基于RBF神经网络的湿蒸汽干度软测量模型

RBF Neural Network Based Soft Measurement Model for Dryness of Wet Steam

针对汽驱采油中湿蒸汽干度测量精度过低的问题 ,采用RBF神经网络来建立湿蒸汽干度软测量模型 ,采用最小正交二乘法确定网络隐层节点数以及训练网络输出数值 ,并在实际运行中采用在线校正环节。所建立模型具有良好的逼近精度 。

In order to resolve the low accuracy problem of dryness measurement for wet steam in gas driving oil extraction,the soft measurement model is established based on RBF neural network.By using least orthogonal square method the node number of implicit layer of network and output value of training network are determined.In practical operation,online correction element is adopted.The model offers excellent accuracy of proximity.The effectiveness of this method is approved in Liaohe Oil field.