基于深度神经网络和支持向量机的海底管线水合物生成预测模型  

Prediction model of submarine pipeline hydrate formation based on deep neural network and support vector machines

作　　者：郑秋梅[1] 商振浩 王风华[1] 林超 ZHENG Qiumei;SHANG Zhenhao;WANG Fenghua;LIN Chao(College of Computer Science and Technology in China University of Petroleum(East China),Qingdao 266580,China;Information Construction Department in China University of Petroleum(East China),Qingdao 266580,China)

机构地区：[1]中国石油大学(华东)计算机科学与技术学院,山东青岛266580 [2]中国石油大学(华东)信息化建设处,山东青岛266580

出　　处：《中国石油大学学报（自然科学版）》2020年第5期46-51,共6页Journal of China University of Petroleum(Edition of Natural Science)

基　　金：国家自然科学基金项目(51274232);山东省自然科学基金项目(ZR2018MEE004)。

摘　　要：针对目前天然气水合物预测方法对实际生产数据预测精度不高问题,分析水合物形成因素,引入支持向量机(SVM)和深度神经网络(DNN)理论,建立新的天然气水合物预测模型。模型通过深度神经网络提取生产数据的网络特征,将提取的特征与生产数据进行融合以增强数据区分度,使用非线性支持向量机对融合数据进行水合物预测,并采用东海CXB至CX平台混输海管近6年的生产数据进行实验。结果表明,与传统方法和已有BP神经网络方法相比,模型预测精度显著提高,且模型结构简单,具有较好推广价值。Aiming at the problem that current natural gas hydrate prediction methods have low prediction accuracy for actual production data,this paper analyzes the hydrate formation factors,introduces support vector machines(SVM)and deep neural network(DNN)theory,and establishes a new gas hydrate prediction model.The model extracts features of production data through a deep neural network,which are fused with production data to enhance the data discrimination.Non-linear support vector machines are used to predict hydrates from the fused data.Then,we verify the production data to the East China Sea CXB to CX platform mixed submarine pipeline for nearly 6 years.Compared with the traditional and the existing BP neural network methods,the model prediction accuracy is significantly improved.Furthermore,the model has a simple structure and is easily applicable to other regions.

关 键 词：天然气水合物 支持向量机 深度神经网络 BP算法 

分 类 号：TE88[石油与天然气工程—油气储运工程]