Research On Extraction And Recognition Of Low Voltage Fault Arc Characteristic

Arc is very easy to cause fire due to its high energy and high temperature characteristics,and it is very harmful.Arc fault is the most common safety hazard in electrical circuits.Although traditional circuit breakers can protect against leakage,overcurrent,and short circuit,many serious fire accidents are caused by arc faults below the rated current.Traditional circuit breakers cannot play a protective role at all,so fault arcs have become a major problem in the field of electrical fire prevention and control.In this context,the theoretical research and application of low-voltage series arc fault detection technology has broad development potential.This paper takes arc fault detection in 220V/50 Hz electrical environment as the research object,finds the method of fault feature extraction,and conducts research on the method of arc fault detection and recognition based on variational modal decomposition sample entropy extraction and extreme learning machine.The main research content of the thesis as follows:(1)According to the UL1699-2008 AFCI standard,build an arc fault simulation experiment platform,select 5 different loads for experiment,simulate the arc fault of distribution lines under different loads and collect current signals,and observe and analyze the collected current signals to find faults Features,lay the foundation for follow-up research.(2)In view of the advantages of the variational modal decomposition(VMD)algorithm and sample entropy extraction in the field of signal processing,it is applied to the arc fault feature extraction,and the K value and penalty parameters that are difficult to determine in the algorithm are scientifically analyzed to obtain the optimal value.The characteristic vector formed by the sample entropy of the arc fault current signal is obtained.(3)For the combination of pattern recognition and artificial intelligence,by comparing the advantages and disadvantages of neural networks,support vector machines,and extreme learning machines based on the hidden layer output matrix,it is determined that the extreme learning machines are performing faults on series fault arcs.The advantages of diagnosis,the results show that the method has better results.