International Conference on Sensing, Measurement & Data Analytics in the era of Artificial Intelligence (ICSMD2023)

162 / 2023-10-22 10:27:08

A Novel General Time-Frequency Energy Concentration Operator for Time-Frequency Representation and Application to Bearing Fault Diagnosis

Time-frequency analysis, energy concentration, double derivation extraction operator, fault diagnosis

Special Sessions > SS13 - Signal Decomposition-Integrated Mechanical Fault Detection and Remaining Useful Life Prediction

Time-frequency (TF) analysis (TFA) is currently a hotly discussed topic on account of its powerful ability in capture the inherent non-stationary features exist in real word. It has always been a key research direction in the field of TFA to accurately present the changing trend of the characteristic component of the signal with a higher energy-concentrated time-frequency representation (TFR) result by improving the energy divergence of the time-frequency spectrum. Aiming at this aim, this paper proposed a novel general energy concentration operator. The operator is obtained by locating the position of TF points where the spectrum amplitude energy is the largest via the first order derivation and Hessian matrix, so it is called double derivation extraction operator (DDEO). The DDEO does not need to be embedded in the original algorithm and can be directly processed on the TF spectrum results, which shows convenience and generality. In numerical simulation, its generality and effectiveness as the TF energy concentration tool were verified by integrating with three TFA methods. Finally, through the bearing fault application case, its potential for integrating with TFA technology for fault diagnosis is further verified.