Showing papers in "Transportation Research Part C-emerging Technologies in 1996"

Combining kohonen maps with arima time series models to forecast traffic flow

Abstract: A hybrid method of short-term traffic forecasting is introduced; the KARIMA method. The technique uses a Kohonen self-organizing map as an initial classifier; each class has an individually tuned ARIMA model associated with it. Using a Kohonen map which is hexagonal in layout eases the problem of defining the classes. The explicit separation of the tasks of classification and functional approximation greatly improves forecasting performance compared to either a single ARIMA model or a backpropagation neural network. The model is demonstrated by producing forecasts of traffic flow, at horizons of half an hour and an hour, for a French motorway. Performance is similar to that exhibited by other layered models, but the number of classes needed is much smaller (typically between two and four). Because the number of classes is small, it is concluded that the algorithm could be easily retrained in order to track long-term changes in traffic flow and should also prove to be readily transferrable.

A microscopic traffic simulator for evaluation of dynamic traffic management systems

Abstract: A MIcroscopic Traffic SIMulator (MITSIM) has been developed for modeling traffic networks with advanced traffic control, route guidance and surveillance systems. MITSIM represents networks in detail and simulates individual vehicle movements using car following, lane changing, and traffic signal responding logic. A probabilistic route choice model is used to capture drivers' route choice decisions in the presence of real time traffic information provided by route guidance systems. The simulator is a component of a larger system for evaluating traffic management systems and interacts with a surveillance module that can represent a wide variety of sensors (e.g. loop detectors, area sensors, probe vehicles, etc.) and a traffic management module which sets traffic signals and signs, routing recommendations, etc. MITSIM is coded in C+ + using object-oriented design and supports distributed implementation. It includes a graphical user interface for animating vehicle movements in the network and displaying aggregate traffic information such as speed and density.

The travel and emissions impacts of telecommuting for the state of California Telecommuting Pilot Project

Abstract: The impacts of home-based telecommuting on travel behavior and personal vehicle emissions for participants in the State of California Telecommuting Pilot Project are analyzed using the most advanced emissions modeling tools currently available. A comparison of participants' telecommuting day travel behavior with their before-telecommuting behavior shows a 27% reduction in the number of personal vehicle trips, a 77% decrease in vehicle-miles traveled (VMT), and 39% (and 4%) decreases in the number of cold (and hot) engine starts. These decreases in travel translate into emissions reductions of: 48% for total organic gases (TOG), 64% for carbon monoxide (CO), 69% for nitrogen oxide (NOx), and 78% for particulate matter (PM). Although the authors developed the methodology to investigate the emissions impacts of telecommuting, the analysis technique can be applied to any demand management or other transportation strategy where all of the necessary model inputs are available. An analysis of the number of personal vehicle trips and VMT partitioned into commute-related and non-commute-related purposes revealed that non-commute personal vehicle trips increased by 0.5 trips per person-day on average, whereas the non-commute VMT decreased by 5.3 miles. This important finding supports (for one indicator, the number of trips) the hypothesis that non-commute travel generation is a potential negative impact of telecommuting. This finding demonstrates the need to monitor these changes as telecommuting moves into the mainstream. In this study, however, the small increase in non-commute trips has a negligible impact compared to the overall travel and emissions savings.

Transportation policy analysis with a geographic information system: the virtual network of freight transportation in europe

Abstract: This paper presents a multi-modal freight transportation model based on a digitized geographic network. A systematic analysis and decomposition of all the transport operations i.e. moving, loading and unloading, transshipping and transiting, leads to the development of a virtual network where each virtual link corresponds to a specific operation, and all transportation modes and means are inter-linked. Software, called NODUS, automatically generates the virtual network so that the model can be conveniently applied to large networks. The analytical structure of the links notation makes it easy to attach specific cost functions to each virtual link. The model is applied to the trans-European freight network of roads, railways and inland waterways for the transportation of wood. Cost functions are built up for each operation by each mode/means combination. A detailed point-to-point origin-destination matrix, calibrated on Eurostat statistics, is generated by a Monte-Carlo technique. Then, the total transportation cost is minimized with respect to the choices of routes, modes and means. This provides estimations of transportation services demands as well as modal splits, to the extent that the two hypotheses of demand based on generalized cost minimization and market contestability are accepted. A sensitivity analysis on the relative road cost is made, which provides measures of arc-elasticities.

