Reports de recerca
http://hdl.handle.net/2117/3943
2016-07-31T05:31:39ZThe general traffic assignment problem: a proximal point method for equilibrium computation with application to the demand adjustment problem
http://hdl.handle.net/2117/89274
The general traffic assignment problem: a proximal point method for equilibrium computation with application to the demand adjustment problem
Montero Mercadé, Lídia; Codina Sancho, Esteve
An adaptation of the proximal algorithm for the traffic assignment problem under a user
equilibrium formulation for a general asymmetric traffic network is presented in this
paper, following the recently published results of Pennanen regarding convergence under
nonmonotonicity. As is well known the problem can be formulated as a variational
inequality and the algorithmic solutions developed uptodate guarantee convergence
only under too restrictive conditions which are difficult to appear in practice. In this
paper new conditions guaranteing convergence are developed and the possibility of
including the algorithm on a bilevel scheme is discussed
Presented at the SEIO (Sociedad Española de Estadística e Investigació Operativa) Congress 2003, Lleida 8 to 11th April.
2016-07-27T13:24:24ZMontero Mercadé, LídiaCodina Sancho, EsteveAn adaptation of the proximal algorithm for the traffic assignment problem under a user
equilibrium formulation for a general asymmetric traffic network is presented in this
paper, following the recently published results of Pennanen regarding convergence under
nonmonotonicity. As is well known the problem can be formulated as a variational
inequality and the algorithmic solutions developed uptodate guarantee convergence
only under too restrictive conditions which are difficult to appear in practice. In this
paper new conditions guaranteing convergence are developed and the possibility of
including the algorithm on a bilevel scheme is discussedIncident prediction: a statistical approach to dynamic probability estimation : application to a test site in Barcelona
http://hdl.handle.net/2117/89273
Incident prediction: a statistical approach to dynamic probability estimation : application to a test site in Barcelona
Montero Mercadé, Lídia; Barceló Bugeda, Jaime; Perarnau, Josep
Real-time models for estimating incident probabilities (EIP models) are innovative methods for predicting the potential occurrence of incidents and improving the effectiveness of incident management policies devoted to increasing road safety. EIP models imbedded in traffic management systems can lead to the development of control strategies for reducing the likelihood of incidents before they occur. This paper presents and discusses the design, implementation and off-line testing of an EIP model in the PRIME (Prediction of Congestion and Incidents in Real Time for Intelligent Incident Management and Emergency Traffic Management) Project of the “Information Societies Technology Programme” of the EU. A statistically-oriented approach based on Generalized Linear Regression models with polytomous responses is developed: geometry, traffic and weather conditions are taken as explanatory variables at a road section level and a binary variable related to incident occurrence or otherwise for the prevailing conditions is taken as a response variable on the first level of decision. Once the probability of a generic incident has been predicted, the lower level models in the selected hierarchical approach will predict the probabilities of incidents in a set of categories defined at a design level.
The EIP model has been incorporated in the AIMSUN microscopic simulation environment (developed by TSS ). AIMSUN is able to emulate a traffic management system, since it simulates traffic evolution including the replication of observed incidents and incorporates different modules of incident and traffic management in such a way that the impact of traffic management strategies can be evaluated by simulation.
A test site in Barcelona, located in a 15-km portion of the Ronda de Dalt ring road provided the data for calibrating and testing the EIP module. The selected site is equipped with 12 CCTV cameras for traffic monitoring, 18 local controllers, 12 detection stations, 10 variable message panels and 13 variable speed signals. Detection stations provide measures of different traffic variables in lane detail every minute.
DR 2002/08 Departament d'EIO - Research Supported by PRIME European Project
2016-07-27T13:16:59ZMontero Mercadé, LídiaBarceló Bugeda, JaimePerarnau, JosepReal-time models for estimating incident probabilities (EIP models) are innovative methods for predicting the potential occurrence of incidents and improving the effectiveness of incident management policies devoted to increasing road safety. EIP models imbedded in traffic management systems can lead to the development of control strategies for reducing the likelihood of incidents before they occur. This paper presents and discusses the design, implementation and off-line testing of an EIP model in the PRIME (Prediction of Congestion and Incidents in Real Time for Intelligent Incident Management and Emergency Traffic Management) Project of the “Information Societies Technology Programme” of the EU. A statistically-oriented approach based on Generalized Linear Regression models with polytomous responses is developed: geometry, traffic and weather conditions are taken as explanatory variables at a road section level and a binary variable related to incident occurrence or otherwise for the prevailing conditions is taken as a response variable on the first level of decision. Once the probability of a generic incident has been predicted, the lower level models in the selected hierarchical approach will predict the probabilities of incidents in a set of categories defined at a design level.
