Looking for breakthrough ideas for innovation challenges? Try Patsnap Eureka!

Fault identification method for flexible DC grid based on convolutional neural network

A convolutional neural network, flexible DC technology, applied in fault location, fault detection by conductor type, information technology support system, etc., can solve the problem of insufficient resistance to transition resistance, selectivity and sensitivity can not meet DC fault detection and identification requirements, single-threshold-type judgment method, etc., to achieve the effect of improving the resistance to transition resistance, meeting the quickness, and improving the accuracy

Active Publication Date: 2019-11-01
SOUTHEAST UNIV
View PDF11 Cites 22 Cited by
  • Summary
  • Abstract
  • Description
  • Claims
  • Application Information

AI Technical Summary

Problems solved by technology

These methods are all based on short-time window sampling, relying only on the original sampling signal or the judgment after simple mathematical processing of the original signal, and are essentially a single threshold-type judgment method
Although such methods can meet the requirements of quick action, they often have the problem of insufficient resistance to transition resistance, resulting in selectivity and sensitivity that cannot meet the requirements of DC fault detection and identification.

Method used

the structure of the environmentally friendly knitted fabric provided by the present invention; figure 2 Flow chart of the yarn wrapping machine for environmentally friendly knitted fabrics and storage devices; image 3 Is the parameter map of the yarn covering machine
View more

Image

Smart Image Click on the blue labels to locate them in the text.
Viewing Examples
Smart Image
  • Fault identification method for flexible DC grid based on convolutional neural network
  • Fault identification method for flexible DC grid based on convolutional neural network
  • Fault identification method for flexible DC grid based on convolutional neural network

Examples

Experimental program
Comparison scheme
Effect test

specific Embodiment 1

[0036] A method for identifying faults in flexible DC power grids based on convolutional neural networks proposed by the present invention, step S1 constructs an initial model of convolutional neural networks for fault identification in flexible DC power grids as follows: figure 1 shown.

[0037] The initial model of the convolutional neural network includes: the first branch structure, the second branch structure, the connection layer, the third convolution layer, the third pooling layer, the random deactivation layer, the first fully connected layer, the second fully connected layer, And the output layer; wherein, the first branch structure includes: the first input layer, the first convolution layer, the first pooling layer; the second branch structure includes: the second input layer, the second convolution layer, the second pooling layer.

[0038] After the first input data is input to the first input layer, it passes through the first convolutional layer and the first p...

specific Embodiment 2

[0043] figure 2 It is a topological structure diagram of a four-terminal flexible DC power grid based on MMC. The topology structure includes: the first converter 1, the second converter 2, the third converter 3, and the fourth converter 4. These four converters all adopt a pseudo-bipolar structure, and the voltage levels are ±200kV. A first line is connected between the first converter 1 and the second converter 2, a second line is connected between the second converter 2 and the third converter 4, and the third converter 4 and the first converter A third line is connected between the four converters 3 , and a fourth line is connected between the fourth converter 3 and the first converter 1 . Both ends of each line are equipped with smoothing reactors and DC circuit breakers near the side of the converter station, and each line adopts a phase-domain frequency-varying model. The fourth line between the first converter 1 and the third converter 3 is used as the test protect...

specific Embodiment 3

[0048] The data of the training data set is normalized and used as the input data of the initial model of the convolutional neural network, and the structure and structure of the initial model of the convolutional neural network are adjusted according to the recognition accuracy and loss function of the output data of the initial model of the convolutional neural network. Parameters, after training and adjustment, when the recognition accuracy reaches 100% and the loss function is less than 0.1, save the model structure and parameters to form a convolutional neural network training model.

[0049] Adjusting the structure of the convolutional neural network initial model specifically includes: adjusting the number of the first convolutional layer, the second convolutional layer and the third convolutional layer, adjusting the first pooling layer, the second pooling layer and the third convolutional layer The number of three-pooling layers, adjusting the number of the first fully...

the structure of the environmentally friendly knitted fabric provided by the present invention; figure 2 Flow chart of the yarn wrapping machine for environmentally friendly knitted fabrics and storage devices; image 3 Is the parameter map of the yarn covering machine
Login to View More

PUM

No PUM Login to View More

Abstract

The present invention relates to the field of fault detection and identification of a flexible DC grid, and particularly relates to a fault identification method for a flexible DC grid based on a convolutional neural network. The method comprises: constructing a convolutional neural network model with a branch structure; simulating to obtain fault case data, training the model and adjusting the model parameters; saving the model structure and parameters with high identification accuracy and a small loss function in the training verification; setting a fault detection and identification starting criterion, starting a fault identification program, sampling a data window in 2ms at a sampling signal detection point, and collecting the positive and negative voltages and currents of the line under the actual working condition; and performing normalized processing on the data and identifying an actual fault type by using the model. According to the method provided by the present invention, window information is sampled by fully using the 2ms, the comprehensive utilization of various dimension fault features can be realized by using the branch structure of the model, the accuracy of faultidentification in the flexible DC grid can be improved, the resistance to transition can be improved, and quickness, selectivity, and sensitivity requirements of fault identification can be satisfied.

Description

technical field [0001] The invention relates to the field of fault detection and identification of flexible DC power grids, in particular to a method for fault identification of flexible DC power grids based on convolutional neural networks. Background technique [0002] With the development of modular multilevel converter (MMC) technology, high-voltage direct current transmission (VSC-HVDC) based on voltage source converter technology has gradually replaced traditional high-voltage direct current transmission (LCC-HVDC) based on thyristor technology. , has the characteristics of independent control of active and reactive power, no commutation failure, and low harmonic level. In order to make full use of the advantages of the DC power grid, realize the asynchronous interconnection of the AC system, multi-energy complementarity, and improve the level of new energy consumption, the multi-terminal flexible DC power grid has become the trend of the future development of DC power...

Claims

the structure of the environmentally friendly knitted fabric provided by the present invention; figure 2 Flow chart of the yarn wrapping machine for environmentally friendly knitted fabrics and storage devices; image 3 Is the parameter map of the yarn covering machine
Login to View More

Application Information

Patent Timeline
no application Login to View More
IPC IPC(8): G01R31/08
CPCG01R31/086Y04S10/52
Inventor 梅军葛锐范光耀王冰冰朱鹏飞严凌霄
Owner SOUTHEAST UNIV
Who we serve
  • R&D Engineer
  • R&D Manager
  • IP Professional
Why Patsnap Eureka
  • Industry Leading Data Capabilities
  • Powerful AI technology
  • Patent DNA Extraction
Social media
Patsnap Eureka Blog
Learn More
PatSnap group products

Browse by: Latest US Patents, China's latest patents, Technical Efficacy Thesaurus, Application Domain, Technology Topic, Popular Technical Reports.

© 2024 PatSnap. All rights reserved.Legal|Privacy policy|Modern Slavery Act Transparency Statement|Sitemap|About US| Contact US: help@patsnap.com