Forecasting of wind speeds and directions with artificial neural networks

In this master’s thesis, wind speeds and directions were modeled with the aim of developing suitable models for hourly, daily, weekly and monthly forecasting. Artificial Neural Networks implemented in MATLAB software were used to perform the forecasts. Three main types of artificial neural network were built, namely: Feed forward neural networks, Jordan Elman neural networks and Cascade forward neural networks. Four sub models of each of these neural networks were also built, corresponding to the four forecast horizons, for both wind speeds and directions. A single neural network topology was used for each of the forecast horizons, regardless of the model type. All the models were then trained with real data of wind speeds and directions collected over a period of two years in the municipal region of Puumala in Finland. Only 70% of the data was used for training, validation and testing of the models, while the second last 15% of the data was presented to the trained models for verification. The model outputs were then compared to the last 15% of the original data, by measuring the mean square errors and sum square errors between them. Based on the results, the feed forward networks returned the lowest generalization errors for hourly, weekly and monthly forecasts of wind speeds; Jordan Elman networks returned the lowest errors when used for forecasting of daily wind speeds. Cascade forward networks gave the lowest errors when used for forecasting daily, weekly and monthly wind directions; Jordan Elman networks returned the lowest errors when used for hourly forecasting. The errors were relatively low during training of the models, but shot up upon simulation with new inputs. In addition, a combination of hyperbolic tangent transfer functions for both hidden and output layers returned better results compared to other combinations of transfer functions. In general, wind speeds were more predictable as compared to wind directions, opening up opportunities for further research into building better models for wind direction forecasting.

Exit times for some processes with normally distributed noise

Denna avhandling handlar om metoder för att hitta begränsningar för det asymptotiska beteendet hos en förväntad uthoppstid från ett område omkring en xpunkt för processer som har normalfördelad störning. I huvudsak behandlas olika typer av autoregressiva processer. Fyra olika metoder används. En metod som använder principen för stora avvikelser samt en metod som jämför uthoppstiden med en återkomsttid ger övre begränsningar för den förväntade uthoppstiden. En martingalmetod och en metod för normalfördelade stokastiska variabler ger undre begränsningar. Metoderna har alla både förtjänster och nackdelar. Genom att kombinera de olika metoderna får man de bästa resultaten. Vi får fram gränsvärdet för det asymptotiska beteendet hos en uthoppstid för den multivariata autoregressiva processen, samt motsvarande gränsvärde för den univariata autoregressiva processen av ordning n.

Artificial Ribonucleases: Oligonucleotides Conjugated with Metal Ion Chelates of Azacrowns

Ribonucleic acid (RNA) has many biological roles in cells: it takes part in coding, decoding, regulating and expressing of the genes as well as has the capacity to work as a catalyst in numerous biological reactions. These qualities make RNA an interesting object of various studies. Development of useful tools with which to investigate RNA is a prerequisite for more advanced research in the field. One of such tools may be the artificial ribonucleases, which are oligonucleotide conjugates that sequence-selectively cleave complementary RNA targets. This thesis is aimed at developing new efficient metal-ion-based artificial ribonucleases. On one hand, to solve the challenges related to solid-supported synthesis of metal-ion-binding conjugates of oligonucleotides, and on the other hand, to quantify their ability to cleave various oligoribonucleotide targets in a pre-designed sequence selective manner. In this study several artificial ribonucleases based on cleaving capability of metal ion chelated azacrown moiety were designed and synthesized successfully. The most efficient ribonucleases were the ones with two azacrowns close to the 3´- end of the oligonucleotide strand. Different transition metal ions were introduced into the azacrown moiety and among them, the Zn2+ ion was found to be better than Cu2+ and Ni2+ ions.

Fitoplâncton (exceto Chlorophyceae) de um lago artificial urbano no Sul do Brasil

O presente estudo tem como objetivo conhecer a composição do fitoplâncton no Lago das Tartarugas, situado no Jardim Botânico da cidade de Porto Alegre, Estado do Rio Grande do Sul. As amostragens foram realizadas mensalmente, no período de junho de 2007 a maio de 2008, em uma estação em três diferentes níveis de profundidade, na zona pelágica. Um total de 49 táxons específicos e infraespecíficos pertencentes a sete classes foram registrados. Cyanobacteria apresentou maior número de táxons (35% dos táxons identificados) seguida de Bacillariophyceae (33%) e Euglenophyceae (16,3%). São apresentadas descrições, medidas e ilustrações dos táxons, assim como a distribuição dos mesmos durante o ciclo anual.

