بهینه سازی عددی و تحقیق تجربی بر روی محیط گوش دادن خدمه مبتنی بر شبکه عصبی

With greater noises in the cabin, a greater impact would be acted on the listening system of crew who lived in the cabin for a long time. The neural network and the listening environment of the cabin were optimized in the paper to improve the listening capacity of crew. Firstly, the periodic sound insulation package was proposed to make a comparison with the traditional sound insulation package to verify its advantage for sound insulation performance of the structure in mid-low frequency. Later, numerical simulation was applied to analyze the influence of duty cycle and periodic number on the transmission loss of the structure. In order to obtain the structure with an optimal transmission loss, the neural network was adopted to predict and optimize the periodic sound insulation package. In the meanwhile, boundary element method was adopted to predict the interior noise in the cabin. In order to analyze noises in the cabin, panel contribution analysis was conducted for the panel of the cabin. Finally, the periodic sound insulation package was applied on these panels to improve the noise in the cabin. It was found that the periodic sound insulation package can effectively control the vibration energy of these panels at these frequency points and reduce noise in the cabin. Finally, the noise in the improved cabin was tested to compare with the result of the numerical computation and obtain good consistency. It indicated that the numerical simulation was reliable. The low frequency noise in the cabin could really be reduced effectively through laying the periodic sound insulation package on the cabin panel. Finally, the listening capacity and environment of crew were obviously improved within the cabin.