Research On Series Arc Fault Detection Based On Higher-Order Statistics

Electric fire posed a huge threat to people's lives and property security,has always been a critical content of Chinese fire prevention work.Arc fault is an important reason for the occurrence of electrical fires,especially the series arc fault caused by decreasing of wire insulation or poor connection,it is difficult to make reliable detection due to its characteristics of concealment and complexity,and the protection equipment such as breakers can not provide effective protection.Therefore,researching the characteristics of series arc fault and identify it correctly,then developing a detection prototype of arc fault has important theoretical research significance and engineering application value to the safety of electricity.Aiming at the detection method of main circuit current which is susceptible to the nonlinear load,caused the problem of misjudgment and missed,the magnetic coupling signal of high frequency current during the process of extinction and reignition is used as the detection object in this paper.The main research contents are as follows:1.Based on the generation of high frequency arc current,took the space electromagnetic radiation and the placement of the transformer into account,the method of coupling signal acquisition is experimentally verified by using the series arc fault experiment platform.On the above basis,the data of main line current and coupling signal under different working conditions are collected.Through detailed comparisons and analysis,it is concluded that the method of coupling signal acquisition will not affected by the singularity rated current signal of non-linear load,and the arc fault signal on the small branch can be detected effectively,which showing the feasibility and superiority.2.Proposed an arc fault identification method based on higher-order statistics.Taking the complexity of the actual operation into account,a variety of situations including single load and its combined load are studied.The kurtosis value of the coupling signal in different states are calculated,and appropriate threshold is extracted to identify the arc fault.In addition,it is proposed that the coupling signal to be judged by multiple cycles so as to eliminate the interference of the load starting and switching effectively.3.The dsPIC is took as the core of the hardware and the higher-order statistical algorithm as the software foundation,modular ideas is used to design the arc fault detection device.The hardware includes power supply module,signal conditioning module and output module.The software includes AD sampling interrupt program,higher-order cumulant calculation program and arc fault diagnosis program.In order to ensure the independence of both acquisition and processing processes,while improving the real-time of detection,double sampling array are used to store the collected signal alternately,and the program is optimized by replaced simple.4.According to national standard,the detection performance of the prototype was tested,and the test result shows that the method of coupling signal acquisition and the detection algorithm of higher-order statistics proposed in this paper can identify the arc fault effectively and will not cause missed and misjudgment.The expected effect of arc fault detection is achieved by the prototype designed in this paper.