基于灰色神经网络建模的水质参数预测

针对水质参数预测过程中样本数据少的特点,结合灰色新陈代谢GM(1,1)模型和BP神经网络模型,提出灰色新陈代谢BP神经网络组合模型。用灰色新陈代谢模型群的数据集作为BP神经网络的学习测试样本,解决了BP网络需要大量样本才能较好地逼近非线性函数的问题。实验表明,与普通BP网络、灰色新陈代谢模型比较,灰色新陈代谢BP神经网络组合模型的预测精度更高,能够应用于水质参数的预测。

Water quality parameter forecast based on grey neural network modeling

Owing to the lack of enough monitoring data in water quality parameter prediction, combining information renewal GM(1,1) model and BP neural network model, the grey information renewal neural network model was proposed. Using the data of information renewal model sets as the training and testing samples, the problem that BP network needs a mass of samples to approach nonlinear function better was solved. Experimental results indicate that, compared to the grey information renewal model and BP model, this forecast model is superior in water quality parameter prediction.