Civil structures, such as buildings and bridges, are constantly at risk of failure due to extensive environmental loads caused by earthquakes or strong winds. In order to minimize this risk, the application of control systems for civil infrastructure stabilization has been proposed. However, implementation challenges including communication latencies, computation inundation at the actuation node, and data loss have been impeding large-scale deployment. In order to overcome many of these challenges, inspiration can be drawn from the signal processing techniques employed by the biological central nervous system. This work uses a bio-inspired wireless sensor node, capable of real-time frequency decomposition, to simplify computations at an actuating node, thus alleviating both communication and computation inundation and enabling real-time control. The simplistic control law becomes 𝐅 = 𝐰𝐍, where 𝐅 is the control force to be applied, 𝐰 is a weighting matrix that is specific to the structure, and 𝐍 is the displacement data from the wireless sensor node. There is no empirical solution for deriving the optimal weighting matrix, 𝐰, and in this study the particle swarm optimization technique was used as a means for determining values for this matrix. Multiple parameters of this optimization method were explored in order to produce the most effective control. This bio-inspired approach was applied in simulation to a five story benchmark structure and using performance metrics it was concluded that this method performed similar to more traditional control method.
Wireless sensor networks (WSNs) have emerged as a reliable, low-cost alternative to the traditional wired sensing paradigm. While such networks have made significant progress in the field of structural monitoring, significantly less development has occurred for feedback control applications. Previous work in WSNs for feedback control has highlighted many of the challenges of using this technology including latency in the wireless communication channel and computational inundation at the individual sensing nodes. This work seeks to overcome some of those challenges by drawing inspiration from the real-time sensing and control techniques employed by the biological central nervous system and in particular the mammalian cochlea. A novel bio-inspired wireless sensor node was developed that employs analog filtering techniques to perform time-frequency decomposition of a sensor signal, thus encompassing the functionality of the cochlea. The node then utilizes asynchronous sampling of the filtered signal to compress the signal prior to communication. This bio-inspired sensing architecture is extended to a feedback control application in order to overcome the traditional challenges currently faced by wireless control. In doing this, however, the network experiences high bandwidths of low-significance information exchange between nodes, resulting in some lost data. This study considers the impact of this lost data on the control capabilities of the bio-inspired control architecture and finds that it does not significantly impact the effectiveness of control.
Damage detection on engineered systems is a challenging task that has been explored by numerous researchers. In recent years wireless sensors systems have arisen as a vehicle for low-power, low-cost, and localized damage detection that can be applied to various structural systems. Such sensors, however, are limited in their computational capacity and as a result, careful consideration must be taken as to which algorithms can be effectively embedded so as to balance energy constraints with computational efficiency. In this study, two classifier algorithms (least squares classifier and Fisher's linear discriminant analysis) are explored for detecting damage on a cooling system test bed. In particular, the algorithms are used to determine the valve configuration of the system and to verify if damage exists within the valves. To validate the efficiency of the algorithms in the embedded domain, the algorithms are implemented on a wireless sensing network and used to classify the system state of the test bed.
While sensing technologies for structural monitoring applications have made significant advances over the last several decades, there is still room for improvement in terms of computational efficiency, as well as overall energy consumption. The biological nervous system can offer a potential solution to address these current deficiencies. The nervous system is capable of sensing and aggregating information about the external environment through very crude processing units known as neurons. Neurons effectively communicate in an extremely condensed format by encoding information into binary electrical spike trains, thereby reducing the amount of raw information sent throughout a neural network. Due to its unique signal processing capabilities, the mammalian cochlea and its interaction with the biological nervous system is of particular interest for devising compressive sensing strategies for dynamic engineered systems. The cochlea uses a novel method of place theory and frequency decomposition, thereby allowing for rapid signal processing within the nervous system. In this study, a low-power sensing node is proposed that draws inspiration from the mechanisms employed by the cochlea and the biological nervous system. As such, the sensor is able to perceive and transmit a compressed representation of the external stimulus with minimal distortion. Each sensor represents a basic building block, with function similar to the neuron, and can form a network with other sensors, thus enabling a system that can convey input stimulus in an extremely condensed format. The proposed sensor is validated through a structural monitoring application of a single degree of freedom structure excited by seismic ground motion.
Wireless sensing technologies have recently emerged as an inexpensive and robust method of data collection in a variety
of structural monitoring applications. In comparison with cabled monitoring systems, wireless systems offer low-cost
and low-power communication between a network of sensing devices. Wireless sensing networks possess embedded
data processing capabilities which allow for data processing directly at the sensor, thereby eliminating the need for the
transmission of raw data. In this study, the Volterra/Weiner neural network (VWNN), a powerful modeling tool for nonlinear
hysteretic behavior, is decentralized for embedment in a network of wireless sensors so as to take advantage of
each sensor's processing capabilities. The VWNN was chosen for modeling nonlinear dynamic systems because its
architecture is computationally efficient and allows computational tasks to be decomposed for parallel execution. In the
algorithm, each sensor collects it own data and performs a series of calculations. It then shares its resulting calculations
with every other sensor in the network, while the other sensors are simultaneously exchanging their information.
Because resource conservation is important in embedded sensor design, the data is pruned wherever possible to eliminate
excessive communication between sensors. Once a sensor has its required data, it continues its calculations and computes
a prediction of the system acceleration. The VWNN is embedded in the computational core of the Narada wireless
sensor node for on-line execution. Data generated by a steel framed structure excited by seismic ground motions is used
for validation of the embedded VWNN model.