全矢AR-Kalman滤波的机械故障趋势预测方法研究

Research of Full Vector AR and Kalman Filtering on Fault Trend Prediction Method of Machinery

旋转机械故障趋势预测中,为提高预测结果的可靠性和精确性,提出一种基于全矢谱技术的AR-Kalman滤波方法。该方法采用全矢谱技术对故障特征进行双通道提取,相比于传统单通道信号提取,该技术保证了信息提取的全面性和完整性,采用Burg算法对AR模型参数进行估计,采用卡尔曼滤波对参数进行修正,同时,对预测结果进行实时修正。实验结果表明,全矢AR-Kalman滤波方法能有效地对旋转机械的故障趋势进行预测,新方法明显提高了传统AR模型在中长期预测的精度。

To improve the reliability and accuracy of the prediction in vibration fault trend prediction of a rotating machinery,a new prediction method,combining with Full vector spectrum and AR model and the Kalmanfiltering,is proposed.The new method uses the full vector spectrum technology to extract dual channel fault features,which can ensure the comprehensiveness and integrity of information extraction compared to the traditional single channel signal extraction.And it uses the Burg algorithm to estimate parameters of AR model.It is that the Kalman filter is used as a real-time corrector to correct not only the parameters estimated but also the prediction results of AR model.The experimental results show the proposed method can predict fault trend of a rotating machinery effectively and improve the prediction accuracy of traditional AR model in the medium and long term forecasting obviously.