Open AccessPower Grid Fault Diagnosis Based on Improved Deep Belief Network
Author(s) -
Yihui Zhang,
Yuansheng Zhang,
Wen Li,
Zhimei Cui,
Yini He,
Guangshi Liu
Publication year - 2020
Publication title -
journal of physics. conference series
Language(s) - English
Resource type - Journals
SCImago Journal Rank - 0.21
H-Index - 85
eISSN - 1742-6596
pISSN - 1742-6588
DOI - 10.1088/1742-6596/1585/1/012021
Subject(s) - fault (geology) , deep belief network , computer science , artificial intelligence , electric power system , set (abstract data type) , fault coverage , deep learning , fault model , node (physics) , power (physics) , grid , data mining , engineering , physics , geometry , electrical engineering , structural engineering , mathematics , quantum mechanics , seismology , electronic circuit , programming language , geology
It is of great significance to quickly and accurately identify power system faults. This paper introduces the idea of deep learning into power system fault diagnosis research and proposes a fault diagnosis model based on improved deep confidence network. Construct a set of 30-dimensional original features that can reflect the fault characteristics of the power system as the model input, and the fault diagnosis result is the model output. Use multi-layer Boltzmann machines to train the mapping relationship between grid faults and system features. Finally, the extreme learning machine is used to supervise the labeled samples to modify the network parameters. Different system failure scenarios were set up on the New England 10-machine 39-node system to test the diagnosis ability of the model. Simulation results show that the improved deep belief network has a strong feature extraction capability. The improved deep belief network has higher accuracy and faster speed in fault categories, fault areas, and fault locations than common artificial intelligence methods.