Genetic Algorithm-based Solver Research Articles

An Automatic Parameter Setting Variational Mode Decomposition Method for Vibration Signals

Variational mode decomposition (VMD) provides a feasible approach to decompose vibration signals obtained from complex machinery for further applications. The mode frequency bandwidth control parameter and the total number of modes are critical parameters for VMD. Thus, optimally and automatically setting the two parameters is an essential issue for VMD for various practical vibration signal sources. To this end, this work proposes an automatic parameter setting VMD for vibration signals. First, we use the bandwidth evaluation criterion and the mean mode-location distance to evaluate the sparsity of modes; we use the energy loss to evaluate the reconstruction from modes. Then, we synthesize the three criteria into a novel optimization model using logarithmic transformation. Accordingly, a genetic algorithm-based solver is developed for the optimization model. Finally, the artificial multi-component signal and the practical rolling bearing vibration signal from CWRU Laboratory are used to verify the performance of the proposed optimization method.

Study of the wavelength dependence of the absorption function of diesel soot by LII modeling integrating the effect of multiple scattering within aggregates

This work represents a first attempt at proposing a refined LII model suitable for simulating signals collected with excitation wavelengths (EWs) of 266, 355, 532 and 1064 nm. In this context, we implemented a comprehensive version of laser-irradiated soot heat and mass balance equations integrating terms representing the saturation of linear, single- and multi-photon absorption processes, cooling by sublimation, conduction, radiation and thermionic emission, as well as mechanisms depicting soot oxidation and annealing, non-thermal photodesorption of carbon clusters and corrective factors accounting for the shielding effect and multiple scattering (MS) within aggregates. This simulation tool was fully parameterized, by coupling design of experiments with a genetic algorithm-based solver, against data collected at different heights above the burner (HAB) in a diesel spray flame. This allowed to assess values of different model parameters involved in absorption and sublimation terms, which, to date, have never been reported in LII modeling studies for UV EWs (e.g., multi-photon absorption cross sections for C2 photodesorption and saturation coefficients for linear- and multi-photon absorption). The simulation work proposed herein then enabled to infer information regarding the evolution of the absorption function of soot (Eleft(m,lambda right)). Eleft(m,266right), E(m,355) and E(m,532) ranging from 0.25 to 0.51, 0.20 to 0.38 and 0.18 to 0.30, respectively, were notably estimated as a function of the HAB. Alternatively, values ranging from 0.31 to 0.53, 0.26 to 0.44 and 0.22 to 0.38 were assessed while neglecting the effect of MS and provided constant E(m,266)/E(m,1064), E(m,355)/E(m,1064) and E(m,532)/E(m,1064) ratios of 1.4, 1.2 and 1.0, regardless of the HAB. In addition, the obtained results showed that the wavelength dependence of the soot absorption function was quite negligible for EWs higher than 532 nm, irrespective of whether the effect of MS was considered or neglected. On the other hand, aggregate properties were proven to substantially influence the E(m,lambda )/E(m,1064) ratios for decreasing lambda and increasing HAB, thus illustrating the significant effect of aggregation on the optical properties of soot. Finally, the results issued from the different analyses we conducted on the diesel flame investigated in this paper led to the conclusion that values of Eleft(m,lambda right), falling within the 0.38 ± 0.15 and 0.29 ± 0.11 range at 266 and 355 nm, versus 0.25 ± 0.09 at 532 and 1064 nm could be considered as suitable for simulating LII signals of progressively aging aggregated particles.

Parameterization of a Refined Model Aimed at Simulating Laser-Induced Incandescence of Soot Using a Visible Excitation Wavelength of 532 nm

Laser-induced incandescence (LII) is one of the most powerful techniques for soot detection in combustion media. It is therefore commonly used to perform experiments in lab-scale flames and industrial combustors with a view to elucidating the formation mechanisms leading to combustion-generated fine carbonaceous particles while assessing their intrinsic properties. Quantitatively interpreting LII measurements, however, requires a firm knowledge of the optical properties of soot, including their wavelength-dependent absorption function (). Among the approaches used to evaluate such a crucial parameter, one can implement a LII model to derive the value which has to be set to reproduce a series of LII signals measured in a well-characterized environment. In this context, the present work aims at parameterizing a refined LII model built upon a comprehensive version of soot heat- and mass-balance equations for assessment when using a visible excitation wavelength of 532 nm. The proposed model integrates terms representing the saturation of linear, single- and multi-photon absorption processes, cooling by sublimation, conduction, radiation and thermionic emission, in addition to mechanisms depicting soot oxidation and annealing, non-thermal photodesorption of carbon clusters, as well as corrective factors accounting for the shielding effect and multiple scattering (MS) within aggregates. To parameterize this advanced simulation tool, an optimization procedure coupling design of experiments with a genetic algorithm-based solver was implemented. Doing so allowed to estimate the values of different factors involved in absorption and sublimation terms, including the multi-photon absorption cross-section for C2 photodesorption, the saturation coefficients for linear- and multi-photon absorption, as well as the value. Obtained parameters turned out to be well-suited to reproduce a set of LII signals acquired in a Diesel flame. While leading to predictions merging on a single curve with measured data, the so-parameterized model notably led to infer values of 0.3 and 0.38 when considering or neglecting MS within aggregates, respectively. Finally, the / ratio estimated based on data collected herein and in a former modeling work was found to be consistent with results issued from two-excitation wavelength LII measurements previously reported in the literature.

