Research On Fault Diagnosis And Condition Predication Of Rotating Machinery Based On Order Tracking And AR Model

Vibration signal testing takes an important part in faulty diagnosis and status monitoring for rotating machinery. In the engineering application, the rational signal processing methods are adopted to extract the virtual information from the acquired signal. Aiming at the rotating machinery, the diagnosis technique based on order analysis and the state prediction method based on Autoregressive (AR) model are proposed.The non-stationary phenomenon is one of the most significant features in the raising and reducing speed process. This paper introduces the quadratic curve fitting theory to calculate the re-sampling time, as well as the interpolation filter to get the angle-domain signal in the order re-sampling course.To obtain the value of the rotational speed in the order analysis, a novel method based on first order numeral differential, Lagrange interpretation and the piecewise least square technique with cubic spine curve fit is put forward. This method can eliminate the errors and faults of computing rotating speed, so precision of rotating speed computation is largely improved.After the mechanism study of Autoregressive Movingaverage (ARMA) model, Movingaverage (MA) model and AR model, the parameter estimation of AR model based on generalized inverse matrix is proposed. The method estimates the parameters of AR model under the non-solution circumstance of linear equation set and fixes the model order through the Akaike information criterion (AIC). The application has presented its advantages in the operating condition prediction.As the carrier of the key technology, the on-line monitoring system of the machining center based on Visual C++ is developed. The above algorithm research on order analysis and AR model state prediction has provided the technology support and guarantee for equipment status monitoring.