基于AR模型的磨机振动信号特征提取方法

摘要

To study the fill-up rate and material-to-ball ratio of the ball-mill in grinding process, AR model for feature extraction of the bearing's vibration signals of the mill was proposed. With a testing mill as the object, the ball-mill bearing vibration signals were collected. By analyzing the correlation of time domain signal characteristics and parameters of the mill load identify the type of mill load (underload, normal, overload);to get the frequency domain character reflect the mill load ,the vibration signal of different mill load state were estimated into power spectrum based on AR model, which provide the evidence for the mill load forecasting.%磨矿过程中的球磨机筒体内部负荷(填充率、料球比)研究是选矿设备节能降耗的重要内容.以试验球磨机为对象,通过采集轴承座振动信号,采用AR模型对振动信号进行特征提取和进行功率谱估计,研究了5种充填率条件下的磨机负荷参数与信号时域特征的相关性,得出随着磨矿过程中筒体中钢球、物料的变化,低频段、高频段的频谱能量值曲线的相应变化规律.研究表明,有量纲时域特征参数可以很好地表征特征信号与负荷状态参数的相关性;无量纲时域特征参数可以很好地解释振动时域信号的波形分布以及振动冲击特征,根据频谱能量分布变化规律,可提取能够表征其对应磨机负荷状态的振动特征,为磨机负荷预测提供依据.