Various and multilevel of wavelet transform for classification misalignment on induction motor with quadratic discriminant analysis

Abstract

Induction motors have become a major part of the industry because of strong construction, cheap in purchasing and maintenance, high efficiency, and easy to operate. Preventive maintenance must always be carried out on all industrial equipment, including induction motors to last long and prevent further damage. Based on research in the industry, around 42%-50% or almost 50% is bearing damage. One reason is the occurrence of misalignment during the installation of the load on the induction motor. This study tries to identify the condition of the motor and classify the level of misalignment damage that occurs. In the process, the mother wavelet like as Daubechis, Symlet and Coiflet discrete wavelet transform (DWT) are selected as tools in processing motor vibration data. The level of DWT applied is 1st to 3rd level. Then, the three types of signal extraction, namely sum, range, and energy, which are obtained from a high-frequency signal of DWT, are used as input to Quadratic and Linear Discriminant Analysis. Then, discriminant analysis analyzes and classifies them into normal operation and two misalignments conditions. The simulation shows that 1st level of Daubechis DWT combined with quadratic discriminant analysis generates the best classification. It results 0% error of classification with Db3, Db4 and Db5, 4.17% error with Db1 and 8.33% error with Db2.

Hardware implementation of series DC arc fault protection using fast Fourier transform

“…Method often be applied for arc fault detection is frequency domain analysis [9]- [11]. Many methods play in frequency domain analysis, one of them is fast Fourier transform (FFT) [12]- [14]. FFT is used because easy implementation and fast computation [15]- [17].…”