Structural optimization of Fibonacci micro-perforated plate broadband absorber by artificial neural networks

In response to the low-frequency broadband sound absorption problem of micro-perforated plates, a structure coupling a micro-perforated plate with a Fibonacci spiral cavity is proposed. The expression for the surface acoustic impedance of this structure is derived, and the reliability of the model is verified through finite element simulation. In structural design, a neural network model has been developed to predict the mapping relationship between structural parameters and sound absorption coefficients, with a mean squared error of less than 1.5 × 10⁻⁴. Additionally, an inverse design method for sound absorbers based on a tandem neural network has been proposed. The optimization time of this method reaches the millisecond level, significantly improving the design efficiency. The results indicate that the broadband sound absorber with structural thickness of 49.65 mm achieves an average sound absorption coefficient of 0.95 within the frequency range of 500−1500 Hz. Finally, the feasibility of this method has been verified through experiment.