Signal analysis of rolling mill vibration based on morphological undecimated wavelets and S-transform

A algorithm was proposed to distinguish the initial vibration of rolling mills in tandem rolling.Firstly, a new morphological un-decimated wavelet(MUDW) based on a multi-scale average filter is developed according to the MUDW's general structure.Then the relationship between the new MUDW scale and real frequency is found by simulation experiment.Finally the time information of fault frequency from rolling mill vibration signals is extracted using the algorithm based on the new MUDW and S-transform.