Forecasting Minor Machine Failure by DDS Spectrum Analysis

Abstract Machine failures are either major or minor. The former usually requires the machine to be stopped immediately, while the latter occurs gradually and can be present before the machine fails. This paper describes the development of an automatic diagnostic system to detect minor machine failure before machine breakdown of computer controlled robotic drilling machines. By using the Dynamic Data System approach, a set of discrete time series data taken from the continuous vibration signal of the machine is analyzed to forecast machine failure and then to distinguish failures due to the spindle, gears, bearings, and air motors of the machine drive system. Compared to the real time analyzer (RTA) approach based on the FFT technique, the DDS power spectrum is smoother and less prone to errors; it successfully identifies defects in the robotic drilling machine.