Abstract

Um dos objetivos dos projetos e estudos na área de transporte público é atrair o maior número possível de viagens. Um primeiro passo para estimular uma maior utilização do transporte público pode ser a captação de pessoas que já têm predisposição para utilizá-lo, mas que não o fazem por alguma deficiência específica no serviço. Este é o contexto no qual mapas que representam o potencial de utilização dos transportes públicos podem desempenhar um papel importante, como discutido neste estudo. A pesquisa tem como objetivo principal a aplicação e avaliação de duas técnicas destinadas a identificar potenciais usuários de transporte público e como estes se distribuem geograficamente em uma cidade brasileira selecionada para o estudo. Nas técnicas aqui exploradas, o município em análise é dividido em áreas em função do código de endereçamento postal. Estas áreas são caracterizadas pelos atributos socioeconômicos da sua população e do sistema de transporte. Diante da hipótese de melhoria na qualidade do transporte público, dois segmentos de usuários de automóvel foram determinados: usuários que trocariam para ônibus e usuários que ainda preferem o carro. Com isso, foi construído um modelo capaz de representar o comportamento de escolha dos usuários de cada área. A metodologia proposta envolve quatro passos: i) comparação de modelos Logit elaborados com dados de São Carlos (Brasil) e Wageningen (Holanda), ii) ajustes no modelo de São Carlos, iii) elaboração de um modelo de escolha modal por redes neurais artificiais e iv) elaboração dos mapas potenciais. As duas últimas etapas foram concebidas tanto para análises independentes, como também para comparação com o modelo Logit. Assim, a construção de cenários futuros permitiu identificar e localizar espacialmente os potenciais usuários de transporte público. Foi possível verificar também qual a influência de alguns atributos sobre a escolha do modo de transporte urbano. Por exemplo, usuários de domicílios com três ou quatro pessoas têm menor probabilidade de vir a utilizar o ônibus regularmente. Por fim, um cenário futuro tornou possível destacar áreas onde é esperado um aumento do potencial de uso do transporte público devido a mudanças nos valores de densidade populacional. One of the objectives of projects and studies on public transport is to attract the largest possible number of trips. A first step for increasing transit ridership may be the attraction of those individuals who already have a predisposition to use the service, but do not use it because of any specific inadequacies. This is the context in which maps displaying the potential use of public transport may play an important role, as discussed in this study. The research aims at the application and evaluation of two techniques used to identify potential users of public transport and to show how they are geographically distributed in a Brazilian city selected for the study. In the techniques discussed here, the municipality under analysis is divided into areas according to the postal codes of the streets. These areas are characterized by socioeconomic attributes of the population and of the transport system. Two segments of automobile users were determined, under the assumption that the quality of public transport would be improved: users who would switch to buses and users who still prefer the car. Based on that, a model designed to represent the choice behavior of users in each urban area was built. The proposed methodology involved four steps: i) the comparison of Logit models built with data of São Carlos (Brazil) and Wageningen (The Netherlands), ii) adjustments in the model of São Carlos, iii) the development of a mode choice model based on artificial neural networks, and iv) construction of potential maps. The third and fourth steps were meant for both independent analysis, and also for comparison with the Logit model. Thus, the construction of future scenarios allowed the identification and spatial location of potential users of public transport. It was also possible to learn about the influence of some attributes on urban transportation choice. For example, users living in households with three or four persons are less likely to become bus riders. Finally, a future scenario was able to highlight the areas where the potential for public transport could be increased due to changes in population density values.

Abstract

Public transportation sectors have played significant roles in accommodating passengers and commodities efficiently and effectively. The modes of public transportation often follow pre-defined operation schedules and routes. Therefore, planning these schedules and routes requires extensive efforts in analyzing the built environment and collecting demand data. Once a transit route is operational as an example, collecting and maintaining real-life information becomes an important task to evaluate service quality using different Key Performance Indicators (KPIs). One of these KPIs is transit travel time along the route. This paper aims to develop a transit travel time prediction model using an artificial intelligence approach. In this study, 12 public bus routes serving the Greater City of Doha were selected. While the ultimate goal is to predict transit travel time from the start to the end of the journeys collected over a period of one-year, routespecific inputs were used as inputs for this prediction. To develop a generalized model, the input variables for the transit route included the number and type of intersections, number of each type of turning movements and the built environment. An Artificial Neural Networks (ANN) model is used to process 78,004 valid datasets. The results indicate that the ANN model is capable of providing reliable and accurate transit travel time estimates, with a coefficient of determination (R2) of 0.95. Transportation planners and public transportation operators can use the developed model as a tool to estimate the transit travel time.

