主元分析的振动频域特征识别与磨机负荷建模研究

Feature Recognition for Principal Component Analysis' Vibration Frequency Domain and Mill Load Modeling

球磨机是矿石破磨的关键设备,由于运行中筒体内负荷(ML)难以有效检测,运行状态参数的预报控制无法进行。研究通过检测筒体轴承座产生的振动信号,采用Welch法对其进行功率谱估计,提取信号的特征频谱段的能量值,分析信号功率谱与磨机筒体内负荷之间的关系,采用主元分析法(PCA)对振动谱能量值进行降维,得到与负荷高度相关的能量谱成分,最后利用支持向量机(SVM)建立磨机负荷参数分类模型,实现磨机负荷参数(填充率、料球比)的预测。试验结果表明:该方法可以准确识别磨机筒体内负荷,为状态参数的预报及控制技术的优化提供理论指导。

Ball mill is the key equipment of ore grinding. The forecast operating condition parameters can not be controlled for the operation of the cylinder body load （ML） is difficult to be detected effectively. By detecting the vibration signal bearing barrel produced, the signal characteristic spectrum segments are extracted using the energy value Welch power spectrum estimation method. The relationship between the signal power spectrum and mill shell load is analyzed. The prediction of mill load parameters is realized by establishing mill load parameters model by SVM. The results showed that the technology can effectively identify mill load, which provides prediction for state parameters and control technology optimization.