用于天然气管网腐蚀速率预测的WT-LSTM网络

WT-LSTM Network for Corrosion Rate Prediction in Natural Gas Pipeline Networks

中文关键词:  天然气净化  腐蚀探针  WT-LSTM网络  db10小波  预警  预测

英文关键词:natural gas purification  corrosion probes  WT-LSTM network  db10 wavelet  warning  prediction

中文摘要:

      在天然气净化过程中，强腐蚀性气体会导致反应容器出现不同程度的损耗。尽管传统腐蚀探针可以提供腐蚀预警，但其存在一定的滞后性。为解决这一问题，提出了基于小波变换-长短期记忆（Wavelet Transform-Long Short Term Memory,WT-LSTM）网络的腐蚀数据趋势预测方法。对原始数据采用db10小波进行分解，获取各细节分量后，利用LSTM网络对各分量进行预测并重构信号。实验结果显示，WT-LSTM模型的均方根误差(Root Mean Square Error,RMSE)为0.002 562，低于仅使用LSTM模型的RMSE值0.003 178，表明WT-LSTM网络在趋势预测上更加精准。基于WT-LSTM网络的预测方案能够有效跟踪腐蚀数据的变化，尤其在数据突变时效果显著，从而增强了腐蚀探针的在线监测能力，实现了对腐蚀情况的预测和预警，确保了天然气净化过程的安稳运行。

英文摘要:

      During the natural gas purification process,strong corrosive gases lead to varying degrees of attrition of the reaction vessel.Although traditional corrosion probes can provide corrosion warning,they have a certain lag.To solve this problem,a corrosion data trend prediction method based on wavelet transform-long short term memory(WT-LSTM) network is proposed.The original data are decomposed using db10 wavelet,and after obtaining each detail component,the LSTM network is used to predict each component and reconstruct the signal.The experimental results show that the root mean square error(RMSE) of the WT-LSTM model is 0.002 562,which is lower than the RMSE value of 0.003 178 for the LSTM-only model,indicating that the WT-LSTM network is more accurate in trend prediction.The prediction scheme based on WT-LSTM network can effectively track the changes of corrosion data,especially the effect is significant when the data change abruptly,so as to enhance the online monitoring capability of the corrosion probe,realize the prediction and warning of corrosion,and ensure the stable operation of the natural gas purification process.