PREDICTIVE MAINTENANCE IN ROTATING MACHINES WITH VIBRATION ANALYSIS AND FUZZY LOGIC
Abstract
This work provides a detection method for faults in the operation of rotating machines based on a change of system vibration standard and in the operation status diagnosis by fuzzy logic. These changes are analyzed and used as parameters for predicting incipient faults, as well as their evolution in operation condition allowing predictive maintenance tasks. A mechanic structure (developed to insert the faults) called Rotating System (see Fig 1) has been used. The data acquisition of structure vibration has been made by a biaxial solid state accelerometer in a low-power chip. The outputs are directly measured by a microcontroller-based system without needing an A/D converter. The acquisition system is composed by two subsystems: a personal computer and a microcontroller-based front-end (Microchip PIC16C73A). The acquired data (stored in *.dat files) are computed and analyzed in frequency domain with a signal processing tool (called SPTOOL) of the Matlab5 package. The faults analyzed in this work are due to the unbalancing of axle_wheel by insertion of unbalanced elements. The relation of mass between the wheel and the smallest unbalanced element is 1 : 10,000. The fuzzy system was calibrated to detect and diagnose the normal, incipient fault, maintenance, and danger conditions of the Rotating System.
Event details
- Event
- TC4 Symposium 2001
- Technical Committee
- TC4
- Place
- Lisbon, PORTUGAL
- Time
- 13 September 2001 - 14 September 2001