Farshid Abbasi

New structural damage-identification method using modal updating and model reduction

Early prediction of damages using vibration signal is essential in avoiding the failure in structures. Among different damage-detection approaches, the finite-element model updating and modal analysis-based methods are of most importance due to their applicability and feasibility. Owing to some restrictions in nodal measurements in experimental cases, finite-element model reduction is an indispensable part of fault-detection methods. Even though model reduction of dynamic systems leads to the less complicated models, an improved convergence rate and acceptable accuracy are highly required for a successful structural health monitoring of the real complex systems. In this paper, the aim is to design a damage-detection algorithm based on a new model updating method, which has a faster rate of convergence and higher accuracy. Then the proposed method is applied on a simulated damaged beam considering different noise levels to see how capable the method is in dealing with noise-corrupted data. Finally, the experimentally extracted data from a cracked beam in a real noisy condition are used to evaluate the efficiency of the proposed method in identifying the damages in a beam-like structure. It is concluded that the identification of the damages by the proposed method is encouraging and robust to the noise compared with the traditional method. Also, the proposed method converges faster and is more accurate in identifying damage than the traditional method.

Fault diagnosis using noise modeling and a new artificial immune system based algorithm

A new fault classification/diagnosis method based on artificial immune system (AIS) algorithms for the structural systems is proposed. In order to improve the accuracy of the proposed method, i.e., higher success rate, Gaussian and non-Gaussian noise generating models are applied to simulate environmental noise. The identification of noise model, known as training process, is based on the estimation of the noise model parameters by genetic algorithms (GA) utilizing real experimental features. The proposed fault classification/diagnosis algorithm is applied to the noise contaminated features. Then, the results are compared to that obtained without noise modeling. The performance of the proposed method is examined using three laboratory case studies in two healthy and damaged conditions. Finally three different types of noise models are studied and it is shown experimentally that the proposed algorithm with non-Gaussian noise modeling leads to more accurate clustering of memory cells as the major part of the fault classification procedure.

Power line robotic device for overhead line inspection and maintenance

Purpose This paper aims to focus on the design and testing of a robotic device for power line inspection and cleaning. The focus for this design is on simplicity and compactness with a goal to create a device for linemen and other power line workers to keep in their toolbox. Design/methodology/approach The prototype uses V-grooved wheels to grip the line and can pass obstacles such as splices. It is equipped with a video camera to aid in line inspection and a scrub brush to clean debris from the line. The operator controls the device remotely from a laptop through a wireless connection. The novel way in which this device moves down the power line allows compactness while still being able to overcome in-line obstacles up to a certain size. Findings The device has been tested on a test bed in the lab. The device is able to move down a line and expand to overcome in-line obstacles as it travels. Testing proved the mechanical feasibility and revealed new requirements for a future prototype. Practical implications The device can be used for power line asset management by power companies; line inspection can lead to preventative repairs, leading to less downtime. Social implications It stands to reduce costs related to maintenance and mitigates down time and emergency repairs. Originality/value Innovative features include its size, mobility and control methods. Overall, the impact of this work extends to the utility maintenance sector and beyond.

A support vector machine-based method for LPV-ARX identification with noisy scheduling parameters

In this paper, we present a method that utilizes support vector machines (SVM) to identify linear parameter-varying (LPV) auto-regressive exogenous input (ARX) models corrupted by not only noise, but also uncertainties in the LPV scheduling variables. The proposed method employs SVM and takes advantage of the so-called “kernel trick” to allow for the identification of the LPV-ARX model structure solely based on the input-output data. The objective function, as defined in this paper, allows to consider uncertainties related to the LPV scheduling parameters, and hence results in a new formulation that provides a more accurate estimation of the LPV model in the presence of scheduling uncertainties. We further demonstrate the viability of the proposed LPV identification method through numerical examples, where we show that higher best fit rate (BFR) can be achieved under realistic noise conditions using the proposed method compared to the method initially proposed in [6].

