神经网络技术在加氢裂化产品收率预测模型中的应用  

Application of neural network technology in prediction models for the yield of hydrocracking products

作　　者：王国荣[1] 

机构地区：[1]中国石化扬子石油化工有限公司芳烃厂,江苏南京210048

出　　处：《石化技术与应用》2015年第4期349-353,共5页Petrochemical Technology & Application

摘　　要：利用BP神经网络技术建立了加氢裂化产品收率的预测模型;以模型预测计算结果为指导,根据实际工况优化调整了工艺参数。结果表明,工艺参数优化调整后,重石脑油、航煤、尾油总收率由75.8%提高至83.1%,经济效益增加674.5万元。The prediction models for the yield of hydrocracking products were established by BP （Back Propagation） neural network technology. Under the guidance of model prediction results, the process parameters were optimized and adjusted according to the actual operation conditions. The results showed that after optimizing and adjusting the process parameters, the total yield of heavy naphtha, jet fuel and tail oil increased from 75.8% to 83.1% and the economic benefits increased by ￥6. 745 million.

关 键 词：加氢裂化 神经网络 产品收率 模型 

分 类 号：TE624.432[石油与天然气工程—油气加工工程]