神经网络模型在环境现状评价中应用研究

人工神经网络技术具有很强的非线性映射能力和并行性、自适应、容错性及自学能力,已广泛应用于包括环境在内的多学科领域。文章将人工神经网络技术应用于环境影响评价中,通过案例研究用人工神经网络解决环境评价中的问题。以山西吕梁地区环境影响评价为案例,选择人工神经网络中的BP网络,径向基网络和自组织竞争网络等三种网络模型对其进行环境影响现状评价,并对评价结果作对比分析,通过网络设计、网络训练和模拟,结果说明BP神经网络模拟结果比其它两种方法更贴近环境质量现状。同时对不同参数选取得到的结果进行分析,并经过网络参数的不断调整提高评价结果的精度,总结出各种神经网络模型在环境评价应用过程中参数的选取方法。尝试用神经网络解决环境评价中的问题,使环境工作中的方法技术更科学。

Application of Neural Network Model to Environmental Assessment

Taking environmental impact assessment of Lvliang,Shanxi Province as a case,3 artificial neural networks including BP network,RBF network and self-organizing competitive network as the model were selected to conduct the atmospheric environmental impact assessment,and through the comparison and analysis,results showed that the simulation results of BP network was closer to the status quo of environmental quality than the other two methods.Different parameters were selected to analyze results and through the continuous adjustment of network parameters to improve the accuracy of evaluation results,the method of parameters selection of various neural networks were reviewed,which would apply to the environmental assessment process.