人工神经网络在轻板隔墙隔声预计中的应用

轻钢龙骨薄板墙(简称轻板隔墙)的隔声性能由于其构造的复杂性,没有简单公式可以预计其隔声量。在大量实验结果的基础上,利用人工神经网络对一些常见的轻板墙组合构造进行计权隔声量(RW)的预计。根据相关分析和主成份分析的结果,选用板材面密度、墙体总厚度、腔内吸声层厚度、龙骨型式和垫层等因子作为网络的输入变量,由此进行隔声性能的预计,可达到较满意的结果。

Application of Artificial Neural Networks for Predicting the Sound Insulation of Lightweight Panel Walls

It is difficult to predict the sound insulation of the lightweight panel wall with a simple equation due to the complicity of the wall construction. Based on numerous measuring results of sound insulation, artificial neutral network analysis can be used for predicting the weighted sound insulation index(RW) of different lightweight panel walls constructions. Results from correlation analysis and main components analysis show that the major factors related the weighted sound insulation index are surface density of the panel, total thickness of the panel wall, thickness of absorbent in the cavity, type of stud, and type of sound bridge between stud and panel. It is shown that predictions are well agreed with the measurements.