可拓神经网络在水质评价中的应用

Application of extension neural network in water quality assessment

可拓神经网络是可拓学与人工神经网络的有机结合,能够更好地模拟人脑神经系统思维等智能行为。讨论了神经网络物元模型、神经网络的物元可拓性及基本物元变换,并利用可拓学的扩缩变换,通过在输出空间中用一个区域来代替BP神经网络的训练停止区域,极大地提高了神经网络的训练速度。以几个主要指标作为衡量水质优劣的标准并作为神经网络的输入样本,建立可拓神经网络训练水质的模型,并与普通BP神经网络进行训练速度和训练效果比较,实验表明,用可拓神经网络对水质的评价效果更为明显。

The extension neural network is the organic synthesis of extension theory and the artificial neural network,which can better simulate intelligent behaviors such as thoughts of human neural system.We introduce the neural network element model and analyze the neural network element extension,and the basic element transformation.By extension transformation,we use a region to replace a training stopped region in the output space,so the training speed is extraordinarily improved.The extension neural network model is established for water quality assessment by several main water quality indexes which are also taken as the input samples of the model.The comparison of training speed and training effectiveness between established model and the BP network shows that the extension neural network model is more effective for water quality assessment.