| When the line series fault arc occurs in the home environment,the line will produce a high temperature of more than 3000 degrees Celsius.It can ignite the line insulation easily,and cause the serious electrical fire.However,it is difficult to detect the series fault arc because of its concealment,diversity and randomness.The main detection method of series fault arc is based on current signal,but the current signal is greatly affected by the load.In this paper,the characteristic detection of the signals of power supply terminal voltage and the load terminal voltage were carried out,and the series fault arc detection device is researched based on this:Firstly,the voltage characteristics experimental platform of series fault arc was built.When the classical household load is normal and series fault arc occurs,the signals of power supply and load voltage and current are collected and analyzed by the experimental platform;Then,the collected power terminal voltage and load terminal voltage signals are extracted through cosine similarity,Pearson correlation coefficient and Hausdorff distance.The circuit voltage signal is received by subtracting the two signals mentioned above.And the time domain feature,frequency domain feature and time-frequency domain feature of the circuit voltage signal are extracted to eight feature quantities.The eight featrue quantities are used as inputs of neural network,and the criterion of series fault arc is calculated by neural network;Secondly,after testing the data of different synchronization accuracy and sampling frequency,the results showed that the synchronization accuracy was within 0.2ms and the sampling frequency is above 5k Hz,which can achieve the requirements of the algorithm;Thirdly,the fault arc detection device was designed which included the time synchronization system and data acquisition system.The time synchronization system adopts BDS with wireless IRIG-B scheme,and the data acquisition system adopts AD7606 chip.The hardware part of the series fault arc detection device was designed.And the fault feature recognition algorithm and neural network algorithm were successfully improved to embedded algorithm,which is applied to the MCU of ARM core;Finally,the series fault arc detection device is tested.The test results show that the series fault arc detection device can detect the series fault arc in the line well,and the accuracy is more than 98%.When the series fault arc occurs,the circuit can be cut off in time to prevent early electrical fire.This paper has 51 pictures,13 tables and 58 references. |