【摘要】 水质预测物理模型在水环境保护中起着十分重要的作用,然而由于模型的参数识别问题,使其应用受到很大局限.对人工神经网络(A rtific ia l N eura l N etw ork,简称ANN)水质预测建模作了初步研究.用试错法,以训练时间和测试误差两项指标为依据,对比分析三层、四层网络结构,认为ANN模型适应于水质预测建模,并提出了适合的模型结构、学习速率、传递函数.更多还原

【Abstract】 Physical models for water quality prediction play a very important role in water environment protection,but their application has been greatly limited due to the problems in identification of model parameters.An approach to the modeling of water quality using the artificial neural network was studied.The error and trial method was employed to analyze the artificial network model for water quality prediction,and a 3-layer network structure and a 4-layer network structure were compared based on training time and error.The results indicated that the artificial neural network is applicable to modeling of water quality prediction.The suitable model structure,learning rate and transfer function are also presented.更多还原