| The performance of intelligent substation relay protection device directly affects the safe and stable operation of the substation,and the performance test of relay protection device is usually completed by a special relay protection tester,including factory dispatching and on-site debugging.A large number of feedback information from the commissioning site,especially for the commissioning feedback information of the protection unit,shows that using the existing digital relay protection tester to complete the commissioning of the protection unit often consumes a lot of energy and time,and using the existing digital relay protection tester cannot complete the communication consistency check of the protection device.In order to improve the test efficiency and effect of the digital relay protection tester,the standardization and automation of the test process is its development direction.In 2016,led by Sichuan Electric Power Research Institute,through the integration of protection test function and station control layer function,and the combination of protection test model and station control layer model,the first generation of intelligent substation automatic relay protection tester was successfully researched and manufactured,during the test,the tester is not required to go to the protection device in person to read and modify the setting value,switch on / off the soft pressing plate,reset and other work,which truly realizes the automatic test of the protection device.Although the first generation of intelligent substation relay protection automatic tester improves the automation level and intelligence level compared with the original digital relay protection tester,and basically realizes and meets the requirements of automatic test of relay protection device,but the test efficiency does not reach the expected effect in the actual field debugging process,this is because of the occasional faults in the test process.It takes a lot of time to check manually,which makes the test efficiency decrease.Aiming at the problems of the above-mentioned automatic tester,this paper focuses on the fault diagnosis of relay protection test in intelligent substation,and the research object is 220 k V intelligent substation.Firstly,this paper classifies the alarm information.According to the field operation and maintenance experience and relevant technical specifications,the above data information is divided into sampling fault,tripping fault,ontology fault and communication fault according to the service object,it is convenient for the operation and maintenance personnel of the intelligent substation to recognize the alarm information and improve the test efficiency;moreover,it combs the faults and possible causes in the test process of the single bay protection of the intelligent substation.Secondly,aiming at the fiber link failure in the test process,a graph-based relay protection test fiber fault location algorithm is proposed,by using the correlation matrix of the graph and the link alarm information to perform related operations,the link fault can be located relatively simply,and the reliability,accuracy and computational efficiency of the overall fault location are increased.Finally,aiming at the other faults in the single-interval protection test of intelligent substation,a fault location algorithm for relay protection test based on deep learning is proposed,the Recurrent Neural Network(RNN)is used for deep learning of fault data which with supervised,and the Back-Propagation Trough Time(BPTT)algorithm is used to optimize the training parameters.Taking the 220 k V line interval of the 220 k V intelligent substation as an example for the model performance test,the test shows that the fault diagnosis algorithm has 92.31% accuracy and the training time is about 10 min,which proves that the algorithm has high reliability and calculation efficiency,can effectively diagnose the problems in the test process,and indirectly improve the test efficiency of the relay protection device. |
PDF Full Text Request