Abstract The main object of this thesis was to research injection mouldable 0–3 type ceramic polymer composites and their dielectric and magnetic properties in the GHz frequency region. The work has been divided into three sections. In the first section, two–phase ceramic polymer composites containing dielectric and magnetic fillers have been investigated and their characteristics analysed by reference to pre–existing mixing rules. The exploitation of these composites in miniaturizing devices, such as antennae, is presented and discussed. The second part describes three phase composites containing different nanosize additives (silver, silicon and alumina fibres) towards improving their dielectric properties. In the third part, some periodical and multilayer structures for ceramic polymer composite layers are proposed. In the case of two–phase ceramic polymer composites, with 37 vol.% of dielectric filler (Barium Strontium Titanate, BST) embedded into a thermoplastic polymer (ER140) matrix, the highest measured relative permittivity was 15 with a dielectric loss value of 0.008 at 1 GHz. With 43 vol.% of magnetic filler (hexaferrite, CO2Z) in ER182 matrix, the highest achieved relative permeability was 1.8 with a magnetic loss value of 0.077 at 1 GHz. Composites with Co2Z filler provide a 77% size reduction, and could thus be used advantageously in antennae. It was found that a 2–6 vol.% nanoaddition in BST–ER140 composites enhanced the relative permittivity drastically with only a minor effect on the dielectric losses. In particular, with only 2 vol.% addition of nanosize silver particles into the BST–ER140 composite, a 52% increase in the relative permittivity was obtained, with no significant change in the dielectric losses (tan δε = 0.004). Vertically and horizontally periodical dielectric composite structures comprising layers of different dielectric properties have been fabricated as well as multilayered structures containing dielectric and magnetic layers. The measurement results indicate that such multimaterial multilayer structures are good candidates for components with reduced dielectric and magnetic losses. / Tiivistelmä Väitöstyön tavoitteena oli tutkia ruiskuvalettavien 0–3 –liitännäisten keraami-polymeerikomposiittien ominaisuuksia erityisesti niiden GHz-taajuusalueen dielektristen ja magneettisten ominaisuuksien kannalta. Työ on jaettu kolmeen osaan. Ensimmäisessä osassa on tutkittu kaksikomponenttisia keraami-polymeerikomposiitteja, joissa täytemateriaali on joko dielektristä tai magneettista materiaalia. Komposiittien ominaisuuksia on analysoitu jo olemassa olevien seosmallinnuskaavojen avulla. Komposiittien hyödyntämistä erilaisten sovellusten, kuten antennien, minityrisoinnissa on myös käsitelty. Toinen osa käsittelee kolmikomponenttisia komposiitteja, joissa lisäaineena on käytetty pieniä määriä nanomateriaaleja (hopea- ja piipartikkelit sekä alumiinioksidikuitu) tarkoituksena parantaa komposiitin dielektrisiä. Kolmannessa osassa on tutkittu periodisia ja monikerroksisia keraami- polymeerikomposiittirakenteita rakenteita. Kaksikomponenttisten keraami-polymeerikomposiittien tapauksessa suurin permittiivisyyden arvo 15 dielektristen häviöiden ollessa 0.008 (mittaustaajuus 1 GHz) saatiin komposiitille, jossa dielektristä täytemateriaalia (Barium Strontium Titanaatti, BST) oli 37 tilavuus-% termoplastisessa polymeerimatriisissa (ER140). Korkein saavutettu permeabiliteetin arvo 1.8 magneettisten häviöiden ollessa 0.077 (mittaustaajuus 1 GHz) saatiin komposiitille, jossa magneettista täyteainetta (hexaferriitti, Co2Z) oli 43 tilavuus-% ER182 -matriisissa. Tämä täyteaine mahdollistaa nykyistä jopa 77 % pienempien antennielementtien kehittämisen. Tukimuksessa todettiin 2–6 tilavuus-% nanomateriaalin lisäyksen BST-ER140 -komposiitteihin kasvattavan permittiivisyyttä merkittävästi juurikaan vaikuttamatta dielektrisiin häviöihin. Erityisesti 2 tilavuus-% hopeananopartikkeleiden lisäys BST-ER140 -komposiitteihin kasvatti permittiivisyyttä 52 % dielektristen häviöiden (tan δε =  0.004) kasvamatta. Työssä on myös tutkittu periodisesti (vertikaali ja horisontaali) koostettuja dielektrisiä komposiittirakenteita, jossa eri kerroksissa on erilaiset dielektriset ominaisuudet sekä monikerrosrakenteita, joissa vuorottelevat dielektriset ja magneettiset kerrokset. Mittaukset osoittivat, että monimateriaaliset monikerrosrakenteet ovat hyviä kandidaatteja komponentteihin, jotka vaativat pieniä dielektrisiä ja magneettisiä häviöitä.