Coverage control of moving sensor networks with multiple regions of interest

This paper addresses the coverage control problem in environments where several regions of interest exist. To this purpose, a heterogeneous group of robots are deployed to minimize a cost function defined with respect to various spatial probability density functions, each of which describes a desired area for a different group of robots. Each region of interest is assigned to a group of robots with respect to their dynamics and sensing capabilities. A distributed coverage scheme is proposed to allow adjusting to the environment with several important areas in a collaborative way. The regions with higher importance would be covered with an appropriate number of robots. The proposed method also allows for a better allocation of robots to guarantee the desired coverage over the region. Two numerical examples are finally given to examine the proposed coverage approach in case of multiple regions of interest that may need to be covered by a certain number of robots.

A New Voronoi-Based Blanket Coverage Control Method for Moving Sensor Networks

This brief addresses the blanket coverage problem, in which it is desired to cover a long region by moving the blanket within the boundaries representing the main region. To this purpose, a group of autonomous mobile sensors are deployed aiming at maximizing the sensing performance. The blanket coverage area, which is considered to be a region with changing boundaries, is directed to move along the boundaries of the region. Throughout this process, the agents adapt to the varying coverage area by imposing the dynamics of the boundaries on their respective control law. The presented control law ensures that the agents move toward the centroid of their respective Voronoi cell while taking into account the effect of the moving boundaries. The proposed coverage method deploys the agents within the boundaries of coverage area and ensures the (locally) optimal partitioning for the moving coverage area. Performance of the proposed blanket coverage method is examined via numerical examples that use sections of Ohio river and a border buffer zone.

A team-based approach for coverage control of moving sensor networks

The present paper proposes a new team-based approach that allows for forming multiple teams of agents within the coverage control framework. The objective function defined for this purpose tends to minimize the accumulative distance from each agent while reckoning with the given density function that defines the probability of events in the environment to be covered. The proposed team-based approach via the defined optimization problem allows for forming teams of agents when for a variety of reasons, e.g., heterogeneity in their embedded communication capabilities or the dynamics, it is required to keep the similar agents together in the same team. To realize this, the overall objective function is defined as the accumulated sensing cost of individual agents belonging to different teams. The defined collective cost function captures the interdependency of the teams Voronoi cells on the position of the agents that can be viewed as the impact of the dynamic boundaries on the agents. A gradient descent-based controller is designed to ensure the locally optimum configuration of the teams and agents within each team. The convergence of the proposed method is studied to ensure the stability of the implemented controller in both teams and agents final configuration. In addition, a new formation control approach is proposed using the team-based framework to impose either the same or different formation structures while performing the underlying coverage task. It is shown that maintaining the desired configuration through the proposed formation control is achieved at the cost of sacrificing the sensing performance. Finally, the proposed coverage and formation methods are examined via a numerical example where multiple heterogeneous teams of agents with potentially different number of agents are deployed.

State-space LPV model identification using kernelized machine learning

This paper presents a nonparametric method for identification of MIMO linear parameter-varying (LPV) models in state-space form. The states are first estimated up to a similarity transformation via a nonlinear canonical correlation analysis (CCA) operating in a reproducing kernel Hilbert space (RKHS). This enables to reconstruct a minimal-dimensional inference between past and future input, output and scheduling variables, making it possible to estimate a state sequence consistent with the data. Once the states are estimated, a least-squares support vector machine (LS-SVM)-based identification scheme is formulated, allowing to capture the dependency structure of the matrices of the estimated state-space model on the scheduling variables without requiring an explicit declaration of these often unknown dependencies; instead, it only requires the selection of nonlinear kernel functions and the tuning of the associated hyper-parameters.

A Bayesian approach for model identification of LPV systems with uncertain scheduling variables

This paper presents a Gaussian Process (GP) based Bayesian method that takes into account the effect of additive noise on the scheduling variables for identification of linear parameter-varying (LPV) models in input-output form. The proposed method approximates the noise-free coefficient functions by a local linear expansion on the observed scheduling variables. Therefore, additive noise on the scheduling variables is reconstructed as a corrective term added to the output noise that is proportional to the squared gradient obtained from the posterior of the Gaussian Process. An iterative procedure is given so that the obtained solution converges to the best estimation of the coefficient functions according to the given measure of fitness. Moreover, the expectation and covariance functions estimated by GP are modified for the noisy scheduling variable case to include the noise contribution on the estimated expectation and covariance functions. The model training procedure identifies noise level in the measurements including outputs and scheduling variables by estimating the noise variances, as well as other defined hyperparameters. Finally, the performance of the proposed method is compared to the standard GP approach through a numerical example.