基于人工神经网络方法的河北省近海沉积物重金属污染综合评价

Evaluation of the offshore sediment heavy metal pollution in Hebei province based on the artificial neural network method

利用误差反相传播神经(BP)网络对河北省近海沉积物中的铅、镉、锌、汞、砷5种重金属元素的污染水平进行分析,利用自组织特征映射(SOFM)网络对上述重金属元素分布特征进行分类,通过分类与污染水平量化值的结合,进行综合评价。SOFM把52个沉积物样品分别划分为3、4、6类和9类。对比各种分类,分为3类的物理意义较明确,每个类别分别对应高中低不同的污染物浓度水平,差异显著、分类方式比较合理。通过此种分类可以判断河北省近海的沉积物重金属污染在不同海域存在一定的差别,整体上是离海岸越远,沉积物的重金属污染水平越高,距海岸较近的海域内,沉积物的重金属污染水平较低,但渤海湾内的重金属污染水平高于其他海域。

By means of two artificial neural networks, SOFM network and BP network, this paper makes a comprehensive assess- ment to five kinds of heavy metal pollution （lead, cadmium, zinc, mercury and arsenic） existing in offshore sediments in Hebei prov- ince. 52 sediment samples are used for evaluation. We set up SOFM network to classify and BP network to grade, respectively train- ing 1 000 times. Thus, pollutions are classified into 3, 4, 6 and 9 classifications. Comparing the classifications of SOFM network and grades of BP network, there are good correlations. Especially, the results of 3 classifications are corresponding to high, middle and low levels of comprehensive pollution, which should be more reasonable. The result also shows the distribution characteristic of heavy metal pollution in offshore sediments： The farther offcoastal line, the more serious the pollution is. Among different sea areas, the Bohai Bay has higher heavy metal pollution than others.