Route choice and advanced traveler information systems on a capacitated and dynamic network

Abstract: Much of the push behind Intelligent Transportation Systems (ITS) has come from the hope that providing travelers with better information will result in reduced travel time and traffic congestion. Phase 1 of the United States' IVHS National System Architecture project, for instance, made ATIS (Advanced Traveler Information Systems) the centerpiece of its benefits evaluation, and ATIS has been the subject of numerous traffic simulation studies. The objective of this paper is to examine the ‘informational’ assumptions embedded in traffic simulations, and to assess how these assumptions affect simulation results. Most importantly, this paper examines the hypothesis that increasing market penetration can lead to a decrement in network performance. The paper proves, for a simple parallel network, that increasing the market penetration of accurate information cannot harm network performance. For this same network, the paper shows that increasing the penetration of instantaneous travel time estimates might degrade network performance. The paper also asserts that the existence, or non-existence, of an optimal market penetration is moot. The suggestion is that ATIS should not be viewed as a strategy for achieving system optimal traffic distributions. ATIS should instead be viewed first as a service to the public, to improve their confidence and comfort in using the system, and second as a means for steering traffic away from dis-equilibrium behavior and toward user optima that utilize alternate routes where feasible.

Some general results on the optimal ramp control problem

Abstract: In an effort to relieve peak hour congestion on freeways, various ramp metering algorithms have been employed to regulate the inputs to freeways from entry ramps. In this paper, we consider a freeway system comprised of a freeway section and its entry/exit ramps, and formulate the ramp control problem as a dynamic optimal process to minimize the total time spent in this system. Within this framework, we are able to show when ramp metering is beneficial to the system in terms of total time savings, and when it is not, under the restriction that the controlled freeway has to serve all of its ramp demand, and the traffic flow process follows the rules prescribed by the LWR theory with a triangular flow-density relationship. We also provide solution techniques to the problem and present some preliminary numerical results and empirical validation.

Modelling inter-urban transport flows in Italy: A comparison between neural network analysis and logit analysis

Abstract: In the present paper a modal split problem is analysed by means of two competing statistical models, the traditional logit model and the new technique for information processing, viz. the feedforward neural network model. This study aims to explore the modal split between rail and road transport modes in Italy in relation to the introduction of a new technological innovation, the new High-Speed Train. The paper is sub-divided into two major parts. The first part offers some general considerations on the use of neural networks in the light of the increasing number of empirical applications in the specific area of transport economics. The second part describes the Italian case study by using the two above mentioned statistical models. The results highlights the fact that the two adopted models, although methodologically different, are both able to provide a reasonable spatial forecasting of the phenomenon studied. In particular, the neural network model turns out to have a slightly better performance, even though there are still critical problems inherent in its application.

A link-based variational inequality formulation of ideal dynamic user-optimal route choice problem

Abstract: The ideal dynamic user-optimal (DUO) route choice problem is to determine vehicle flows on each link at each instant of time resulting from drivers using actual minimal-time routes. Actual route time is the travel time incurred while driving along the route. In a previous paper, we presented a route-based optimal control model for the ideal DUO route choice problem. However, this model is not appropriate for large-scale transportation networks because some degree of route enumeration is necessary to solve the model. In this paper, we first present the traffic network constraints and link-based DUO route choice conditions. Then, we introduce a link-based variational inequality (VI) formulation for the ideal DUO route choice problem so that route enumeration can be avoided in both the formulation and the solution procedure. By proving the necessity and sufficiency of this VI, we demonstrate that the VI formulation is equivalent to the link-based DUO route choice conditions.

