Title :
Flue Gas Turbine Condition Trend Prediction Based on Improved Echo State Network
Author :
Shaohong, Wang ; Tao, Chen ; Xiaoli, Xu
Author_Institution :
Sch. of Mech. Eng., Beijing Inst. of Technol., Beijing, China
Abstract :
Fault prediction is the key technology for ensuring safe operation and scientific maintenance of large equipment. As the running of flue gas turbine has nonlinear characteristics, echo state network (ESN) was introduced to predict the condition trend of the turbine. Singular value decomposition was used to improve the linear regression algorithm of ESN, and the prediction workflow was given. Condition trend prediction results showed the effectiveness of the proposed method.
Keywords :
condition monitoring; fault location; flue gases; gas turbines; maintenance engineering; mechanical engineering computing; recurrent neural nets; regression analysis; singular value decomposition; condition trend prediction; fault prediction; flue gas turbine; improved echo state network; linear regression algorithm; nonlinear characteristics; operation safety; prediction workflow; scientific maintenance; singular value decomposition; Linear regression; Matrix decomposition; Nonlinear dynamical systems; Prediction algorithms; Singular value decomposition; Training; Turbines; condition trend prediction; echo state network; flue gas turbine;
Conference_Titel :
Measuring Technology and Mechatronics Automation (ICMTMA), 2011 Third International Conference on
Conference_Location :
Shangshai
Print_ISBN :
978-1-4244-9010-3
DOI :
10.1109/ICMTMA.2011.348