Cianobactérias planctônicas de um lago artificial urbano na cidade de Goiânia, GO

Os criatórios de peixe do estado de Goiás são inúmeros e de intensa atividade recreativa. No entanto, estudos sobre as cianobactérias nesses ambientes são escassos, fato preocupante, uma vez que é comum notar-se intensa proliferação do fitoplâncton em pesqueiros, principalmente devido a ações antrópicas. O perigo consiste na formação de florações de espécies potencialmente tóxicas, principalmente de cianobactérias. Este trabalho visa inventariar as espécies planctônicas de cianobactérias ocorrentes em um pesqueiro (lago Jaó - um lago artificial raso) da área municipal de Goiânia (GO) (16º39'13" S-49º13'26" O). As amostragens foram realizadas nos períodos de seca (2003 a 2008) e chuva (2009), quando visualmente era evidente a ocorrência de florações. Foram aferidas variáveis climatológicas, morfométricas e limnológicas. O período de seca foi representativo nos anos amostrados apresentando no máximo 50 mm de precipitação mensal em 2005. Foram registrados 31 táxons de cianobactérias pertencentes aos gêneros Dolichospermum (5 spp.), Aphanocapsa (4 spp.), Microcystis (3 spp.), Pseudanabaena (3 spp.), Radiocystis (2 spp.), Oscillatoria (2 spp.), Bacularia, Coelosphaerium, Cylindrospermopsis, Geitlerinema, Glaucospira, Limnothrix, Pannus, Phormidium, Planktolyngbya, Planktothrix, Sphaerocavum e Synechocystis, esses últimos com uma espécie cada. Nos anos de 2003 a 2005 ocorreu predomínio de florações de espécies de Dolichospermum e em 2006 predominaram espécies de Microcystis, Radiocystis e Aphanocapsa. Das espécies inventariadas neste estudo, 21 são primeiras citações para o estado de Goiás e 13 foram constadas na literatura como potencialmente tóxicas.

Plant height reduction in populations of triticale (X triticosecale Wittmack) by induced mutations and artificial crosses

Induced mutations by gamma radiation (0, 5, 10, 20 and 40 kR doses) and reciprocal crosses were tested as mechanisms of enhancing genetic variability for plant height in two triticale cultivars, BR4 and EMBRAPA18. The reciprocal crosses and all doses of radiation showed similar increase in genetic amplitude for this trait, being suitable for increasing variability in breeding programs. Genotypes showed different responses as the gamma ray doses were increased, expressing shorter plant height. The decision of using induced mutations or artificial crosses depends on the resources available and the selection method to be used

Lämpövuoanturin mittauselektroniikan suunnittelu ja toteutus

Energian kulutuksen vähentäminen ja sen tutkiminen on kasvavan kiinnostuksen kohteena. Syntyneen lämmön mittaaminen on yksi tapa mitata energian siirtymistä. Lämpötilan mittaaminen on yleistä, vaikka usein on merkittävämpää selvittää missä ja miten lämpöenergia on siirtynyt. Tästä syystä tarvitaan lämpövuoantureita, jotka reagoivat suoraan lämpövuohon eli lämpöenergian siirtymiseen. Tässä tutkimuksessa suunnitellaan ja toteutetaan lämpövuoanturin mittauselektroniikka vaativaan käyttöympäristöön. Työssä käytettävän gradienttilämpövuoanturin tuottama jännitesignaali on mikrovolttiluokkaa ja ympäristön aiheuttama kohina voi olla huomattavasti suurempi. Tämän takia anturin tuottamaa signaalia on vahvistettava, jotta sitä voidaan mitata luotettavasti. Tutkimuksessa keskitytään vahvistimen suunnitteluun, mutta suunnittelussa on otettava huomioon koko järjestelmä. Anturin sähköiset ominaisuudet ja ympäristö asettavat rajoitteita vahvistimelle. Tavoitteena on selvittää miten voidaan mitata mikrovolttien jännitesignaalia mahdollisimman suurella taajuuskaistalla vaativassa käyttöympäristössä. Työn tuloksena syntyi mittalaite, jota voidaan käyttää vaativassa ympäristössä lämpövuon mittaamiseen. Suunnitteluparametrien mukainen vahvistus ja päästökaista sekä offset-jännitteen ryömintä saavutettiin suunnitellulla mittalaitteella, mutta offsetjännite ja kohina olivat hieman suunniteltua suuremmat. Mittalaitteella ja lämpövuoanturilla havaittiin selvästi lämpövuon muutoksia keinotekoisilla herätteillä.

