Application Of Fault Diagnosis Technology And Monitoring Method In Building Electrical System

With the continuous development of modern building systems toward automation and intelligence,the importance of building electrical systems is becoming more and more prominent,and its operating status and operating reliability directly affect the normal operation of the building system.Therefore,the research and analysis of the fault diagnosis of the building electrical system is of great value to the development of the building industry.Most building electrical systems are neutral-grounded systems with large fault currents.Once a grounding or short-circuit fault occurs,the entire system will be unstable.Therefore,when a fault occurs,it will cause greater harm to the power distribution system and electrical equipment.In this thesis,combined with the actual operating conditions of the building electrical system,the fault characteristics and characteristics of 9 different faults including interphase short circuit,high resistance grounding,arc grounding,arc short circuit,phase loss,and overload are studied,and their effects on power distribution are analyzed respectively.The impact caused by the network and electrical equipment.Next,by summarizing the characteristics and characteristics of these 9 types of building electrical faults,extracting the corresponding 3 phases and 4 types of 12-dimensional fault feature vectors,and on this basis,draw lessons from traditional fault identification ideas and neural network algorithms to propose a neural network-based building Electrical fault identification method,so as to realize automatic identification and judgment of low-voltage power distribution system faults.This thesis or this dissertation studies the principles of traditional BP neural network and genetic algorithm to optimize BP neural network and realizes the calculation examples.Combining the advantages and disadvantages of traditional neural network to realize fault identification and the characteristics of building electrical system,it proposes the use of extreme learning machine(ELM)algorithm for building electrical System fault identification,by comparing the speed and accuracy of the three algorithms to diagnose faults,the conclusion is finally reached:ELM neural network can accurately identify under any type of fault,with an accuracy rate of 97.56% and a time-consuming only 0.202 s.Its speed and accuracy are the best.This thesis has 37 pictures,9 tables and 81 references...