Research On Condition Monitoring System Of Large Rotating Machinery

Machinery abruptly breakdown is a problem that need be solved urgently.Preventing machinery temporarily shutdown from failures not only beneficial toimproving production efficiency, but also avails to reducing maintenance costs andachieving equipment lifecycle management. Condition monitoring has become animportant research. This dissertation aims at the problem of cement production linemust operation continuous in case of huge losses, and makes research of the typicalfailure modes of the main reducer of rotary kiln. By analyzing in-depth of vibrationfeatures, identified the vibration signal frequency range of reducer. Applying digitalfilters to filtering strong coupling signal, this was collecting from targeted equipmentsin workshop. Researching vibration signal spectrum analysis techniques based onmulti-sensor data fusion. Design hardware and software systems for on-line monitoringof machinery.This paper analyses failure mode of major transmission parts, such as gears,bearings and shafts, comes to a conclusion that the vibration signal components aremainly distributed in speed, mesh, natural vibration frequencies and harmonics.According to the structure size and operating parameters of specific reducer, we cancalculate the frequency range as technical indicators of condition monitoring system.To extract useful signal from raw signal, mitigate interference of externalenvironment and system volatility, provide signal of high noise ratio for subsequentlyspectral analysis. In this paper, both analog low-pass filter and digital filter are used,optimize methods of designing digital filter are studied in depth; simplify designschedules of multi-band filter.Condition monitoring is the essence of feature recognition; vibration signal isdrawn into map and then compared with maps of normal and typical failure signal todetermine the status of machinery. To compensate defects for the traditionalsingle-source signal spectral analysis, this paper put forward multi-sensor data fusionmethod of spectral analysis, fully exploit the implicit signal for status information.Research integer period synchronous data sampling techniques, obtain the vibrationsignal period starting and ending points by differentiating the phase signal, resample thevibration signal in these periods, this method is effective to prevent the spectrumleakage and verified data fusion superiority at the same time.Finally, select monitoring points and parameters for the ‘JH800C-306W-31.5’model reducer. Develop the condition monitoring hardware system with sensors anddata acquisition card. Program the condition monitoring software system with C++andMATLAB. Experimental results show that the system performance meets the design requirements; it is useful for the transitional of maintenance methods from downtimemaintenance and planned maintenance to condition based maintenance.