Induction heating is one of the efficient and modern technique for heating raw materials for hot forging process. The induction heating furnaces use electro-magnetic field to transfer energy to the metal workpiece and heat is generated inside the material. The magnetic field can be provided by using induction coil. The power supplied to induction coil, the moving speed of the billet that is called conveyor speed and the coil box hole diameter are the factors affecting the resultant temperature of the heated billet. In this study, AISI 1045 type steel billets with a diameter of &Oslash / 30 mm and length of 100 mm have been heated in a particular induction heater. During heating, effects of different levels of power, conveyor speed and the coil boxes with different hole diameters are investigated. The 125 KW 3000 Hz induction heater which is available in METU-BILTIR Research and Application Center Forging Laboratory is used in experiments. The heating experiments are designed according to 23 Factorial Design of Experiment Method. Multiple linear regression technique is used to derive a mathematical formula to predict the temperature of the heated billet. A good correlation between the measured temperatures that are the results of different sets of induction heating parameters and the predicted temperatures that are calculated by using temperature prediction formula has been observed.

The economic evaluation of a new airport investment requires the use of estimated future air passenger demand.Today it is well known that air passenger demand is basicly dependent on various socioeconomic factors of the country and the region where the planned airport would serve. This study is focused on estimating the future air passenger demand for planned airports in Turkey where the historical air passsenger data is not available.For these purposses, neural networks and multi-linear regression were used to develop forecasting models. As independent variables,twelve socioeconomic parameters are found to be significant and used in models. The available data for the selected indicators are statistically analysed and it is observed that most of the data is highly volatile, heteroscedastic and show no definite patterns. In order to develop more reliable models, various methods like data transformation, outlier elimination and categorization are applied to the data.Only seven of total twelve indicators are used as the most significant in the regression model whereas in neural network approach the best model is achieved when all the twelve indicators are included. Both models can be used to predict air passenger demand for any future year for Or-Gi and Zafer Airports and future air passenger demand for similar airports. Regression and neural models are tested by using various statistical test methods and it is found that neural network model is superior to regression model for the data used in this study.

Abstract Travel time reliability is defined as the consistency or dependability in travel times during a specified period of time under stated conditions, and it can be used for evaluating the performance of traffic networks based on LOS (Level of Service) of the HCM (Highway Capacity Manual). Travel time reliability is also one of the most understood measures for road users to perceive the current traffic conditions, and help them make smart decisions on route choices, and hence avoid unnecessary delays (Liu & Ma, 2009). Therefore, travel time reliability on urban arterials has become a major concern for daily commuters, business owners, urban transportation planners, traffic engineers, MPO (Metropolitan Planning Organization) members as congestion has grown substantially over the past thirty (30) years in urban areas of every size. Many studies have been conducted in the past on travel time reliability without a full analysis or explanation of the fundamental traffic and geometric components of the corridors. However, a generalized model which captures the different factors that influence travel time reliability such as posted speed, access density, arterial length, traffic conditions, signalized intersection spacing, roadway and intersection geometrics, and signal control settings is still lacking. Specially, there is a need that these factors be weighted according to their impacts. This dissertation by using a linear regression model has identified 10 factors that influence travel time reliability on urban arterials. The reliability is measured in term of travel time threshold, which represents the addition of the extra time (buffer or cushion time) to average travel time when most travelers are planning trips to ensure on-time arrival. "Reliable" segments are those on which travel time threshold is equal to or lowers than the sum of buffer time and average travel time. After validation many scenarios are developed to evaluate the influencing factors and determine appropriate travel times reliability. The linear regression model will help 1) evaluate strategies and tactics to satisfy the travel time reliability requirements of users of the roadway network--those engaged in person transport in urban areas 2) monitor the performance of road network 3) evaluate future options 4) provide guidance on transportation planning, roadway design, traffic design, and traffic operations features.

