My research journey: from fundamental research to

industry[excerpt]

Gabriel Preda

ELECTRICAL ENGINEERING DEPARTMENTElectrical Engineering Faculty

POLITEHNICA University of Bucharest

Research path – categoriesComputational electromagnetics

• Integral FEM solution for nonlinear moving media @ POLITEHNICA University

• FEM-BEM formulation for nonlinear media @ Tokyo University

• FEM multi-physics professional solution @ SSIL Inc., Tokyo

Artificial intelligence based inverse problem solution in Nondestructive

Testing

• Artificial defects geometry reconstruction using Neural Networks @ Tokyo University

• Material degradation evaluation using Neural Networks @ Tokyo University

• Deep defects geometry evaluation using pulse eddy currents, Neural Networks or Genetic Algorithms @ POLITEHNICA University

Image processing and medical imaging

• Object recognition, position and orientation detection using Hough transform and pattern recognition @ SSIL Inc., Tokyo

• Fringe pattern analysis for interferometry images @ SSIL Inc., Tokyo

• Medical imaging for virtual endoscopy @ Advantec Solutions, Bucharest

Research path – an historical approach

Tokyo University JSAEM SSIL Inc.

Kisano Medical Systems

1992 1999 2001,March

2001,Sep

2005,May 2014

20122006 2009

Advantec Solutions

UPB

Defend PhDStart Post-Doc

Enroll PhD

1993

ENSIMAG

1994

- Integral FEM solution for nonlinear moving media media

UPB

Main PhD work published

1993 1997

- FEM-BEM for nonlinear media - Material degradation evaluation- Artificial defects geometry reconstruction using NN

ECT & MFL based NDT

- FEM solver- Image processing

- Pulse ECT for deep defects- Medical imaging- Telemedicine

- Programming & computer science training

2011

IXI Mobile, Telmap, Endava

1999-2001: Post-doc, Tokaimura, Todai Eddy-currents testing using FEM-BEM linear code (Kurokawa). Application to

detection of manufactured and natural defects in stainless steel (Inconel 600) tubing of Steam Generator (SG) of Pressurized Water Reactors (PWR);

Inverse problem solution (reconstruction of defects geometry) using an in-house method based on Principal Component Analysis (PCA), Artificial Neural Networks and Shifting Aperture method (Popa): solution for fully-isolating defects and partial-conductivity defects;

Preda, G., Popa, R.C., Demachi, K., Miya, K., Neural network for inverse mapping in eddy current testing, Neural Networks, 1999, IJCNN'99, International Joint Conference on, 6(1), pp. 4033-4036, 1999.

1999-2001: Post-doc, Tokaimura, Todai Material degradation characterization using magnetic measurements:

austenitic phase of non-magnetic materials changed to martensitic phase thus detection of magnetic flux density distribution outside of the material probe (Preda, Takaya);

Shigeru Takaya, Gabriel Preda, Kazuyuki Demachi, Tetsuya Uchimoto and Kenzo Miya, Reconstruction of Magnetization from Magnetic Flux Leakage for Evaluation of Material Degradation, Electromagnetic Nondestructive Evaluation V, 1(1), pp. 291-298, 2001.

1999-2001: Post-doc, Tokaimura, Todai Eddy-currents testing using FEM-BEM nonlinear code

(Preda, after Kurokawa, Kurtz & Fetzer);

Pulse eddy-currents testing using FEM-BEM nonlinear code and NN (Popa).

Preda, G., Cranganu-Cretu, B, Hantila, F.I., Mihalache, O., Zhenmao Chen, Miya, K, Nonlinear FEM-BEM formulation and model-free inversion procedure for reconstruction of cracks using pulse eddy currents, Magnetics, IEEE Transactions on, 38(2), pp. 1241-1244, 2002.