Moineddini, Amirbahador; (2024) Development and Evaluation of Multichannel Nano-Fibre Piezoelectric Acoustic Transducers: Leveraging Neural Networks And Exploring Novel Structural Fabrications. Masters thesis (M.Phil), UCL (University College London).

Text Moineddini_thesis.pdf - Other Access restricted to UCL open access staff until 1 August 2025. Download (5MB)

Abstract

This MPhil research project focuses on developing a test platform for the characterization of a novel class of acoustic sensors developed by the Division of Surgery and Interventional Sciences at UCL. Additionally, the project explores fabrication methods and materials to enhance the performance of these sensors. To achieve this objective, an automated test instrument (testbed) was designed and manufactured in the lab for precise and automated data collection. Furthermore, a data analysis algorithm was developed to standardize data collection and automate the testing and analysis of data collected from the devices. The hardware and software for this project were tightly integrated with bioinspired piezoelectric nanocomposite nanofiber-based acoustic sensors, showing promising results for the next generation of self-powered cochlear implants. The testbed developed in this project serves as a normalized test platform that enables testing various sensors, recording data, and analysing the performance of different iterations of these sensors using a new neural network algorithm for speech and spatial recognition (sound source localisation). The device can automatically collect and process data from multiple sensor channels and train neural networks for testing these devices. These acoustic sensors have been systematically characterized and demonstrated high-frequency selectivity and multifunctional capabilities in speech recognition and localization. This is attributed to the specific geometry of the sensors electrodes and the piezoelectric properties of highly aligned radially polymer nanofibers made of poly(vinylidene fluoride-co-trifluoroethylene) (PVDF-TrFE) doped with Barium Titanate (BaTiO3) nanoparticles. Moreover, using the testbed, it was demonstrated that a single multichannel asymmetrical device could localize sound sources, in contrast to human hearing, which requires both ears for localization. This attribute results from the asymmetrical nature of the sensor combined with the design of the neural network used for the sensors. Additionally, this report proposes a new method of fabricating the sensors using Electrohydrodynamic direct printing instead of the electrospinning process currently used. This method offers more control over the diameter, orientation, and placement of the fibres on the electrode, making the fabrication process slower but more repeatable. Finally, the report discusses how the fibre structure could be modified using coaxial printing to improve the endurance and performance of the sensors.

Download activity - last month

Export as JSONXMLCSV

Download activity - last 12 months

Export as XMLCSVJSON

Downloads by country - last 12 months

Export as CSVXMLJSON

Archive Staff Only

View Item