The present study is concerned with developing some statistical models to evaluate and analyze software reliability. We have developed the analytical structure of the logistic model to be used for testing and evaluating the reliability of a software package. The proposed model has been shown to be useful in the testing and debugging stages of the developmental process of a software package. It is important that prior to releasing a software package to marketing that we have achieved a target reliability with an acceptable degree of confidence. The proposed model has been evaluated and compared with several existing statistical models that are commonly used. Real software failure data was used for the comparison of the proposed logistic model with the others. The proposed model gives better results or it is equally effective. The logistic model was also used to model the mean time between failure of software packages. Real failure data was used to illustrate the usefulness of the proposed statistical procedures. Using the logistic model to characterize software failures we proceed to develop Bayesian analysis of the subject model. This modeling was based on two different difference equations whose parameters were estimated with Bayesian regressions subject to specific prior and mean square loss function.

The paper consists of two parts. In the first part the linear regression method is described, also employment possibilities and pupil achievement research are discussed. The second one deals with some factors and what influence they have on the eight class pupil’s mathematical achievement analysis.

Dissolved oxygen (DO) is an important water quality parameter that is used to assess the health of aquatic ecosystems. Typically physically-based numerical models are used to predict DO, however, these models do not capture the complexity and uncertainty seen in highly urbanised riverine environments. To overcome these limitations, an alternative approach is proposed in this dissertation, that uses a combination of data-driven methods and fuzzy numbers to improve DO prediction in urban riverine environments. A major issue of implementing fuzzy numbers is that there is no consistent, transparent and objective method to construct fuzzy numbers from observations. A new method to construct fuzzy numbers is proposed which uses the relationship between probability and possibility theory. Numerical experiments are used to demonstrate that the typical linear membership functions used are inappropriate for environmental data. A new algorithm to estimate the membership function is developed, where a bin-size optimisation algorithm is paired with a numerical technique using the fuzzy extension principle. The developed method requires no assumptions of the underlying distribution, the selection of an arbitrary bin-size, and has the flexibility to create different shapes of fuzzy numbers. The impact of input data resolution and error value on membership function are analysed. Two new fuzzy data-driven methods: fuzzy linear regression and fuzzy neural network, are proposed to predict DO using real-time data. These methods use fuzzy inputs, fuzzy outputs and fuzzy model coefficients to characterise the total uncertainty. Existing methods cannot accommodate fuzzy numbers for each of these variables. The new method for fuzzy regression was compared against two existing fuzzy regression methods, Bayesian linear regression, and error-in-variables regression. The new method was better able to predict DO due to its ability to incorporate different sources of uncertainty in each component. A number of model assessment metrics were proposed to quantify fuzzy model performance. Fuzzy linear regression methods outperformed probability-based methods. Similar results were seen when the method was used for peak flow rate prediction. An existing fuzzy neural network model was refined by the use of possibility theory based calibration of network parameters, and the use of fuzzy rather than crisp inputs. A method to find the optimum network architecture was proposed to select the number of hidden neurons and the amount of data used for training, validation and testing. The performance of the updated fuzzy neural network was compared to the crisp results. The method demonstrated an improved ability to predict low DO compared to non-fuzzy techniques. The fuzzy data-driven methods using non-linear membership functions correctly identified the occurrence of extreme events. These predictions were used to quantify the risk using a new possibility-probability transformation. All combination of inputs that lead to a risk of low DO were identified to create a risk tool for water resource managers. Results from this research provide new tools to predict environmental factors in a highly complex and uncertain environment using fuzzy numbers. / Graduate / 0543 / 0775 / 0388

