Fault Diagnosis For Intelligent Substation With Fault Tolerance

Intelligent substation is the foundation and support of the Smart Grid,the intelligence of substation is emdbodied by advanced application.Unlike traditional substation,the monitoring system of intelligent substation receives a large number of alarms in failure.In order to prevent power outages,improve the safety and reliablity of the power system,the fault diagnosis for intelligent substation need take full advantage of multi-source information,diagnosises quickly and accurately,then gives a reasonable action evalution.Firstly,the article uses Kohonen Network for initial diagnosis for substation.Kohonen is a self-organizing network,it has a good ability for clustering.This paper uses Kohonen Network for initial diagnosis by changing the structure of the function in order to adapt to the characteristics of the fault information.The network locates the fault equipment in a specific region.This method does not miss nodes,computes quickly,and it has fault tolerance,so the method is suitable for the initial fault diagnosis of intelligent substation.Then the article uses Fuzzy Causal Network for fault diagnosis with the uncertain information.It gives fuzzy values of causal relationship between actions of equipments by the uncertain information.The information including equipment defects,judgement of run history and judgement of maintenance.It corrects the error message with timing information and electrical parametres then it identifies the accurate fault by Fuzzy Causal Network.The article gives actual samples of fault diagnosis by causal network and it's explanation.This method is simple,running very fast,it has improved fault tolerance by using uncertain information,the method is suitable for substations with simple causal relationship.Then this paper estabilishes a fusion method of fault diagnosis based on Cellular Neural Network and Fuzzy Integral.It builds cellular RBF neural network in the center of each equipment,uses fuzzy value of actions as inputs of neuron,obtains the fault probability of each equipment,then fuses the fault probabilities of ralated equipments,get the global fault diagnosis result of the entire substation at last.This method can adapt to the complex substation.It can distinguish complex fault correctly with multi-source information,the computing speed can meet the requirements.The fault of intelligent substation can be well diagnosised from the methods recited herein.Using the uncertainty substation fault information it can give a reasonable judgment and a scientific evaluation.These methods can easily improve fault tolerance and have practical values.