Les MOSFETs ultra-thin body UTB ont été fabriqués avec une technologie auto-alignée. Le canal conducteur est constitué d’InGaAs à 75% de taux d’indium ou d’un composite InAs/In0,53Ga0,47As. Une fine couche d'InP (3 nm) a été insérée entre le canal et l'oxyde, afin d’éloigner les défauts de l’interface oxyde-semiconducteur du canal. Enfin, une épaisseur de 4 nm d'oxyde de grille (Al2O3) a été déposée par la technique de dépôt des couches atomiques. Les contacts ohmiques impactent les performances des MOSFETs. La technologie UTB permet difficilement d’obtenir des contacts S/D de faibles résistances. De plus, l’utilisation de la technique d’implantation ionique pour les architectures UTB est incompatible avec le faible budget thermique des matériaux III-V et ne permet pas d’obtenir des contacts ohmiques de bonne qualité. Par conséquent, nous avons développé une technologie auto-alignée, basée sur la diffusion du Nickel « silicide-like » par capillarité à basse température de recuit (250°C) pour la définition des contacts de S/D. Finalement, nous avons étudié et analysé la résistance de l'alliage entre le Nickel et les III-V. A partir de cette technologie, des MOSFET In0,75Ga0,25As et InAs/In0,53Ga0,47As ont été fabriqués. On constate peu de différences sur les performances électriques de ces deux composants. Pour le MOSFET InAs/InGaAs ayant une longueur de grille LG =150 nm, un courant maximal de drain ID=730 mA/mm, et une transconductance extrinsèque maximale GM, MAX = 500 mS/mm ont été obtenu. Le dispositif fabriqué présente une fréquence de coupure fT égale à 100 GHz, et une fréquence d'oscillation maximale fmax de 60 GHz, pour la tension drain-source de 0,7 V. / Abstract : Silicon-based devices dominate the semiconductor industry because of the low cost of this material, its technology availability and maturity. However, silicon has physical limitations, in terms of mobility and saturation velocity of the carriers, which limit its use in the high frequency applications and low supply voltage i.e. power consumption, in CMOS technology. Therefore, III-V materials like InGaAs and InAs are good candidates because of the excellent electron mobility of bulk materials (from 5000 to 40.000 cm2 /V.s) and the high electron saturation velocity. We have fabricated ultra-thin body (UTB) InAs/InGaAs MOSFET with gate length of 150 nm. The frequency response and ON-current of the presented MOSFETs is measured and found to have comparable performances to the existing state of the art MOSFETs as reported by the other research groups. The UTB MOSFETs were fabricated by self-aligned method. Two thin body conduction channels were explored, In0,75Ga0,25As and a composite InAs/In0,53Ga0,47As. A thin upper barrier layer consisting of InP (3nm) is inserted between the channel and the oxide layers to realized a buried channel. Finally, the Al2O3 (4 nm) was deposited by the atomic layer deposition (ALD) technique. It is well known that the source and drain (S/D) contact resistances of InAs MOSFETs influence the devices performances. Therefore, in our ultra-thin body (UTB) InAs MOSFETs design, we have engineered the contacts to achieve good ohmic contact resistances. Indeed, for this UTB architecture the use of ion implantation technique is incompatible with a low thermal budget and cannot allow to obtain low resistive contacts. To overcome this limitation, an adapted technological approach to define ohmic contacts is presented. To that end, we chose low thermal budget (250°C) silicide-like technology based on Nickel metal. Finally, we have studied and analyzed the resistance of the alloy between Nickel and III-V (Rsheet). MOSFET with two different epilayer structures (In0,75Ga0,25As and a composite InAs/In0,53Ga0,47As) were fabricated with a gate length (LG) of 150 nm. There were few difference of electrical performance of these two devices. We obtained a maximum drain current (ION) of 730 mA/mm, and the extrinsic transconductance (GM, MAX) showed a peak value of 500 mS/mm. The devices exhibited a current gain cutoff frequency fT of 100 GHz and maximum oscillation frequency fmax of 60 GHz for drain to source voltage (VDS) of 0.7 V.