Neste trabalho procurou-se avaliar o processo de construção e aplicação de modelos preditivos no meio florestal. Para tanto, no primeiro artigo parte-se de uma amostra destrutiva de 200 indivíduos de dez espécies arbóreas distintas, originárias do bioma Atlântico, testando-se três modelos teóricos comumente usados na predição de volume e biomassa, sendo a esses adicionados preditores informativos da densidade básica da árvore. Para a avaliação os modelos ajustados foram simuladas três situações preditivas distintas. Os resultados demonstraram que aplicar o modelo em situações distintas a da amostra de ajuste gera viés nas predições que, no entanto, é reduzido com a entrada dos referidos preditores. O segundo artigo apresenta aplicações da calibração do modelo linear de efeito misto na predição do volume em plantios de Eucalyptus grandis em primeira e segunda rotação. Para tanto, partiu-se do modelo de Schumacher e Hall, em sua forma linearizada, para o desenvolvimento modelo de efeitos mistos, que considerou alguns de seus parâmetros como sendo aleatórios ao longo das diferentes fazendas. A calibração foi realizada em nível de fazenda partindo-se de um pequeno número de árvores-amostra. A abordagem foi desenvolvida para modelos univariados de primeira rotação, além de modelos bivariados de duas rotações. Os resultados mostraram que o procedimento de calibração fornece predições mais confiáveis que a dos modelos tradicionais de efeitos fixos em ambas as rotações. O terceiro artigo apresenta aplicações da calibração do modelo linear de efeito misto na predição da biomassa de árvores de espécies nativas numa floresta Ombrófila densa. Partiuse do modelo de potência, em sua forma linearizada, para o desenvolvimento modelo de efeitos mistos e dois níveis: parcela e espécie, O ajuste do modelo foi feito considerando esses dois níveis, mas a calibração foi realizada em cada nível ignorando o efeito do outro, nível. Os resultados mostraram que o procedimento de calibração fornece predições mais confiáveis em nível de espécie que os modelos tradicionais. Em nível de parcela, a calibração não foi efetiva. / In this study we tried to evaluate the process of construction and application of predictive models in forestry. Therefore, in the first paper we started from a destructive sample of 200 individuals from ten different tree species, originating from the Atlantic biome. We tested three theoretical models commonly used to predict volume and biomass, which was added predictors related to tree basic density. To evaluate the models were simulated three different predictive situations. The results showed that applying the model in different situations from the sample generates bias on predictions; however, it is reduced by adding the referred predictors. The second article presents applications of linear mixedeffects models and calibration to predict the volume in Eucalyptus grandis plantations in first and second rotation. Therefore, we started with the model of Schumacher and Hall, in their linearized form to develop the mixed-effects model, which considered some of its parameters as random throughout the different farms. The calibration was performed at the farm level and starting from a small number of sample trees. The approach was developed to first rotation univariate models, and a bivariate model of both rotations. The results showed that the calibration procedure provides more reliable predictions than the traditional fixed effects models in both rotations. The third article presents applications of linear mixedeffects model and calibration to predict the biomass in a rain forest. We started from the power model, in its linearized form, for developing the mixed-effects model considering two levels of grouping: plot and species, Model fitting was made considering these two levels, but the calibration was performed on each level ignoring the other level effect. The results showed that the calibration procedure provides more reliable predictions at species level than traditional models. On the plot level, the calibration was not effective.

Conselho Nacional de Desenvolvimento Científico e Tecnológico / The production of sunflower (Helianthus annuus L.) in Brazil is booming, in emphasis there is the south with an acreage of 17% with the oilseed. The performance of sunflower is directly related to the choice about many factors like sowing date, genotype, the appropriate management and soil fertility, crop rotation, crop succession and especially environmental factors. This culture adapts to different soil and edaphoclimatic conditions and can be cultivated in almost all over Brazil. On this basis, the aim of this work was to evaluate the cultivation environment influence and sowing date on phonology and morphology traits of sunflower. Besides identifying the main stages of culture development within sowing dates tested. It also aimed to build a model based on multiple linear regression equations that express the growth of sunflower under field conditions. And this way can be used as a tool to estimate the potential yield of sunflower to the Northwestern of RS, based on the projection of meteorological variables for the period and management since the establishment to harvest this important oilseed. For this, was evaluated the genotypes Hélio 250, 251, 358, 884 and 885 in the 2008-09 harvest and genotypes Hélio 250, 251, 253, 360, 211, and HLA 211, and Paraíso 33 in the 2009-10 harvest in three seasons (August, October and December). The experiment was conducted in Random Block Complete with four replications on the UFSM Campus in Frederico Westphalen, in Northwestern Brazil. The morphological traits measured were plant height, chapter size, a thousand seeds weight and yield. Phenological characters were