Automated grading of left ventricular segmental wall motion by an artificial neural network using color kinesis images

The present study describes an auxiliary tool in the diagnosis of left ventricular (LV) segmental wall motion (WM) abnormalities based on color-coded echocardiographic WM images. An artificial neural network (ANN) was developed and validated for grading LV segmental WM using data from color kinesis (CK) images, a technique developed to display the timing and magnitude of global and regional WM in real time. We evaluated 21 normal subjects and 20 patients with LVWM abnormalities revealed by two-dimensional echocardiography. CK images were obtained in two sets of viewing planes. A method was developed to analyze CK images, providing quantitation of fractional area change in each of the 16 LV segments. Two experienced observers analyzed LVWM from two-dimensional images and scored them as: 1) normal, 2) mild hypokinesia, 3) moderate hypokinesia, 4) severe hypokinesia, 5) akinesia, and 6) dyskinesia. Based on expert analysis of 10 normal subjects and 10 patients, we trained a multilayer perceptron ANN using a back-propagation algorithm to provide automated grading of LVWM, and this ANN was then tested in the remaining subjects. Excellent concordance between expert and ANN analysis was shown by ROC curve analysis, with measured area under the curve of 0.975. An excellent correlation was also obtained for global LV segmental WM index by expert and ANN analysis (R² = 0.99). In conclusion, ANN showed high accuracy for automated semi-quantitative grading of WM based on CK images. This technique can be an important aid, improving diagnostic accuracy and reducing inter-observer variability in scoring segmental LVWM.

Artificial reflecting structures based on metamaterials

This thesis studies metamaterial-inspired mirrors which provide the most general control over the amplitude and phase of the reflected wavefront. The goal is to explore practical possibilities in designing fully reflective electromagnetic structures with full control over reflection phase. The first part of the thesis describes a planar focusing metamirror with the focal distance less than the operating wavelength. Its practical applicability from the viewpoint of aberrations when the incident angle deviates from the normal one is verified numerically and experimentally. The results indicate that the proposed focusing metamirror can be efficiently employed in many different applications due to its advantages over other conventional mirrors. In the second part of the thesis a new theoretical concept of reflecting metasurface operation is introduced based on Huygens’ principle. This concept in contrast to known approaches takes into account all the requirements of perfect metamirror operation. The theory shows a route to improve the previously proposed metamirrors through tilting the individual inclusions of the structure at a chosen angle from normal. It is numerically tested and the results demonstrate improvements over the previous design.

Synergistic control of forearm based on accelerometer data and artificial neural networks

In the present study, we modeled a reaching task as a two-link mechanism. The upper arm and forearm motion trajectories during vertical arm movements were estimated from the measured angular accelerations with dual-axis accelerometers. A data set of reaching synergies from able-bodied individuals was used to train a radial basis function artificial neural network with upper arm/forearm tangential angular accelerations. The trained radial basis function artificial neural network for the specific movements predicted forearm motion from new upper arm trajectories with high correlation (mean, 0.9149-0.941). For all other movements, prediction was low (range, 0.0316-0.8302). Results suggest that the proposed algorithm is successful in generalization over similar motions and subjects. Such networks may be used as a high-level controller that could predict forearm kinematics from voluntary movements of the upper arm. This methodology is suitable for restoring the upper limb functions of individuals with motor disabilities of the forearm, but not of the upper arm. The developed control paradigm is applicable to upper-limb orthotic systems employing functional electrical stimulation. The proposed approach is of great significance particularly for humans with spinal cord injuries in a free-living environment. The implication of a measurement system with dual-axis accelerometers, developed for this study, is further seen in the evaluation of movement during the course of rehabilitation. For this purpose, training-related changes in synergies apparent from movement kinematics during rehabilitation would characterize the extent and the course of recovery. As such, a simple system using this methodology is of particular importance for stroke patients. The results underlie the important issue of upper-limb coordination.

