Damage Detection of Bridge Network With Spatiotemporal Pattern Network

The alarmingly degrading state of transportation infrastructures combined with their key societal and economic importance calls for automatic condition assessment methods to facilitate smart management of maintenance and repairs. In particular, scalable data-driven approaches is of great interest, because it can deal with large volume of streaming data without requiring models that can be inaccurate and computationally expensive to run. Properly designed, a data-driven methodology could enable fast and automatic evaluation of infrastructures, discovery of causal dependencies among various sub-system dynamic responses, and inference and decision making with uncertainties and lack of labeled data. In this work, a spatiotemporal pattern network (STPN) strategy built on symbolic dynamic filtering (SDF) is applied to explore spatiotemporal behaviors in bridge network. Data from strain gauges installed on two bridges are simulated by finite element method, and the causality among strain data in spatial and temporal resolutions is analyzed. Case studies are conducted for truck identification and damage detection from simulation data. Results show significant capabilities of the proposed approach in: (i) capturing spatiotemporal features to discover causality between bridges (geographically close), (ii) robustness to noise in data for feature extraction, and (iii) detecting and localizing damage via the comparison of behaviors within the bridge network.

Detecting Behavioral Anomaly in Social Networks Using Symbolic Dynamic Filtering

This paper investigates the use of Symbolic Dynamic Filtering (SDF) algorithms in detecting anomalous behavior trends in social networks. Data is generated from an agent-based discrete choice model, which relies on a Markov Decision Process framework for stochastic simulation of decision-making in a social setting, where choices and decisions by individuals are influenced by social interactions. We show that such collective imitative behavior leads to rapid unstable fluctuations in the society, the fluctuation statistics being a weak function of the number of extremist nodes present in the network as well as the prevailing political climate. In this paper, using a time-trace of global opinions in the said society, we investigate the effectiveness of SDF in estimating the number of extremist nodes in a network, and studying the role of unpopular government policies as an enabler of political instability. Spread of influence and ‘recruiting’ by extremist groups through social networks has become an important political issue in recent years. This study is a step in the direction of building tools to preempt and intervene such efforts.

Investigation of Combustion Instability in a Swirl-Stabilized Combustor Using Symbolic Time Series Analysis

A laboratory-scale swirl-stabilized combustor is experimentally characterized for various configurations involving variable air flow rates and different fuel injection locations. Unsteady pressure and heat release rate measurements were obtained simultaneously in order to determine the stability map of the combustor for the experimented configurations. It is observed that a sharp rise in pressure amplitude coincides with a break in the dominant spectral content variation with the inlet Reynolds number. The time series data were analyzed by using the tools of symbolic dynamic filtering and the divergences among the outputs of each sub-class of observations were obtained as anomaly measures. In the proposed method, symbol strings are generated by partitioning the (finite-length) time series to construct a special class of probabilistic finite state automata (PFSA) that have a deterministic algebraic structure. The anomaly measures are defined based on the probabilistic state vectors distribution across each sub class. The method which is based on representing a given time series data as a set of PFSA is observed to be capable of predicting an impending combustion instability as well as to distinguish between the symbol-state distribution among various instability conditions. The measure also successfully captures changes in the thermoacoustic regime as a function of the fuel injection location.

On the Use of CFD and Symbolic Dynamic Filtering for Health Monitoring of Rotating Machinery

This paper presents an approach to use Computational Fluid Dynamics (CFD) analysis for the development of a health monitoring system based on Symbolic Dynamic Filtering (SDF) for rotating machinery. A simplified model of a turbomachinery (single rotor) is analyzed using commercial CFD software with and without blade damages. Virtual pressure sensors are placed on the case of the turbomachinery directly above the rotating blades to measure the dynamic pressure pulse generated by the rotating blades. The pressure pulse profiles from the rotating rotor blades with and without blade damages are processed using SDF to determine the presence and magnitude of the fault. Various degrees of damage and effect of measurement noise are examined.