基于全矢谱-ARMA模型的机械振动强度预测研究

摘要

Single channel prediction method due to the imperfect defects of vibration information, lead to prediction results having poor consistency, thus it cannot realize fault prediction.In view of the imperfect defects of the single channel information, combing time-series forecasting method ARMA model with full vector spectrum technology based on the homologous information fusion technique, full vector spectrum-ARMA model prediction method was proposed.Full vector spectrum was made up of information fusion of two mutually perpendicular channel, and the spectrum structure of the same cross section could not change with a pair of sensors installation location.The spectrum structure based on full vector spectrum has uniqueness, so this method can realize the prediction and predictive results have good consistency, and so that they can realize the fault prediction.Experiments show that this method has good prediction results and practical.%单通道预测方法由于获取振动信息不完善的缺陷,导致预测结果一致性差,从而不能实现故障的预测.针对单通道信息不完善的缺陷,将时序预测重要方法ARMA模型结合基于同源信息融合技术的全矢谱技术,提出了全矢谱-ARMA模型预测方法.全矢谱是由互相垂直的两个通道获取的信息融合而成,在同一截面其频谱结构不会随着一对传感器安装的方位不同而变化.通过全矢谱获得的频谱结构具有唯一性,因此该方法能够实现预测并且结果一致性好,从而能够实现对故障的预测.实验表明:该方法预测结果好,并且比较实用.