In recent years a strong interest has been expressed to produce lighter structures.One possible solution to reduce the weight is to utilize high strength steels and use welding as the joining method. Many components experience fatigue loadingduring all or part of their life time and welded connections are often the prime location of fatigue failure. This becomes more critical in welded high strength steels as fatigue strength of welds does not increase by increasing the steel strength. A possible solution to overcome this issue is to use fatigue improvement methods.The main objectives of this project are, therefore, to increase understanding of the factors that control fatigue life and to investigate how the fatigue strength improvement methods; high frequency mechanical impact (HFMI) treatment and use of Low Transformation Temperature (LTT) consumables will affect fatigue properties of welds in high strength steels. In this regard, Gas Metal Arc Welding(GMAW) was used to produce butt and fillet welds using LTT or conventional fillers in steels with yield strengths ranging from 650-1021 MPa and T-joint weldsin a steel with 1300 MPa yield strength. The effect of HFMI on fatigue strength of the welds in 1300 MPa yield strength steels was also investigated. Butt and fillet welds in 650-1021 MPa steels were fatigue tested under constant amplitude tensile loading with a stress ratio of 0.1 while T-joints were fatigue tested under constant amplitude fully reversed bending load with a stress ratio of -1. The nominal stress approach was used for fatigue strength evaluation of butt and fillet welds whereas the effective notch stress approach was used in case of T-joints. Relative effectsof the main parameters such as residual stress and weld toe geometry influencing fatigue strength of welds were evaluated. Residual stresses were measured using X-ray diffraction for as-welded and HFMI treated welds. Neutron diffraction was additionally used to investigate the near surface residual stress distribution in 1300 MPa LTT welds.Results showed that use of LTT consumables increased fatigue strength of welds in steels with yield strengths ranging from 650-1021 MPa. For butt welds, the vii characteristic fatigue strength (FAT) of LTT welds at 2 million cycles was up to46% higher when compared to corresponding welds made with conventional fillermaterials. In fillet welds, a maximum improvement of 132% was achieved when using LTT wires. The increase in fatigue strength was attributed to the lower tensile residual stresses or even compressive stresses produced close to the weldtoe in LTT welds. Weld metals with martensite transformation start temperatures around 200 °C produced the highest fatigue strength. In 1300 MPa yield strength steel, similar FAT of 287 MPa was observed for LTT welds and 306 MPa for conventional welds, both much higher than the IIW FATvalue of 225 MPa. The relative transformation temperatures of the base and weldmetals, specimen geometry and loading type are possible reasons why the fatigue strength was not improved by use of LTT wires. Neutron diffraction showed that the LTT consumable was capable of inducing near surface compressive residual stresses in all directions at the weld toe. It was additionally found that there arevery steep stress gradients both transverse to the weld toe line and in the depth direction, at the weld toe. Due to difficulties to accurately measure residual stresses locally at the weld toe most often in the literature and recommendations residual stresses a few millimetre away from the weld toe are related to fatigue properties. However, this research shows that caution must be used when relating these to fatigue strength, in particular for LTT welds, as stress in the base materiala few millimetre from the weld toe can be very different from the stress locally at the weld toe.HFMI increased the mean fatigue strength of conventional welds in 1300 MPa steels about 26% and of LTT welds by 13%. It increased the weld toe radius slightly but produced a more uniform geometry along the treated weld toes. Large compressive residual stresses, especially in the longitudinal direction, were introduced adjacent to the weld toe for both LTT and conventional treated welds. It was concluded that the increase in fatigue strength by HFMI treatment is due to the combined effect of weld toe geometry modification, increase in surface hardness and introduction of compressive residual stresses in the treated region.It was concluded that the residual stress has a relatively larger influence than the weld toe geometry on fatigue strength of welds. This is based on the observation that a moderate decrease in residual stress of about 15% at the 300 MPa stress level had the same effect on fatigue strength as increasing the weld toe radius by approximately 85% from 1.4 mm to 2.6 mm, in fillet welds. Also, a higher fatigue strength was observed for HFMI treated conventional welds compared to as welded samples having similar weld toe radii but with different residual stresses.

