Sie befinden Sich nicht im Netzwerk der Universität Paderborn. Der Zugriff auf elektronische Ressourcen ist gegebenenfalls nur via VPN oder Shibboleth (DFN-AAI) möglich. mehr Informationen...
An Improved Convolutional LSTM Network with Directional Convolution Layer for Prediction of Water Quality with Spatial Information
Ist Teil von
Advances in Swarm Intelligence, p.94-105
Ort / Verlag
Cham: Springer International Publishing
Quelle
Alma/SFX Local Collection
Beschreibungen/Notizen
The prediction of water quality indicators is an important topic in environmental protection work. For the prediction of water quality data with multi-site data, this paper proposes an improved model based on ConvLSTM, which achieves the introduction of multi-site spatial relationships in water quality indicators prediction. On the basis of ConvLSTM, a directional convolutional layer is introduced to deal with the spatial dependence of multiple information collection stations with upstream and downstream relationship of a river to improve the prediction accuracy. The model proposed in this paper is applied to a dataset from three data collection stations to predict several indicators. Experiments on real-world data sets and results demonstrate that the improvements proposed in this paper make the model perform better compared to both the original and other common models.