Analisi di serie temporali riguardanti dati energetici mediante architetture neurali profonde

Il presente lavoro di tesi riguarda lo studio e l'impiego di architetture neurali profonde (nello specifico stacked denoising auto-encoder) per la definizione di un modello previsionale di serie temporali. Il modello implementato è stato applicato a dati industriali riguardanti un impianto fotovoltaico reale, per effettuare una predizione della produzione di energia elettrica sulla base della serie temporale che lo caratterizza. I risultati ottenuti hanno evidenziato come la struttura neurale profonda contribuisca a migliorare le prestazioni di previsione di strumenti statistici classici come la regressione lineare multipla.

Energy harvesting from body motion using rotational micro-generation

Autonomous system applications are typically limited by the power supply operational lifetime when battery replacement is difficult or costly. A trade-off between battery size and battery life is usually calculated to determine the device capability and lifespan. As a result, energy harvesting research has gained importance as society searches for alternative energy sources for power generation. For instance, energy harvesting has been a proven alternative for powering solar-based calculators and self-winding wristwatches. Thus, the use of energy harvesting technology can make it possible to assist or replace batteries for portable, wearable, or surgically-implantable autonomous systems. Applications such as cardiac pacemakers or electrical stimulation applications can benefit from this approach since the number of surgeries for battery replacement can be reduced or eliminated. Research on energy scavenging from body motion has been investigated to evaluate the feasibility of powering wearable or implantable systems. Energy from walking has been previously extracted using generators placed on shoes, backpacks, and knee braces while producing power levels ranging from milliwatts to watts. The research presented in this paper examines the available power from walking and running at several body locations. The ankle, knee, hip, chest, wrist, elbow, upper arm, side of the head, and back of the head were the chosen target localizations. Joints were preferred since they experience the most drastic acceleration changes. For this, a motor-driven treadmill test was performed on 11 healthy individuals at several walking (1-4 mph) and running (2-5 mph) speeds. The treadmill test provided the acceleration magnitudes from the listed body locations. Power can be estimated from the treadmill evaluation since it is proportional to the acceleration and frequency of occurrence. Available power output from walking was determined to be greater than 1mW/cm³ for most body locations while being over 10mW/cm³ at the foot and ankle locations. Available power from running was found to be almost 10 times higher than that from walking. Most energy harvester topologies use linear generator approaches that are well suited to fixed-frequency vibrations with sub-millimeter amplitude oscillations. In contrast, body motion is characterized with a wide frequency spectrum and larger amplitudes. A generator prototype based on self-winding wristwatches is deemed to be appropriate for harvesting body motion since it is not limited to operate at fixed-frequencies or restricted displacements. Electromagnetic generation is typically favored because of its slightly higher power output per unit volume. Then, a nonharmonic oscillating rotational energy scavenger prototype is proposed to harness body motion. The electromagnetic generator follows the approach from small wind turbine designs that overcome the lack of a gearbox by using a larger number of coil and magnets arrangements. The device presented here is composed of a rotor with multiple-pole permanent magnets having an eccentric weight and a stator composed of stacked planar coils. The rotor oscillations induce a voltage on the planar coil due to the eccentric mass unbalance produced by body motion. A meso-scale prototype device was then built and evaluated for energy generation. The meso-scale casing and rotor were constructed on PMMA with the help of a CNC mill machine. Commercially available discrete magnets were encased in a 25mm rotor. Commercial copper-coated polyimide film was employed to manufacture the planar coils using MEMS fabrication processes. Jewel bearings were used to finalize the arrangement. The prototypes were also tested at the listed body locations. A meso-scale generator with a 2-layer coil was capable to extract up to 234 µW of power at the ankle while walking at 3mph with a 2cm³ prototype for a power density of 117 µW/cm³. This dissertation presents the analysis of available power from walking and running at different speeds and the development of an unobtrusive miniature energy harvesting generator for body motion. Power generation indicates the possibility of powering devices by extracting energy from body motion.

