Real-time sinusoidal parameter estimation for damage growth monitoring during ultrasonic very high cycle fatigue tests

Ultrasonic fatigue tests (UFT) are used to study the life behavior of metallic components undergoing a very high number cycles (typically ≥107−109 cycles) under relatively low mechanical loads. By soliciting specimens at 20 kHz, ultrasonic machines indispensable for monitoring damage growth and failures in reasonable amount time. As accumulates specimen, specimen’s free-end exhibits nonlinear dynamic response. The resulting quasi-stationary, harmonic signals have sinusoidal parameters (frequency amplitude) which slowly time-varying with respect excitation frequency. discrete Fourier transform (DFT) is typically extract these evolving from window finite data vibration signal. Alternative spectral estimation methods, specifically line spectra estimators (LSEs), exploit priori information signal via their modeling basis overcome limitations seen by DFT. Many LSEs known state-of-the-art results when benchmarked on purely stationary unit amplitudes. However, performances unknown context typical UFT, leading widespread use Thus, this paper benchmarks classical modern against specific synthetic arise UFTs. Adequate algorithms then recommended made publicly available process experimental coming depending performance metrics restraints.