As the operational hub of an automated high-bay warehouse, the operating status of a stacker crane affects the efficiency and stability of warehousing operations. Stacker cranes are prone to malfunctions due to their high operating frequency and wide range of load variations, and damage to their mechanical components requires diagnosis using condition-monitoring signals. Therefore, this paper proposes a periodic optimization entropy deconvolution and adaptive matching method for bearing fault diagnosis. First, to address pulse distortion caused by inaccurate period information, the decoupling harmonic period estimation (DHPE) method is designed. The spectrum is weighted based on health signals to eliminate periodic interference and obtain the decoupled spectrum (DS). Based on the DS, low-amplitude redundancy removal and harmonic power spectrum analysis are implemented to avoid misidentifying pseudo-periodicity. Second, to restore the periodic shock components and filter out redundant components, multipoint optimal minimum entropy deconvolution adjusted (MOMEDA) is used for time-domain filtering. The periodicity information obtained from DHPE guides MOMEDA in constructing a filter to remove periodic disturbances. Finally, to address subjective bias in manual classification results, the adaptive harmonic energy accumulation matching (AHEAM) method is designed. Using the spectral line amplitude of the envelope signal as a metric, a set of frequency candidates is constructed and subjected to harmonic product spectral analysis, enabling automatic matching and identification of fault types. Simulated signals and stacker signals are used to validate the performance of the proposed method.

fault diagnosis, period estimation, multipoint optimal minimum entropy deconvolution, harmonic product spectrum analysis