Active surgical positioning device for a cochlear implant electrode array

Cochlear implants have been of great benefit in restoring auditory function to individuals with profound bilateral sensorineural deafness. The implants are used to directly stimulate auditory nerves and send a signal to the brain that is then interpreted as sound. This project focuses on the development of a surgical positioning tool to accurately and effectively place an array of stimulating electrodes deep within the cochlea. This will lead to improved efficiency and performance of the stimulating electrodes, reduced surgical trauma to the cochlea, and as a result, improved overall performance to the implant recipient. The positioning tool reported here consists of multiple fluidic chambers providing localized curvature control along the length of the attached silicon electrode array. The chambers consist of 200μm inner diameter PET (polyethylene therephthalate) tubes with 4μm wall thickness. The chambers are molded in a tapered helical configuration to correspond to the cochlear shape upon relaxation of the actuators. This ensures that the optimal electrode placement within the cochlea is retained after the positioning tool becomes dormant (for chronic implants). Actuation is achieved by injecting fluid into the PET chambers and regulating the fluidic pressure. The chambers are arranged in a stacked, overlapping design to provide fluid connectivity with the non-implantable pressure controller and allow for local curvature control of the device. The stacked tube configuration allows for localized curvature control of various areas along the length of the electrode and additional stiffening and actuating power towards the base. Curvature is affected along the entire length of a chamber and the result is cumulative in sections of multiple chambers. The actuating chambers are bonded to the back of a silicon electrode array.

Graphite exfoliation by supercritical carbon dioxide extraction

Supercritical carbon dioxide is used to exfoliate graphite, producing a small, several-layer graphitic flake. The supercritical conditions of 2000, 2500, and 3000 psi and temperatures of 40°, 50°, and 60°C, have been used to study the effect of critical density on the sizes and zeta potentials of the treated flakes. Photon Correlation Spectroscopy (PCS), Brunauer-Emmett-Teller (BET) surface area measurement, field emission scanning electron microscopy (FE-SEM), and atomic force microscopy (AFM) are used to observe the features of the flakes. N-methyl-2-pyrrolidinone (NMP), dimethylformamide (DMF), and isopropanol are used as co-solvents to enhance the supercritical carbon dioxide treatment. As a result, the PCS results show that the flakes obtained from high critical density treatment (low temperature and high pressure) are more stable due to more negative charges of zeta potential, but have smaller sizes than those from low critical density (high temperature and low pressure). However, when an additional 1-hour sonication is applied, the size of the flakes from low critical density treatment becomes smaller than those from high critical density treatment. This is probably due to more CO2 molecules stacked between the layers of the graphitic flakes. The zeta potentials of the sonicated samples were slightly more negative than nonsonicated samples. NMP and DMF co-solvents maintain stability and prevented reaggregation of the flakes better than isopropanol. The flakes tend to be larger and more stable as the treatment time increases since larger flat area of graphite is exfoliated. In these experiments, the temperature has more impact on the flakes than pressure. The BET surface area resultsshow that CO2 penetrates the graphite layers more than N2. Moreover, the negative surface area of the treated graphite indicates that the CO2 molecules may be adsorbed between the graphite layers during supercritical treatment. The FE-SEM and AFM images show that the flakes have various shapes and sizes. The effects of surfactants can be observed on the FE-SEM images of the samples in one percent by weight solution of SDBS in water since the sodium dodecylbenzene sulfonate (SDBS) residue covers all of the remaining flakes. The AFM images show that the vertical thickness of the graphitic flakes can ranges from several nanometers (less than ten layers thick), to more than a hundred nanometers. In conclusion, supercritical carbon dioxide treatment is a promising step compared to mechanical and chemical exfoliation techniques in the large scale production of thin graphitic flake, breaking down the graphite flakes into flakes only a fewer graphene layers thick.

