Title / Author(s) / Keywords

Publication  Date 
201608 EWSHM 2016 Diagnostics and Prognostics for Damage Assesment Damage Localization from Free Vibration Signals D. Bernal^{1}^{5}, Y. Zhang^{2} ^{1}^{}Civil and Environmental Engineering; Northeastern University^{12}, Boston, MA, USA ^{2}^{}School of Aerospace Engineering and Applied Mechanics; Tongji University ^{24}, Shanghai, China defect characterization, subspace technique, damage localization, free vibration, random decrement
Close
An approach to localize damage that operates without compressing the data into a modal model is reviewed. The metric used to provide localization information measures how well a vector of stacked Fourier transforms of free vibration at different frequencies in the damaged state fits in a subspace obtained from a model of the reference state that depends on the postulated damage distribution. It is recognized that recording of freevibration is not feasible in many instances so the extraction of freevibration signals from forced vibration response using the random decrement technique is reviewed. Robustness in the approach is promoted by aggregating results obtained for different initial conditions which are realized by adjustments of the triggering condition in the random decrement scheme. While the method does not impose constraints relating the number of sensors to the model degrees of freedom, the number of sensors, to be within the confines of the theory, must be no less than the rank of the change in the transfer matrix resulting from damage. The performance of the procedure, which has been designated as the FreeVibration Damage Localization (FVDL) approach, is illustrated in simulations that include measurement noise and model error and is shown to operate well.
^{}
 EWSHM 2016 Session: Diagnostics and Prognostics for Damage Assesment  201608 
Sensor Placement for Modal Parameter Subset Estimation: A Frequency Responsebased Approach M. Ulriksen^{1}^{6}, D. Bernal^{2}^{5}, L. Damkilde^{1}^{4} ^{1}^{}Department of Civil Engineering; Aalborg University (Campus Esbjerg)^{4}, Esbjerg , Denmark ^{2}^{}Civil and Environmental Engineering; Northeastern University^{12}, Boston, MA, USA civil engineering, vibration analysis, Modal parameter estimation, Fisher information, Sensor placement approaches
Close
The present paper proposes an approach for deciding on sensor placements in the context of modal parameter estimation from vibration measurements. The approach is based on placing sensors, of which the amount is determined a priori, such that the minimum Fisher information that the frequency responses carry on the selected modal parameter subset is, in some sense, maximized. The approach is validated in the context of a simple 10DOF massspringdamper system by computing the variance of a set of identified modal parameters in a Monte Carlo setting for a set of sensor configurations, whose anticipated effectiveness are ranked according to the noted criterion. It is contended that the examined maxmin criterion satisfies the objective of optimizing the accuracy of an identification setup more effectively than the more commonly used trace or determinant of the Fisher information matrix (FIM). It is shown that the widely used Effective Independence (EI) method, which uses the modal amplitudes as surrogates for the parameters of interest, provides sensor configurations yielding theoretical lower bound variances whose maxima are up to 30 % larger than those obtained by use of the maxmin approach.
^{}
 EWSHM 2016 Session: Optimization sensors topology  201608 
Algorithm Fusion in Novelty Detection D. Bernal^{5} ^{}Center for Digital Signal Processing; Northeastern University^{12}, Boston, MA, USA Data fusion, Parametric identification, Damage detection and localization
Close
Algorithm fusion has received significant attention in the machine learning community in supervised learning mode but it appears little has been done at this point in a novelty detection framework. This paper examines the merit of a fusion strategy wherein metrics from multiple algorithms are treated as entries of a vector whose probability density is subsequently estimated and used for detection. In the present paper the framework is investigated using two algorithms: 1) a robust version of a whiteness test on Kalman filter innovations and 2) a robust version of a scheme that operates with residuals obtained from an orthogonality test. The density estimation part of the process is replaced by the Kernel PCA algorithm which provides a decision boundary without having explicit density estimates. The fused scheme is implemented in a change detection format and is show to provide notable improvements over the use of either algorithm independently.
^{}
 EWSHM 2014 Session: Probabilistic SHM  201502 
On the Stability of Sequential Deconvolution D. Bernal^{5} ^{}Center for Digital Signal Processing; Northeastern University^{12}, Boston, MA, USA signal processing, deconvolution, Inverse problems, source localization, input reconstruction
Close
The time domain estimation of inputs from knowledge of the kernel and the outputs is a deconvolution operation. For finite dimensional systems treated in discrete time the operation is tantamount to solving a set of linear equations. When performing a deconvolution an issue that must be dealt with is the fact that the dimension of the system of equations grows with duration and becomes prohibitively large when the inputs are long. For this reason, and sometimes because it is of interest to estimate the inputs with the smallest possible delay, deconvolution must often be implemented on a moving window. This paper shows that a sequential deconvolution is a conditionally stable process and derives the expression that governs numerical stability.
^{}
 EWSHM 2014 Session: Signal Processing  201502 
Examination of Two Sensor Placements Schemes in Damage Detection D. Bernal^{1}^{5}, M. Döhler^{2}^{7}, D. Parker^{3} ^{1}^{}Center for Digital Signal Processing; Northeastern University^{12}, Boston, MA, USA ^{2}^{}Campus de Beaulieu; Institut national de recherche en informatique et en automatique (INRIA)^{12}, Rennes cedex, France ^{3}^{}Department of Mechanical Engineering; AVNIK Defense Solutions Inc., Huntsville, AL , USA Sensor systems and networks, Damage detection and localization, Statistical methods
Close
Deciding on the position of sensors by optimizing the utility of the monitoring system over a structureÕs lifetime is typically forbidden by computational cost. Sensor placement strategies are, instead, usually formulated for a preselected number of sensors and are based on cost functions that can be evaluated for any arrangement without the need for simulations. This paper examines the performance of two such schemes, the first one is derived directly from a technique that detects damage from the shift of a chisquare distribution from central to noncentral and takes the optimal arrangement as the one that maximizes the sensitivity of the noncentrality to all parameter changes of equal norm. The second scheme selects the sensor arrangement as that which maximizes a weighted version of the norm of the sensitivity of the covariance of the output to all feasible changes in system parameters. The performance of the two schemes is tested in simulations.
^{}
 EWSHM 2014 Session: Statistical Approaches  201502 
