Domestic research projects


Research projects (co)funded by the Slovenian Research Agency.


  • Member of University of Ljubljana: UL Faculty of Mechanical Engineering
  • Project code: J2-1730
  • Project title: High-speed-camera based high-spatial-density sensing of 3D vibrations with applications in digital-twins and remote sensing
  • Period: 01.07.2019 - 30.06.2022
  • Range on year: 0,72 FTE
  • Head: prof. dr. Janko Slavič
  • Research activity: Engineering sciences and technologies 
  • Research Organisation: Link
  • Researchers: Link
  • Citations for bibliographic records: Link

In mechanical engineering validation of digital models (i.e. digital-twins) requires experimental measurements. 3D dynamics response with high spatial density is hard to measure and requires multi-channel simultaneous measurements, long-lasting and expensive preparation. Using hundreds of accelerometers still results in relatively low spatial-density of experimental data. A non-contact, high spatial-density, high accuracy and fast 3D dynamic response measurement approach is not available today. 


The grand challenge of this research is to develop a new method for fast 3D vibration response measurement which will be non-contact, will have high-accuracy and will provide high spatial-density. 

The grand challenge will approached using 3 research objectives:

·        RO1: Theoretical development of the hybrid frequency domain triangulation method. 
This RO will be reached by developing a multi-view extension to the hybrid modal identification method which will include frequency domain triangulation (camera calibration will be based on a digital 3D model which will be later in digital-twin model updating)
·        RO2: Implementation of automatic masking of moving objects in the high-speed camera footage.       
The information in the background of the high-speed footage which is not vibrating (e.g. moving truck on a bridge) will be automatically masked. Due to the automated masking the identification of 3D deflection shapes in real-life environments will be possible and enable advanced model updating and vibration monitoring. 
·        RO3: Development of high-accuracy, high spatial-density model updating for rapid prototyping. 
With the high-accuracy, high spatial-density deflection shapes (RO1, RO2) will be related to the sensitivity and Bayesian based model updating techniques. It is expected that the proposed model updating technique will show the spatial uncertainty of the model updating; consequently, the rapid-updating will also provide statistical measure of certainty.

The new 3D experimental vibration identification method will (due to the high spatial density, high resolution and the fast 3D identification) have a significant scientific impact on the development of the rapidly developing scientific field of high-speed camera based experimental modal analysis and model updating. Further, as the new method will enable accurate, fast, dense and remote vibration monitoring it will open new possibilities for real-time remote 3D vibration monitoring of civil structures (e.g. bridges and wind turbines).

The phases of the project and their realization:

Task 1.1: Implementation of the hybrid method for high-speed image based deflection shape identification for single projection view. Here, the recently published hybrid method was adopted for the single view deflection shape identification. The projection from the camera view to global coordinate system was researched and implemented in our own numerical code.  

Task 1.2: Development of camera calibration based on the digital twin. Due to multi-view, different perspectives need to be calibrated. The camera calibration was based on the numerical model (digital-twin). With this, each perspective was relatively easy to calibrate and the multi-view identification was made possible. The method was implemented in our own numerical code.  

Task 1.3: Development of the frequency domain triangulation for multi-view projections. The calibrated multi-view frequency domain deflection amplitudes obtained using Tasks 1.1 and 1.2 were here used for the frequency-domain triangulation. With this the 3D amplitudes of deflection shapes were identified. The method was published with MSSP (IF 5.006):