TIME LAPSE SEISMIC OBSERVATIONS AND EFFECTS OF RESERVOIR COMPRESSIBILITY AT TEAL SOUTH OIL FIELD

One of the original ocean-bottom time-lapse seismic studies was performed at the Teal South oil field in the Gulf of Mexico during the late 1990’s. This work reexamines some aspects of previous work using modern analysis techniques to provide improved quantitative interpretations. Using three-dimensional volume visualization of legacy data and the two phases of post-production time-lapse data, I provide additional insight into the fluid migration pathways and the pressure communication between different reservoirs, separated by faults. This work supports a conclusion from previous studies that production from one reservoir caused regional pressure decline that in turn resulted in liberation of gas from multiple surrounding unproduced reservoirs. I also provide an explanation for unusual time-lapse changes in amplitude-versus-offset (AVO) data related to the compaction of the producing reservoir which, in turn, changed an isotropic medium to an anisotropic medium. In the first part of this work, I examine regional changes in seismic response due to the production of oil and gas from one reservoir. The previous studies primarily used two post-production ocean-bottom surveys (Phase I and Phase II), and not the legacy streamer data, due to the unavailability of legacy prestack data and very different acquisition parameters. In order to incorporate the legacy data in the present study, all three poststack data sets were cross-equalized and examined using instantaneous amplitude and energy volumes. This approach appears quite effective and helps to suppress changes unrelated to production while emphasizing those large-amplitude changes that are related to production in this noisy (by current standards) suite of data. I examine the multiple data sets first by using the instantaneous amplitude and energy attributes, and then also examine specific apparent time-lapse changes through direct comparisons of seismic traces. In so doing, I identify time-delays that, when corrected for, indicate water encroachment at the base of the producing reservoir. I also identify specific sites of leakage from various unproduced reservoirs, the result of regional pressure blowdown as explained in previous studies; those earlier studies, however, were unable to identify direct evidence of fluid movement. Of particular interest is the identification of one site where oil apparently leaked from one reservoir into a “new” reservoir that did not originally contain oil, but was ideally suited as a trap for fluids leaking from the neighboring spill-point. With continued pressure drop, oil in the new reservoir increased as more oil entered into the reservoir and expanded, liberating gas from solution. Because of the limited volume available for oil and gas in that temporary trap, oil and gas also escaped from it into the surrounding formation. I also note that some of the reservoirs demonstrate time-lapse changes only in the “gas cap” and not in the oil zone, even though gas must be coming out of solution everywhere in the reservoir. This is explained by interplay between pore-fluid modulus reduction by gas saturation decrease and dry-frame modulus increase by frame stiffening. In the second part of this work, I examine various rock-physics models in an attempt to quantitatively account for frame-stiffening that results from reduced pore-fluid pressure in the producing reservoir, searching for a model that would predict the unusual AVO features observed in the time-lapse prestack and stacked data at Teal South. While several rock-physics models are successful at predicting the time-lapse response for initial production, most fail to match the observations for continued production between Phase I and Phase II. Because the reservoir was initially overpressured and unconsolidated, reservoir compaction was likely significant, and is probably accomplished largely by uniaxial strain in the vertical direction; this implies that an anisotropic model may be required. Using Walton’s model for anisotropic unconsolidated sand, I successfully model the time-lapse changes for all phases of production. This observation may be of interest for application to other unconsolidated overpressured reservoirs under production.

