Name
Abstract | ||
|---|---|---|
Traction load prediction is the key part in operation planning of electrified railway. In order to better dig effective information from massive traction load data and improve the prediction accuracy and training speed, a new traction load prediction method is designed according to the characteristics of time sequence, random fluctuation and no-load happens frequently of traction load. In this method, a CNN-TCN joint model is proposed by combining Convolutional Neural Networks (CNN) and Temporal Convolutional Networks (TCN). The abundant traction load data are adopted as the input of the joint model. First, CNN is used to extract local features and generate continuous feature maps. The output of the last feature layer of CNN is taken as the input of TCN, and then the TCN is used for traction load prediction. The proposed method is used to conduct load prediction experiments with the traction load data of the power supply arms in a certain traction substation. The experimental results show that the proposed method has faster training speed and higher prediction accuracy than the current prediction methods in the field of prediction algorithm. |
Year | DOI | Venue |
|---|---|---|
2020 | 10.1109/IECON43393.2020.9255259 | IECON 2020: THE 46TH ANNUAL CONFERENCE OF THE IEEE INDUSTRIAL ELECTRONICS SOCIETY |
Keywords | DocType | ISSN |
electrified railway, traction load prediction, Convolutional Neural Networks, Temporal Convolutional Networks | Conference | 1553-572X |
Citations | PageRank | References |
0 | 0.34 | 0 |
Authors | ||
6 | ||