Abstract

Transportation systems play a fundamental role in nowadays society. Indeed, every developed or countries undergoing development have invested and keep investing to build a safe and functional transportation network. The main concern nowadays, particularly for developed countries that already have a very complete network, is to keep it operational under all conditions. However, due to the network extension and increased budget constraints, such task is difficult to accomplish. In the framework of transportations networks, particularly for railway, slopes are perhaps the element for which their failure can have a strongest impact at several levels. Although there are some models and systems to detect slope failures, most of them were developed for natural slopes, presenting some constrains when applied to engineered (human-made) slopes. They have limited applicability as most of the existing systems were developed based on particular case studies or using small databases. Moreover, another aspect that can limit its applicability is related with the information used to feed them, such as data taken from complex tests or from expensive monitoring systems. Aiming to overcome this drawback, we took advantage of the high flexible learning capabilities of Artificial Neural Networks (ANNs) and Support Vector Machines (SVMs), which have been used in the past to model complex nonlinear mappings. Both data mining algorithms were applied in the development of a classification tool able to identify the stability condition of a rock cutting slope, keeping in mind the use of information usually collected during routine inspections activities (visual information) to feed them. For that, two different strategies were followed: nominal classification and regression. Moreover, to overcome the problem of imbalanced data, three training sampling approaches were explored: no resampling, SMOTE and Oversampling. The achieved results are present and discussed, comparing the performance of both algorithms (ANN and SVM) according to each modeling strategy as well as the effect of the sampling approaches. This work was supported by FCT – “Fundação para a Ciência e a Tecnologia", within ISISE, project UID/ECI/04029/2013 as well Project Scope: UID/CEC/00319/2013 and through the post-doctoral Grant fellowship with reference SFRH/BPD/94792/2013. This work was also partly financed by FEDER funds through the Competitivity Factors Operational Programme - COMPETE and by national funds through FCT within the scope of the project POCI-01-0145-FEDER-007633. This work has been also supported by COMPETE: POCI01-0145-FEDER-007043. A special thanks goes to Network Rail that kindly make available the data (basic earthworks examination data and the Earthworks Hazard Condition scores) used in this work. info:eu-repo/semantics/publishedVersion

Abstract

Modeling vehicle delay has been an interesting subject for traffic engineers and urban planners. Determination of vehicle delay is a complex task and the delay is influenced by many variables that have uncertainties and vagueness, especially for non-uniform or over-saturated conditions. In this study, vehicle delay is modeled using new approaches such as Fuzzy Logic (FL) and Artificial Neural Networks (ANN) to deal with all conditions. The Neuro Fuzzy Delay Estimation (NFDE) model and Artificial Neural Networks Delay Estimation (ANNDE) model are developed. The overall delay data required for the model were collected from ten signalized intersections in Turkey. The results of the developed models are compared with the Highway Capacity Manual (HCM), Akqelik's methods and the delay data collected from intersections. The results showed that delay estimations by the ANNDE and NFDE model are promising. It is also inferred that the NFDE model results are the best fitted. The Average Relative Error (ARE) rates of NFDE model are determined as 7% for under-saturated and 5% for over-saturated conditions. The results reflect the fact that the neuro-fuzzy approach may be used as a promising method in vehicle delay estimation. (c) 2006 Elsevier Ltd. All rights reserved. C1 Pamukkale Univ, Fac Engn, Dept Civil Engn, TR-20070 Kinikli, Denizli, Turkey.

Abstract

In recent years there has been extensive development of the fire computer models, and its use in the study of the fire safety, fire investigation, etc. has been increased. The most important types of fire computer models are the field model and the zone model. The first model reaches a better approximation to fire dynamics, but the second one requires less computational time. Additionally, in the last years, it should be noted the great advances in information processing using artificial neural networks, and it has become a useful tool with application in very diverse fields. This paper analyzes the possibilities of develop a new fire computer model using artificial neural networks. In the first approach to this objective, a simple compartment was analyzed with a field model. After that, simulations employing General Regression Neural Network were performed. This method achieves similar results that the field model employing computational times closer to the zone models. The neural network has been trained with FDS field model and validating the resulting model with data from a full scale test. In later stages other phenomena and different types of networks will be evaluated.

Abstract

The objective of this paper was to predict the failure load in single lap adhesive joints subjected to tensile loading by using artificial neural networks. Experimental data obtained from the literature cover the single lap adhesive joints with various geometric models under the tensile loading. The data are arranged in a format such that two input parameters cover the length and width of bond area in single lap adhesive joints and the corresponding output is the ultimate failure load. An artificial neural network model was developed to estimate relationship between failure loads by using geometric dimensions of bond area as input data. A three-layer feedforward artificial neural network that utilized Levenberg–Marquardt learning algorithm model was used in order to train network. It was observed that artificial neural network model can estimate failure load of single lap adhesive joints with acceptable error. Mean absolute percentage error and Nash–Sutcliffe coefficient of efficiency values of both training and testing data were 3.523 and 3.524 and 0.997 and 0.992, respectively. The results showed that the artificial neural network is an efficient alternative method to predict the failure load of single lap adhesive joints. Also estimated results are in very good agreement with the experimental data.

Abstract

The artificial neural network (RNA) is a computational model that emulates the biological neural system in information processing. The originating models are suitable for the purpose of describing long-term specifics, in addition to nonlinear relationships. This tool is used to predict physical chemical and microbiological parameters that influence water quality. The United States National Sanitation Foundation proposed a water quality index, known as the NSF WQI. This article describes the design, training and use of the three-layer neural perceptron neural model for the calculation of the NSF WQI of the Utcubamba River and its tributaries. Using the Matlab software and applying the Levenberg-Marquardt training algorithm, the optimal RNA architecture was found to be 6-12-1, plus the percentage for the training, validation, and test sets of 70 %, 10 %, and 20 % respectively. RNA performance has been evaluated using the root of the root mean square error (RMSE) and the correlation coefficient (R). High correlations (greater than 0.94) were made between the measured and predicted values. Finally, the RNA proposal offers a useful alternative for the calculation and prediction of the water quality index in relation to dissolved oxygen (DO), biochemical demand for oxygen (BOD), nitrates, fecal coliforms, potential for hydrogen ions (pH) and turbidity.