Oscillation based permutation entropy calculation as a dynamic nonlinear feature for health monitoring of rolling element bearing

Aiming at overcoming the limitations of permutation entropy (PE) in dynamic bearing health monitoring (Healthy-initial fault inception-severe fault stage) , in this paper, PE calculation based on tunable Q factor wavelet transform (TQWT) is implemented to quantify bearing health. Firstly, weaknesses of PE in depicting bearing health are studied through numerically simulated signals. Secondly, based on the biasness of PE towards bearing fundamental signal component (FSC), oscillation based method TQWT is used to separate the FSC from the bearing signal to calculate its PE. Since, the proposed PE calculation method includes an oscillation based procedure, the resultant feature value is termed as oscillation based PE (OBPE). Selection of the parameters needed for OBPE calculation are discussed in details. Two public bearing run to failure experiments have confirmed that OBPE provides a much better result than direct application of PE in dynamic bearing health monitoring.