Atsuya Oishi

Quantitative nondestructive evaluation with ultrasonic method using neural networks and computational mechanics

This paper describes an inverse analysis method using hierarchical neural networks and computational mechanics, and its application to the quantitative nondestructive evaluation with the ultrasonic method. The present method consists of three subprocesses. First, by parametrically changing the location and size of a defect hidden in solid, elastic wave propagation in the solid is calculated with the dynamic finite element method. Second, the back-propagation neural network is trained using the calculated relationships between the defect parameters and the dynamic responses of solid surface. Finally, the trained network is utilized to determine appropriate defect parameters from some measured dynamic responses of solid surface. The accuracy and efficiency of the present method are discussed in detail through the identification of size and location of a defect hidden in solid.

Domain Decomposition Based Parallel Contact Algorithm and Its Implementation to Explicit Finite Element Analysis Code.

Computer simulations are about to replace experiments in various fields related to dynamic problems, such as crashworthiness of vehicles, vibration of a nuclear reactor in earthquake and ultrasonic wave propagation in solid with cracks for nondestructive evaluation. For solving such large scale problems, the parallel processing has become a key technology. In this paper, a parallel contact algorithm suitable for a domain decomposition technique is proposed, and implemented on a dynamic finite element analysis code based on an explicit time integration scheme. Parallel efficiency of the code is tested through sample analyses on a PC cluster.

Defect Identification with Ultrasonics Using Neural Networks and Computational Mechanics. Verification of Accuracy through Laser Ultrasonics Experiment.

This paper describes an application of the neural networks to defect identification with laser ultrasonics. The present method consists of three subprocesses. First, sample data of identification parameters vs. dynamic responses of displacements at several monitoring points on surface are calculated using the dynamic finite element method. Second, the back-propagation neural network is trained using the sample data. Finally, the well trained network is utilized for defect identification. This method is applied to the identification of a surface defect hidden in solid with laser ultrasonics. Its performance in accuracy and robustness is quantitatively verified in detail through both numerical simulations and experiments.

Quantitative Nondestructive Evaluation with Laser Ultrasonics Using Neural Network and Computational Mechanics

Nondestructive Evaluation (NDE) techniques to detect cracks and defects hidden in solid are very important to assure the structural integrity of operating plants and structures, and to evaluate their residual life time. Various NDE techniques using ultrasonic wave, X-ray, magnetic powder, eddy current and so on have been studied so far. In most cases, they detect only the existence of cracks and defects. Then, quantitative nondestructive evaluation (QNDE) techniques to determine sizes, shapes and locations of cracks and defects are strongly desired.