Tens of thousands of the nation?s bridges are structurally deficient. This project proposes to design a self sustaining, wireless structural monitoring system. The novel low-power Flash FPGA-based hardware platform and the corresponding software architecture offer a radically new approach to CPS design. A soft multi-core platform where software modules that run in parallel will be guaranteed to have dedicated single-threaded soft processor cores enables flexible power management by running only the necessary cores at any given time at the slowest clock rate mandated by the observed/controlled physical phenomena. As bridges tend to vibrate due to wind and dynamic load conditions, we are developing a novel vibration-based energy harvesting device that is capable of automatically adjusting its resonant response in order to capture much more energy than the current techniques can. Moreover, the PIs are developing structural health assessment techniques involving quantitative analysis of signals to determine crack type, location and size.
The technology will indicate structural problems before they become critical potentially saving human lives and averting late and extensive repairs. The impact of the vibration harvesting technique and the soft multi-core architecture will go beyond structural monitoring. A separate soft core dedicated to each software component that interacts with the physical world will make CPS more responsive while saving power at the same time.
The education plan focuses on outreach toward underrepresented minorities by recruiting such undergraduates to participate in the research. To facilitate the dissemination of our results, all hardware designs and software developed under this project will be open source.
There are close to 33,000 steel railroad bridges and 600,000 highway bridges in the United States, over quarter of which are structurally deficient or functionally obsolete. One of the most comprehensive bridge testing methods is based on acoustic emissions, however current AE measurement systems based on hardwired centralized data collection are expensive, power hungry and heavy, thus these systems are used for the occasional inspection of a few selected bridges. This grant supported the design, prototype development and experimental verification of a self sustaining, autonomous, wireless structural monitoring system based on stress, vibration and acoustic emission sensing. Our research effort was based on three pillars: (1) a low-power multi-core Flash FPGA-based hardware platform, (2) a vibration based self-adaptive energy harvesting technology and (3) low-power signal processing and event classification algorithms based on acoustic emission signals to determine crack type, location, and growth. In addition to the obvious impact of the overall system, the individual technology components are of great value themselves with broad applicability in future CPS designs. Our open-source MarmotE wireless sensor platform provides a proven and refined modular hardware platform for low-power sensing and data collection (https://sites.google.com/site/marmoteplatform/). Built on this hardware infrastructure our team developed a scalable multi-core architecture using variable number of soft microcontroller cores, with integrated communication and synchronization primitives. Software and application development efforts are supported by the developed code generator and static analysis tools and an instruction set simulator, which was extended with the multi-core capabilities of the MarmotE platform. Our software model and tools are natural extensions of TinyOS, a well established software framework targeting wireless sensor networks. The need for continuous and permanent deployment of such sensors resulted in the development of a new modeling and electrical load analysis approach for electromagnetic vibration energy harvesters. The ultimate goal of this effort is to utilize the vibrating motion of the bridge as a perpetual energy source, with little or no a priori knowledge on the frequency and displacement of the vibration. This technique was validated experimentally using a custom-built electromagnetic energy harvester. Electrical tuning was performed using purely resistive electrical loads in an attempt to increase harvester power output in cases when the excitation frequency to which the harvester was subjected did not match the harvester’s mechanical natural frequency. The project concluded with the PhD dissertation defense of one graduate student sponsored by this award. His thesis provides a detailed summary of the efforts and results of the project. The grant also partially supported the research of two other PhD candidate students. Our results were published and resulted in six conference papers, one book chapter and five journal publications.
