基于TS-FNN模型的常压塔塔顶汽油干点预测  

Prediction of Gasoline Dry Point at Atmospheric Tower Top Based on TS-FNN Model

作　　者：郭惠娟 李奇安[1] GUO Huijuan;LI Qi an(School of Information and Control Engineering,Liaoning Shihua University,Fushun 113001,China)

机构地区：[1]辽宁石油化工大学信息与控制工程学院,辽宁抚顺113001

出　　处：《沈阳大学学报（自然科学版）》2022年第2期119-125,共7页Journal of Shenyang University：Natural Science

摘　　要：为了实现汽油干点的直接在线测量,提出了一种基于T-S模糊神经网络(TS-FNN)的汽油干点预测方法。以某炼油厂常减压装置常压塔塔顶汽油干点为背景,确定影响因素,分析采集的数据,依据样本数据建立BP神经网络(BPNN)模型及TS-FNN模型,验证模型的预测有效性。结果表明:基于TS-FNN模型预测的汽油干点与观测值基本一致,平均相对误差为0.68%,该网络模型通过调整参数来修正隶属度函数,具有很强的自适应能力。基于T-S模糊模型的改进神经网络软测量模型跟踪效果较好,预测结果具有较高的准确性。In order to realize the direct on-line measurement of gasoline dry point,a gasoline dry point prediction method based on T-S fuzzy neural network(TS-FNN)was proposed.Taking the gasoline dry point at the top of the atmospheric tower of an atmospheric and vacuum unit in a refinery as the background,the influencing factors were determined,the collected data was analyzed,and the BP neural network(BPNN)model and the TS-FNN model were established according to the sample data,and the validity of the model prediction was verified.The results show that the predicted gasoline dry point based on the TS-FNN model is basically consistent with the observed value,and the average relative error is 0.68%.The network model modifies the membership function by adjusting the parameters,and has a strong adaptive ability.The improved neural network soft-sensor model based on T-S fuzzy model has better tracking effect and higher prediction results.

关 键 词：软测量 T-S模型 模糊神经网络 汽油干点 预测 

分 类 号：TP29[自动化与计算机技术—检测技术与自动化装置]