Vibration signal analysis in comminution

Abstract: Comminution accounts for a significant part of the energy consumption and metal wear in powder manufacturing industries and as a result, it is a major cost factor. To maintain high efficiency, the operating parameters must be monitored and adjusted promptly to optimal values. In practice, it is either difficult or laborious and time consuming to measure most of the parameters involved, therefore, a quick and economic way to monitor the parameters of interest is essential for system control. Vibration signals from comminution devices (ball mills) were picked up using piezoelectric accelerometers andlor an acoustic microphone. The source signals were amplified and digitised for an IBM compatible PC. The time-domain waveforms were transformed into frequencydomain power spectra. Finally, the vibration signals were reduced, using multivariate data analysis, to a few "latent" variables which could then be related to the physical parameters. Signal analyses were performed with the DSP4ME toolbox (MatLab) on a PC. Since the vibration signal pattern varies correspondingly with the operating state of an equipment, a technique based on the vibration signal analysis have been widely used in fault-detection for heavy machines. Most of the analytical instruments for this purpose are based on hardware, which make them inflexible and hard to adopt for other purposes. Moreover, there is no routine for signal identification and model building. To overcome all these disadvantages, a strategy based on a simple hardware, but a complex computer software was used for the vibration signal analysis both in the laboratory scale and industrial grinding operations. Various operating parameters from the comminution process were predicted by measuring and processing vibration signals. Since the signal measurement can be made on-line, this technique provides accurate and fast information for the control system to adjust the operating parameters to optimal values.