TUNING, AVO, AND FLAT-SPOT EFFECTS IN A SEISMIC ANALYSIS OF NORTH SEA BLOCK F3

Reflection seismic data from the F3 block in the Dutch North Sea exhibits many large-amplitude reflections at shallow horizons, typically categorized as “brightspots ” (Schroot and Schuttenhelm, 2003), mainly because of their bright appearance. In most cases, these bright reflections show a significant “flatness” contrasting with local structural trends. While flatspots are often easily identified in thick reservoirs, we have often occasionally observed apparent flatspot tuning effects at fluid contacts near reservoir edges and in thin reservoir beds, while only poorly understanding them. We conclude that many of the shallow large-amplitude reflections in block F3 are dominated by flatspots, and we investigate the thin-bed tuning effects that such flatspots cause as they interact with the reflection from the reservoir’s upper boundary. There are two possible effects to be considered: (1) the “wedge-model” tuning effects of the flatspot and overlying brightspots, dimspots, or polarity-reversals; and (2) the stacking effects that result from possible inclusion of post-critical flatspot reflections in these shallow sands. We modeled the effects of these two phenomena for the particular stratigraphic sequence in block F3. Our results suggest that stacking of post-critical flatspot reflections can cause similar large-amplitude but flat reflections, in some cases even causing an interface expected to produce a ‘dimspot’ to appear as a ‘brightspot’. Analysis of NMO stretch and muting shows the likely exclusion of critical offset data in stacked output. If post-critical reflections are included in stacking, unusual results will be observed. In the North Sea case, we conclude the tuning effect was the primary reason causing for the brightness and flatness of these reflections. However, it is still important to note that care should be taken while applying muting on reflections with wide range of incidence angles and the inclusion of critical offset data may cause some spurious features in the stacked section.

USING AUTO- AND CROSS-CORRELATIONS FROM SEISMIC NOISE TO MONITOR VELOCITY CHANGES AT VILLARRICA VOLCANO, CHILE

We used the Green's functions from auto-correlations and cross-correlations of seismic ambient noise to monitor temporal velocity changes in the subsurface at Villarrica volcano in the Southern Andes of Chile. Campaigns were conducted from March to October 2010 and February to April 2011 with 8 broadband and 6 short-period stations, respectively. We prepared the data by removing the instrument response, normalizing with a root-mean-square method, whitening the spectra, and filtering from 1 to 10 Hz. This frequency band was chosen based on the relatively high background noise level in that range. Hour-long auto- and cross-correlations were computed and the Green's functions stacked by day and total time. To track the temporal velocity changes we stretched a 24 hour moving window of correlation functions from 90% to 110% of the original and cross correlated them with the total stack. All of the stations' auto-correlations detected what is interpreted as an increase in velocity in 2010, with an average increase of 0.13%. Cross-correlations from station V01, near the summit, to the other stations show comparable changes that are also interpreted as increases in velocity. We attribute this change to the closing of cracks in the subsurface due either to seasonal snow loading or regional tectonics. In addition to the common increase in velocity across the stations, there are excursions in velocity on the same order lasting several days. Amplitude decreases as the station's distance from the vent increases suggesting these excursions may be attributed to changes within the volcanic edifice. In at least two occurrences the amplitudes at stations V06 and V07, the stations farthest from the vent, are smaller. Similar short temporal excursions were seen in the auto-correlations from 2011, however, there was little to no increase in the overall velocity.

Topo IIIalpha and BLM act within the Fanconi anemia pathway in response to DNA-crosslinking agents

The Bloom protein (BLM) and Topoisomerase IIIalpha are found in association with proteins of the Fanconi anemia (FA) pathway, a disorder manifesting increased cellular sensitivity to DNA crosslinking agents. In order to determine if the association reflects a functional interaction for the maintenance of genome stability, we have analyzed the effects of siRNA-mediated depletion of the proteins in human cells. Depletion of Topoisomerase IIIalpha or BLM leads to increased radial formation, as is seen in FA. BLM and Topoisomerase IIIalpha are epistatic to the FA pathway for suppression of radial formation in response to DNA interstrand crosslinks since depletion of either of them in FA cells does not increase radial formation. Depletion of Topoisomerase IIIalpha or BLM also causes an increase in sister chromatid exchanges, as is seen in Bloom syndrome cells. Human Fanconi anemia cells, however, do not demonstrate increased sister chromatid exchanges, separating this response from radial formation. Primary cell lines from mice defective in both Blm and Fancd2 have the same interstrand crosslink-induced genome instability as cells from mice deficient in the Fancd2 protein alone. These observations demonstrate that the association of BLM and Topoisomerase IIIalpha with Fanconi proteins is a functional one, delineating a BLM-Topoisomerase IIIalpha-Fanconi pathway that is critical for suppression of chromosome radial formation.