Effect of scanner acoustic background noise on strict resting-state fMRI

Functional MRI (fMRI) resting-state experiments are aimed at identifying brain networks that support basal brain function. Although most investigators consider a ‘resting-state’ fMRI experiment with no specific external stimulation, subjects are unavoidably under heavy acoustic noise produced by the equipment. In the present study, we evaluated the influence of auditory input on the resting-state networks (RSNs). Twenty-two healthy subjects were scanned using two similar echo-planar imaging sequences in the same 3T MRI scanner: a default pulse sequence and a reduced “silent” pulse sequence. Experimental sessions consisted of two consecutive 7-min runs with noise conditions (default or silent) counterbalanced across subjects. A self-organizing group independent component analysis was applied to fMRI data in order to recognize the RSNs. The insula, left middle frontal gyrus and right precentral and left inferior parietal lobules showed significant differences in the voxel-wise comparison between RSNs depending on noise condition. In the presence of low-level noise, these areas Granger-cause oscillations in RSNs with cognitive implications (dorsal attention and entorhinal), while during high noise acquisition, these connectivities are reduced or inverted. Applying low noise MR acquisitions in research may allow the detection of subtle differences of the RSNs, with implications in experimental planning for resting-state studies, data analysis, and ergonomic factors.

Artificial neural network models for predicting 1-year mortality in elderly patients with intertrochanteric fractures in China

The mortality rate of older patients with intertrochanteric fractures has been increasing with the aging of populations in China. The purpose of this study was: 1) to develop an artificial neural network (ANN) using clinical information to predict the 1-year mortality of elderly patients with intertrochanteric fractures, and 2) to compare the ANN's predictive ability with that of logistic regression models. The ANN model was tested against actual outcomes of an intertrochanteric femoral fracture database in China. The ANN model was generated with eight clinical inputs and a single output. ANN's performance was compared with a logistic regression model created with the same inputs in terms of accuracy, sensitivity, specificity, and discriminability. The study population was composed of 2150 patients (679 males and 1471 females): 1432 in the training group and 718 new patients in the testing group. The ANN model that had eight neurons in the hidden layer had the highest accuracies among the four ANN models: 92.46 and 85.79% in both training and testing datasets, respectively. The areas under the receiver operating characteristic curves of the automatically selected ANN model for both datasets were 0.901 (95%CI=0.814-0.988) and 0.869 (95%CI=0.748-0.990), higher than the 0.745 (95%CI=0.612-0.879) and 0.728 (95%CI=0.595-0.862) of the logistic regression model. The ANN model can be used for predicting 1-year mortality in elderly patients with intertrochanteric fractures. It outperformed a logistic regression on multiple performance measures when given the same variables.

HYBRID ARTIFICIAL NEURAL NETWORK APPLIEDTO MODELING SCFE OF BASIL AND ROSEMARY OILS

This work presents the results of a Hybrid Neural Network (HNN) technique as applied to modeling SCFE curves obtained from two Brazilian vegetable matrices. A series Hybrid Neural Network was employed to estimate the parameters of the phenomenological model. A small set of SCFE data of each vegetable was used to generate an extended data set, sufficient to train the network. Afterwards, other sets of experimental data, not used in the network training, were used to validate the present approach. The series HNN correlates well the experimental data and it is shown that the predictions accomplished with this technique may be promising for SCFE purposes.

Previsão de estabilidade de óleos vegetais através da rede de inteligência artificial

Redes Neurais Artificiais são técnicas computacionais que se utilizam de um modelo matemático capaz de adquirir conhecimentos pela experiência; esse comportamento inteligente da rede provém das interações entre unidades de processamento, denominadas de neurônios artificiais. O objetivo deste trabalho foi criar uma rede neural capaz de prever a estabilidade de óleos vegetais, a partir de dados de suas composições químicas, visando um modelo para a previsão da shelf-life de óleos vegetais, tendo como parâmetros apenas dados de suas composições químicas. Os primeiros passos do processo de desenvolvimento da rede consistiram na coleta de dados relativos ao problema e sua separação em um conjunto de treinamento e outro de testes. Estes conjuntos apresentaram como variáveis dados de composição química, que incluíram os valores totais em ácidos graxos, fenóis, tocoferóis e a composição individual em ácidos graxos. O passo seguinte foi a execução do treinamento, onde o padrão de entrada apresentado à rede como parâmetro de estabilidade foi o índice de peróxido, determinado experimentalmente por um período de 16 dias de armazenagem na ausência de luz, a 65ºC. Após o treinamento foi testada a capacidade de previsão adquirida pela rede, em função do parâmetro de estabilidade adotado, mas com um novo grupo de óleos. Seguindo o teste, foi determinada a correlação linear entre os valores de estabilidade previstos pela rede e aqueles determinados experimentalmente. Com os resultados obtidos, pode-se confirmar a viabilidade de previsão da estabilidade de óleos vegetais pela rede neural, a partir de dados de sua composição química, utilizando como parâmetro de estabilidade o índice de peróxido.