A hybrid approach combining genetic algorithm and sensitivity information extracted from a parallel layer perceptron

Vieira, Daniel; Vieira, D.A.G.; Caminhas, Walmir M.; Vasconcelos, J.A.

doi:10.1109/tmag.2005.846039

Cited by 8 publications

(3 citation statements)

References 12 publications

Supporting

Mentioning

Contrasting

Order By: Relevance

“…As it is well known, stochastic methods tend to be slow even though robust. The authors of this paper believe that the combination of both techniques must be investigated trying to extract the strengths of each one [8].…”

Section: Discussionmentioning

confidence: 99%

Characterization of Inclusions in a Nonhomogeneous GPR Problem by Artificial Neural Networks

Travassos¹,

Vieira²,

Ida³

et al. 2008

IEEE Trans. Magn.

Self Cite

View full text Add to dashboard Cite

This paper aims at detecting and characterizing inclusions in concrete structures by inverting ground-penetrating radar (GPR) data. First, the signal is preprocessed using the principal component analysis (PCA) and then used to train an artificial neural network (ANN). The GPR data consists of 1200 time steps. Using PCA, the data can be compressed to 286 dimensions without losing any information. Moreover, with 99.99% of the original variance the data needs only 139 dimensions. This dimensional reduction makes the ANN training easier and faster. The ANN were trained to find the buried inclusions characteristicsand -considering a nonhomogenous host medium by inverting the preprocessed data. The results show that the expected maximum error was kept under 1%, which is a remarkable result, since the host medium is nonhomogenous.Index Terms-Artificial neural network (ANN), buried objects, ground-penetrating radar (GPR), inverse problem.

show abstract

Section: Discussionmentioning

confidence: 99%

Characterization of Inclusions in a Nonhomogeneous GPR Problem by Artificial Neural Networks

Travassos¹,

Vieira²,

Ida³

et al. 2008

IEEE Trans. Magn.

Self Cite

View full text Add to dashboard Cite

show abstract

“…In [18], a new fuzzy genetic algorithm (FGA) is developed to optimize the ground plane with dielectric electromagnetic bandgap (D-EBG) structure. Some improved genetic algorithms have been proposed successively to accelerate the convergence speed, improve the result accuracy, and expand the band gap bandwidth [19][20][21]. Although these optimize strategies have good application in the structural design of EBG structures in multi-layer packaging systems, there are a few designs for EBG in smallsize packaging [22][23][24][25].…”

Section: Introductionmentioning

confidence: 99%

Optimizing 1D Dielectric Electromagnetic Bandgap (D-EBG) Structures Using Multistage Genetic Algorithm (MS-GA) and Considering Parameter Variations

Guo,

Hu,

et al. 2023

PIER B

View full text Add to dashboard Cite

An optimization method utilizing a multistage genetic algorithm (MS-GA) and considering parameter variations has been proposed to obtain optimal design of one-dimensional dielectric bandgap (1D D-EBG) structures with a few periods in small packaging power distribution networks. Onedimensional finite method (1D FEM) is used to improve computational efficiency and iteration speed. MS-GA consists of 3 stages: In stage 1, the population was initialized by Hamming distance, and the fitness was calculated to determine the number of EBG period. In stage 2, genetic manipulation and sensitivity analysis were used to improve local search ability and obtain preliminary results. In stage 3, cubic spline interpolation and local integral were used to reconstruct the fitness evaluation function considering parameter deviation, adjust the results, and obtain the optimal parameters. Three optimized target frequency bands with center frequencies of 2.4 GHz, 3.5 GHz, and 28 GHz were optimized, and Pearson coefficient was used to analyze the correlation between the parameters to better understand the influence of parameter deviation on the optimization results. The achieved results meet the optimization object within the allowable range of parameter errors, and the parameter constraints were successfully met for all three designs, with their final dimensions below 20 mm. Three-dimensional full-wave simulation software was used to simulate and analyze the stopband bands, and the simulation results were consistent with the calculation results.

show abstract

“…Sensitivity extraction of electromagnetic devices using neuro-fuzzy and neural models is addressed in [1]- [3]. Even though these results have proven to be very useful to decrease the computational effort in the design of electromagnetic devices [4], the aforementioned methodologies are strongly dependent of some user's defined parameters as the number of neurons. In this work, the Parallel Layer Perceptron network [5] trained with the Minimum Gradient Method (PLP-MGM) [6], is applied to the problem of sensitivity extraction of electromagnetic devices.…”

Section: Introductionmentioning

confidence: 99%

Improving Neural Networks Sensitivity Extraction of Electromagnetic Devices

Vieira¹,

Vasconcelos²,

Caminhas³

2006 12th Biennial IEEE Conference on Electromagnetic Field Computation

Self Cite

View full text Add to dashboard Cite

This paper applies the Parallel Layer Perceptron network trained with the Minimum Gradient Method (PLP-MGM) to the problem of sensitivity extraction of electromagnetic devices. The networks trained with the MGM are less dependent of user's defined parameters, as, for instance, the number of neurons. Some results are presented considering the sensitivity extraction of a Loudspeaker Magnet Assembly, and they show the effectiveness of the proposed approach.

show abstract

A hybrid approach combining genetic algorithm and sensitivity information extracted from a parallel layer perceptron

Cited by 8 publications

References 12 publications

Characterization of Inclusions in a Nonhomogeneous GPR Problem by Artificial Neural Networks

Characterization of Inclusions in a Nonhomogeneous GPR Problem by Artificial Neural Networks

Optimizing 1D Dielectric Electromagnetic Bandgap (D-EBG) Structures Using Multistage Genetic Algorithm (MS-GA) and Considering Parameter Variations

Improving Neural Networks Sensitivity Extraction of Electromagnetic Devices

Contact Info

Product

Resources

About