days of sowing emergence (S-E), days from emergence to early flowering (E-FI) days from initial flowering to full flowering (FI-FP) and days of full flowering to physiological maturity (FP-MF). Individual analysis and combined and the averages compared by Tukey test at 5% were realized. The models estimated form multiple linear regression equations, using, the method Backward in the 5% level of error probability. The independent variables in models of entry were: TM, Tmin, Tmax, PP and GD of each phonologic stage. The phonologic stage and yield were the dependent variables. Based on results, concludes that the morphological traits plant height, is influenced by the environment and sowing dates, for the chapter size there is the environment influence and was found that the yield has positively answers to environmental conditions and sowing times. For the studied phonological traits also occurs influence of environmental conditions and sowing date. With the exception onset flowering to full flowering witch presents no interference form the sowing time. When evaluated the agrometeorological variables interference, we can say that there was no interaction between them and the dependent variables, it can be said that the thermal plus and the maximum and minimum temperatures are crucial for the phonological sub-stages of sunflower, been the achene yield influenced just for the thermal plus for the sub-stages form sowing to emergence and since this with the initial flowering and full flowering. About the sowing stages, the August present an increase of the plant cycle, while in the December suggests a cycle reduction, and that the sowing in October is the culture preferred phase to Northwestern region of the RS. / A produção de girassol (Helianthus annuus L) no Brasil está em expansão, destaca-se o Sul com uma área cultivada de 17% com a oleaginosa. O desempenho do girassol está diretamente relacionado à escolha da época de semeadura, do genótipo, do manejo adequado do solo e fertilidade, sistema de rotação, da sucessão de culturas e especialmente dos fatores ambientais. A cultura que se adapta a diferentes condições edafoclimáticas, podendo ser cultivada em quase todo o Brasil. Com base nisto, o objetivo deste trabalho foi avaliar a interferência do ambiente de cultivo e da época de semeadura sobre caracteres morfológicos e fenológicos da cultura do girassol. Além de identificar as principais fases do desenvolvimento da cultura dentro das épocas de semeadura testadas. Também se objetivou construir um modelo com base em equações de regressão linear múltipla que expressem o crescimento da cultura do girassol nas condições de campo. E, desta forma possa ser empregado como ferramenta para se estimar o potencial produtivo da cultura do girassol para a região Noroeste do RS, com base na projeção das variáveis meteorológicas para o período e do manejo desde a implantação até a colheita desta importante oleaginosa. Para isto, foram avaliados os genótipos Hélio 250, 251, 358; 884 e 885 na safra 2008-09 e os genótipos Hélio 250, 251, 253; 360, HLA 211 e Paraíso 33 na safra 2009-10 em três épocas (agosto, outubro e dezembro). O experimento foi conduzido em blocos completos com quatro repetições no Campus da UFSM em Frederico Westphalen, na região noroeste do RS. As variáveis avaliadas foram altura de planta, tamanho do capítulo, massa de mil aquênios e o rendimento. As variáveis morfológicas avaliadas foram altura de planta (AP), tamanho do capítulo (TC), massa de mil aquênios (MMA) e o rendimento (REND). Os caracteres fenológicos foram dias da semeadura a emergência (S-E), dias da emergência a floração inicial (E-FI), dias da floração inicial a floração plena (FI-FP) e dias da floração plena a maturação fisiológica (FP-MF). Realizaram-se análises individuais e conjuntas e as médias comparadas pelo teste de Tukey a 5%. Os modelos estimados a partir de equações de regressão linear múltipla, através do método Backward, em nível de 5% de probabilidade de erro. As variáveis independentes de entrada nos modelos foram: a TM, TMin, TMax, PP, e a GD de cada fase fenológica. As variáveis dependentes foram às fases fenológicas e o rendimento. A variável morfológica altura de planta é influenciada pelo ambiente e por épocas de semeadura, para o tamanho do capítulo ocorre influência do ambiente e para o rendimento verificou-se que responde positivamente a condições ambientais e a épocas de semeadura. Para os caracteres fenológicos estudados ocorre também influência da época de semeadura, com exceção da variável início da floração a floração plena que não apresenta interferência da época de semeadura. Quando avaliado a interferência das variáveis agrometeorológicas, pode-se afirmar que não houve interação entre estas e as variáveis dependentes, podendo-se dizer que a soma térmica e as temperaturas máximas e mínimas são determinantes para os subperiodos fenológicas da cultura do girassol, sendo o rendimento de aquênios influenciado somente pela soma térmica dos subperiodos da semeadura a emergência, desta com a floração inicial e a floração plena. Quanto às épocas de semeadura, a de agosto acarreta aumento do ciclo da planta, enquanto que a de dezembro determina redução do ciclo, e que, a época de semeadura de outubro é a fase preferencial da cultura para a região Noroeste do Rio Grande do Sul.

