https://repositorio.ufba.br/handle/ri/9682 2026-04-17T03:56:51Z https://repositorio.ufba.br/handle/ri/44289 Calibração das medições de energia em um detector de partículas utilizando informações de assimetria do perfil de deposição no calorímetro e técnicas de inteligência artificial Figueredo, Caroline RIbeiro Farias, Paulo César Machado de Abreu The ATLASdetector(AToroidalLHCApparatuS),partoftheLargeHadronCollider(LHC)at CERN, isoneofthelargestandmostcomplexparticlephysicsexperimentseverbuilt.Itiscomposed of severalsubsystems,includingthecalorimetrysystem,whichconsistsoffinelysegmenteddetectors responsible formeasuringtheenergyandpositionofparticles.ATLASwasdesignedtodetectandclassify subatomic particlesproducedinhigh-energycollisions.Oneofthechallengesfacedintheexperiment is thecalibrationoftheenergyofthedetectedparticles,whichisessentialtoensuretheaccuracyof the analysesperformed.ThisstudyinvestigatedtheenergycalibrationoftheATLAScalorimeterusing regressorsbasedonGradientBoostedDecisionTrees(GBDT)andfeedforwardneuralnetworksofthe Multilayer Perceptron(MLP)type,whichreceivedasinputstructurescalledStandardRingsandQuarter Rings. Thesestructuresarebuiltfromcalorimeterinformationandorganizedinawaythatpreserves the spatialcharacteristicsofparticleshowers.Fromthedefinitionofaregionofinterestaroundthe particle’sinteractionpoint,thesignalsfromthesensorsarearrangedinconcentricrings.TheStandard Rings encodetheenergyofthecellscontainedineachringforeachcalorimeterlayer,whiletheQuarter Rings areobtainedbydividingeachStandardRingintofourparts—exceptforthefirstring(HotCell) — allowingthecaptureofasymmetriesintheenergydepositionprofile.Oneofthemainchallenges in calorimetercalibrationisdealingwithvariationsinthedetectorresponse,whichcanbesignificant depending ontheparticles’transverseenergy(ET ) andpseudorapidity(η), ageometricparameterthat describes theparticle’spositionrelativetothebeamaxis.Thesevariationsbecomeparticularlycritical at lowenergyvalues,whereelectronidentificationismoredifficultduetothecharacteristicenergy deposition profile.Thecalibrationfactor(α), definedastheratiobetweentheMonteCarlosimulated energyandtheenergyestimatedbytheHighLevelTrigger(HLT),isappliedtotherawenergyprovided by theFastCaloalgorithminordertocorrectdistortionsresultingfromenergylossesinthedetector. Another aspectconsideredinthisstudyistheimpactofpile-up—multiplesimultaneousinteractionsina single collision—whichaffectsthepreciseidentificationandreconstructionofparticles.Therefore,the analysis wasconductedunderdifferentscenarios,withandwithoutthepresenceofpile-up,andusing differentinputstrategies:rawdataanddataderivedfromtheextractionofenergeticandasymmetric variables.Theresultsindicatethatalthoughpile-uppartiallyreducescalibrationeffectiveness—especially in energy-basedestimates—thetestedmodelsmaintainedrobustperformance.TheQuarterRingsstood out byshowingmoresignificantgainsincertainpseudorapidityregions,whiletheapplicationoffeature extractioncontributedtoimprovingtheestimates,althoughsomeconfigurationsprovedmoresensitiveto complexexperimentalenvironments.TheuseofQuarterRingsledtoimprovementsofupto18.08%in pseudorapidity-based calibrationandupto15.2%inenergy-basedcalibration,showingaclearadvantage overtheuncalibratedscenarioandperformancesimilartothatobtainedwithStandardRings.Theseresults reinforce thefeasibilityofusingspatiallyrefinedstructures,suchastheQuarterRings,combinedwith machine learningmodelstoimproveATLAScalorimetercalibration,evenunderadverseconditionssuch as thepresenceofpile-up.Theproposedapproachoffersapromisingalternativeforfutureimprovements in fastreconstructionsystemsandeventanalysisintheexperiment. Universidade Federal da Bahia Dissertação 2025-07-22T00:00:00Z https://repositorio.ufba.br/handle/ri/44131 Análise das sobretensões decorrentes de descargas atmosféricas em estruturas de turbinas eólicas offshore com fundação do tipo monopile Souza, Fellipe Meira Souza e Moreira, Fernando Augusto This work presents a detailed electromagnetic model of offshore wind turbines whose monopile foundation acts as the grounding system, aiming to evaluate transient overvoltages resulting from direct lightning strikes (upward and downward), based on real data from Morro do Cachimbo and Monte San Salvatore. Protecting offshore wind turbines against lightning strikes is a major technical challenge to ensure safe and continuous operation in marine environments. Due to their height and isolation, such structures are highly vulnerable to severe transient overvoltages. The developed model consists of (i) a multicomponent representation of the return stroke current using Heidler functions fitted to measured lightning currents, (ii) a modeling of blades and tower through distributed transmission-line segments whose electrical parameters vary according to the actual geometry of the structure, and (iii) a grounding impedance of the monopile foundation calculated using classical equations and subsequently fitted via Vector Fitting, and then represented by an equivalent lumped-element network in ATP/ATPDraw. The Ground Potential Rise (GPR), structural overvoltages, and the sensitivity to seawater resistivity and lightning-channel modeling are systematically assessed. The main results show: (a) strong dependence of the GPR on seawater resistivity and the representation of the lightning channel; (b) high dielectric stress on glass-fiber-reinforced