Title :
Experimental study of performance monitoring for centrifugal fan and fault diagnosis for pipe network in power plant
Author :
Songling, Wang ; Junhu, Hou ; Yuan Xiongjun ; Liansuo, An ; Tongxin, Liu
Author_Institution :
Dept. of Power Eng., North China Electr. Power Univ., Hebei, China
Abstract :
In this paper, accurate fitting on performance curves of fan by RBF networks has been performed and then flow-monitoring model of a fan is derived. The experimental study on the fan of 4-73No8D is taken on the condition of the change of rotation speed, resistance of pipe network and angle of regulator, and then the accurate monitoring results of flow are obtained. In the process of experimental study of jam-leak trouble diagnosis for pipe network, the resistance coefficient of pipe network and the static pressure of inlet and outlet of fan are defined as the eigenvector for fault diagnosis, which act as the inputs of the neural networks. The jam and air leak in different positions of the pipe network are simulated in the laboratory, some experimental data for training and others for test, as a result the troubles of pipe network are diagnosed accurately.
Keywords :
computerised monitoring; eigenvalues and eigenfunctions; fault diagnosis; flow measurement; power engineering computing; power stations; radial basis function networks; RBF networks; centrifugal fan; eigenvector; fault diagnosis; flow-monitoring model; jam-leak trouble diagnosis; monitoring results; neural networks; performance monitoring; pipe network; power plant; regulator angle; resistance coefficient; rotation speed; static pressure; Chemical industry; Condition monitoring; Curve fitting; Fans; Fault diagnosis; Intelligent networks; Neural networks; Power generation; Surface fitting; Ventilation;
Conference_Titel :
TENCON '02. Proceedings. 2002 IEEE Region 10 Conference on Computers, Communications, Control and Power Engineering
Print_ISBN :
0-7803-7490-8
DOI :
10.1109/TENCON.2002.1182737
