With the development of society, the audible noise in substation has been paid more and more attention, so it is necessary to study the substation’s audible noise. Firstly, the noise of 110 kV Heshangshan Substation in Chongqing was measured with sound level meter, and the audible noise in substation was predicted by using GM(1,1) model, with which the monthly variation of the noise can be obtained. Then, the noise in different temperatures and distances was predicted by using radical basis function (RBF) neural network. Finally, the audible noise in different temperatures and operation loads of transformer was calculated with using RBF neural network, which could provide references for the design of new substations.