We found a match
Your institution may have access to this item. Find your institution then sign in to continue.
- Title
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.
- Authors
郑秋梅; 商振浩; 王风华; 林 超
- Abstract
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.
- Subjects
UNDERWATER pipelines; SUPPORT vector machines; GAS hydrates; FORECASTING; NATURAL gas
- Publication
Journal of China University of Petroleum, 2020, Vol 44, Issue 5, p46
- ISSN
1673-5005
- Publication type
Article
- DOI
10.3969/j.issn.1673-5005.2020.05.006