La mécanique cellulaire a gagné un intérêt croissant en raison de son implication fondamentale dans des nombreux processus cellulaires, notamment la migration, la division, la différentiation et l’apoptose. Entre autres techniques, la microscopie à force atomique (AFM) s’est avérée particulièrement utile pour la caractérisation mécanique des cellules vivantes. Dans cette thèse, deux aspects différents ont été étudiés par AFM. Dans un premier temps, l’élasticité des cellules épithéliales étalées sur des micropatterns adhésifs a été cartographiée. Cette étude montre que l’élasticité d’une cellule varie en fonction de sa géométrie d’adhésion à la fois au niveau global et subcellulaire. La deuxième partie de cette thèse est dédiée à la caractérisation de la réponse viscoélastique d’une cellule à un stimulus mécanique oscillatoire à haute fréquence. Des études précédentes montrent que la réponse des cellules est dominée par un stress élastique et suive une loi de puissance faible à basse fréquence. Une réponse cellulaire essentiellement visqueuse est attendue à haute fréquence, mais jusqu’à présent les limitations techniques ont empêché l’évaluation de cette propriété. Dans ma thèse, ces limitations ont été dépassées grâce à la modification d’un AFM à grande vitesse (HS-AFM). Des mesures de rhéologie active sur fibroblastes ont été réalisées entre 1Hz et 120 kHz, permettant d’étendre de deux ordres de grandeur l’échelle de fréquences explorée. Ce travail montre une réponse cellulaire aux stimulations à haute fréquence plus visqueuse qu’à basse fréquence, mais suggèrent aussi une réponse bien plus complexe qu’attendue. / The field of cell mechanics gained a growing interest because of its fundamental implication in several cellular processes, such as migration, division, differentiation and apoptosis. Among other techniques, atomic force microscopy (AFM) demonstrated particularly useful for the mechanical characterization of living cells. In this thesis, two different aspects were investigated by AFM. In the first part, the elastic properties of epithelial cells grown on adhesive micropatterns were mapped. This study shows that the elasticity of a cell varies as a function of the geometry of its adhesive environment on both global and subcellular scales. The second part of this thesis focuses on the characterization of the viscoelastic response of a cell subjected to an oscillatory mechanical stimulus at high frequency. Previous studies show that the response of cells to such stimuli is mainly dominated by elastic stress and follows a weak power law at low frequency. Instead, a predominantly viscous behavior is expected at high frequency. Up to now, technical limitations prevented the experimental validation of this property. In this thesis, these limitations were overcome thanks to the modification of a high-speed AFM (HS-AFM). With this setup, active rheological measurements of living fibroblasts could be performed from 1 Hz to 120 kHz, extending of two orders of magnitude the frequency scale explored until now. This work highlights a response of cells to high-frequency stimuli which is more viscous than at low frequency, but also suggests a more complex response than expected.

L'épilepsie affecte en France environ 500 000 personnes, parmi lesquelles environ un tiers présente une pharmaco-résistance. Pour ces patients, une chirurgie peut être envisagée, qui consiste en la résection de la région cérébrale à l'origine des crises, également appelée « zone épileptogène » (ZE). Le succès du traitement chirurgical repose à la fois sur la possibilité de bien circonscrire la zone nécessaire et suffisante à la survenue des crises, mais aussi sur l'identification du potentiel rôle fonctionnel du tissu cérébral situé dans ou à proximité de cette zone, afin d'éviter que l'intervention engendre des déficits cognitifs.Dans ce travail de thèse, nous nous sommes penchés sur des signaux électrophysiologiques entre 80 et 500 Hz, appelés oscillations haute fréquence (HFO), qui ont été récemment désignés comme pouvant être de nouveaux marqueurs de la ZE, et sur leurs analogues, les HFO physiologiques, qui semblent quant à eux être impliqués dans des processus cognitifs. Nous avons tenté d'identifier des paramètres permettant de distinguer ces deux classes de HFO. Alors que ces signaux sont à l'heure actuelle essentiellement enregistrés en EEG intracérébral (SEEG pour stéréo-électro-encéphalographie), nous avons également exploré la possibilité de les étudier en Imagerie par Résonnance Magnétique fonctionnelle simultanément à la SEEG, ce qui pourrait permettre à l'avenir d'en obtenir une cartographie sur l'ensemble du cerveau. La vocation générale de ces travaux est de participer à la compréhension et l'amélioration de l'étude des HFO pathologiques et physiologique, afin d'optimiser leur utilisation pour la chirurgie des épilepsies pharmaco-résistantes / Epilepsy affects 500 000 persons in France, among which about 30% will continue to have seizure despite medication. For those patients, a surgical treatment can be proposed, that consists in removing the brain region causing seizures, namely the “epileptogenic zone” (EZ). To achieve a good surgical outcome, it is necessary to identify the zone that is both necessary and sufficient for seizure start, but also, to identify the function of the brain tissue that is located close to this zone, to avoid eventual cognitive deficits.In this thesis, we were interested in the electrophysiological activities between 80 and 500 Hz called high frequency oscillations (HFO), that were recently identified as new markers of the EZ, and the analogous physiological HFO that seem to be implicated in cognitive processes. We tried to identify features that could enable to distinguish between this to class of HFO. While these activities are still mainly recorded with intra-cerebral electro-encephalography (icEEG), we also explored the possibility to study them with functional Magnetic Resonance Imaging, simultaneously with icEEG. This method could, in the future, enable to get a HFO mapping over the whole brain. The aim of this work is to participate in the comprehension and exploration improvement of pathological and physiological HFO, to enhance their use in intractable epilepsy surgery

No description available.

