Multiparameter fault identification in elastic elements of constructions based on artificial neural networks

• Autorzy: Krasnoschekov A.
• Konferencja: Conference on Fracture Mechanics (XIII ; 5-7.09.2011 ; Opole ; Polska)
• Języki publikacji: EN

Abstrakty

EN

Identification of defects is a critical aspect in the operational safety and func- tionality of elastic structures. It’s well known that changes in a structure of the material, entail a change in its dynamic properties. One of the most commonly used methods for determination of these properties is modal analysis, with witch we obtain spectrum of natural frequencies and forms of investigated object. This data can be used to identify presence of the defect and it's quantative parameters (like location or size). In recent years many studies in this direction were held and artificial neural networks (ANN) are increasingly used for reconstruction of inhomogeneities [1,2]. In this article we propose a method that includes the complete cycle of designing ANN for identification of defects in a given 3-D object: starting from a model and ending with location of sensors on the object's surface and trained ANN. The problems of constructing simulation model of the object, optimal placement of sensors and subsequent identification of the defect were solved. We've also investigated possibilities of application of different architectures and training algorithms of ANN, and analyzed the influence of errors on the accuracy of determination of the defect's parameters. Simulation model in the purposes of the study in ANSYSŽ we've built 3-dimensional FE mod- el of the beam of rectangular cross section with a through crack which is coming out on top of the face (Fig. 1). The left end is rigidly clamped, right - free.

Słowa kluczowe

EN

neural networks; elastic elements; elastic structure

• Wydawca: Oficyna Wydawnicza Politechniki Opolskiej
• Czasopismo: Zeszyty Naukowe. Mechanika / Politechnika Opolska
• Rocznik: 2011
• Tom: z. 99
• Strony: 33--35
• Opis fizyczny: Bibliogr. 3 poz., rys.

Twórcy

autor

Krasnoschekov A.

• Don State Technical University, Rostov-on-Don, Russia,

Bibliografia

• [1] WASZCZYSZYN Z., ZIEMIANSKI L.: Neural networks in mechanics of structures and materials, new results and prospects of applications - Elsevier Science Ltd.: Computers and Structures 79, 2001, pp. 2261- 2276
• [2] FANG L., LUO H., TANG J.: Structural damage detection using neural network with learning rate improvement - Elsevier Science Ltd. : Com- puters and Structures 83 , 2005, pp. 2151-2152
• [3] WORDEN K., BURROWS A.: Optimal sensor placement for fault de- tection, Elsevier Science Ltd.: Engineering Structures 23, 2006, pp. 885– 901