Modeling laser-induced incandescence of Diesel soot\u2014Implementation of an advanced parameterization procedure applied to a refined LII model accounting for shielding effect and multiple scattering within aggregates for {varvec{alpha}}_{{varvec{T}}} and {varvec{E}}left( {varvec{m}} right) assessment

An extensive set of LII signals measured in a Diesel spray flame has been simulated using a refined LII model built upon a comprehensive version of soot heat- and mass-balance equations. This latter includes terms standing for saturation of linear, single- and multi-photon absorption processes, cooling by sublimation, conduction, radiation and thermionic emission in addition to mechanisms depicting soot oxidation and annealing, non-thermal photodesorption of carbon clusters as well as corrective factors allowing considering shielding effect and multiple scattering (MS) within aggregates. A complete parameterization of the so-proposed model has been achieved by means of an advanced optimization procedure coupling design of experiments with a genetic algorithm-based solver. Doing so, the values of different factors involved in absorption and sublimation terms have been assessed for a 1064-nm laser excitation wavelength including the multi-photon absorption cross section for C2 photodesorption and the saturation coefficients for linear- and multi-photon absorption, among others. This parameterized model has then been demonstrated to effectively reproduce signals measured in different combustion media including a CH4/O2/N2 premixed flat flame and a diffusion ethylene flame. As a result of the data derived from the analysis of the Diesel flame, a thermal accommodation coefficient value of 0.49 has been assessed against 0.34 when neglecting the shielding effect. In addition, values of the soot absorption function (Eleft( m right)) comprised between 0.18 and 0.31 have been inferred depending on the particle maturation stage. On the other hand, Eleft( m right) 24% higher on average have been estimated when neglecting MS thus illustrating the importance of aggregate characteristics on soot properties derived through LII modeling. Eventually, the Eleft( m right) evolution observed herein has been compared with results issued from studies conducted with varied hydrocarbons which led to highlight the crucial role played by the soot maturity level over the nature of the burnt fuel as far as optical properties are concerned.

Analysis of the Influence of the Conduction Sub-Model Formulation on the Modeling of Laser-Induced Incandescence of Diesel Soot Aggregates.

Laser-induced incandescence (LII) is a powerful diagnostic technique allowing quantifying soot emissions in flames and at the exhaust of combustion systems. It can be advantageously coupled with modeling approaches to infer information on the physical properties of combustion-generated particles (including their size), which implies formulating and solving balance equations accounting for laser-excited soot heating and cooling processes. Properly estimating soot diameter by time-resolved LII (TiRe-LII), nevertheless, requires correctly evaluating the thermal accommodation coefficient driving the energy transferred by heat conduction between soot aggregates and their surroundings. To analyze such an aspect, an extensive set of LII signals has been acquired in a Diesel spray flame before being simulated using a refined model built upon expressions accounting for soot heating by absorption, annealing, and oxidation as well as cooling by radiation, sublimation, conduction, and thermionic emission. Within this framework, different conduction sub-models have been tested while a corrective factor allowing the particle aggregate properties to be taken into account has also been considered to simulate the so-called shielding effect. Using a fitting procedure coupling design of experiments and a genetic algorithm-based solver, the implemented model has been parameterized so as to obtain simulated data merging on a single curve with experimentally monitored ones. Eventually, values of the thermal accommodation coefficient have been estimated with each tested conduction sub-model while the influence of the aggregate size on the so-inferred has been analyzed.

A Generalized Genetic Algorithm-Based Solver for Very Large Jigsaw Puzzles of Complex Types

In this paper we introduce new types of square-piece jigsaw puzzles, where in addition to the unknown location and orientation of each piece, a piece might also need to be flipped. These puzzles, which are associated with a number of real world problems, are considerably harder, from a computational standpoint. Specifically, we present a novel generalized genetic algorithm (GA)-based solver that can handle puzzle pieces of unknown location and orientation (Type 2 puzzles) and (two-sided) puzzle pieces of unknown location, orientation, and face (Type 4 puzzles). To the best of our knowledge, our solver provides a new state-of-the-art, solving previously attempted puzzles faster and far more accurately, handling puzzle sizes that have never been attempted before, and assembling the newly introduced two-sided puzzles automatically and effectively. This paper also presents, among other results, the most extensive set of experimental results, compiled as of yet, on Type 2 puzzles.

Optimal multi-objective distribution system reconfiguration with multi criteria decision making-based solution ranking and enhanced genetic operators

In electrical distribution system optimisation, the presence of multiple conflicting objectives is effectively addressed by using Pareto front analysis. This paper deals with optimal reconfiguration considering network losses and energy not supplied as multi-objectives. A set of original contributions are provided with reference to the construction and updating of the best-known Pareto front using a genetic algorithm-based solver. The crossover operator is extended to address multi-objective solutions. The mutation operator is extended to handle a broader number of cases. Multi-objective solution ranking is applied by resorting to multi criteria decision making methods during the creation of the offsprings in the crossover operator, as well as to provide an automatic support for the decision maker to identify the preferable solution in the final Pareto front. The proposed approach is applied on two reference test networks, for which the complete Pareto front is calculated from the entire set of multi-objective solutions. The resulting best-known Pareto front is compared with the complete Pareto front using a metric based on geometrical considerations. This comparison framework is helpful to assess the performance of the multi-objective optimisation solvers.