MONITORING SYSTEM OF ROTATING MACHINES FOR MAINTENANCE USING ARTIFICIAL NEURAL NETWORK

• Main Author: Fadillah, Kelvin
• Format: Final Project
• Language: Indonesia
• Online Access: https://digilib.itb.ac.id/gdl/view/62653
• Institution: Institut Teknologi Bandung

Every moving machine must have the vibration incident. This vibration is natural things that happen to all machine and each of them have their own charactheristic. From that charactheristics, we will be able to know the machine condition by comparing it to the standard characteristics of the broken machine. As for that machine characteristics could be known using vibration measurement. That measurement will be done by a specific acquisition instrument that can read time domain vibration signal to be transform to frequency domain vibration signal to be analyzed later on. In this undergraduate paper, vibration measurement will be made using data acquisition LabVIEW that will be represented by merging two signals from signal generator sine, square, triangle and sawtooth with simple mathematic operation which are addition, subtraction, multiplication and division. That signal will be filter using lowpass filter to reduce the noise from vibration measurement which data can be saved. Fourier transformation will be done on the signal so frequency domain vibration also phase could be shown. There are 4 types of high vibration simulation on this paper which are mass unbalance from additing sine signal with triangle signal, bent shaft and misalignment from additing sine signal with sawtooth signal and mechanical looseness from additing sawtooth signal with triangle signal that have control parameter which are amplitude, phase, frequency, offset, sample frequency and total samples. The result from this program are frequency and phase when high amplitude happening on frequency domain vibration or can be known as vibration spectrum.