Stochastic iterative modal identification algorithm and application in wireless sensor networks

Abstract

SUMMARY Computational capability of wireless sensor network (WSN) significantly facilitates application of dense sensory arrays, which is increasingly important in health monitoring of large-scale structural systems. As the wireless sensor technology improves, more complicated tasks can be assigned to sensing units, and the communication between sensing nodes and their base station can be minimized, utilizing in-network processing. This strategy should be used to address WSN challenges, such as limited communication bandwidth and prohibitive power consumption, associated with wireless communication and battery power. An iterative modal identification algorithm is proposed in this paper, which uses the on-board processors for estimation of system parameters through iteration cycles. The iterative algorithm was originally developed such that each individual sensor, having an initial estimate of the system parameters, its local measurement, and the excitation signal, updates the estimated model and passes it through the network until convergence. This study further improves the algorithm to eliminate its limitations in need for availability of excitation load and initial estimate of the system parameters. As a result, the algorithm is applicable for modal identification of structural systems under ambient loading without need for prior information about the system parameters. The development of the algorithm is presented in this paper and validated through implementation on a numerically simulated example and a laboratory experiment. Furthermore, its performance is evaluated using data from an ambient vibration test of the Golden Gate Bridge using a WSN. Results of these implementations verify the functionality of the algorithm in monitoring of real-life structural systems. Copyright © 2012 John Wiley & Sons, Ltd. Received 10 May 2012; Revised 27 August 2012; Accepted 31 August 2012