Research Of Arc Fault Detection Technology Based On Wavelet Transform

In the many factors causing electrical fire, the arc fault is one of the most important incentives.And in the domestic low-voltaged power distribution system, the traditional circuit interrupter is notwith the function of arc fault detection, so the research of arc fault detection technology is of greatsignificance.Arc fault detection is a circuit protection technology, whose main function is to identify arcfault and to take certain measures to protect the circuit timely and accurately when the arc appearsin the circuit. And the device using this technology of protection was called Arc Fault CircuitInterrupter (AFCI). At present, arc fault detection technology in America has reached a high level ofmaturity and AFCI products have been gradually promoted, while it started late in China and AFCIproducts is almost blank in the domestic market.Arc fault detection methods and testing standards, widely used in America, can not be appliedmechanically in China because of the difference of low-voltage distribution environment indifferent place. This thesis researches the arc fault detection technology and provides a new arcfault detection method based on the study of the arc characteristics, which turns to be a relativelyaccurate recognition rate on the normal load circuit when the arc appears and a lower falseoperation rate on the normal load circuit in a normal situation.The thesis focus on research of arc fault detection technology based on the wavelet transform,providing the criteria and method of the arc fault detection as well as test validation according to thearc characteristics and the results of the analysis of wavelet transform. Its main work is as follows:1. The thesis introduces the research status and implementation methods of arc fault detectiontechniques, and points the key issues which need to be solved;2. The thesis builds the fault arc experimental platform according to UL1699, the standard ofAFCI in America, then collects and sorts out fault arc data to establish an effective arc database;3. Through time domain and frequency domain analysis of the arc data, the thesis figures outsome conclusions that characteristics of arc in time domain and its limitation on arc fault detection,and also finds that the characteristic frequency range of the arc is50kHz~100kHz, which turns outto be the foundation of arc fault detection in the way of frequency domain;4. The thesis analyzes the principles and advantages of wavelet transform for arc faultdetection firstly, then designs experiment to select db4wavelet as the optimal wavelet basis for arcdetecting, then extracts characteristic quantity in the wavelet coefficients cD3based oncharacteristic frequency range of arc, then determines three characteristic quantities——variance, the number of pulses, energy and formulates the corresponding judging criterions of arc fault infrequency domain, then verificates the judging criterions and analyzes the their malfunction withexperiments, and finaly gives out the complete algorithm flow of the arc fault detection methodbased on wavelet transform.5. The thesis designs experiments according to the UL1699standards as well as the existingexperimental conditions, to test the basic function and malfunction of our arc fault detection methodbased on wavelet transform. The test results show that this method can recognize the arc faultaccurately and it also has a good anti-interference ability.In this thesis, all research, experiments and testing are based on data of common loads whichare plentiful and reasonable, so the conclusion of this thesis has good universality.