基于声发射信号与BP神经网络的煤粉粒径识别研究

煤粉粒径的测量是燃煤电站一项重要的工作。针对目前筛分法存在的缺点,提出了一种结合声发射信号与BP神经网络在线识别煤粉粒径的方法。在频域中对噪声信号与煤粉声发射信号进行比较,确定了信号中反映煤粉粒径的频率区间,并利用小波包置零方法对信号进行去噪,在信噪比与信号平滑度方面比较了几种常用小波函数的去噪效果。通过功率谱分析发现了信号能量随煤粉粒径的变化特征。最后提取信号能量特征,利用BP神经网络对煤粉粒径进行识别。研究结果表明,结合声发射信号与BP神经网络识别煤粉粒径,可以获得良好的效果。

Coal particle size recognition based on acoustic emission signal and BP neural network

The measurement of coal particle size is an important task for coal-fired power stations. Aiming at the shortcomings of the current sieving method, a method combining on-line recognition of coal particle size with acoustic emission (AE) signal and BP neural network is proposed. The characteristics of the background noise and AE signals were compared in the frequency domain, and the frequency interval related to the particle size was confirmed in the signal. The wavelet packet zeroing method was used to de-noise the AE signal, and the de-noising performance of different wavelet function was compared in terms of signal-to-noise ratio and signal smoothness. Through the power spectrum analysis, the characteristics of signal energy with the particle size were found. Finally, the signal energy characteristics were extracted, and BP neural network was used to recognize the particle size. The research indicates that the acoustic emission technology and BP neural network can be used to monitor the coal particle size.