The EIP model has been incorporated in the AIMSUN microscopic simulation environment (developed by TSS ). AIMSUN is able to emulate a traffic management system, since it simulates traffic evolution including the replication of observed incidents and incorporates different modules of incident and traffic management in such a way that the impact of traffic management strategies can be evaluated by simulation.
A test site in Barcelona, located in a 15-km portion of the Ronda de Dalt ring road provided the data for calibrating and testing the EIP module. The selected site is equipped with 12 CCTV cameras for traffic monitoring, 18 local controllers, 12 detection stations, 10 variable message panels and 13 variable speed signals. Detection stations provide measures of different traffic variables in lane detail every minute.Some proposals for incident prediction models in in-response project
http://hdl.handle.net/2117/88148
Some proposals for incident prediction models in in-response project
Montero Mercadé, Lídia; Barceló Bugeda, Jaime
This research is a part of the authors collaboration in the European project IN-RESPONSE (INcident RESPonse with ON-line innovative Sensing) in Drive Program. This project is at this point a 2 years project that began in January 1.996. The goal of the project consists on the physical development of a system for Incident Management in urban motorways, that might be available at the local Traffic Control Center. Sites involved in the definition of the tool are: Valencia, Thessaloniki, Munchen, Eindhoven, Paris and Oslo.
2016-06-17T16:57:06ZMontero Mercadé, LídiaBarceló Bugeda, JaimeThis research is a part of the authors collaboration in the European project IN-RESPONSE (INcident RESPonse with ON-line innovative Sensing) in Drive Program. This project is at this point a 2 years project that began in January 1.996. The goal of the project consists on the physical development of a system for Incident Management in urban motorways, that might be available at the local Traffic Control Center. Sites involved in the definition of the tool are: Valencia, Thessaloniki, Munchen, Eindhoven, Paris and Oslo.An approach to use cooperative car data in dynamic OD matrix estimation
http://hdl.handle.net/2117/87124
An approach to use cooperative car data in dynamic OD matrix estimation
Montero Mercadé, Lídia; Barceló Bugeda, Jaime
Traffic management applications are supported by dynamic models whose input should be realistic real-time OD demand matrices in order to find efficient network state estimates and fore-cast their short-term evolution. OD matrices have been so far usually estimated from historic and/or real-time data collection and prior matrices. Some of those methods developed by the authors in pre-vious works provide realistic matrices to cope with day-to-day demand variability and real-time traffic conditions. Off-line time-sliced OD matrices estimation based on a simulation-optimization Bilevel-DUE approach has proved to provide appropriate initializations for on-line and real-time dynamic OD estimation methods based on specific versions of Kalman filtering whose input data requirements are traffic counts collected from traffic detection stations and other data supplied by ICT (Information and Communication Technologies) sensors, as for instance travel times between pairs of fixed ICT sensors . In this work, we present a review of the contributions to the on-line/off-line estimation of Dynamic OD matrices and we examine how new data provided by cooperative vehicles as the track-ing along trajectories giving travel times between intermediate points of OD trips can be incorporated into the Kalman filtering equations for on-line dynamic OD estimation. Cooperative vehicles can be considered as mobile sensors, generating data from any point of the network, the computational bur-den of the adapted Kalman approach to cope with tracking data is examined in depth in order to guar-antee the on-line applicability of the proposed approach for a mid-sized urban network.