A theory of traffic flow in automated highway systems

Abstract: This paper presents a theory for automated traffic flow, based on an abstraction of vehicle activities like entry, exit and cruising, derived from a vehicle's automatic control laws. An activity is represented in the flow model by the space and time occupied by a vehicle engaged in that activity. The theory formulates Traffic Management Center (TMC) plans as the specification of the activities and velocity of vehicles, and the entry and exit flows for each highway section. We show that flows that achieve capacity can be realized by stationary plans that also minimize travel time. These optimum plans can be calculated by solving a linear programming problem. The theory permits the study of transient phenomena such as congestion, and TMC feedback traffic rules designed to deal with transients. We propose a “greedy” TMC rule that always achieves capacity but does not minimize travel time. We undertake a microscopic study of the “entry” activity, and show how lack of coordination between entering vehicles and vehicles on the main line disrupts traffic flow and increases travel time. We conclude by giving some practical indication of how to obtain the space and time usage of activities from vehicle control laws. Finally, we illustrate the concepts presented in this paper with two examples of how the model is used to calculate the capacities of a one-lane automated highway system. In one example we study market penetration of adaptive cruise control and in the second example we study the effect of platooning maneuvers in a platooning architecture for AHS.

Use of vehicle positioning data for arterial incident detection

Abstract: This paper investigates the effectiveness of individual vehicle movement measures in the detection of incidents on urban arterial road segments. Several measures of varying measurement complexity, including average speed, running time and speed, and coefficient of variation of speed, are calculated from vehicle positioning data recorded at 1-s time intervals. The sensitivity of measures to differences in traffic condition is investigated, and the most sensitive measures are used as predictor variables in discriminant classification models. The results of these analyses are presented and discussed, including a simulation of the implementation of the classification models.

The freeway service patrol evaluation project: Database support programs, and accessibility

Abstract: This paper discusses the Freeway Service Patrol (FSP) Evaluation Project conducted under the California PATH program. The goal of this project was to determine the cost effectiveness of the FSP program along a 10-mile “beat” on I-880, near Hayward, California. We collected data from three different sources: loop detector data, probe vehicle data and incident log data. We describe these sources in detail and the problems that were encountered with them. We also discuss the support programs developed to process this data, and the efforts to place the data and the programs on-line for use by researchers around the country.

Automatic control of variable message signs in Aalborg

Abstract: This paper presents the design, implementation, and partial evaluation work performed by a European consortium for the development of a Variable Message Sign (VMS) information and guidance system in the city of Aalborg, Denmark. The employed control strategy is based on simple automatic control concepts with decentralized feedback loops aiming at approximating a user optimal traffic flow distribution in the mixed network, that comprises a motorway axis and an urban component. Simulation studies demonstrate the potential improvements achievable with this kind of control measures and control strategies. The implementation concept and first field results are outlined.

Neural network analysis of travel behavior : Evaluating tools for prediction

Abstract: This article explores the application of neural networks to a behavioral transportation planning problem. The motivation for adding neural networks as a new modeling methodology stems from its apparent relevance to problems requiring large scale, highly dimensional, data analysis, such as travel related behavior. Neural networks provide a tool to analyze the data in which we can model our intuition, and they provide that capability without the complication of having to formalize all the complex causal variables and relationships which other models require. The transportation issue explored, upon which the neural network methodology is tested, is a comparison of travel demand patterns of men and women in Israel. The information base is the Traveling Habits Survey (Central Bureau of Statistics, Israel, 1984, Statistical of Israel , No. 35) commissioned by the Israel Ministry of Transport; combined with demographic and socioeconomic data of the 1983 Population and Housing Census. As extensive as such surveys are, the neural networks imply that additional categories of data are necessary to predict how these elements relate to travel behavior. This article concentrates on the extent to which neural networks can combine the relative simplicity of aggregate transportation models, with the theoretical advantages and level of detail of disaggregate transportation models, without the latter's complexity. We describe the various directions we took in analyzing complex travel related data with feed forward, backpropagation trained, neural networks.

