基于多神经网络结构的常压塔侧线产品质量软测量

The Multiple Neural Network Structure Based Soft Measurement of the Quality of Siddeline Products of Normal Pressure Tower

根据常压塔的原油进料及产品多变 ,提出采用多神经网络结构建立侧线产品质量软测量模型。利用基于马氏距离的数据分类技术 ,对输入样本分类。利用产品质量化验分析值 ,对软测量模型进行校正。实际应用表明多神经网络结构的软测量精度高。

The paper stated that the soft measurement model for the quality of sideline products is established by using multiple neural networks structure to meet variable crude feedings and products of normal pressure tower. By adopting data sorting technology based on Mars distance the input samples are sorted. The soft measurement model is corrected by analyzing values of chemical examination on products. The result of practical application shows the accuracy of soft measurement based on multiple neural network structure is high.