期刊
NEURAL COMPUTING & APPLICATIONS
卷 33, 期 16, 页码 10293-10310出版社
SPRINGER LONDON LTD
DOI: 10.1007/s00521-021-05791-4
关键词
Fuzzy sets; B-spline; Dissolved oxygen; Particulate matter
资金
- IT Academy programme
A computing scheme variant of granular recurrent neural network, deduced from a system of partial differential equations, is able to capture spatiotemporal variability and has shown promising results in predicting dissolved oxygen and particulate matter concentration.
We present a computing scheme as a variant of a recently proposed granular recurrent neural network. Being deduced from a generic system of partial differential equations, this variant is able to capture the spatiotemporal variability of some datasets and problems. The convergence of the computing scheme has been formally discussed. Some preliminary numerical experiments were first performed by using synthetic datasets, inferring some particular partial differential equations. Then two application examples were considered (by using publicly available datasets), namely the prediction of dissolved oxygen in surface water simultaneously at different depths (unlike the current literature) and the prediction of the concentration of particulate matter less than 2.5 mu m in diameter at different sites. The numerical results show the potential of the approach for forecast against well-known techniques such as Long Short-Term Memory networks.
作者
我是这篇论文的作者
点击您的名字以认领此论文并将其添加到您的个人资料中。
推荐
暂无数据