Application of adaptive filtering for weak impulsive signal recovery for bearings local damage detection in complex mining mechanical systems working under condition of varying load

The paper shows application of an adaptive filter as a pre-processor for impulsive cyclic weak signal recovery from raw vibration signals captured from complex mechanical systems used in the industry (namely bearings used in pulleys parts of driving units for belt conveyors). Periodic/cyclic impulses are related to local faults which cause impulse/concentric forces/stresses in kinematic pairs. Typical examples of such local faults which cause mechanical system condition change are spall/pitting on bearings elements: outer/inner races and/or rolling elements. For analyzed objects, impulses associated with local faults are masked by other signal sources. In the first part of the paper objects are presented for the better understanding of mechanical phenomena that exist in the system, then preliminary signal analysis will be performed (in time, frequency and time-frequency domain) for the identification of signal nature. Next the idea of an adaptive system and the brief description of Normalized Least Mean Square (NLMS) algorithm will be presented. Application of NLMS is better than classical LMS due to stability of the adaptation. In the last section the results of adaptive filtering for signals from bearings is discussed. The authors show the application of NLMS (for the first time in literature) for the case when signals are received from machines working in industrial condition. There were made only trails when the machines were investigated in laboratory conditions.