Anton Duca

Diagnosis of real cracks from eddy current testing signals using parallel computation

Abstract The paper presents a novel approach for three-dimensional diagnosis of partially conductive cracks from two-dimensional eddy current testing signals by means of the tabu search stochastic method. A new testing probe driving uniformly distributed eddy currents is employed for the inspection. Three spatial components of the perturbation electromagnetic field due to partially conductive cracks are sensed as the response signals in order to enhance information level of the inspection. The signals are simulated by a fast forward FEM-BEM solver using a database. Cracks are modeled as defects with a complex shape and homogeneous conductivity. The length, depth, width and conductivity of a detected crack are unknown in the inversion process. Numerical results of the three-dimensional reconstruction of partially conductive cracks from simulated two-dimensional signals are presented and discussed in the paper. Artificial white noise is added to the simulated signals and robustness of the algorithm is tested. Moreover, parallel computation is employed to reduce the computation time.

Neighborhood Strategies for QPSO Algorithms to Solve Benchmark Electromagnetic Problems.

Several neighborhood strategies for QPSO algorithms are proposed and analyzed in order to improve the performances of the original methods. The proposed strategies are applied to some of the most well known QPSO algorithms such as the QPSO with random mean, the QPSO with Gaussian attractor and of course the basic QPSO. To prevent the premature convergence and to avoid being trapped in local minima the neighborhoods are dynamically changed during the optimization process. For testing the efficiency of the neighborhood techniques two benchmark optimization problems from the electromagnetic field computation have been chosen, Loney’s solenoid and TEAM22.

SPSO Parallelization Strategies for Electromagnetic Applications

Two parallelization techniques, GPGPU and Pthreads for multiprocessor architectures, are used to implement a SPSO algorithm in order to solve electromagnetic optimization problems. Several configurations for the GPGPU implementation are tested and a new full parallel minimum branching implementation is proposed. The best GPGPU approaches are then compared with a Pthreads implementation in terms of speed up and solution quality. To test the efficiency of the parallelization techniques two electromagnetic optimization problems were chosen, namely the TEAM22 benchmark and Loney’s solenoid. In the end the paper provides suggestions regarding what parallelization technique should be used considering the implementation features of the optimization function.

Advanced PSO algorithms and local search strategies for NDT-ECT inverse problems

This paper studies the efficiency of some advanced Particle Swarm Optimization (PSO) algorithms, such as Standard PSO and Quantum-behaved PSO, to solve a nondestructive eddy current (NDT-ECT) inverse problem. The inverse problem is formulated as an optimization problem, aiming to minimize the distance between the true 2D ECT signal, produced by the real crack and the simulated signal, produced by a potential solution (crack). To increase the efficiency of the PSO based methods, enhancements based on problem specific local search strategies are also tested in order to speed up the optimization process. The inversion schemes are compared in the three-dimensional reconstruction of partially conductive cracks, which have a uniform conductivity smaller than the conductivity of the base material and a cuboid shape.

Particle Swarm Optimization with social exclusion and its application in electromagnetics

The behavior of Particle Swarm Optimization (PSO), a population based optimization algorithm, depends on the movements of the particles and the attractions among them. This behavior was extracted from the observations of the swarms in nature. Every swarm desires to remain powerful in order to survive in nature and to protect its descendants. Therefore, the weakest members in the swarm are isolated, and generally abandoned to live on their own resources. This act is known as social exclusion. In this research, this phenomenon is incorporated to PSO. At the early phase of time-line, the swarm is divided into two groups based on their cost/fitness values. Each group proceeds their own journey without the knowledge of other group. This new algorithm is named as Social Exclusion-PSO (SEPSO). First, the performance of this new algorithm was evaluated/compared with an inertia weight PSO via unimodal, multimodal, expended benchmark functions, and then, it is applied to the circular antenna array design problem. For each implementation, the performance of two sub-populations and the undivided population are presented to demonstrate and compare the behaviour of the socially excluded swarm. The results show that excluding the members with the worst cost values from the population increases the performance of the algorithm in terms of global best solution with approximately 20% smaller number of function evaluations.

Advanced procedure for non-destructive diagnosis of real cracks from eddy current testing signals

The paper presents an advanced procedure for tree-dimensional diagnosis of real cracks from two-dimensional simulated eddy current testing response signals by means of the tabu search stochastic method. A new testing probe driving uniformly distributed eddy currents is employed for the inspection. All three spatial components of the perturbation electromagnetic field are acquired during two-dimensional scan of the probe. The ECT signals due to partially conductive cracks are simulated by a fast forward FEM-BEM solver using a database. Two crack models are proposed for the inversion. The first one has a cuboid shape and the later one reflects a more complex geometry. Both the cracks models consider uniform distribution of the partial conductivity. Numerical results of the three-dimensional reconstruction of partially conductive cracks from simulated two-dimensional signals are presented and discussed in the paper.

Enhancement of Particle Swarm Optimization algorithms for the 3D crack reconstruction from ECT signals

This paper presents some strategies to enhance the efficiency of Particle Swarm Optimization (PSO) algorithms applied for the three-dimensional reconstruction of cracks from eddy current testing (ECT) signals. A fast forward FEM-BEM solver using database was adopted to simulate the ECT signals due to cracks with parallelepiped shape for various lengths, widths and depths. The inverse problem is formulated as an optimization problem, aiming to minimize the distance between the signal due to of the real crack and the signal due to a potential solution (crack). A novel technique which combines the features of PSO and local search methods is used in the inversion of ECT signals. The local search methods are specific to such kind of inverse problems. The points used in the local search are chosen in order to avoid local minimum in PSO algorithms. Numerical results of the 3D reconstruction of cracks from simulated 2D signals are presented and discussed.

Parametric reduced order models in static multiphysics analysis of MEMS switches

This contribution proposes a method to extract parametric reduced models that describe the coupled structural-electric behavior of RF MEMS switches. The equivalent capacitance coefficients and the effective elastic coefficients are extracted from coupled structural-electrostatic analysis. Parametric models are built based on the sensitivities of the extracted equivalent coefficients. The method is validated on two benchmarks: one for which experimental values are available, and the other one from the literature. The results show that, for the tested configurations, even a model of order 1 or 2 can catch accurately enough (e.g. relative error of less than 5 %) the pull-in voltage of the full order model, for a variation of the investigated parameter of less than 20 %. Such a reduced parametric model is useful in early stages of the design.

High performance computing techniques for fast solving a NDET forward problem

The present paper studies two different high performance computing (HPC) approaches to speed up a non-destructive electromagnetic testing (NDET) forward problem, namely general purpose computing for graphics processing units (GPGPU) and multiprocessor programming. The two methods are studied and compared for a set of cracks, non-conductive or partial conductive with uniform conductivity. Both HPC techniques use professional libraries such as MAGMA or Intel MKL.

GPGPU vs multiprocessor SPSO implementations to solve electromagnetic optimization problems

This paper studies two parallelization techniques for the implementation of a SPSO algorithm applied to optimize electromagnetic field devices, GPGPU and Pthreads for multiprocessor architectures. The GPGPU and Pthreads implementations are compared in terms of solution quality and speed up. The electromagnetic optimization problems chosen for testing the efficiency of the parallelization techniques are the TEAM22 benchmark problem and Loneys solenoid problem. As we will show, there is no single best parallel implementation strategy since the performances depend on the optimization function.