The invention discloses a mine hoist fault acoustic analysis method based on MFCC-CNN. The method comprises the following steps of carrying out pre-emphasis, framing, threshold endpoint detection and windowing preprocessing on a collected audio signal during operation of a hoist; performing short-time Fourier transform, Mel filter filtering and discrete cosine transform to complete MFCC feature parameter extraction of the audio signal; constructing an eight-layer CNN model including an input layer; and training the CNN model and hoisting fault diagnosis. The the method is advantaged in that the characteristics of high acoustic signal acquisition speed, simple acquisition equipment and the like are utilized, real-time online monitoring and analysis of the state of the elevator can be completed, and intelligence of the fault diagnosis of the elevator is realized in combination with a neural network analysis method.
本发明公开了一种基于MFCC‑CNN的矿井提升机故障声学分析方法,包括对采集的提升机运行时音频信号进行预加重、分帧、门限端点检测和加窗的预处理;进行短时傅里叶变换、Mel滤波器滤波和离散余弦变换,完成音频信号的MFCC特征参数提取;构建包括输入层在内的八层CNN模型;对CNN模型进行训练以及提升机故障诊断等步骤,本发明具有诸多优势,利用了声学信号采集速度快,采集设备简单等特点,能够完成提升机状态实时的在线监测与分析,结合神经网络分析方法实现了提升机故障诊断的智能化。
MFCC-CNN-based mine hoist fault acoustic analysis method
一种基于MFCC-CNN的矿井提升机故障声学分析方法
2021-07-30
Patent
Elektronische Ressource
Chinesisch
CS-MUSIC-based mine hoist fault source positioning analysis method
Europäisches Patentamt | 2022
|Engineering Index Backfile | 1949
|