Research On Fault Trend Prediction Method Of Rotating Machinery Based On Full Vector AR Model

For fault trend prediction of large-scale rotating machinery, the traditional AR prediction model is based on single-source information. In fact, due to the whirling motion characteristics of the rotor, the structure of the frequency spectrum at the same cross section is different in the different directions. As a result, according to fault trend prediction to single-source information, the result is different. To ensure the reliability and uniqueness of the predicting result, full vector AR prediction model,combining with full vector spectrum technology and the AR prediction model, is constructed. Because AR model and full vector AR model have the characteristics of high prediction accuracy in short term and low prediction accuracy in long term, the Kalman filtering is introduced into AR model and full vector AR model to improve the prediction accuracy. Finally, their effectiveness is verified by examples.The main research work is as follows:(1) Research on AR Model based on full Vector Spectrum and its application in the fault trend prediction of the rotor system of large-scale rotating machinery. The specific theoretical calculation formula and its corresponding fault trend prediction flowchart are given. The experimental results show full Vector AR Model can predict fault tendency in the rotor system of large-scale rotating machinery effectively and ensure the prediction result’s reliability effectively.(2) Research on AR Model based on the Kalman Filter and its application in the fault trend prediction of the rotor system of large-scale rotating machinery. The AR-Kalman prediction(ARKF) method, combining with the Kalman Filter and the AR model, is proposed. The specific theoretical calculation formula and its corresponding fault trend prediction flowchart are given. The experimental results show the proposed method can predict fault tendency of large-scale rotating machinery effectively and improve the linear prediction accuracy of traditional AR model obviously.(3) Research on full vector AR Model based on the Kalman Filter and its application in the fault trend prediction of the rotor system of large-scale rotating machinery. The full vector AR Kalman forecast(FARKF) Model and the full vector AR Kalman correction(FARKC) Model, combining with full vector spectrum and AR model and the Kalman filter, are proposed. The specific theoretical calculation formula and its corresponding fault trend prediction flowchart are given. The experimental results show both of the proposed methods can predict fault tendency of large-scale rotating machinery effectively and improve the linear prediction accuracy of full vector AR model obviously.