OPTICAL METHODS FOR VIBRO-ACOUSTIC MEASUREMENTS

The accurate characterisation of structural vibrations and their associated acoustic radiation represents a key task in vibro-acoustic research, with direct implications for design, diagnostics, and control. Conventional approaches rely on accelerometers, laser vibrometers, and microphone arrays, which, while accurate, are costly, intrusive, and impractical for in situ applications. This doctoral research develops vision-based methodologies as alternative routes to reconstruct vibration and sound radiation fields, emphasising total sound power radiation. Three case studies are presented, addressing simple laboratory structures and realistic closed shells. The potential of event-based cameras is also explored to enable lightweight vibration image processing. The objective is to advance the feasibility, accuracy, and applicability of vision-based vibro-acoustic measurements. The first contribution concerns the optical reconstruction of vibration and acoustic radiation from a thin-walled cylindrical shell. A framework integrating high-speed camera acquisitions with digital image correlation and photogrammetry recovers the displacement field. The reconstructed vibration fields are validated against laser vibrometer data, while radiated acoustic fields are benchmarked against a microphone array, with deviations within 1–3 dB. A formulation derives total radiated sound power from surface-integrated sound intensity, approximated by a Riemann summation over elemental radiators. This demonstrates that camera-based methods can achieve accuracies comparable to traditional measurements while requiring fewer sensors and offering non-intrusive, full-field acquisitions. The second contribution extends the methodology to a machine-like vibro-acoustic source, a thin-walled metallic box, shifting from laboratory models to a realistic scenario. The study confirms the suitability of vision-based techniques for structures with discontinuities and geometric complexities. A methodological advancement replaces the classical far-field integration of sound intensity prescribed by ISO standards with a formulation collapsing the Kirchhoff–Helmholtz integral onto the radiating surface. Implemented via the Boundary Element Method, it allows the derivation of sound power directly from reconstructed vibration fields. Results highlight both the accuracy and advantages of the proposed approach for in situ applications where conventional facilities may not be available. The third contribution explores event-based cameras for vibration measurement. Unlike frame-based cameras, they asynchronously record pixel-level brightness changes, achieving microsecond temporal resolution with minimal data redundancy. Event-to-frame reconstruction using the E2VID neural network, multi-camera calibration, subpixel tracking, and 3D triangulation enable accurate reconstruction of flexural deflection shapes on a cantilever beam, with RMS errors below 4% when compared with laser vibrometer measurements. Parametric analysis shows increased robustness with four synchronised cameras and sub-microsecond alignment. Beyond structural validation, the study highlights the potential of event cameras as compact, high-speed, and data-efficient sensors for vibro-acoustic applications. Overall, this doctoral work consolidates and expands the role of vision-based methods in vibro-acoustics, showing that optical sensing, whether frame-based or event-driven, can provide dense, accurate, and non-intrusive measurements of vibration fields and radiated sound power. The formulations and experimental validations indicate that such techniques can match or exceed traditional measurements while offering scalability, flexibility, and applicability in real environments. This thesis provides a foundation for future research at the intersection of structural dynamics, acoustics, and optical measurements, with promising implications for laboratory studies and in situ monitoring.