Optimization of railway operations using neural networks

Abstract: Railroad operations involve complex switching and classification decisions that must be made in short periods of time. Optimization with respect to these decisions can be quite difficult due to the discrete and non-linear characteristics of the problem. The train formation plan is one of the important elements of railroad system operations. While mathematical programming formulations and algorithms are available for solving the train formation problem, the CPU time required for their convergence is excessive. At the same time, shorter decision intervals are becoming necessary given the highly competitive operating climates of the railroad industry. The field of Artificial Intelligence (AI) offers promising alternatives to conventional optimization approaches. In this paper, neural networks (an empirically-based AI approach) are examined for obtaining good solutions in short time periods for the train formation problem (TFP). Following an overview, and formulation of railroad operations, a neural network formulation and solution to the problem are presented. First a training process for neural network development is conducted followed by a testing process that indicates that the neural network model will probably be both sufficiently fast, and accurate, in producing train formation plans.

A multiattribute analysis of goals for intelligent transportation system planning

Abstract: Transportation planning in general, and planning for Intelligent Transportation Systems (ITS) in particular, are notable both for multiple goals and for multiple constituencies. In response to this policy environment, multicriteria decision analysis has often been utilized to evaluate alternative transportation investments. This approach is extended here to assess stakeholder valuation of broad goals of an ITS planning process, the FAST-TRAC operational field test in Oakland County, a suburban region of metropolitan Detroit, Michigan, U.S.A. Representatives of stakeholder groups, ranging from emergency response firm employees to city managers to environmental groups, were interviewed. Using a modified Analytical Hierarchy Process, implicit preference weights for transportation planning goals were derived, and inter- and intragroup comparisons made. Overall, collision reduction emerged as a dominant goal accounting for nearly 35% of the overall valuation of all goals. In contrast, travel time reduction and energy/environmental impacts each accounted for about 20% of the total valuation. Stakeholder group affiliation appeared to affect transportation system preferences most strongly with regard to environmental preferences and reduction in commercial travel time; with regard to other goals, individual interests seemed to dominate those of the ostensible stakeholder group. In an environment such as that of ITS, in which policy goals are diverse and potentially conflicting, the methodologies presented here can aid in policy and system design by gauging the relative preferences of strongly interested individuals and groups. While the specific findings presented here are not generalizable to other regions, they underscore the relative importance of a range of ITS goals apart from simple reductions in travel times.

Design by Petri nets of an intersection signal controller

Abstract: Currently, traffic responsive control strategies of an isolated intersection gather available information about traffic conditions and then, in a real-time context, determine a stage (set of simultaneous green signals) by optimizing an arbitrary criterion. This optimal control is then executed by the intersection controller, which transforms the stage in traffic signal colors. In this paper, an optimal signal control device for any generic intersection is constructed just from the functioning rules: a timed color sequence for each signal, clearing times between conflicting signals and a maximum service delay for any user waiting in the intersection. The design takes into account two constraints: the controller must always respect the specified safety operating rules, and it must give the maximum freedom to the upper level. Petri nets formalism is a graphical and mathematical tool adapted to the modeling of the main features of discrete event systems; it can be used to translate specifications into operational systems. Each of the previous three safety rules can be translated in this formalism, furnishing an exact intersection controller, described by a Petri net. The application of the methodology to an actual intersection is presented. It shows that the intersection controller description is fully achieved by Petri nets, providing both a model for any upper control level and automation for any real implementation.

Case-based reasoning: A planning tool for intelligent transportation systems

Abstract: This paper develops a Case-Based Reasoning (CBR) methodology for PLANiTS (Planning and Analysis Integration for Intelligent Transportation Systems). To address a current transportation planning situation, CBR presents similar historical cases. Specifically, it estimates the impacts of proposed transportation improvement actions, including Intelligent Transportation Systems, based on previous experiences with similar actions. In this paper, a hierarchical structure for representing historical cases is developed. All historical cases consist of transportation improvement actions, performance measures and environments defined in terms of their spatial, temporal and user/traveler dimensions. In addition, cases contain information about lessons learnt such as inferences regarding their success or failure, prescriptions and case quality. Information about historical cases can be synthesized. Specifically, the reasoner contains mechanisms for 1. (a) matching at various levels of stringency, 2. (b) ranking with alternative distance and weight measures 3. (c) analyzing similar past cases with statistical operations. Also discussed are the limitations of CBR applications to transportation planning. Overall, the structure for the CBR is flexible and incorporates different stakeholder preferences for alternative transportation improvement actions and evaluation criteria. Further, the application of CBR to transportation planning, discussed in this paper, formalizes the use of case knowledge in transportation. This can lead to improving the transportation planning process.