A modular LHC built on the DNA three-way junction

A light-harvesting complex composed of a p-stacked multichromophoric array in a DNA three-way junction is described. The modular design allows for a ready exchange of non-covalently attached energy acceptors

Enhancement of excess electron transfer efficiency in DNA containing a phenothiazine donor and multiple stable phenanthrenyl base pairs

EET grown ohm: Excess electron transfer (EET) was observed within a DNA duplex containing π-stacked phenothiazine as an electron donor, phenanthrenes as electron carriers and 5-bromouracil as an electron trap. Increasing the number of phenanthrenyl base pairs increased EET efficiency.

Charge transport through perylene bisimide molecular junctions: An electrochemical approach

Single molecular junction conductances of a family of five symmetric and two unsymmetric perylene tetracarboxylic bisimides (PBI) with variable bay-area substituents were studied employing a scanning tunneling microscope (STM)-based break junction technique. The stretching experiments provide clear evidence for the formation of single molecular junctions and π–π stacked dimers. Electrolyte gating demonstrates a distinct gating effect in symmetric molecular junctions, which strongly depends on molecular structure and properties of the solvent/electrolyte. Weak π–π-coupling in the unsymmetric dimers prevents rectification.

Two-dimensional supramolecular polymers

One of the biggest issues of modern materials science is developing of strategies to create large and ordered assemblies in the form of discrete nanoscale objects. Oligopyrenotides (OPs) represent novel class of amphiphilic molecules which tend to self-assemble forming highly ordered structures. As has been already shown OPs are able to form 1D («rod-like») supramolecular polymer [1]. Since programmed arraying of polyaromatic hydrocarbons in structurally defined objects could offer enhanced performance over the individual components, prediction and controlling of their spatial arrangement remains challenging. Herein we demonstrate that certain changes to design of pyrene’s molecular core allow Py3 form 2D supramolecular assemblies («nanosheets») instead of 1D. Two dimensional supramolecular polymers are attractive objects due to their exceptional properties which originate from in-plan alignment of molecular units in the sheets with constant thickness ~ 2 nm [2]. These assemblies have high degree of internal order: the interior consists of hydrophobic pyrenes and alkyl chains, whereas the exterior exists as a net of hydrophilic, negatively charged phosphates. The Py3 units are hold up by non-covalent interactions what makes these assemblies totally reversible. Moreover the polymerization occurs via nucleation-elongation mechanism. To study Py3 self-assembly, we carried out whole set of spectroscopic (UV/vis, fluorescence, DLS) and microscopic experiments (AFM)

Two-dimensional supramolecular polymers

One of the biggest issues of modern materials science is developing of strategies to create large and ordered assemblies in the form of discrete nanoscale objects. Oligopyrenotides (OPs) represent novel class of amphiphilic molecules which tend to self-assemble forming highly ordered structures. As has been already shown OPs are able to form 1D («rod-like») supramolecular polymer [1]. Since programmed arraying of polyaromatic hydrocarbons in structurally defined objects could offer enhanced performance over the individual components, prediction and controlling of their spatial arrangement remains challenging. Herein we demonstrate that certain changes to design of pyrene’s molecular core allow Py3 form 2D supramolecular assemblies («nanosheets») instead of 1D. Two dimensional supramolecular polymers are attractive objects due to their exceptional properties which originate from in-plan alignment of molecular units in the sheets with constant thickness ~ 2 nm [2]. These assemblies have high degree of internal order: the interior consists of hydrophobic pyrenes and alkyl chains, whereas the exterior exists as a net of hydrophilic, negatively charged phosphates. The Py3 units are hold up by non-covalent interactions what makes these assemblies totally reversible. Moreover the polymerization occurs via nucleation-elongation mechanism. To study Py3 self-assembly, we carried out whole set of spectroscopic (UV/vis, fluorescence, DLS) and microscopic experiments (AFM)

