Surge Detection Approach for Magnetically Suspended Centrifugal Compressors Using Adaptive Frequency Estimator

The active surge suppression is a challenging topic in the industrial compressor systems. This paper addresses the two issues on the surge detection approach for magnetically suspended centrifugal compressors (MSCCs) utilizing the signal monitoring capabilities of magnetic bearings. An adaptive frequency estimator is proposed to extract the surge signal, and a judgment criterion is establish to the surge state. First, the derivation of a surge disturbance transfer model for the fluctuating pressure and rotor displacement was presented, and the characteristics of surge signal in the MSCC system is analyzed. Then, a conventional notch filter (NF) is used to eliminate the synchronous disturbance in the surge signal, and a modified adaptive notch filter (ANF) method is used to estimate the frequency of surge signal. The convergence analysis of the modified ANF shows that the linearized system is asymptotically stable. The mild surge detection technology and judgement criterion are also discussed. Experimental results on both the 100 kW and 315 kW MSCC systems show that the proposed method can rapidly detect the surge before the MSCC systems develops into the modified surge.