Coordenação de Aperfeiçoamento de Pessoal de Nível Superior / The growing world population requires a higher demand for food and one of the techniques to meet this demand is irrigation. One of the best alternatives for the storage of water for use in irrigation are earth dams. The accurate determination of cohesion and angle of internal friction is an essential and of great concern in the drafting of earth dams process criteria, but their determination is expensive and also time consuming process. The objective of this study is to evaluate a model that allows an estimation of soil shear strength using two different techniques (multivariate analysis and artificial neural networks) to obtain the strength parameters (cohesion and angle of internal friction) as a function of textural composition , density, Atterberg limits (plasticity, liquidity and plasticity index) and the degree of soil moisture. Different database were searched in the literature with the dependent and independent variables needed to conduct the study. 6 dataset were totaled. PTFs were generated through multiple linear regression (MLR), stepwise, and artificial neural networks (ANN) with each data set. Through MLR were estimated friction angle and cohesion separately since the RNAs were estimated jointly and separately maintaining these two parameters form the architecture (one hidden layer) and varying the topology of the networks (10, 20, 30, 40 , 50 and 70 neurons in the hidden layer). After the performance index (Id) and subsequent classification of each FPT was calculated. The results demonstrated the inefficiency in MLRs to estimate parameters and the superiority of ANN to predict the cohesion and friction angle. Estimation of the parameters together shows different results than when estimated separately. Thus, the estimated shear the soil parameters RNAs can be effective for a given set of data, in this case belonging to RNAs 3, 5 and 6. / O crescimento da população mundial requer uma maior demanda de alimentos e uma das técnicas para suprir essa demanda é a irrigação. Uma das melhores alternativas para o armazenamento da água para utilizar na irrigação são as barragens de terra. A determinação precisa da coesão e do ângulo de atrito interno do solo é um critério essencial e de grande preocupação no processo de elaboração de projetos de barragens de terra, porém sua determinação é de alto custo e também um processo demorado. O objetivo deste trabalho é avaliar um modelo que permita uma estimativa da resistência do solo ao cisalhamento utilizando duas técnicas diferentes (análise multivariada e redes neurais artificiais) para a obtenção dos parâmetros de resistência (coesão e ângulo de atrito interno) em função da composição textural, densidade do solo, limites de Atterberg (plasticidade, liquidez e índice de plasticidade) e do grau de umidade do solo. Foram pesquisados na literatura diferentes banco de dados contendo as variáveis dependentes e independentes necessárias para realizar o estudo. Foram totalizados 6 conjunto de dados. Foram geradas FPTs por meio de regressão linear múltipla (RLM), método Stepwise, e redes neurais artificiais (RNA) com cada conjunto de dados. Por meio de RLM foram estimados coesão e ângulo de atrito separadamente, já pelas RNAs foram estimados de forma conjunta e de forma separada esses dois parâmetros mantendo a arquitetura (uma camada oculta) e variando a topologia das redes (10, 20, 30, 40, 50 e 70 neurônios na camada escondida). Após foi calculado o índice de desempenho (Id) e posterior classificação de cada FPT. Os resultados demonstraram a ineficiência nas RLMs para estimativa dos parâmetros e a superioridade das RNAs na predição da coesão e ângulo de atrito. A estimativa dos parâmetros conjuntamente mostra diferença nos resultados do que quando estimados de forma separada. Assim, a estimativa dos parâmetros cisalhantes do solo pelas RNAs, podem ser eficazes, para determinado conjunto de dados, nesse caso pertencentes às RNAs 3, 5 e 6.