Optimized lane assignment on an automated highway

Abstract: Highway automation entails the application of control, sensing and communication technologies to road vehicles, with the objective of improving highway performance It has been envisioned that automation could increase highway capacity by a factor of three To attain this capacity, it will be important to minimize the amount of lane-changing and optimally assign vehicles to lanes This paper develops and applies a linear programming based lane assignment model The highway system is modeled as a multi-commodity network, where the commodities represent trip destinations (ie exit ramps on highways) An unusual feature of the model is that capacities are defined by bundle constraints, which are functions of the flow entering, leaving, continuing and passing through lanes in each highway segment The objective is to maximize total flow, subject to a fixed origin/destination pattern, expressed on a proportional basis The model is tested for highways with up to 80 segments, 20 destinations and 5 lanes, and parametric analyses are provided with respect to the time-space requirement for lane-changes, number of lanes, number of segments and origin/destination pattern

Geographic information system design for network equilibrium-based travel demand models

Abstract: Travel demand analyses are useful for transportation planning and policy development in a study area. However, travel demand modeling faces two obstacles. First, standard practice solves the four travel components (trip generation, trip distribution, modal split and network assignment) in a sequential manner. This can result in inconsistencies and non-convergence. Second, the data required are often complex and difficult to manage. Recent advances in formal methods for network equilibrium-based travel demand modeling and computational platforms for spatial data handling can overcome these obstacles. In this paper we report on the development of a prototype geographic information system (GIS) design to support network equilibrium-based travel demand models. The GIS design has several key features, including: (i) realistic representation of the multimodal transportation network, (ii) increased likelihood of database integrity after updates, (iii) effective user interfaces, and (iv) efficient implementation of network equilibrium solution algorithms.

Quality of information given by advanced traveler information systems

Abstract: A number of studies have evaluated the services provided by Advanced Traveler Information Systems [ATIS] under the assumption that information supplied to drivers would be, in some sense, perfect. However, lack of sufficiently useful data and system design constraints can lead to information that is less than useful to the ATIS user. This paper examines the effects of such imperfection through a simulation-based model that was applied over a part of a large metropolitan area. The model has four basic components: 1. (i) an ATIS structure (that specifies the information-gathering, processing and disseminating aspects of the system) 2. (ii) traveler behavior 3. (iii) network characteristics 4. (iv) vehicle movement logic. Using a ‘yoked driver’ concept, a number of different route guidance strategies are examined. The results indicate that some strategies that would appear to be desirable are not so. Conversely, under high-congestion situations, strategies can be constructed that come close to ‘rectifying’ completely the effects of information imperfection. Overall the paper reiterates the potential of ATIS if information-giving strategies are designed carefully.

Environmental protection, economic efficiency and intermodal competition in freight transport

Abstract: The awareness of the consequences of a further rise in transport for the environment has not only been a matter of concern for scientific researchers but also for planners and policymakers. In fact, the environment is now an ever present factor in the new political agenda and issues of excessive traffic congestion and global atmospheric pollution are increasingly attracting administrators' attention. One of the most important scenarios proposed for the protection of the environment, taking into account the adverse effects of traffic, is the redistribution of freight transport demand. In this paper the Italian situation has been tested, evidencing productive sectors and regions really benefiting from a more effective redistribution of trade flows among existing links on the freight network. This pattern is estimated by evaluating substitution elasticities before and after the introduction of a pollution tax. Numerical simulations, in terms of reduction of pollution emissions and transportation costs, are also provided.

