Research on compound fault diagnosis of rolling bearing of CNC machine tools based on parameter optimization MOMEDA

The characteristic of vibration signal of bearing compound fault of CNC machine tool is that the fault are not clear, and it is hard to separate and extract the error. An excellent approach for extracting periodic pulses from signals is the multi-point optimal minimum entropy deconvolution adjustment (MOMEDA). Nevertheless, the chosen pulse cycle and filter length strongly influence the deconvolution performance of MOMEDA. To address some of these shortcomings of MOMEDA, a method of optimizing MOMEDA parameters based on multi-point kurtosis and composite index is proposed. Firstly, multi-point kurtosis is used to ensure that MOMEDA algorithm selects accurate fault period. Secondly, the optimum filter length is determined by a comprehensive index. This removes the interference of filter length with the separation performance. The emulation and experiment results demonstrate the approach could isolate and refine the eigenfrequency of the compound error vibration signal of CNC machine tool accurately and effectively.