讨论了BP神经网络模型的特性和建模条件,并给出建立合理BP神经网络模型的基本原则和步骤。针对实际水质评价问题,建立了渭河地面水环境质量综合评价的BP神经网络模型,并与单因子法、主成分分析法进行了分析比较。实验结果表明,BP神经网络可以较好地实现水质综合评价,且具有较高的实用性和客观性。
The shortcomings and the conditions of BP neural networks modeling are discussed, and the basic principles and steps of establishing a reasonable BP neural network model are given. Aimed at actual question of water quality assessment, the BP neural network model is established which synthetically assesses the surface water quality of the Wei river. Then it is compared with water quality assessment methods of single factor assessment and principal components analysis (PCA). The results show that BP neural networks is used to achieve comprehensive evaluation of water quality, and it is more practical and objective.