Research On Fault Location Of Secondary System In Smart Substation Based On Deep Learning

The smart substation is the core part of promoting the construction of the smart grid.Its networked secondary system realizes equipment information sharing and interactive operation,and also brings a large amount of secondary system status data.The current secondary system fault location methods have many shortcomings.Deep learning is a research hotspot of artificial intelligence,and its powerful data mining capabilities meet the needs of quickly and accurately locating fault points in the secondary system.Based on the current research,this paper proposes a fault location method for the secondary system of smart substations based on deep learning.The main work is as follows:(1)A method for locating secondary equipment faults in smart substations based on recurrent neural networks is proposed.First,starting from the characteristic information of different modules of the secondary equipment when faults occur,the fault location reasoning knowledge base is sorted out.Secondly,combining the abnormal switch value alarm,abnormal sampling alarm,and abnormal device alarm of secondary equipment,the characterization method of fault characteristic information is proposed.Finally,using the deep learning training method,a secondary equipment fault location model based on the recurrent neural network is established and the fault location steps are given.The results of calculation examples show that this fault location model can process high-dimensional fault feature sets and accurately locate faults.In the case of loss of feature information or false alarms,the fault location model has good fault tolerance performance.(2)A method of fault location for the smart substation communication network based on the deep neural network is proposed.First,based on the redundancy monitoring of the communication network,the characteristic information obtained by different monitoring nodes at different locations is analyzed,and the characterization method of the fault characteristic information is proposed.Secondly,fault samples are generated based on the emergence principle,which expands the training sample set.Finally,combined with the training rules in the deep learning theory,the fault location model of the communication network based on the deep neural network is established and the fault location steps are given.The results of calculation examples show that this fault location model can handle high-dimensional fault feature sets,and can accurately locate faults in various environments.In the case of feature information loss or false alarms,the fault location model has good anti-interference ability.(3)A distributed fault location method for secondary systems based on deep learning is proposed.First,the fault characteristics of the secondary equipment and communication network at different intervals are analyzed,and the characterization method of the fault characteristic information is proposed based on the research in the previous chapters.Secondly,considering that it is difficult to quickly process data and accurately locate faults using a single-point model,a distributed fault location method for the secondary system is proposed.Finally,combined with the existing neural network,a distributed secondary system fault location model is established.The results of calculation examples show that the use of distributed fault location model can accurately locate the fault point of the secondary system.