Fault Diagnosis of Fusion Power Station Based on Neural Network

Fault diagnosis is an important software facility for ensuring the safety of substations. The fault diagnosis system can accurately detect and locate damaged power components, thus ensuring that the substation can provide reliable and stable power. In this article, a fault diagnosis model based on a deep fully connected neural network is proposed. For all kinds of power modules in the substation, the fault diagnosis model is established, respectively. The input of the model is the fault feature vector of the power module, and the output is the fault probability of the power module. Based on the historical data of the EAST 110 kV substation, the sample generation method based on probability statistics and random sampling is used to generate a large amount of sample data to successfully train the fault diagnosis model. The test results show that the model can correctly and quickly diagnose faulty power elements. In addition, the model can be easily ported to other substations.

I. Introduction

The main function of the EAST substation is to connect the EAST device to the national grid and to transmit and distribute electric energy to each power system of EAST through the feeder line [1].