Development of vibration signature analysis using multiwavelet systems

In this paper the vibration analysis for local faults and transient phenomena detection, using multiwavelet systems, is developed. Unlike the scalar wavelet systems, in which their coefficients are scalar parameters, the transformation coefficients of multiwavelet systems are vector valued, and their calculation requires specialized techniques. In this investigation, having considered the technique used to obtain the scalar wavelet system coefficients, the transformation coefficients of the multiwavelet system are calculated, and by applying the method to artificial vibration signals, decomposition of the signal into different multiscale and multiwavelet functions (as introduced by Donovan, Geronimo, Hardin and Massopust) is examined, as well as the capability of this multiwavelet system for transient phenomena detection. By analyzing the vibration signal of a faulty gearing system the applicability of Donovan, Geronimo, Hardin and Massopust multiwavelet system for local fault detection of the mechanical systems is shown. The results confirm that using the multiwavelet system, not only can the fault in the gearing system be diagnosed, but also its location can be determined precisely.