基于双重降噪和改进SVM的煤岩界面识别方法

Coal-rock Interface Identification Method Based on Double Noise Reduction and Improved SVM

针对采煤机在实际开采过程中所采集的信号含噪量较高且煤岩界面识别困难的问题,提出了一种基于双重降噪和改进支持向量机(SVM)的多传感信息融合的煤岩界面识别方法。首先构建模拟采煤机,采集采煤机截割不同比例煤岩时产生的5种振动信号作为样本信息库;其次利用小波阈值和互补集合经验模态分解(CEEMD)2种方法完成传感器信号的降噪过程,并进一步实现信号多维特征向量的提取和降维;将特征阵输入基于改进SVM和DS证据理论的煤岩界面识别模型中,最终获得理想的识别精度,实现对煤岩界面煤岩比的分类识别。

Aiming at the problem that the signal collected by the shearer in the actual mining process has high noise content and the coal-rock interface identification is difficult, a coal-rock interface identification method based on double noise reduction and improved support vector machine(SVM) was proposed. Firstly, built a simulated shearer and collected five kinds of vibration signals generated when the shearer cuts different proportions of coal and rock as the sample information database. Secondly,the noise reduction process of the sensor signal was completed by two methods of wavelet threshold and complementary ensemble empirical mode decomposition(CEEMD), and the extraction and dimension reduction of the multi-dimensional feature vector of the signal were further realized. The characteristic array was input into the coal-rock interface recognition model based on the improved SVM and DS evidence theory, and finally the ideal recognition accuracy was obtained, and the coal-rock ratio classification and recognition of the coal-rock interface were realized.