Application of artificial neural networks to acoustic composites: A review

Artificial neural networks (ANN) provide a method for accurately predicting acoustic properties of materials. The most commonly used artificial neural network scheme in acoustic research is the backpropagation neural network (BPNN), but there are several other types of neural networks, including radial basis function neural networks (RBFNN) and deep neural networks (DNN). This paper reviews the connection between acoustics and neural networks using VOSviewer and briefly introduces the concepts of neural networks and acoustic mechanisms. Some acoustical materials, such as porous acoustic absorbers, acoustic insulators and damping and vibration damping materials, are listed. Finally, the applications of neural networks in acoustics are categorized and the future prospects of neural network acoustic prediction are discussed for further improvement of neural network based acoustic prediction in the future.