Research on envelope demodulation of reducer shock signals based on peak retention and down-sampling method

Timely and effective reducer fault diagnosis is essential to ensuring their stability, reliability, and performance. In this paper, we design a Peak Retention and Down-sampling (PRD)-based approach to the envelope demodulation of reducer shock signals. PRD is an effective way of extracting the shock features in reducer fault diagnosis systems and is easy to integrate into real-time online diagnosis. PRD acquires the envelope waveform of an original signal by using a peak extraction algorithm. This algorithm focuses on identifying signal differences and on their multiplication and comparison. After the envelope waveform has been acquired, PRD is used to down-sample the envelope waveform to reduce the analyzed frequency. Finally, the spectrum characteristics of the fault signal are obtained by using a Fourier transform on the down-sampled data, so that the relevant fault information can be obtained. Reducer fault diagnosis experiments were undertaken that compared the PRD method with three traditional methods: Hilbert Transforms (HT); Generalized Envelope Detection (GED); and Peak Detection (PD). The results show that PRD is particularly effective at solving the following problems in envelope demodulation: the need for prior parameters; poor real-time performance; the loss of peak waveform information; the loss of the original signal’s attenuation and harmonic characteristics; and distortion of the envelope waveform after demodulation.