Surge detection of magnetically suspended high-speed centrifugal blower

For the surge problem existing in a magnetically suspended high-speed centrifugal blower, a surge detection method based on a magnetic bearing was proposed. A universal co-frequency wave trap was adopted to filter the co-frequency components in the process of rotor displacement for eliminating the influences of co-frequency perturbation on surge detections. By analyzing effects of different convergence factors on estimated functions of surge frequency, an adaptive estimation algorithm with variable convergence factors was proposed to reduce the interference of low frequency signals and to improve the quick reaction capacity to surge signal detection. Finally, the convergence of the proposed detection method was analyzed, and a verification experiment was performed on a test system for the 100 kW magnetically suspended centrifugal blower. The experimental result shows that the improved surge detection algorithm detects premonitory signals of surge rotating stall before the surge occurs and detects surge signals within 0.23 s after the surge occurs, which shortens the detection time by 2.6 s as compared with that of common algorithms. The proposed algorithm is characteristics by simple, high velocity, less calculation and can be implemented without other auxiliary systems. Moreover, it can reflect the changes of frequency and amplitude during the surges.