The measurement of resistive current for metal oxide arrester（MOA） is affected by environmental interference

high-frequency noise interference

and pulse interference. The measured resistive current signal inevitably contain a variety of strong interference components. In order to cope with the difficulty of traditional methods for eliminating resistive current’s interference in a combination of many interference

this paper proposes a new method for eliminating this interference which can affect the resistive current of MOA based on EMD-WPT-SVD and exponentially weighted average.Firstly

according to whether the fundamental frequency

sampling frequency and characteristic frequency of resistive current are immersed by inter-harmonic

the optimal numbers of decomposition layers and wavelet basis function are selected. Secondly

the MOA’s resistive current is performed by EMD

and then the wavelet packet transform is used to make each sub-signal to only contain one characteristic frequency

so that the characteristic components of all sub-signals can be accurately extracted by SVD. Finally

when the periodic interference of the resistive current is eliminated accurately from the original signal

the exponential weighted average algorithm is used to obtain the changing trend of resistive current that can truly reflect the MOA working condition. The method of this paper is compared with many classical methods in simulation and instance

respectively. It is proved that the proposed method can eliminate many kinds of interference and reflect accurately the real-time condition of MOA. The processing result meets the requirements of practical engineering measurement.