基于图像识别的矿井煤与瓦斯突出预警研究

Study on Early-warning of Coal and Gas Outburst in Mines Based on Image Recognition

针对现有矿井煤与瓦斯突出预警方法存在的不足,提出了一种基于图像识别的矿井煤与瓦斯突出预警方法。首先,基于VGG16网络建立瓦斯浓度、声发射强度和电磁辐射强度变化的煤与瓦斯突出危险性图像识别模型。然后,基于Dempster-Shafer(D-S)证据理论对3种模型的突出危险性识别结果进行融合处理,消除预警的不确定性,实现准确预警。最后,利用来自生产现场的数据所生成的瓦斯浓度、声发射强度、电磁辐射强度变化图像进行实验。实验结果表明,所提方法的预警准确率高、预警速度快,其预警效果优于峰谷比值法和Res Net50方法。实验结果证明了所提方法的有效性。

In view of the problems existing in the early-warning methods of coal and gas outburst in mines,an early-warning method of coal and gas outburst based on image recognition is proposed.Firstly,based on VGG16 network,the outburst risk image recognition models of gas concentration,acoustic emission intensity and electromagnetic radiation intensity are established respectively.Then,the Dempster-Shafer(D-S)theory of evidence is used to fuse the outburst risk recognition results of the three models for reducing the uncertainty and achieving accurate early-warning.Finally,the images of gas concentration,acoustic emission intensity and electromagnetic radiation intensity generated by the data from the production site are used for experiments.The experimental results show that the proposed method has high early-warning accuracy and fast warning speed,and its warning effect is better than that of the peak-valley ratio method and the Res Net50 method.The experimental results demonstrate the effectiveness of the proposed method.