Diagnosis Research And Application On The Low Voltage Seriesarc Fault

With the development of modern industry, electric fire has become a great hiddendanger causing personal danger and property loss. According to the2010Fire StatisticalYearbook in China, the fire caused by the electrical faults every year accounted for30.7%in total fire cases. Based on the NEC standards, in the low voltage power system client, theloop currents of parallel arc and grounding arc are greater than75A generally, the existingprotection circuit breaker has been able to isolate and protect. For the series arc fault,however, the effective value of loop current in series arc fault limited by the line loads isusually approches to the normal, that is, between5to30A. The traditional protectionmethods cannot achieve to isolate and protect the series arc fault effectively which wouldlead to electrical fires. Therefore, the detection and diagnosis of the series arc fault areimportant for the safe operation of the electrical equipment.According to the UL1699-2008AFCI standard, the arc fault experimental platform isestablished to obtain the electrical parameters of the arc fault. Under the experimentalplatform, the different types of typical load for the arc fault simulation experiments aretaken and the current data both in normal and fault conditions are sampled to establish thearc fault diagnostic database for analyzing the arc fault characteristics.When the arc fault occurs, limited by the arc resistance and the line impedance, thetraditional current protection device cannot work effectively. According to themathematical model of the arc theory, the Mayr, Cassie and Stokes arc model in lowvoltage have been modeled and simulated. Compared with the simulation and theexperimental data for the waveform and wavelet energy entropy, the simulation model areanalyzed and identified for discribing the arc behavior.In order to study the series arc fault characteristics and to achieve the series diagnosisidentification for the arc fault, two methods are taken to diagnose the series arc fault withBP neural network. One is based on the wavelet transform modulus maximum theory bytaking the ratio of the wavelet bands modulus maximum as the network input vector.The other is according to the singularity, uncertainties and energy characteristics of the signalby identifying three characteristics signal characteristic distribution related to determinethe occurrence of series arc fault.Based on the STC microcontroller, the third-harmonic amplitude detection andenergy detection are combined to achieve arc fault detection. The hardware and softwareparts of arc fault detection device have been designed and the prototype has been realized.And based on LabVIEW virtual instrument platforms, the arc fault detection system hasbeen designed by G-language programming with the MPS-010601data acquisition card.