Below is my PhD research proposal, submitted to a university, but I have had no luck. I posted it in my personal blog; maybe someone needs that as inspiration or for other purposes. Freely to copy, change, or improve my original research idea.
Abstract
Road and rail tunnels face risks from structural degradation (cracks, corrosion, etc.) and environmental stressors (flooding, seismic activity, etc.), yet conventional SHM systems often react post damage, lacking preventive capabilities. Conventional SHM systems, reliant on grid power, require extensive maintenance, increasing costs and limiting reliability in remote sections. This research proposes a “smart ear” system for preventive monitoring, using audio sensors (piezoelectric transducers) and predictive AI to enhance tunnel resilience. AI models, including RNNs and transformers, forecast risks before escalation, enabling proactive maintenance. Piezoelectric transducers harvest energy from traffic vibrations, ensuring sustainability. A digital twin integrates audio data to simulate tunnel behavior, optimizing preventive strategies. Targeting seismic and flood-prone regions globally, the system will be validated in lab and real-world international cases. Objectives include system design, AI development, and evaluation of safety, cost, and sustainability impacts. The approach is novel for its acoustic-based monitoring, predictive AI, and self-powered design. Hypothesis: The system will achieve 95% precision, 85% recall, and 88% mAP in risk prediction, with a 10 cm² piezoelectric transducer generating 100–500 µW at 1–5 V and 100–500 µA, able to power the monitoring device continuously for 30 days without external charging.
System Architecture
State of the Art
Road and rail tunnels are critical infrastructures, facing risks from structural degradation (cracks, corrosion, etc.) and environmental stressors (flooding, seismic activity, etc.). Structural Health Monitoring (SHM) Systems traditionally use sensors like accelerometers and strain gauges to detect damage via vibration or strain analysis (Farrar & Worden, 2013). These systems often react to damage post-occurrence, limiting preventive capabilities. Preventive monitoring, which predicts risks before they manifest, is an emerging need but underdeveloped for tunnels due to complex environments (e.g., noise, humidity) and data challenges.
Acoustic-based SHM, using Acoustic Emission (AE) or environmental sound analysis, offers potential for early detection. AE captures high-frequency signals from micro-cracks, while piezo transducer detect environmental changes (water ingress, ventilation anomalies) (Grosse & Ohtsu, 2008). Studies like Zhang et al. (2020) applied AE for bridges, but tunnel applications are limited by noise from traffic and echoes. Predictive AI, particularly deep learning models like Recurrent Neural Networks (RNNs) and Transformers, has advanced SHM by forecasting degradation trends (Azimi et al., 2020). Wang et al. (2023) used AI for AE in composites, but predictive models for tunnel audio data are scarce due to noise interference and limited datasets.
Power supply is a key challenge for SHM in tunnels. Traditional systems rely on grid power, which requires maintenance and are impractical in remote tunnel sections (Lynch & Loh, 2006). Energy harvesting, particularly via piezoelectric transducers, offers a sustainable solution by converting mechanical vibrations from traffic into electrical power. Piezoelectric harvesting can yield 100–500 µW/cm² from traffic vibrations (10–50 Hz), sufficient for low-power sensors (10–100 mW) and IoT modules (Erturk & Inman, 2011). A 10 cm² transducer array could power a single node, making it ideal for tunnel environments where vibrations are abundant. While applied in pavements and bridges, its integration with predictive SHM in tunnels is novel, especially for dual-purpose monitoring and power generation. Digital twins, virtual models integrating real-time data, enable predictive maintenance by simulating infrastructure behavior (Ye et al.,2021). Their use in audio-based SHM for tunnels is rare, offering untapped potential for preventive strategies.
Global standards like Eurocode and ITA emphasize tunnel resilience but lack guidance on preventive acoustic monitoring or predictive AI. This research addresses these gaps by developing a self-powered “smart ear” system for preventive monitoring, using predictive AI to analyze audio data from piezoelectric transducers, integrated with energy harvesting and digital twins.
Research Objectives
The goal is to develop a self-powered “smart ear” system for preventive monitoring of road and rail tunnels using predictive AI and acoustic sensors, enhancing resilience through proactive risk mitigation. The objectives are:
- Design an Acoustic Monitoring System: Develop a network of audio sensors (piezoelectric transducers and microphones) to capture structural acoustic emissions (micro-cracks) and environmental sounds (water ingress, ventilation issues), integrated with IoT for real-time data in seismic and flood-prone regions.
- Develop Predictive AI Algorithms: Create AI models, including RNNs and Transformers, to forecast structural damage and environmental risks. These models will predict degradation trends and anomalies, enabling preventive maintenance by filtering noise in tunnel environments.
- Incorporate Energy Harvesting: Utilize piezoelectric transducers for dual-purpose monitoring and energy harvesting, powering the system with traffic vibrations to ensure sustainability, a novel feature for preventive SHM.
- Implement a Digital Twin Framework: Develop a digital twin to integrate audio data with predictive models, simulating tunnel behavior under stressors (seismic events, flooding) to optimize preventive strategies.
- Validate and Assess Preventive Impact: Test the system in a lab-scale model and a real-world setting, evaluating its predictive accuracy and sustainability. Metrics include early risk prediction rate, piezoelectric power generation, and consumption.
This program delivers a predictive, sustainable approach to tunnel resilience, leveraging audio sensors, predictive AI, and digital twins for preventive monitoring over three year timeline.
Here is my presentation
a Technopreneur – writer – Enthusiastic about learning AI, IoT, Robotics, Raspberry Pi, Arduino, ESP8266, Delphi, Python, Javascript, PHP, etc. Founder of startup Indomaker.com