论文部分内容阅读
为了能够客观地对海水水质进行综合评价,在分析人工神经网络概念和原理的基础上,从阈值角度出发,通过对各类海水水质污染指标浓度生成样本的方法,生成了适用于BP人工神经网络模型训练的样本,并应用基于误差反向传播原理的前向多层神经网络,建立了用于海水水质评价的BP人工神经网络模型。将该模型用于渤海湾近岸海域水环境评价,通过模型的计算,得到该海域的水质类别。结果表明,2004-2007年,渤海湾近岸海域污染指标总体上在河流丰水期时比枯水期时高,2005年和2006年污染较为严重,2007年有所好转。经训练的评价模型应用于实例的评价结果表明,该模型设计合理、泛化能力强,对海水水质评价具有较好的客观性、通用性和实用性。
In order to objectively evaluate the quality of seawater, based on the analysis of the concept and principle of artificial neural network, from the threshold point of view, through the method of generating samples for all kinds of seawater pollution indicators concentration, a model suitable for BP artificial neural network Model training samples, and the application of forward multilayer neural network based on error back propagation theory, a BP artificial neural network model for seawater quality evaluation is established. The model was applied to the assessment of water environment in the coastal area of Bohai Bay. The water quality of the area was obtained through the calculation of the model. The results show that in 2004-2007, the pollution index of coastal areas in the Bohai Bay is generally higher than that in the dry season in the river wet season, with serious pollution in 2005 and 2006 and a slight improvement in 2007. The evaluation results of the trained evaluation model applied to the examples show that the model has reasonable design and generalization ability, and has good objectivity, versatility and practicability for the evaluation of seawater quality.