Real-time stall detection of centrifugal fan based on symmetrized dot pattern analysis and image matching

Abstract

Abstract This paper presents a stall detection method based on symmetrized dot pattern (SDP) analysis and image matching, which can timely and accurately detect therotating stall of centrifugal fan in real time. Firstly, the experiments of gradual development of rotating stall were conducted on the G4-73No.8D centrifugal fan and the aerodynamic pressure signals of the air flow inside the fan were recorded. Then, the SDP technique was used to analyze the pressure signals and the SDP pattern templates under normal and stall operation state were established. After that, the gradual development pressure signal within 80 sampling points was analyzed by the SDP technique to obtain the real-time SDP pattern, which was matched with the templates. By repeating the above steps every 16 sampling points until the fan operated from normal to full stall state, the real-time fan stall detection was achieved. Finally, in order to strengthen the individual features of operation state and weaken the common features of interference factors, two optimization approaches were proposed. For one thing, the wavelet noise reduction was utilized to denoise the pressure signal before it was analyzed by the SDP technique. For another, the principal component analysis (PCA) was utilized to extract the important information and remove the up-important information of the SDP pattern before image matching. Besides, to verify the accuracy of the detection results, the tested signals were analyzed by wavelet transform to off-line detect the rotating stall. It was shown that the method based on the SDP analysis and image matching, which was optimized by the PCA technique, could detect the starting point of rotating stall within 21 sampling points (0.0656\u202fs), which is 96 sampling points (0.3\u202fs) faster than the method optimized by wavelet noise reduction. This confirms that the proposed method can meet the need for real-time rotating stall detection of centrifugal fan timely and accurately.