Application Of Artificial Neural Network In Wide-area Backup Protection System

Relay protection plays a vital role in the power system.However,traditional relay protection also has shortcomings.Three-stage distance protection is the main protection for relay protection,especially when it is applied to high-voltage transmission lines,protection often does not work.With the continuous development of Chinese smart power networks,the proportion of EHV transmission lines in transmission lines will become larger and larger.At the same time,the Wide Area Measurement System(WAMS)has also formed a certain scale in Chinese power system.For the problems of traditional relay protection,and combing with the development trend of modern power systems,a new type of backup protection for the future smart grid was proposed –wide-area backup protection.Wide-area backup protection has the advantages of wide protection range and long-distance protection.It means the wide-area backup protection to get a lot of information from the measuring devices within the protection range.Therefore,it also brings problems to the wide-area backup protection : the limited communication capabilities of current communication systems.When a large amount of real-time data of the power system is concentrated in one decision-making unit of the wide-area backup protection,it may cause delay or loss of the information.It is not allowed for some devices with higher protection requirements.Based on a large number of experimental studies by predecessors,this paper studies the above problems.Firstly,this paper built a power system model of wide-area backup protection.The analytic hierarchy process(AHP)is applied to wide-area backup protection firstly.The devices are divided into different levels according to the importance of the protected device.According to devices level,the information of different level devices is uploaded to the wide-area backup protection of the central station,which reduces the probability of communication delays and loss.Then,the artificial neural network is used to comprehensively judge the information and identify the fault information.In this thesis,I built a powersystem model using MATLAB simulation software.It simulates the various faults of the power system under different operating modes.Then I used the large amount of simulation data to train artificial neural networks.Finally,I conducted a fault simulation experiment of 35 KV transmission line using the background monitoring system(CBZ-8000),WXH-820 series microcomputer line protection and monitoring device and GMX-600 type fault test box in the laboratory of power system and automation.The experimental results suggest that the artificial neural networks can accurately identify the normal operating status information and fault information of the power system.The research of related problems broad application field of artificial neural networks and promote to solve reliability problems of wide-area backup protection systems.