Large-capacity high-frequency transformers (HFT) are typically under the excitation of high-frequency square wave voltage

and a large number of voltage harmonic components will cause a large number of high-order vibration components in the core vibration process of HFT

which in turn will cause a significant increase in the sound pressure level at high-order harmonics

resulting in serious noise pollution. To solve this issue

a new multi-layer resonance composite sound-absorbing structure was first designed. The sound absorption performance of six different sound-absorbing structures was compared and analyzed by using the transfer matrix method. Subsequently

a three-dimensional electromagnetic-structure-acoustic simulation model for HFT was established to analyze characteristics of the sound field distribution in the presence and absence of the composite sound-absorbing device. On this basis

a method combining central composite experimental design with finite element simulation was adopted to obtain the sound field simulation results of the sound-absorbing device under different structural parameters. Then

a radial basis function (RBF) neural network model was established to predict sound pressure levels

and global sensitivity analysis technology is used to study the influence degree of the structural parameters of the sound absorbing device on the sound pressure level. Finally

the sparrow search optimization algorithm fused with Sine-cosine and Cauchy mutation was applied to obtain the optimal design parameters of sound-absorbing device and its was verified through simulation. The simulation results show that the noise suppression effect of the optimized sound-absorbing structure is significant

and the sound pressure level is decreased by 10.176 dB compared to the non-optimized structure. Therefore

it is verified that the multi-layer resonance composite sound-absorbing structure has a good noise suppression effect on high-frequency transformers.