Abstract-Those motion parameters that cannot be recovered from image measurements are unobservable in the visual dynamic system. This paper studies this important issue of singularity in the context of kernel-based tracking and presents a novel approach that is based on a motion field representation which employs redundant but sparsely correlated local motion parameters instead of compact but uncorrelated global ones. This approach makes it easy to design fully observable kernel-based motion estimators. This paper shows that these highdimensional motion fields can be estimated efficiently by the collaboration among a set of simpler local kernel-based motion estimators, which makes the new approach very practical.

In order to predict the physical characteristics of the large vibrating screen from its scale-down model, the similarity ratios of the frequency response functions, mode shapes, and dynamic stresses between the prototype and the scale model screen are built according to the similarity theory. The natural frequencies and modal shapes are extracted from the frequency response function by means of modal tests, in which the relative error of the natural frequencies is less than 9% and the modal shapes are consistent between the prototype and the model. The operating condition parameters including dynamic stress, displacement, velocity, and acceleration were also measured and conform to the similarity criteria. The results show that the inherent and operating condition parameters of the large vibrating screen can be obtained from the scale-down model conveniently, which provides an effective method for structural optimization and substructure coupling analysis of the large vibrating screen.

This paper presents a novel stall detection method based on symmetrized dot pattern (SDP) analysis, which detects the starting point of rotating stall timely and accurately during operations of centrifugal fans. To demonstrate the proposed method, experiments were first performed on a G4-73 No. 8D centrifugal fan to measure the aerodynamic pressure signals of the air flow inside the fan during the gradual development of rotating stall. Then, the SDP technique was used to analyze the pressure signals and extract the time-domain characteristics of the pressure signal during the gradual development of rotation stall. Finally, a comprehensive autocorrelation coefficient was defined and employed as the index for real-time stall detection. In addition, to verify the accuracy of detection results, the tested signals were also analyzed off-line by wavelet transform to detect the actual starting point of rotating stall. The comparison results show that the stall detection method based on the symmetrized dot pattern (SDP) analysis can accurately detect the starting point of rotating stall in centrifugal fans within a short period of 0.05 s.

Abstract

Abstract-Those motion parameters that cannot be recovered from image measurements are unobservable in the visual dynamic system. This paper studies this important issue of singularity in the context of kernel-based tracking and presents a novel approach that is based on a motion field representation which employs redundant but sparsely correlated local motion parameters instead of compact but uncorrelated global ones. This approach makes it easy to design fully observable kernel-based motion estimators. This paper shows that these highdimensional motion fields can be estimated efficiently by the collaboration among a set of simpler local kernel-based motion estimators, which makes the new approach very practical.