Title :
Signature analysis of acceleration and torque signals of compressor pump assemblies for quality control
Author :
Demirdogen, A. Caner ; Houghton, J. Richard
Author_Institution :
Center for Manuf. Res. & Technol. Utilization, Tennessee Technol. Univ., Cookeville, TN, USA
Abstract :
A description is given of an evaluation of signature analysis techniques for online inspection of every compressor pump assembly from an automatic manufacturing and assembly line. Time-domain and frequency-domain diagnostic methods are tested for the ability to amplify the signal from a flaw in the presence of background signals. A signature analysis method is then developed for which the signals are only evaluated in the time domain. The torque signal must stay within a maximum and minimum pass band to be acceptable. The accelerometer signal is averaged to reduce random-noise signals, then squared and integrated. This is shown to provide a superior method for the detection of transient defects and the angular location of flaws.<>
Keywords :
assembling; computerised pattern recognition; computerised signal processing; manufacturing computer control; pumps; quality control; accelerometer signal; automatic manufacturing; compressor pump assembly; computerised pattern recognition; computerised signal processing; online inspection; quality control; signature analysis; time domain; torque signals; transient defects; Acceleration; Accelerometers; Assembly; Frequency domain analysis; Inspection; Manufacturing automation; Signal analysis; Testing; Time domain analysis; Torque;
Conference_Titel :
System Theory, 1988., Proceedings of the Twentieth Southeastern Symposium on
Conference_Location :
Charlotte, NC, USA
Print_ISBN :
0-8186-0847-1
DOI :
10.1109/SSST.1988.17126