La stimulation à haute fréquence (SHF) du noyau subthalamique (NST) joue un rôle essentiel chez les patients Parkinsoniens dans l'amélioration des troubles moteurs pour lesquels la dopa-thérapie n'est plus satisfaisante. Tout comme l'administration à long terme de L-DOPA, la SHF du NST, peut aussi, selon l'intensité de stimulation, évoquer des mouvements dyskinétiques. Ces dyskinésies sont considérées comme un phénomène d'apprentissage moteur pathologique, secondaire à une altération de la transmission glutamatergique et sont sous-tendues par des modifications durables d'expression génique, notamment dans le striatum. L'objectif de ce travail de thèse est d'étudier et de comparer les mécanismes moléculaires des dyskinésies induites par la L-DOPA à celles induites par la SHF, en se focalisant plus particulièrement sur la sous unité NR2B des récepteurs NMDA. Dans un premier temps, nous avons montré par immunohistochimie que la sous unité NR2B est hyperphosphorylée dans le NST et l'EP suite à l'induction de dyskinésie par la SHF du NST chez l'animal sain. Ces résultats ont été confirmés par la suite dans un modèle animal de la maladie de Parkinson, le rat 6-OHDA. La comparaison de ces modifications avec celles observées chez le rat 6-OHDA rendus dyskinétique par un traitement chronique à la L-DOPA nous permet de suggérer que l'induction des dyskinésies est associée à une hyperphosphorylation de NR2B au sein d'une voie subthalamo-entopédonculaire alors qu'une activation de NR2B dans le striatum semble être impliquée dans l'expression des dyskinésies. Enfin, nos résultats mettent également en évidence une implication différentielle des deux structures de sorties des ganglions de la base dans les processus akinétiques et dyskinésiogènes. / High frequency stimulation of the subthalamic nucleus (STN-HFS) alleviates parkinsonian motor symptoms and indirectly improves dyskinesia by decreasing L-DOPA requirement. However, inappropriate stimulation can also trigger dyskinetic movements Dyskinesia are thought to be a pathological learning process due to an overactive glutamate transmission within the basal ganglia. Moreover, several molecular changes seem to be involved in this process. The aim of the present study is to compare the molecular mechanisms of dyskinesia induced by L-DOPA and by STN-HFS, by focusing more particularly on the NR2B-containing NMDA receptor. We show by immunohistochemistry that NR2B subunit is hyperphosphorylated within the STN and the EP during a dyskinesiogenic STN-HFS in normal rats. Similar results are obtained from 6-OHDA rats, a model of Parkinson disease. Comparison of these results with those observed in 6-OHDA dyskinetic rats chronically treated with L-DOPA suggest that dyskinesia induction is associated with an hyperphosphorylation of NR2B within a subthalamo-entopeduncular network while activation of NR2B within the striatum seem to be involved in the expression of dyskinesia. A different implication of the two output of the basal ganglia in akinetic and dyskinesiogenic process is also demonstrated. STAR Date de soutenance : 8 juillet 2011 Thèse sur travaux: non

Rapid implementation of universal newborn hearing screening programs has exposed a need for a reliable test of middle ear function for timely identification of middle ear pathology and for differentiation between true sensorineural and conductive hearing losses. Use of higher probe tone frequencies for the assessment of immitance measures have proven to be more reliable and accurate in identifying MEE in infants. However a lack of classification-guidelines and age specific normative data exists. This study investigated the characteristics and normative values of high frequency tympanometric and acoustic reflex results for infants (n = 936 ears). Participants were 510 infants (262 male, 248 female) aged 0 – 12 months (mean age = 12.8 weeks) recruited from primary health care and immunization clinics in a South African community. A three-part procedure was performed on each test ear: 1) OAEs were recorded and pass results served as control variable for normal middle ear functioning; 2) 1000 Hz probe tone admittance, susceptance and conductance tympanograms were recorded and analysed in terms of shape, tympanometric peak pressure and maximum (peak) admittance; 3) 1000 Hz probe tone acoustic reflexes, measured with a 1000 Hz ipsilateral stimulus, were recorded and thresholds determined. Significant associations were observed between tympanogram shape, and OAE pass or fail results. 