Characteristic Extraction Research Of Rotating Machinery Vibration Signal

Vibration parameter signal of rotating machinery equipment on operating, especially inthe presence of a fault condition, the vast majority are non-stationary signals, the frequencyof signal characteristics change with time, if only in the time or frequency domain analysisis far not enough, it need to master the non-stationary signals in the frequency-time scalesamplitude or energy distribution.Fault feature extraction process is the most critical and important issues amongrotating machinery fault diagnosis, to consider non-stationary signals characteristics ofrotating mechanical system, compared to other time-frequency analysis method, this paperconducted in-depth research on non-stationary signal failure hotspot feature extractionmethod that based on Hilbert Huang transform time-frequency analysis method, itsummarizes that the empirical mode decomposition algorithm exists the endpoint effect andmodal aliasing problems. According to the field signal of rotating machinery is often mixedmuch random noise and impulse interference problems, research on singular valuedecomposition and morphological filtering theory, combined with singular valuedecomposition can effectively remove the random noise and morphological filter cansuppress the interference pulse characteristics, proposed singular morphological filteringmethod, the method can effectively eliminate the random noise and pulse interference thatmeasured in the field, avoid modal aliasing phenomenon of empirical mode decomposition.According to the empirical mode decomposition in actual exist the end effect and the modemixing problem, proposed suitable for rotating machinery fault feature of non stationarysignal extraction method-ensemble extreme range mean decomposition algorithm, thismethod can effectively solve the limitations of empirical mode decomposition algorithm.Finally, QPZZ-II rotating machinery vibration fault simulation platform to simulate therolling bearing fault forms the outer losses were simulated failure, malfunction and damagethe inner ring rolling rolling bearing fault damage progresses, the proposed method ofexperimental verification, test results show that the proposed method for strongnon-stationary noise vibration signals can effectively analyze the fault characteristicfrequency.