the inputs of the neural network and their herbicidal activities as the outputs. This method may be helpful for QSAR＇ s study and may he effective in predicting other important properties of organics.