松花江有毒有机物污染防治定量构效关系研究

Study on quantitative structure-activity relationship of toxic organic compounds pollution control in Songhuajiang River

本文以分子拓扑学研究为基础,开展了有机毒物分子结构与生物活性的定量研究,采用拓扑指数——分子连接性指数计算,实现了软件化。并对人工神经网络的反向传播法作了深入系统的研究,研制了人工神经网络软件。利用此软件对松花江45种(致癌、致畸、致突变)"三致"污染物进行了生物毒性预测与研究,并与多元逐步回归法进行了比较,取得了三项创新性重大科研成果。

Based on the molecular topology research, the paper researches quantitatively the molecular structure and biological activity of the organic toxicant, the quantitative research has realized the software by adopting the calculation of the molecular connectivity index, a kind of topological index. Based on the studies deeply and systematically for the back propagation method of the artificial neural network, the artificial neural network software is developed. The 45 kinds of pollutions with the carcinogenic, teratogenic and mutagenic actions in Songhuajiang river are forecasted and researched for the biological toxicity by using the software. The results have the more good effect comparing with the results of the multiple stepwise regression method and have made three innovativeness great achievements in scientific research.