This paper presents a systematic approach to the design and implementation of an energy-efficient multi-sensor network. The nodes of the sensor network form the basis of a sectioned Bayesian network that can be used to determine the state of the system being monitored. A key issue in the design of Bayesian networks for monitoring engineering systems is to ensure that reliable inference scheme about the health state of the system can be made by combining information acquired from each sensor in the system into a single Bayesian network. As the size of the network increases, aggregating information made by all the sensors becomes computationally intractable. Hence, sectioning of the Bayesian network based on functional or logical constraints allows computational efficiency in aggregating information and reduces overall communication requirements. Furthermore, an in-network data processing scheme, motivated by the concept of Dynamic Voltage Scheduling, has been investigated to minimize computation energy consumption through dynamically adjusting the voltage supply and clock frequency of the individual sensors. As a result, the processor idle time can be better utilized for prolonged computation latency, leading to significantly reduced energy cost and increased computational efficiency.
This paper presents an adaptive filtering technique for the health diagnosis of mechanical systems, based on the generalized harmonic wavelet transformation. Through selection of two wavelet level parameters, a series of sub-frequency band wavelet coefficients corresponding to equi-bandwidth vibration signals measured from a machine were constructed. The energy and entropy associated with each sub-frequency band were then calculated, and the band with the maximum energy-to-entropy ratio was chosen to form a band-limited filter for the vibration signals. Experimental studies using rolling bearings that contain structural defects have confirmed that, the developed new technique enables high signal-to-noise ratio for effective machine failure detection and health diagnosis.
This paper presents a new approach to on-line health diagnosis of mechanical systems, based on the wavelet packet transform. Specifically, signals acquired from vibration sensors are decomposed into sub-bands by means of the discrete harmonic wavelet packet transform (DHWPT). Based on the Fisher linear discriminant criterion, features in the selected sub-bands are then used as inputs to three classifiers (Nearest Neighbor rule-based and two Neural Network-based), for system health condition assessment. Experimental results have confirmed that, comparing to the conventional approach where statistical parameters from raw signals are used, the presented approach enabled higher signal-to-noise ratio for more effective and intelligent use of the sensory information, thus leading to more accurate system health diagnosis.