摘要
在湖泊富营养化已成为世界性的水污染治理难题的今天,富营养化预测模型应用广泛,已取得较大发展。文章介绍了运用BP人工神经网络预测水体富营养化的计算过程,综合论述了学者们在预测水体富营养化时水体中BP人工神经网络模型联合各种算法的优化情况,由此可以看出,足够多的样本是BP神经网络进行学习训练的关键;各种联合模型比普通BP人工神经网络模型更加准确、有效;多种联合模型并未运用于水体营养化评价方面;联合模型优化的BP人工神经网络必将具有巨大的价值和发展前景。
With lake eutrophication growing as a worldwide water pollution, eutrophication prediction model has found wide application and achieved great progress today. This paper introduces the application process by using Back Propagation (BP) neural network on water eutrophication prediction, and presents a comprehensive review of various improvements made by scholars in their efforts to eliminate the defects with Back Propagation (BP) neural network. It concludes : 1 ) plenty of samples serve as the key to the training of BP neural network. 2) combined models are more accuracy and effective than BP neural network. 3 ) a varie- ty of combined model is not applied in lake eutrophication predicting. 4) there is a valuable prospect for the improved model of combined BP neural network.
出处
《常州工学院学报》
2013年第3期70-77,共8页
Journal of Changzhou Institute of Technology
基金
北京市教育科学"十二五"规划青年专项课题(CGA12100)
北京市高等教育学会"十二五"高等教育科学研究规划课题(BG125YB012)
北京对外文化交流与世界文化研究基地2013-2014年度青年研究项目(BWSK201304)
北京市属高等学校人才强教深化计划中青年骨干人才资助项目( PHR201108319)
关键词
富营养化水体
BP人工神经网络
预测
water eutrophication
Back Propagation (BP) neural network
prediction model