<![CDATA[One way in recognizing mechanical faults of electrical machines is using the vibration signals. This paper studies the recognition of mechanical faults of electrical machines which have faulty bearing, faulty impeller, and faulty foot valve. Electrical machines studied in this paper are household water pumps. The recognition of mechanical faults is conducted using the vibration signals at the frequency domain at various frequency resolution. The recognition method used is template matching for pattern classes. Euclidean distance function is used in this template matching. Based on the results, the highest recognition rate of 100% is optimally obtained using the vibration signal at frequency resolution of 5,9 Hz, and the number of 5 samples/classes in the water pump database. The 5,9 Hz frequency resolution was obtained using a 256 point FFT, where the number of 128 FFT coefficients are used as the feature extraction.]]>
Copyrights © 2015
