Residual Life Prediction of RF Oscillator Under Accelerated Stress Failure Condition

Background: Crystal oscillator plays a major role in the RF system as the frequency source while often encountered in vibration condition. In order to maintain the normal operation of the system, implementing estimation and prediction is necessary as a strategy based on performance degradation. However, there is little research addressing the issues associated with health status evaluation and Residual Useful Life (RUL) prediction of the RF crystal oscillator. Methods: This paper proposes a method to assess the health status and the RUL tested in a case study under vibration condition. Conclusion: Our approach utilizes acceleration sensitivity(AS) as indicator and the M-estimation based optimally-pruned extreme learning machine (M-OPELM) as technology, obtaining ideal predicting results according to the actual RUL.