人工神经网络预测含蜡原油的屈服应力  

Artificial Neural Network Model to Predict Yield Stress of Waxy Crude Oil

作　　者：饶心[1] 张国忠[2] 胡月[3] 吴玮 

机构地区：[1]中国石油天然气股份有限公司西部管道分公司科技服务中心 [2]中国石油大学(华东) [3]中国石油天然气股份有限公司西部管道分公司乌鲁木齐输油气分公司 [4]北京中航油工程建设有限公司

出　　处：《油气储运》2009年第11期17-20,共4页Oil & Gas Storage and Transportation

摘　　要：针对含蜡原油输送管道的停输再启动问题,以室内模拟环道试验数据为基础,利用误差反向传播算法(BP算法)对相同运行工况含蜡原油在不同停输温度条件下的启动屈服应力进行计算,并由此预测了含蜡原油在不同启动温度条件下的启动屈服应力值。计算结果与试验结果对比表明,BP神经网络对启动屈服值的预测值和试验值最大误差为1.9%,最小误差为0.12%。In allusion to the problem of shutting down and restarting of waxy crude oil pipeline and based on the data obtained from simulating experimental loop, a calculation on startup yield stress for waxy crude oil under similar running condition at different shutdown temperatures is made with back propagationalgorithm (BP algorithm), and startup yield value is predicted for waxy crude oil at different startup temperatures based on calculation results. The computational results are compared with the experimental results. The results show that as for the startup yield stress value predicted by the artificial neural network and the experimental value,the maximum value is 1. 9%,and the minimum error is 0. 12%.

关 键 词：含蜡原油 屈服应力 停输再启动 BP算法 人工神经网络 预测 

分 类 号：TP183[自动化与计算机技术—控制理论与控制工程] TQ336.1[自动化与计算机技术—控制科学与工程]