Dr. Mehrdad Shafiei Dizaji Profile

Dr. Mehrdad Shafiei Dizaji

Postdoctoral Research Associate

SPIE Involvement:

Author

I am a PhD candidate at University of Virginia. My primarily research is currently focused on the Computer Vision–based structural health monitoring in classification and localization of damage that is present in the 3D model. I utilize robust non-contact photogrammetric technique such as 3D Digital Image Correlation (3D-DIC) technology and High Resolution Laser Scanner device to find intrinsic/extrinsic properties of the structural elements such as modal frequencies, mode shapes, constitutive properties of materials, geometric description in order to integrate and tune Finite Element Models and evaluating the feasibility of using 3D-DIC for non-contact Structural Identification (St-Id) which is a full-field non-contact manner with minimal service interruption

Proceedings Article | 18 April 2022 Presentation + Paper

Machine-learning to see defects: a hybrid attention-ConvLSTM-based convolutional neural network deep learning architecture for structural damage detection

Mehrdad Shafiei Dizaji, Zhu Mao

Proceedings Volume 12046, 120460L (2022) https://doi.org/10.1117/12.2615542

KEYWORDS: Video, Damage detection, Convolution, Feature extraction, Data modeling, Convolutional neural networks, Visual process modeling, High speed cameras, Signal detection, Sensors

Read Abstract +

Proceedings Article | 1 April 2019 Presentation + Paper

3D InspectionNet: a deep 3D convolutional neural networks based approach for 3D defect detection on concrete columns

Mehrdad Dizaji, Devin Harris

Proceedings Volume 10971, 109710E (2019) https://doi.org/10.1117/12.2514387

KEYWORDS: 3D modeling, 3D image processing, Convolutional neural networks, Inspection, Defect detection, Solid modeling, Data modeling, Convolution, Neurons, Computer aided design

Read Abstract +

Proceedings Article | 27 March 2018 Presentation + Paper

Reframing measurement for structural health monitoring: a full-field strategy for structural identification

Mehrdad Dizaji, Devin Harris, Mohamad Alipour, Osman Ozbulut

Proceedings Volume 10599, 1059910 (2018) https://doi.org/10.1117/12.2296521

KEYWORDS: Digital image correlation, Structural health monitoring, 3D modeling, Data modeling, Finite element methods, Cameras, Sensors, Optimization (mathematics), Mechanical sensors, Beam shaping

Read Abstract +

Structural health monitoring (SHM) describes a decision-making framework that is fundamentally guided by state change detection of structural systems. This framework typically relies on the use of continuous or semi-continuous monitoring of measured response to quantify this state change in structural system behavior, which is often related to the initiation of some form of damage. Measurement approaches used for traditional SHM are numerous, but most are limited to either describing localized or global phenomena, making it challenging to characterize operational structural systems which exhibit both. In addition to these limitations in sensing, SHM has also suffered from the inherent robustness inherent to most full-scale structural systems, making it challenging to identify local damage. These challenges highlight the opportunity for alternative strategies for SHM, strategies that are able to provide data suitable to translate into rich information. This paper describes preliminary results from a refined structural identification (St-ID) approach using fullfield measurements derived from high-speed 3D Digital Image Correlation (HSDIC) to characterize uncertain parameters (i.e. boundary and constitutive properties) of a laboratory scale structural component. The St-ID approach builds from prior work by supplementing full-field deflection and strain response with vibration response derived from HSDIC. Inclusion of the modal characteristics within a hybrid-genetic algorithm optimization scheme allowed for simultaneous integration of mechanical and modal response, thus enabling a more robust St-ID strategy than could be achieved with traditional sensing techniques. The use of full-field data is shown to provide a more comprehensive representation of the global and local behavior, which in turn increases the robustness of the St-Id framework. This work serves as the foundation for a new paradigm in SHM that emphasizes characterizing structural performance using a smaller number, but richer set of measurements.

Proceedings Article | 27 March 2018 Paper

Integrating 3D scanning within a simulation framework for structural mechanics

Shuzhen Yang, Mehrdad Dizaji, Devin Harris

Proceedings Volume 10598, 1059846 (2018) https://doi.org/10.1117/12.2300762

KEYWORDS: 3D modeling, Data modeling, Process modeling, 3D scanning, Finite element methods, Scanners, Laser scanners, Performance modeling, Clouds, Bridges

Read Abstract +

Keywords/Phrases

Search In:

Publication Years