摘要
为了评价不同PM_(2.5)质量浓度的雾霾对人群健康的危害性,采用函数Petri网与人工神经网络的学习机制相结合的方法,构建了结构解析型神经Petri网模型,简称SP_NPN模型。该模型可以继承ANN的自学习功能,具有ANN的联想存储功能,并能够快速寻找最优解;该模型的网络结构可以是任意的,从而具有良好的适应性;该模型对输入信息没有依赖,它将网络内部的动态作为关注焦点,因而特别适用于复杂系统结构的解析与因果关联分析;该模型的内部节点具有明确的物理含义,因而方便对复杂系统建模。应用结果表明:利用SP_NPN模型解析雾霾对人体健康的危害性,一组观测输入和输出会获得多种解析结果,据此可以发现人体系统内部存在的所有致病类型,从而为相关精确诊断方案的制定指明方向;通过固定某些弧的权系数,可以使解析结果达到唯一,从而发现导致解析结果唯一时的关键因果关联关系,依据这些关联关系设置观测方案,能够为精确诊断方案的制定提供依据;此外,利用SP_NPN模型的因果关联关系解析功能,能够揭示雾霾中PM_(2.5)质量浓度变化与人体致病机制之间存在的因果关联关系,并计算出不同PM_(2.5)质量浓度的雾霾对人群健康的危害性程度,从而达到对雾霾危害性进行评价的目的。
In order to evaluate the detrimental effect of the foggy haze in different PM（2.5）concentrations on the human health,the paper tries to establish a structure-parsed neural Petri-net model as a function Petri-net in combination with the learning mechanism of artificial neural network（ ANN） system. The said neural Petri-net model is known as the inheritable SP NPN. SP NPN self-learning of ANN functional model,which can associate the storage capability with the quick-finding optimal solutions.Though the network structure of SPNPN can be arbitrarily chosen and adopted,it enjoys a fine adaptability,for it does not depend on the input information,rather,it takes the system interior dynamics as its focus to make it particularly suitable for the complicated system structure and cause-result relation analysis. In addition,the internal nodes of SPNPN model has its clear physical implications for the convenience in the complex system modeling.The results of the primary application show that,when it is used for the structural analysis of the effect of the haze detriment on the human health,it would be possible to gain a variety of structure parsing results via a set of input and output data. Therefore,it can help to find all kinds of human disease in the internal structure and provide a scientific guide to the accurate diagnosis prescriptions. By fixing some weight values of some arcs,it can be made possible to obtain the uniqueness through analytical results,which can lead to finding the key cause-result relations and make the system of the analytical structure unique. In addition,such key causative-consequence relations can also be used to construct the observation schemes to provide a scientific basis for the relevant diagnosis schemes. Moreover,by use of the causal identification function of SPNPN,it can also help to identify and determine the scientific cause-effect relation between the concentration variations of PM（2.5）and the pathogenic mechanisms of human bodies. It can also be used to calculate the detr
出处
《安全与环境学报》
CAS
CSCD
北大核心
2017年第4期1554-1562,共9页
Journal of Safety and Environment
基金
教育部人文社会科学研究规划基金项目(15YJA910002)
陕西省社会科学基金项目(2014P07)
