A Multiperiodicity-Induced Sparse-Fidelity Representation Model for Compound Fault Diagnosis of Wind Turbine Gearbox

Compound faults are common in wind turbine gearbox due to complex structure and harsh operating condition, which are represented by the bearing-gear fault, gear-gear fault, and so on. It is a challenging task to accurately detect multiple faults from on-site vibration signal potentially contaminated by intensive background noise. To address this issue, a multiperiodicity-induced sparse-fidelity representation (MPSFR) model is proposed in this article. The proposed method is based on the periodicity-induced overlapping group shrinkage (POGS) model with the constraint of the sparsity within and across groups (SWAG). The l_{\mathrm {p}} -norm and the proposed reweighted minimax-concave penalty function (MCP) with adjustable order are adopted for SWAG constraints. The Hanning sequence is utilized as the periodicity-induced sequence in the POGS model. The closed-form solution of the proposed method is deduced using the majorization-minimization (MM) algorithm. Eventually, a weighted strategy for the decomposed results is inferred to prevent overdecomposition issue. The proposed method is validated by simulation signals, experimental signal, and on-site signal of actual wind turbine, which indicate that the characteristics of compound fault can be extracted plainly and the energy loss is reduced effectively.