An alternative formulation of the simultaneous route and departure-time choice equilibrium problem

Abstract: Traditional (static) network equilibrium models have always been formulated in a route-based fashion rather than a vehicle-based fashion. That is, the decision variables have been the number of vehicles using each route rather than the route choices of each vehicle. Given the success of this approach, it is not surprising that recent “dynamic” network equilibrium models have been formulated in a similar way. That is, the decision variables in these models are usually the route-specific departure rates over time. In this paper, we develop a vehicle-based equilibrium model of simultaneous route and departure-time choice and discuss the possible advantages of this approach. We then describe a heuristic for solving this model and demonstrate its effectiveness on several small examples.

Some solution strategies for automated highway exit bottleneck problems

Abstract: Transportation planners are exploring methods to ease congestion on urban freeways. The Automated Highway System (AHS) is a potential solution that could triple current freeway capacity. A three-fold increase in freeway capacity would translate into a significant increase in peak freeway traffic flow. The corresponding increase of flow on urban roads due to exiting AHS vehicles could quickly overwhelm the capacity of urban streets and lead to critical bottlenecking of AHS flow. If exiting AHS vehicles cannot sufficiently clear AHS exits because the local streets are jammed, traffic will spill onto the AHS mainline and reduce AHS capacity. The loss of AHS capacity due to congestion in urban destinations and the resulting back spill of traffic onto the AHS will greatly detract from the AHS system.The following Non-Intermodal Interface (NII) and Intermodal Interface (II) strategies are proposed as solutions to this critical problem: 1. (1) the dispersion of flow through AHS networks; 2. (2) the dispersion of flow at commuter destinations; 3. (3) the consolidation of flow at central locations. Specifically, smart street signals and prioritized street design, intermodal terminals, and automated garages are identified as potential solutions. This paper qualitatively discusses the advantages and disadvantages of each of the three general strategies, and quantitatively discusses the advantages and disadvantages of each of the four specific solutions through simulated scenarios which account for varying ramp volumes, trip attraction patterns, and urban structures. Our investigation shows that creative combinations of these methods can be useful to integrate future AHS with local road networks. Without such integration, the AHS cannot succeed.

Spin glass and the interactions of congestion and emissions: An exploratory step

Abstract: This paper models traffic congestion formation on highways and roads by recognizing the centrality of dynamical systems and using concepts from complexity theory as imbedded in the spin glasses analogue. Further, it explores the concept of how an increase in air pollution caused by vehicle exhaust emission can be traced to traffic congestion, specifically to the acceleration/deceleration of vehicles on the roads. First, spin glass is introduced and then by applying the two-dimensional x − y Ising model and defining a Hamiltonian (based on Edwards-Anderson and Mattis models of spin glass systems) for a system of vehicles on the road, derivations are made of the specific friction of congestion and the bulk modulus of congestion using the Gibbs-Boltzmann statistic. Similarly using the interactions of vehicles with each other and the resulting accelerations and decelerations of vehicles as the basis for exhaust emissions, derivations are made of a specificity of exhaust emissions . These are analogues to the entropy models of thermodynamics. This series of derivations serves as an analytical model for detecting incidents of congestion and increase in air pollution due to exhaust emissions in transportation systems.

Design and management of transport networks. New approaches and experiments from a European and American perspective : An introduction

Abstract: To tackle ongoing transport problems--both at the urban and inter-urban level--such as sudden traffic growth, congestion, lack of capacity, pollution, road accidents, etc., a great deal of attention has recently been paid to new issues emerging in transport systems analysis. These concern complex networks, dynamic (in)stability, and especially sustainability. As a consequence, innovative models/methodologies have been geared to creating new tools and technologies able to come with these spatio-temporal transport changes (from both the demand and the supply side). This special issue on new approaches/experiments--oriented towards the design and management of transport networks--has therefore, as its first aim, to offer some reflections on these new methodological "challenges" and achievements, by trying to answer some actual and fundamental research questions, like those concerning the modeling of traffic (forecasts) in a network of n dimensions, where n is a large number; the real-time design of network flow dynamics; the prediction of traffic congestion/pollution; the route-choice behavior in the presence of uncertainty in the travel costs, etc.