CSR: Medium: Self-organizing Cyber Substrates: Exploring a Modular Computing and Communications Architecture for Structural Health Monitoring

点击翻译按钮获取中文摘要

Many physical structures, including airplane wings, bridges, wind turbine blades, dams, and buildings, are subject to the development of fractures over time. Detecting these problems early is essential to preventing major catastrophes. A promising approach is to install sensors that continuously monitor and report on the structure's integrity. However, adoption has been limited because of several limitations in current technology, including a need for frequent battery replacement for battery-powered sensors, limited interrogation range of passive sensors, and a separate communication subsystem usually involving radio frequency sensor networks. This project aims to remove these barriers by exploring a set of new technologies, including mechanical energy harvesting and ultra-sound communication using the substrate of the structure itself.The long term vision of this research is to develop a transformative self-organizing embedded system design paradigm that facilitates convergence of self-powered computing, self-powered networking and structural health monitoring into "cyber-substrates", that can "sense", "feel" and "diagnose" impending catastrophic failures. The approach is based on a new integrated sensing, signal processing, and communication chipset, which scavenges both strain and vibration energy from the substrate to which it is attached or embedded into. Energy is accumulated in a capacitor for computing and communications. These chips are capable of enough local signal processing to detect events indicative of failures, so that their communication requirement is reduced to sending a single bit per event to a logging device. Key innovations in this project are in the areas of: 1) embedded computing hardware for sensing, event classification, and signal processing with ultra-low power budgets, 2) packet-less and through-substrate pulse networking for extreme energy economy, 3) energy scavenging from the substrate structure, 4) binary computation models for structural health evaluation and event predictions, and 5) smart embedded software for self-organizing integration of all the above elements. All these components are designed around the core theme of binary information framework which is heavily leveraged for power optimization in sensing, communication, and global inference algorithms for structural processing. Education and outreach activities include involvement of diverse graduate, undergraduate and high school students in the research, and development of a new interdisciplinary graduate course in structural health monitoring.