Title :
Application of Improved BP Network in the Flaws Evaluation of Conductive Materials
Author :
Qingjie, Yang ; Yanfeng, Chen ; Xinhua, Mao ; Xiaohong, Kong
Author_Institution :
Sch. of Mechinery & Electron., Henan Inst. of Sci. & Technol., Xinxiang, China
Abstract :
The back-propagation (BP) network is widely recognized as a powerful training tool of the multilayer neural networks (MLNNs). Usually it suffers from a slow convergence rate and often results in local minimums, since it applies the steepest descent method to update the network weights. A variety of related algorithms have been introduced to address that problem. Levenberg-Marquardt algorithm is one of the fastest types of these algorithms. This paper presents an approximation calculation for Hesse matrix to train the neural networks when the second order item can not be omitted, and the improved algorithm is successfully used in the surface flaws evaluation of the conductive materials based on eddy current testing (ECT).
Keywords :
Hessian matrices; approximation theory; backpropagation; conducting materials; convergence; eddy current testing; flaw detection; gradient methods; materials science computing; materials testing; multilayer perceptrons; Hesse matrix; Levenberg-Marquardt algorithm; approximation calculation; backpropagation network; conductive materials; convergence rate; eddy current testing; multilayer neural network training; steepest descent method; surface flaws evaluation; Artificial neural networks; Biological neural networks; Computer networks; Conducting materials; Electrical capacitance tomography; Humans; Multi-layer neural network; Neural networks; Neurons; Signal processing algorithms;
Conference_Titel :
Artificial Intelligence and Computational Intelligence, 2009. AICI '09. International Conference on
Conference_Location :
Shanghai
Print_ISBN :
978-1-4244-3835-8
Electronic_ISBN :
978-0-7695-3816-7
DOI :
10.1109/AICI.2009.471
