基于BP神经网络的模拟冷却水中碳钢的腐蚀速率预测  

Prediction of Corrosion Rate of Carbon Steel in Simulated Cooling Water Based on BP Artificial Neural Network

作　　者：伊帆[1,2] 李德豪[1,3] 郎春燕[2] 朱越平[1] 殷旭东[1] 

机构地区：[1]广东石油化工学院,茂名525000 [2]成都理工大学,成都610059 [3]华东理工大学,上海200237

出　　处：《腐蚀与防护》2012年第11期947-951,共5页Corrosion & Protection

基　　金：广东省重大科技专项(2009A080206002);广东省科技攻关项目(2011B010100045)

摘　　要：针对石油化工企业污水回用于循环冷却水系统时造成的化工设备腐蚀问题,采用BP神经网络算法建立了碳钢在模拟循环冷却水中的平均腐蚀速率预测模型。该模型以温度、pH值、Ca2+、Cl-、SO42-、HCO3-浓度为输入参数,以平均腐蚀速率为输出参数。计算结果表明,该模型具有良好的预测精度,模拟出的平均腐蚀速率与实测值拟合较好,并且能够反映冷却水中各腐蚀性因素与腐蚀速率之间的关系。A model for predicting the average corrosion rate of carbon steel in circulating cooling water was built on the basic of BP neural network theory in order to analyzse the problems of corrosion caused by simulated circulating cooling water. In this model, pH, Cl-, Ca2＋ , SO42- , HCO3 and temperature were set as input parameters while the average corrosion rate was the output parameter. The results show that the model has good forecasting accuracy and it could reflect the corrosion factors effectively.

关 键 词：人工神经网络 碳钢 循环冷却水腐蚀 

分 类 号：TG174[金属学及工艺—金属表面处理] TQ085[金属学及工艺—金属学]