DBD演算法BP神经网络水质综合模拟研究

建立水质模型对水体水质的变化进行模拟是一条有效的途径。由于污染物之间的相互作用以及水环境中物理、化学、生物、水力特征的高度非线性耦合，对其了解是有限的，致使传统的微分方程建模方法和数值求解在应用中遇到许多困难。本文采用BP人工神经网络进行河流综合水质模型的建模，为了加强模型的收敛速度采用学习速率自适应的DBD演算法，通过应用长江干流重庆段实测水质样本对模型进行的训练与检验表明：该模型用于一维水质综合模拟可行，且精度较高，为河流水质模拟提供了一条简便可行的新方法。

General water quality simulation model based on delta-bar-delta algorithm of artificial neural network

Based on the theory of delta-bar-delta algorithm(DBD)of Artificial Neural Networks and concept of general water quality simulation,a model of general water quality simulation of one -dimensional was put forward.After learning of this model with the data of water quality of Yangtze River mainstream in Chongqing region,the adaptability of the model to simulate water quality was examined.The result shows that it is feasible for general water quality simulation and is more accurate than traditional model of water quality;it is better for speed of learn is DBD and simulate of congener water quality.BP neural network inaugurates a new approach for general water quality synthesize simulation.