Research on Prediction Method of Pipe Gallery Environment based on LSTM Circular Convolution Neural Network

Pan Geng ； Xiao Yang ； Yaozhong Zhang ； Xiansong Huang

《International Core Journal of Engineering》 6卷9期 (2020/09) Pp. 340-346

https://doi.org/10.6919/ICJE.202009_6(9).0043

摘要

At present, there are many blind spots in the management of the multi-purpose network for underground pipelines in cities, while its security mechanism is not so improved. A pipe gallery accident can pose a threat to the urban functions, and the security of life as well as that of property, no matter it is a gas leak or explosion in the pipelines entering the gallery, or a failure occurs in the equipment attached to the cabin.To realize the safe operation and efficient management of the underground integrated pipe gallery, our project is based on the data recorded in authentic operations, employs the neural network algorithm to predict its environmental parameters under authentic operations, and also apples LSTM algorithm to detect abnormal states. The designed model has a high accuracy rate in calculating and can evaluate the running state of the pipe gallery.

關鍵字

Neural network ； LSTM ； Pipe gallery

參考文獻

HOCHREITER S, SCHMIDHUBER J. Long short- term memory[J]. Neural Computation, 1997, 9(8): 1735- 1780.

Lecun Y, Bengio Y, Hinton G. Deep learning. [J]. Nature, 2015, 521(7553):436..

Girshick R. Fast r-cnn[C]//Proceedings of the IEEE international conference on computer vision. 2015: 1440-1448.

Ye Wang, Xinxiang Zhang, Mi Lu, Han Wang, Yoonsuck Choe. Attention augmentation with multi-residual in bidirectional LSTM[J]. Neurocomputing, 2020, 385.

Mingming Gao, Jianjing Li, Feng Hong, Dongteng Long. Day-ahead power forecasting in a large-scale photovoltaic plant based on weather classification using LSTM[J]. Energy,2019,187.