Abstract:

The medium draining is considered a crucial link in the coal selection process. In order to mitigate faults triggered by prolonged operation of the medium draining screen under full load conditions, an algorithm based on wavelet transform and energy feature extraction was employed for monitoring and diagnosing the medium draining screen in real time. By employing methods such as adaptive threshold correction and improving the threshold function, the wavelet threshold denoising algorithm was optimized. The sound signals of the medium draining screen in normal and faulty operating states underwent wavelet decomposition. After threshold denoising, the energy features of the reconstructed signals were extracted. The low-frequency sub-band section was analyzed, and the energy ratio of primary frequency to secondary frequency under different states was calculated, enabling the diagnosis of faults in the medium draining screen. Results indicate that the algorithm of improved wavelet threshold denoising increases the correction accuracy by approximately 10%. The energy of the low-frequency sub-band part of the wavelet decomposed reconstructed signal is higher than that of other sub-bands, making its analysis representative. The energy ratio of primary frequency to secondary frequency of the low-frequency sub-band in faulty operating states is significantly higher than that in normal operating states, enabling the diagnosis of faults in the medium draining screen. Compared with traditional fault diagnosis methods, the diagnostic accuracy of the proposed method increases by about 5%.

Key words: wavelet transform, feature extraction, medium draining screen, fault diagnosis, energy features, wavelet threshold denoising