基于支持向量机集成算法的煤矿顶板状态检测

Detection of coal mine roof based on support vector machine ensemble

针对小样本、多类煤矿顶板的状态检测问题,提出了基于支持向量机集成的算法。采用深度优先搜索对支持向量机集成参数进行优化,并结合煤矿顶板敲击声信号的人耳听觉功率谱特征,对煤矿顶板中浮石、剥层等现象进行分类识别。实验表明,该算法能够对多类顶板状态进行有效分类,且识别率较高,能够作为安全开采的保障手段。

In this paper, a method based on support vector machine ensemble about the detection of Many types of coal mine roof is proposed. In optimization process, the depth-first search is used. The classification and recognition of security risk is according to power spectral characteristics of human ear of mine roof percussion signal. The experiments show that the algorithm can classify the security risk of coal mine roof effectively.