Intracoronary Optical Coherence Tomography and Histology of Overlapping Everolimus-Eluting Bioresorbable Vascular Scaffolds in a Porcine Coronary Artery Model

OBJECTIVES: This study sought to assess the vascular response of overlapping Absorb stents compared with overlapping newer-generation everolimus-eluting metallic platform stents (Xience V [XV]) in a porcine coronary artery model. BACKGROUND: The everolimus-eluting bioresorbable vascular scaffold (Absorb) is a novel approach to treating coronary lesions. A persistent inflammatory response, fibrin deposition, and delayed endothelialization have been reported with overlapping first-generation drug-eluting stents. METHODS: Forty-one overlapping Absorb and overlapping Xience V (XV) devices (3.0 × 12 mm) were implanted in the main coronary arteries of 17 nonatherosclerotic pigs with 10% overstretch. Implanted coronary arteries were evaluated by optical coherence tomography (OCT) at 28 days (Absorb n = 11, XV n = 7) and 90 days (Absorb n = 11, XV n = 8), with immediate histological evaluation following euthanasia at the same time points. One animal from each time point was evaluated with scanning electron microscopy alone. A total of 1,407 cross sections were analyzed by OCT and 148 cross sections analyzed histologically. RESULTS: At 28 days in the overlap, OCT analyses indicated 80.1% of Absorb struts and 99.4% of XV struts to be covered (p < 0.0001), corresponding to histological observations of struts with cellular coverage of 75.4% and 99.6%, respectively (p < 0.001). Uncovered struts were almost exclusively related to the presence of "stacked" Absorb struts, that is, with a direct overlay configuration. At 90 days, overlapping Absorb and overlapping XV struts demonstrated >99% strut coverage by OCT and histology, with no evidence of a significant inflammatory process, and comparable % volume obstructions. CONCLUSIONS: In porcine coronary arteries implanted with overlapping Absorb or overlapping XV struts, strut coverage is delayed at 28 days in overlapping Absorb, dependent on the overlay configuration of the thicker Absorb struts. At 90 days, both overlapping Absorb and overlapping XV have comparable strut coverage. The implications of increased strut thickness may have important clinical and design considerations for bioresorbable platforms.

Stratigraphic analysis of lake level fluctuations in Lake Ohrid: an integration of high resolution hydro-acoustic data and sediment cores

Abstract. Ancient Lake Ohrid is a steep-sided, oligotrophic, karst lake that was tectonically formed most likely within the Pliocene and often referred to as a hotspot of endemic biodiversity. This study aims on tracing significant lake level fluctuations at Lake Ohrid using high-resolution acoustic data in combination with lithological, geochemical, and chronological information from two sediment cores recovered from sub-aquatic terrace levels at ca. 32 and 60m water depth. According to our data, significant lake level fluctuations with prominent lowstands of ca. 60 and 35m below the present water level occurred during Marine Isotope Stage (MIS) 6 and MIS 5, respectively. The effect of these lowstands on biodiversity in most coastal parts of the lake is negligible, due to only small changes in lake surface area, coastline, and habitat. In contrast, biodiversity in shallower areas was more severely affected due to disconnection of today sublacustrine springs from the main water body. Multichannel seismic data from deeper parts of the lake clearly image several clinoform structures stacked on top of each other. These stacked clinoforms indicate significantly lower lake levels prior to MIS 6 and a stepwise rise of water level with intermittent stillstands since its existence as water-filled body, which might have caused enhanced expansion of endemic species within Lake Ohrid.