The predominant campestral vegetation in the state of Rio Grande do Sul, as well as all types of vegetal formations, may be considered as a dynamic system subject to several disturbance agents. Fire is frequently mentioned as one of them, and may present natural or anthropic causes. Occasional burns are mainly a result of electric discharges from the atmosphere. Intentional burns may be controlled or not and are usually associated to the management of areas aimed at agriculture and cattle raising activities. Thus, the objective of this research was to perform a study on accidental field burns in order to characterize and to identify places of higher burn incidence in the county of Santa Maria RS, Brazil, and to aid in the planning and control of fires, correlating the number of burns with meteorological elements in order to identify the most propitious conditions for the occurrence of these events. The interest variable (response) in this study was the number of daily calls received by the Santa Maria Fire Brigade obtained from its records within the period from January 1st 1993 to December 31st 2004. This variable was explained by meteorological elements such as: maximum and minimum temperature; relative air humidity measured at 9:00am, 3:00pm and 9:00pm; insolation; rain precipitation and average wind velocity at the day of occurrence and by the number of days without any pluviometric precipitation before the occurrence of the interest variable. It was verified that the Fire Brigade received 1.81 daily calls, on average; that the call was preceded by a dry period of four days on average, and that most burns occurred in the afternoon and at the almost uninhabited RS 287 highway alongside region. The month in which the Fire Brigade received the highest number of calls was August, and the year of 1999 was the one presenting the highest occurrence of field burns. Moreover, the number of calls was equally distributed along the weekdays. Based on quartiles, city districts with high, intermediate and low chances for the occurrence of burns were determined, and regions alongside the highway and the following city districts: Distrito Industrial, Medianeira, Itararé, Tomazzetti and Parque Pinheiro Machado were those presenting the highest chances for the occurrence of burns. Based on the correlation between dependent variable and all independent variables, it was verified that the variable with the highest correlation with the number of calls received by the Fire Brigade was the relative air humidity. The evaluation of the twenty-four models tested revealed that in all of them, the presuppositions were violated, being therefore, inappropriate for the forecast of the independent variable. / A vegetação campestre, predominante no Rio Grande do Sul, bem como todos os demais tipos de formações vegetais, pode ser considerada um sistema dinâmico sujeito a uma série de agentes de perturbações. O fogo costuma ser citado como um destes, podendo ser de causa natural ou provocado pelo homem. As queimadas casuais são resultantes, em especial, de descargas elétricas da atmosfera. Queimas provocadas podem apresentar-se controladas ou não e costumam estar vinculadas ao manejo de áreas utilizadas em fins agropecuários. Desta forma, esta pesquisa teve como objetivos realizar um estudo de queimadas acidentais em campo para identificar, caracterizar e localizar os locais de maiores ocorrências dessas na cidade de Santa Maria - RS com intuito de auxiliar no planejamento e controle de incêndios, relacionar o número de queimadas com os elementos meteorológicos para identificar as condições mais propícias à ocorrência desse evento. A variável de interesse (resposta), neste estudo, foi o número de chamadas recebidas por dia pelo Corpo de Bombeiros de Santa Maria, obtidas dos seus fichários, no período de 1º de janeiro de 1993 a 31 de dezembro de 2004. Essa variável foi explicada por elementos meteorológicos, como: temperatura máxima e mínima, umidade relativa do ar coletada às 9 h, às 15 h e às 21 h, insolação, precipitação e velocidade média do vento do dia de sua ocorrência e pelo número de dias sem precipitação pluviométrica, anteriores ao da ocorrência da variável de interesse. Verificou-se que em média o Corpo de Bombeiros recebia 1,81 chamadas diárias. Antes da ocorrência de uma chamada, não chovia, em média, a quatro dias, e a grande maioria das chamadas eram no período da tarde e para as margens das rodovias que circundam a cidade, principalmente na RS 287, rodovia com margens pouco habitadas. O mês em que ocorreu o maior número de chamadas ao Corpo de Bombeiros foi agosto, sendo o ano de 1999 o que acumulou maior ocorrência de queimadas. Além disso, o número de chamadas distribuem-se equivalentemente nos dias de semana. A partir dos quartis, determinou-se os bairros do município com grande, média e poucas chances de ocorrência de queimadas, sendo que as margens das rodovias e os bairros Distrito Industrial, Medianeira, Itararé, Tomazzetti e Parque Pinheiro Machado foram as áreas com maiores chances de ocorrência de queimadas. Determinada a correlação entre a variável dependente e todas as variáveis independentes, verificou-se que a mais correlacionada com o número de chamadas recebidas pelo Corpo de Bombeiros foi a umidade relativa média. A avaliação dos pressupostos dos vinte e quatro modelos testados, revelou que em todos eles as pressuposições foram violadas, não sendo portanto, adequados à previsão da variável independente.