In light of poor regularity and predictability of RMS value of node voltage in power distribution networks,this paper proposes to apply the ARMA model to predict the RMS value,which mainly comprises of data preprocessing,fitting training of AMRA,fitting review of AMRA,and forecasted application of ARMA.This method is realized through using Python.ARMA fitting training is performed on two randomly selected 10 kV RMS value monitoring sequences,before conducting analysis with the model generated by the training,which demonstrates that root-mean-square errors between the two predicted sequences and actual values are 9.57 and 5.05 respectively.Therefore,method proposed in this paper is applicable in performing voltage RMS value predictions,with reliable effectiveness and practicality.