93% of ears with an OAE pass result displayed peaked tympanograms, while 79% of ears with absent OAE’s displayed flat tympanograms. Single peaked tympanograms were recorded in 782 ears (84%), double peaked tympanograms in 41 (4%) ears and flat sloping tympanograms in 112 (12%) ears. Admittance (Ya) tympanograms for the total sample displayed a mean admittance value of 2.9 mmho, with a standard deviation of 1.1 mmho. The 90th percent range was determined at 1.5 mmho (5th percentile) to 4.9 mmho (95th percentile). Mean tympanometric peak pressure in Ya tympanograms was 0.1 daPa, with a standard deviation of 61 daPa. The 90th percent range was -110 daPa to 90 daPa for the 5th and 95th percentiles respectively. Gender specific norms indicated a higher admittance for male ears. Age specific norms indicate a gradual increase in admittance indicating the need for age specific normative classification systems. Ipsilateral 1000 Hz stimuli acoustic reflex measurement proved successful with a 1000 Hz probe tone and present reflexes were recorded in 84% of ears tested. Significant association between acoustic reflex presence, OAE pass and peaked tympanogram results were observed. The normative tympanometric values derived from the cohort may serve as a guide for identification of middle ear effusion in neonates. High frequency tympanometry in combination with acoustic reflexes proves a useful measure for verifying middle ear functioning in young infants. / Dissertation (M (Communication Pathology))--University of Pretoria, 2006. / Speech-Language Pathology and Audiology / M (Communication Pathology) / unrestricted

Physical health can be measured in several ways both based on subjective experiences and with objective tools. However, mental health can only be measured through subjective experiences and sensations, which can be biased. Therefore, researchers adopted the notion of an objective tool to assess well-being as a complement to existing self-reported scales and suggested that heart rate variability (HRV) might be an indicator of well-being. Hence, this thesis investigates the relationship between subjective well-being (SWB) and HRV, particularly high frequency-HRV (HF-HRV). Three hypotheses, which included different forms of well-being, were developed to test the relationship. And the hypotheses were: Cognitive well-being correlates positively with HF-HRV. Positive affect correlates positively with HF-HRV, and negative affect correlates negatively with HF-HRV. A total of 19 healthy Swedish females aged from 20-35 participated and answered questionnaires measuring SWB. After they completed the SWB-scales, their heart rate was measured and then converted into HF-HRV data. The findings revealed no correlations between the cognitive SWB and HF-HRV, neither to affective SWB.

[pt] O objetivo deste trabalho é inserir variáveis exógenas no modelo Holt-Winters com múltiplos ciclos, genuinamente univariado. Serão usados dados horários de demanda de energia elétrica provenientes de uma cidade da região sudeste do Brasil e dados de temperatura, tanto em sua forma primitiva quanto derivada, por exemplo, indicadores de dias quentes, o chamado cooling degree days (CDD). Com isso, pretende-se melhorar o poder preditivo do modelo, gerando previsões com maior acurácia. / [en] The aim of this thesis is to insert exogenous variables in the model Holt-Winters with multiple cycles, genuinely univariate. Hourly data will be used for electricity demand from a city in southeastern Brazil and temperature data, both in its original form as derived, for example, indicators of hot days, cooling degree days called (CDD). With this, we intend to improve the predictive power of the model, generating predictions with greater accuracy.