| KEYWORD |
Development of a low-power embedded solution for the real-time detection of structural cracks through acoustic emission
keywords ARTIFICIAL INTELLIGENCE, EDGE COMPUTING, EMBEDDED ALGORITHMS, EMBEDDED SIGNAL PROCESSING, INTERNET OF THINGS (IOT), REAL-TIME AUDIO PROCESSING
Reference persons STEFANO DI CARLO, AMEDEO DOMENICO BERNARDO MANUELLO BERTETTO, GIUSEPPE CARLO MARANO, ALESSANDRO SAVINO
External reference persons Alessio Carpegna (DAUIN), Jonathan Melchiorre (DISEG)
Research Groups DAUIN - GR-24 - SMILIES - reSilient coMputer archItectures and LIfE Sci
Thesis type APPLIED RESEARCH, EXPERIMENTAL RESEARCH
Description The identification of structural cracks using acoustic emission signals is increasingly capturing the interest of both researchers and companies due to its effectiveness.
Nowadays, structural health monitoring has become a prominent topic, particularly within the Italian territory. This focus is driven by its capability to facilitate the efficient maintenance of historic structures and infrastructure. The extensive adoption of structural monitoring necessitates the creation of techniques and instrumentation that are not only accurate but also cost-effective. In the acoustic emission technique, a collection of piezoelectric sensors is positioned on the structural surface to capture the acoustic emissions resulting from crack formation. Subsequently, post-processing techniques are employed on the recorded signals to retrieve information about the structural damage.
The goal of the thesis is to develop a system based on commercial embedded boards, like STM Nucleo and Xilinx Pynq-Z2. The system is intended to autonomously identify the formation of a crack, determine its location, and classify its type.
The outcome could be the development of a cost-effective system for continuous building monitoring, facilitating early detection of cracks.
Both standard signal processing techniques and Machine Learning approaches will be explored along the project, to compare them in terms of performance and requirements. Real-world tests will be conducted upon the system's completion, simulating the occurrence of a crack in an actual concrete block. The efficacy of the developed solution will be assessed in terms of real-time detection, location, and classification of the crack.
Required skills Embedded programming, C/C++ programming, python programming, signal processing, embedded electronics
Deadline 21/12/2024
PROPONI LA TUA CANDIDATURA