基于爆炸声音识别的煤矿瓦斯和煤尘爆炸感知方法研究

Research on perception method of coal mine gas and coal dust explosion based on explosion sound recognition

  • 摘要: 分析了煤矿瓦斯和煤尘爆炸特征:气体浓度发生突变;环境温度迅速升高;空气压力突然升高;产生火球和烟尘;产生较强的红外和紫外辐射;产生爆炸冲击波和火焰波;产生爆炸音。基于爆炸声音感知煤矿瓦斯和煤尘爆炸具有以下优点:① 爆炸冲击波和火焰波衰减快,传播距离近;声波衰减慢,传播距离远。远离爆源的矿用拾音设备可用于煤矿瓦斯和煤尘爆炸感知。② 与基于气体浓度和温度等传感器的煤矿瓦斯和煤尘爆炸感知方法相比,具有响应速度快的优点。③ 与基于视频图像的煤矿瓦斯和煤尘爆炸感知方法相比,具有不受粉尘、光照、遮挡等影响的优点。④ 矿用拾音设备成本低、易安装。⑤ 声音传播距离远,受巷道和分支影响小。⑥ 声音处理速度快,可在短时间内从各种声音信号中快速识别瓦斯和煤尘爆炸声音。提出了基于爆炸声音识别的煤矿瓦斯和煤尘爆炸感知方法:利用麦克风阵列拾音器采集监测区域的声音信号,经过归一化、分帧、添加类别标签等预处理后,提取声音信号特征,将特征输入到统计分类器中进行训练,建立煤矿瓦斯和煤尘爆炸声音识别模型;实时采集监测区域的声音信号,将提取的声音信号特征输入训练完成的煤矿瓦斯和煤尘爆炸声音识别模型中,判断是否为煤矿瓦斯和煤尘爆炸声音,若是则进行报警。

     

    Abstract: The characteristics of coal mine gas and coal dust explosion are analyzed. The gas concentration changes suddenly. The ambient temperature rises rapidly. The air pressure rises suddenly. It produces fireballs and smoke. It produces strong infrared and ultraviolet radiation. It generates explosion shock waves and flame waves. It produces explosive sound. The coal mine gas and coal dust explosion perception based on explosion sound has the following advantages. ① Explosion shock waves and flame waves attenuate quickly and travel close distances. Sound waves attenuate slowly and travel over long distances. The mine sound pickup equipment far away from the explosion source can be used for the perception of coal mine gas and coal dust explosion. ② Compared with the coal mine gas and coal dust explosion perception method based on gas concentration and temperature sensors, the proposed method has the advantage of fast response. ③ Compared with the coal mine gas and coal dust explosion perception method based on video images, the proposed method has the advantages of not being affected by dust, light, shelter, etc. ④ Mine sound pickup equipment has low cost and is easy to install. ⑤ The sound travels over a long distance and it is less affected by roadways and branches. ⑥ The sound processing speed is fast. The gas and coal dust explosion sound can be quickly recognized from various sound signals in a short time. A perception method of coal mine gas and coal dust explosion based on explosion sound recognition is proposed. The sound signals in the monitoring area are collected using microphone array pickups. After preprocessing such as normalization, framing, and adding category labels, the sound signal features are extracted. The features are input into a statistical classifier for training. The sound recognition model for coal mine gas and coal dust explosion is established. The sound signal of the monitoring area is collected in real-time. The extracted sound signal features are input into the trained coal mine gas and coal dust explosion sound recognition model. Whether it is the coal mine gas and coal dust explosion sound can be determined. If so, an alarm will be given.

     

/

返回文章
返回