NECSI Resources

 International Conference on Complex Systems (ICCS2007)

Energy Constrained Information Capture Using Adaptive Self-Powered Sensor Nodes

Fred Discenzo
Rockwell Automation

Kenneth A. Loparo
Engineering & Computer Science, Case Western Reserve University

Dukki Chung
Rockwell Automation

Farhad Kaffashi
Engineering & Computer Science, Case Western Reserve University

     Full text: Not available
     Last modified: May 31, 2007

Abstract
Motors, gears, pumps, fans, and compressors are critical elements in many industrial, commercial, and shipboard systems. Continuously monitoring critical machines with remote sensors can provide notice of incipient failure and permit corrective action or an orderly shutdown before a failure occurs. The use of wireless sensor nodes avoids the cost and reliability problems associated with direct-wired sensors but at the expense of routinely replacing batteries. The use of power scavenging techniques promises to eliminate the need for batteries or local line power. However, traditional power scavenging techniques tend to be brittle and are designed for specific environments. An adaptive power scavenging device has been developed and tested that dynamically responds to changes in the environment. The resonant frequency of the piezo-electric generator changes dynamically in response to the changing environment. A linear micro-motor continually re-tunes a cantilever generator to track the changes in vibration frequencies and amplitudes in the environment. The adaptive energy harvesting device operates autonomously with the dual objectives of surviving (insuring adequate power generation and adequate power reserves), and extracting maximum information from the environment to avoid machinery failures or process upsets.

The adaptive self-powered sensor node includes the following elements; sensors, processor, radio, generator, actuator, and power conversion electronics. Together these elements operate as an integrated, tightly coupled system. For example the amount of power extracted from the generator element changes the power conversion characteristics of the generator. Continuously tuning the generator for optimum power generation may leave no power to support the core functions of sensor node. By judiciously re-tuning the generator, perhaps in a sub-optimal manner, we may obtain increased generated power needed for state estimation. In addition, the changed response from the re-tuned generator provides additional environmental information useful for fault detection.

An information theoretic approach may be employed to optimize the operation of the adaptive self-powered sensor node. Information entropy or state uncertainty may be reduced by sensor operation, processing sampled data, adaptively tuning the generator, or by radio communications. Certain regimes may possess greater information content or certainty but require substantial energy or risk to acquire the information. The objective is to maximize the relevant information extracted from the environment by the strategic allocation of available energy. Core system models are augmented with statistical and predictive elements and are extensible to multi-node and agent-based systems. An information-theoretic approach to optimize sensor node operation subject to power constraints promises to provide maximum relevant information while operating in dynamic, unexpected environments.







Maintained by NECSI Webmaster Copyright © 2000-2007 New England Complex Systems Institute. All rights reserved.