基于概率神经网络的城市湖泊生态系统健康评价研究

Investigation of health assessment for urban lakes system based on probabilistic neural networks(PNN)

概率神经网络(PNN)是一种结构简单、训练简捷、应用十分广泛的人工神经网络,并且在水质分类等环境领域已取得一定研究成果.本文选取广州市最大的人工湖——白云湖作为研究对象,结合其水质监测数据及生物监测数据,建立概率神经网络模型对其进行湖泊生态系统健康评价,得到不同监测时间点的湖泊生态系统健康评价结果.分析表明:①白云湖生态系统比较脆弱,目前净化水质的效果有限;②各监测点的评价结果均呈季节性变化,丰水期湖泊生态系统健康状态好于枯水期,年际变化不显著.实验结果表明,利用概率神经网络对湖泊生态系统健康状态进行评价是可行的,与传统评价方法相比,其具有训练时间短、权重确定客观、输出结果稳定等优势,可以运用到更多相关领域.

As one kind of artificial neural networks, probabilistic neural networks （PNN） is simple in structure, easy for training and widely used. Some research results have been obtained in environmental area, for example the classification of water quality. The target of this study was Baiyun Lake, the biggest artificial lake of Guangzhou city. Based on the monitoring data of water quality and biology, PNN model was constructed and applied to assess the ecosystem of Baiyun Lake at different periods. The main assessment results are listed as follows： 1 the ecological system of Baiyun Lake was relatively weak, which was unable to function in purifying water. 2 The seasonal variation of health assessment results at different monitoring points was significant, while the inter-annual variation was insignificant.In summary, it is feasible to assess the health of the lake ecosystem by probabilistic neural network. Compared with traditional evaluation methods, e.g. BP neural networks and attribute recognition method, the PNN model is more objective and stable in evaluating the health of lake ecosystem, thus can be extended to other related fields.