Self-powered sensing in structural health and usage monitoring.

摘要

Fatigue and overload of mechanical, civil and aerospace structures remains a major problem that can lead to costly repair and catastrophic failure. Long term monitoring of mechanical loading for these structures could reduce maintenance cost, improve longevity and enhance safety. However, the powering of these sensors throughout the lifetime of the monitored structure remains a major problem.;The ability to convert ambient energy into electric power would eliminate the problem of drained electrical supply, and would allow indefinite monitoring. This work first answers the key question: can sufficient electrical energy be produced from typical structural motions? Realistic earthquake, wind and traffic loads are used to calculate the theoretical maximum energy levels that can be extracted from these dynamic events. The same dynamic loads are used to calculate and experimentally measure the electrical energy produced by a realistic piezoelectric generator.;The collected energy levels are compared to the energy requirements of various electronic subsystems in a wireless sensor. For a 5 cm3 sensor node (the volume of a typical concrete stone), it is found that only extreme events such as earthquakes can provide sufficient energy to power currently available wireless sensors. For most typical dynamic events, it would be impossible to harvest enough energy to power a 5 cm3 wireless sensor. The results show that achieving continuous, self-powered, monitoring will require the development of a sensor node that can operate within a power budget of 1 microW.;The implementation of a novel self-powered fatigue monitoring sensor is presented. The sensor is based on the integration of piezoelectric transduction with floating gate avalanche injection. The miniaturized sensor enables self-powered, continuous monitoring and time-to-failure predictions of mechanical and civil structures. The sensor exploits a log-linear response of a current starved hot-electron injection process on a floating-gate transistor biased in the weak-inversion region. The measured response is shown to be minimally invariant to device mismatch and temperature fluctuations. By configuring an array of floating-gate transistors to respond to different amplitude levels of the input signal, the proposed circuit implements a level counting algorithm which is widely used in many usage monitoring techniques. Measured results from a fabricated integrated circuit in a 0.5-microm CMOS process demonstrate that the prototype can sense, store and compute over 107 loading cycles. The power dissipation of the prototype is measured to be 800nW which makes it ideal for autonomous long-term operation. The prototype is interfaced with different piezoelectric transducers and is tested in the laboratory to demonstrate its applicability for real-time usage monitoring.