论文部分内容阅读
为了解决综采工作面煤矸界面探测问题,提出了利用煤矸下落冲击钢板的振动特征来探测煤(?)界面的新方法。煤矸振动信号表现出非平稳特征,采用EMD方法可以将复杂矿井环境下的煤矸振动加速度信号分解成固有模态分量,每个模态分量都包含了特有的时间尺度特征。将包含煤矸振动特征的前6个IMF分量的能量,结合均值、方差及峭度等时域特征值,构成9维特征模式,作为SVM分类器的输入进行训练及分类。结果表明,基于Hilbert-Huang变换的IMF分量的能量特征能够反映煤矸振动特征的差异,SVM分类方法能够准确判断煤矸混合状态。
In order to solve the problem of coal gangue interface detection in fully mechanized mining face, a new method to detect the coal (?) Interface by using the vibration characteristics of coal gangue drop impacting steel plate is proposed. The coal gangue vibration signal shows non-stationary characteristics. The EMD method can decompose the coal gangue vibration acceleration signal in complex mine environment into the natural modal components. Each modal component contains the unique time-scale features. The energy of the first six IMF components, which contains the vibration characteristics of gangue, is combined with the time-domain eigenvalues, such as mean, variance and kurtosis to form a 9-dimensional feature model, which is trained and classified as the input of the SVM classifier. The results show that the energy characteristics of IMF components based on Hilbert-Huang transform can reflect the difference of coal gangue vibration characteristics, and the SVM classification method can accurately determine the coal gangue mixing status.