polymer (GFRP) blades under typical lightning scenarios; and (c) significant waveform changes when overhead components are included. The proposed modeling approach provides meaningful contributions toward improving the design of more effective lightning-protection systems in offshore wind farms. Universidade Federal da Bahia Dissertação 2025-12-19T00:00:00Z https://repositorio.ufba.br/handle/ri/44103 Análise do impacto de diferentes modelagens do sistema de aterramento no desempenho de linhas de transmissão submetidas a descargas atmosféricas Mattos, Paulo Fernando Oliveira Paim de Moreira, Fernando Augusto The high incidence of lightning strikes and the specific characteristics of Brazilian soil contribute to a significant number of transmission line outages, mainly due to backflashovers caused by direct strikes. This study addresses the performance of transmission lines under lightning conditions, evaluating the impact of different grounding representation models on the performance of a 230 kV transmission line, with emphasis on overvoltage estimation and backflashover rates. The analyzed models include: (i) the low-frequency resistance model (STM), commonly used in grounding design; (ii) the impulse impedance model (Z_p), which approximates the frequency-dependent behavior of the grounding system at the instant of the lightning current peak; and (iii) the frequency-dependent models over the entire spectrum — LTAEM, based on transmission line theory with electromagnetic coupling, and HEM, grounded in electromagnetic field theory and widely recognized as a benchmark in the literature. The results show that the grounding system modeling plays a decisive role in evaluating transmission line performance under lightning events. Advanced models that consider the frequency-dependent behavior of the soil and grounding electrodes over the entire frequency spectrum—such as the HEM and the LTAEM—provide responses that are more consistent with physical reality, thereby reducing uncertainties in overvoltage estimation and in the prediction of line outages. On the other hand, the Z_p model yielded satisfactory results regarding the estimation of overvoltage levels across the insulator string and the rate of outages per 100 km of line per year due to backflashover. However, a limitation of this approach lies in the fact that the frequency sweep is performed only at the instants corresponding to the peak surge current (I_p2) and the ground potential rise (GPR) developed by the grounding system. Consequently, it is recommended that the performance assessment of transmission lines under lightning conditions be preferably carried out using the HEM or LTAEM models, with particular emphasis on the computational efficiency of the latter. Nevertheless, the Z_p model remains a technically sound and practically attractive alternative. Finally, the results show that the phase with the highest insulator string overvoltages does not necessarily correspond to the phase with the highest number of backflashover outages, underscoring the importance of assessing all transmission line phases jointly. Universidade Federal da Bahia Dissertação 2025-11-10T00:00:00Z https://repositorio.ufba.br/handle/ri/44100 Aprendizado de máquina aplicada em modelagem da eficiência de acoplamento entre guias dielétricos. Mercês, Viviane Oliveira das Rodrigues Esquerre, Vitaly Félix This dissertation aims to use machine learning for the formulation and design of mimetic models of TAPER-type photonic devices intended for coupling waveguides with different geometric structures. The objective is to evaluate the coupling efficiency as a function of specific variations in geometric characteristics in the C-Band. In addition to the theoretical foundation, it was necessary to prepare a database to train the learning networks. This database consists of numerical solutions obtained through a finite element-based numerical method, as well as previously published information. These data were consolidated into a comprehensive set whose attributes correspond to variations in the dimensions of the taper segments (the length denoted as “a”), and the output is the coupling efficiency (represented by “η”). A neural network architecture was then developed with the input parameters: length of each of the 15 taper segments (a), wavelength (λ), refractive indices of the core (n1) and substrate (n2), and as the output parameter: the ratio between the input power (Pin) and the output power (Pout) given by the coupling efficiency (η). For this architecture, variations in training algorithms and activation functions were explored. These variations were used to evaluate the performance of the proposed models, considering criteria such as accuracy, precision, simplicity, and the computational costs involved. As a result, the developed architectures demonstrated performances better than the values defined by the stopping criteria, with a mean squared error less than 10-7 and a regression rate or determination coefficient R2 of 100% in more than 92% of the total of 81 models evaluated with reduced use of computational resources. This study aims to contribute to the improvement of the understanding and design of photonic devices through the synergistic application of machine learning and traditional techniques. Universidade Federal da Bahia Dissertação 2024-08-23T00:00:00Z