提出一种基于FFT谱分解进行噪声倍频程谱分析和插值法求计权声级的算法,以解决传统1/3倍频程谱分析和计权声级算法复杂且精度不高的问题.深入分析了基于FFT频谱求1/3倍频程谱、噪声频谱幅值还原成实际声压值和通过加窗FFT减少幅值误差的方法,研究了计权网络的插值应用方法,并对总声压级求取过程中的易错点进行了剖析.噪声分析试验结果验证了所提出的算法具有精度高、速度快等特性.
To solve the intricate and lower accuracy problem in the conventional 1/3 octave spectrum and weighting sound level algorithm, a new method was proposed, based on the decomposition of spectrum and weighted network interpolations. There was an in-depth analysis on obtaining 1/3 octave spectrum based on FFT spectrum, the relationship between spectral amplitude and time-domain amplitude and the way of reducing amplitude error by windowed FFT. The interpolation method of weighting network was illustrated, and the er- ror-prone points in the entire calculation process were pointed out. The experimental results verified that the developed method was of high-accuracy and high-speed.