Cyclic Band Sparsity Measures Driven Blind Deconvolution and its Application to Wind Turbine Bearing DiagnosisCyclic Band Sparsity Measures Driven Blind Deconvolution and its Application to Wind Turbine Bearing Diagnosis
cyclostationary, Blind deconvolution, cyclic band sparsity measures
Jiahao Li / Beijing University of Civil Engineering and Architecture
Yanxue Wang / Beijing University of Civil Engineering and Architecture;
Owing to the specialized operational conditions of wind turbine bearings, bearing fault feature is frequently distorted by environmental noise. Blind deconvolution methods are a validify method for machinery fault diagnosis, because it is significantly for reducing noise and eliminating the interference of the system transmission path. Among all the blind deconvolution methods, blind deconvolution based on traditional sparsity measures is an effective method for extracting weak periodic impulses related to bearing faults. However, in practical applications, a wind turbine may exist multiple cyclostationary sources and the measured signals may not be Gaussian distributed. Many conventional BD methods are influenced by non-fault cyclostationary component (CSC), leading to inaccurate identification of fault-related CSC. The performance of blind deconvolution based on sparsity measures can be reduced by the above problems. This study overcomes these challenges by proposing the novel indices, cyclic band sparsity measures, for blind deconvolution to extract the bearing fault more robustly. The proposed method, cyclic band sparsity measures (CBSM) driven blind deconvolution (CBSM-BD), can effectively to eliminate the influence of un-relate CSC interference and maximize the sparsity of the envelope spectrum at cyclic band. Examination of experimental results indicates that the proposed approach CBSM-BD is a straightforward yet effective method for enhancing faults exhibiting a period signature.