基于ARMA模型的振动信号建模与预测

振动信号是反映系统故障的重要信息,预测振动信号的趋势是系统状态监测与故障诊断中的一个重要内容。本文提出了一种采用时间序列模型来分析和预测非线性随机振动信号的方法,建立了描述振动信号的数学模型。首先将现场测得的非平稳振动信号转化成标准正态平稳时间序列,然后利用这些参考信号建立时间序列模型,并采用非线性最小二乘法进行模型参数估计,最后利用该模型对信号进行预测。应用结果表明该模型能够很好地拟合振动信号时间序列,并取得了一定的预测精度,可以达到预测要求。

Modeling and forecasting of vibration signals based on ARMA Model

Vibration signals are the important information for system failures.Forecasting the trend of vibration signals is an important content of condition monitoring and fault diagnosis.This article presents a method to analyze and predict nonlinear random vibration signals by time series model and establishes mathematical models to describe vibration signals.Firstly,the non-stationary vibration signals acquired in the field are transformed to stationary time series.Secondly,the time series models are constructed from the selected reference signals and nonlinear least square method is used to estimate models＇ parameters.Finally,the vibration signals are forecast by using the models.The application results show that the models can simulate time series of vibration signals quite well with good prediction accuracy and meet the need of forecasting.