Mechanical compound faults extraction based on improved frequency domain blind deconvolution algorithm

Since the signals which picked up by acoustic sensors are usually mixed with fault source signals and other noise signals due to the strong background noise in real-time working conditions, making it difficult to extract and identify the useful signals. In order to solve this problem, an improved frequency blind deconvolution algorithm based on adaptive generalized morphological filter, improve Block-based Model Algorithm and Two-steps FD-SCA is put forward. Followed by the frequency-domain blind deconvolution flow, the adaptive generalized morphological filter was firstly used to extract modulation features embedded in the observed signals, then the improve Block-based Model Algorithm was employed to do blind separation, finally the Two-steps FD-SCA is employed to estimate the source signals. The experiment results of mechanical compound faults extraction in real working-environment demonstrate the accuracy and reliability of the proposed algorithm.