For the problem of online monitoring parameters of metallic oxide arrester（MOA） have suffered from the external environment interference in power system

a novel MOA parameters modified method by eliminating external environmental factors interference was proposed. Firstly the phase calculated by all phase fast Fourier transform（FFT） method was modified to obtain high precision phase. Then the BP neural network based on Levenberg-Marquardt algorithm was used to build three-dimension surface modeling of MOA parameters with temperature and humidity

and the BP neural network was used to approximate the nonlinear model. Finally

the MOA parameters were modified. The simulated results show that the modified all phase FFT method preserved the property of phase flatness

and avoided the problem of phase deviation in engineering measurement

the modified phase deviations of fundamental and 3 rd harmonicis 7.2° and 21.6°. The calculated fluctuation value of total current

phase difference and resistive current using the engineering measured data were reduced by 78.7%

68.0% and 71.7% respectively

thus maintaining the stationary of MOA parameters and avoiding the parameter fluctuation caused by environmental interference

which verified the accuracy and effectiveness of the proposed modified method. The proposed method can eliminate the actual environmental interference

and realize accurate monitoring of MOA parameters in engineering fields.