Tool Holder Unbalance Monitoring
Unbalance in the tool assembly causes excessive loads on spindle bearings, tool wear, and increased vibration levels. As high-speed machining has become a common practice, situation is worse as unbalance force is proportional to square of the speed. Watchdog Agent provide data-driven methods, which use statistical tool and supervised learning algorithms to detect the presence of unbalance in a tool assembly relative to the tools with known balance levels.
Firstly, Fourier analysis is performed to determine the vibration noise of the floor and uncover resonant frequencies. Total Indicator Run out (TIR) is quantified to determine spindle and tool run-outs. Other factors that may affect vibration readings are identified, which characterize the machine, thus coming up with a customized balance measurement technique for that machine.
Secondly, after performing tests, features that are to be extracted from the vibration signals are identified. The gamut of features may include average (dc) value, RMS value, minimum and maximum peak levels, amplitudes of the fundamental and harmonics of the spindle service speed, 3rd and 4th order moments of the vibration signal, kurtosis etc... Fisher criterion are used to identify the features that influence the system performance.
Finally, a supervised learning algorithm is selected to build the best-fitting model to quantify the relationship between the dichotomous characteristics of the dependent variable with a set of independent variables and estimate the unbalance present in the tool assembly.
Trending and monitoring these values will give valuable insights and forecasting capabilities for the tool assembly in terms of its health and usability for a certain precision machining process.
Firstly, Fourier analysis is performed to determine the vibration noise of the floor and uncover resonant frequencies. Total Indicator Run out (TIR) is quantified to determine spindle and tool run-outs. Other factors that may affect vibration readings are identified, which characterize the machine, thus coming up with a customized balance measurement technique for that machine.
Secondly, after performing tests, features that are to be extracted from the vibration signals are identified. The gamut of features may include average (dc) value, RMS value, minimum and maximum peak levels, amplitudes of the fundamental and harmonics of the spindle service speed, 3rd and 4th order moments of the vibration signal, kurtosis etc... Fisher criterion are used to identify the features that influence the system performance.
Finally, a supervised learning algorithm is selected to build the best-fitting model to quantify the relationship between the dichotomous characteristics of the dependent variable with a set of independent variables and estimate the unbalance present in the tool assembly.
Trending and monitoring these values will give valuable insights and forecasting capabilities for the tool assembly in terms of its health and usability for a certain precision machining process.
Subscribe to:
Posts (Atom)