2016-05-17T15:16:45ZMontero Mercadé, LídiaBarceló Bugeda, JaimeTraffic management applications are supported by dynamic models whose input should be realistic real-time OD demand matrices in order to find efficient network state estimates and fore-cast their short-term evolution. OD matrices have been so far usually estimated from historic and/or real-time data collection and prior matrices. Some of those methods developed by the authors in pre-vious works provide realistic matrices to cope with day-to-day demand variability and real-time traffic conditions. Off-line time-sliced OD matrices estimation based on a simulation-optimization Bilevel-DUE approach has proved to provide appropriate initializations for on-line and real-time dynamic OD estimation methods based on specific versions of Kalman filtering whose input data requirements are traffic counts collected from traffic detection stations and other data supplied by ICT (Information and Communication Technologies) sensors, as for instance travel times between pairs of fixed ICT sensors . In this work, we present a review of the contributions to the on-line/off-line estimation of Dynamic OD matrices and we examine how new data provided by cooperative vehicles as the track-ing along trajectories giving travel times between intermediate points of OD trips can be incorporated into the Kalman filtering equations for on-line dynamic OD estimation. Cooperative vehicles can be considered as mobile sensors, generating data from any point of the network, the computational bur-den of the adapted Kalman approach to cope with tracking data is examined in depth in order to guar-antee the on-line applicability of the proposed approach for a mid-sized urban network.Tècniques de feature weighting per casos no supervisats: Implementació a GESCONDA
http://hdl.handle.net/2117/86169
Tècniques de feature weighting per casos no supervisats: Implementació a GESCONDA
Sánchez Marrè, Miquel; Gómez Villamor, Sergio; Teixidò, F; Gibert Oliveras, Karina
Feature selection and feature weighting methods in supervised domains have been thoroughly discussed in the literature. On the other hand, very little work has been done for unsupervised domains, probably due to the assumed hypothesis that their performance would necessary be substantially worse than the supervised method performance. One method found in the literature, in addition to the new methods proposed are detailed in this paper. The methods have been tested and compared in a data base coming from a Wastewater Treatment plant, with good results. Also, the integration of the new software into the GESCONDA tool is detailed.
2016-04-26T07:39:27ZSánchez Marrè, MiquelGómez Villamor, SergioTeixidò, FGibert Oliveras, KarinaFeature selection and feature weighting methods in supervised domains have been thoroughly discussed in the literature. On the other hand, very little work has been done for unsupervised domains, probably due to the assumed hypothesis that their performance would necessary be substantially worse than the supervised method performance. One method found in the literature, in addition to the new methods proposed are detailed in this paper. The methods have been tested and compared in a data base coming from a Wastewater Treatment plant, with good results. Also, the integration of the new software into the GESCONDA tool is detailed.El bagging en casos no supervisats: Implementació a GESCONDA per algoritmes de clustering
http://hdl.handle.net/2117/86168
El bagging en casos no supervisats: Implementació a GESCONDA per algoritmes de clustering
Gibert Oliveras, Karina; Oliva, Luis; Pinyol, Isaac; Sánchez Marrè, Miquel
Els algorismes de clustering per entorns no supervisats que es basen en una inicialització aleatòria (p. Ex.: tria inicial de llavors en l’algorisme Kmeans), presenten un problema a l’hora d’obtenir solucions fiables.
Una solució per eliminar aquest factor d’aleatorietat seria emprar altres tècniques d’inicialització. Però com es veurà posteriorment en l’article, aquestes tècniques tenen una altre problemàtica, i és la de
trobar solucions òptimes locals o solucions esbiaixades.
La solució que es proposa és la utilització de la tècnica de bagging que s’usa en entorns supervisats, i que a través de la unió de diversos resultats de classificació respecte unes mateixes dades, permet obtenir particions òptimes.
Així mateix, es va implementar tres formes de dur a terme el bagging segons la forma de seleccionar la classificació de referència a partir de la qual s’uneixen la resta de classificacions. Aquestes tres tècniques són: agafant la primera classificació, triant la que presenta una major inèrcia (relació variança entre-classes i intra-classes) i triant la que aporta una major informació (mitjançant el càlcul d’Informació Mútua de Shannon).
Finalment es van provar les tècniques d’inèrcia i informació mútua amb dades ambientals reals preses d’una depuradora d’aigües residuals, per tal de comprovar l’efectivitat dels resultats respecte al mètode tradicional.
Totes les implementacions i proves es van dur a terme sobre el Sistema
Intel·ligent d’Anàlisi de Dades GESCONDA, el qual es descriurà en el pròxim apartat.
L’estudi finalitza amb una breu discussió dels resultats obtinguts i unes conclusions sobre el treball realitzat.
2016-04-26T07:20:17ZGibert Oliveras, KarinaOliva, LuisPinyol, IsaacSánchez Marrè, MiquelEls algorismes de clustering per entorns no supervisats que es basen en una inicialització aleatòria (p. Ex.: tria inicial de llavors en l’algorisme Kmeans), presenten un problema a l’hora d’obtenir solucions fiables.
Una solució per eliminar aquest factor d’aleatorietat seria emprar altres tècniques d’inicialització. Però com es veurà posteriorment en l’article, aquestes tècniques tenen una altre problemàtica, i és la de
trobar solucions òptimes locals o solucions esbiaixades.
La solució que es proposa és la utilització de la tècnica de bagging que s’usa en entorns supervisats, i que a través de la unió de diversos resultats de classificació respecte unes mateixes dades, permet obtenir particions òptimes.
Així mateix, es va implementar tres formes de dur a terme el bagging segons la forma de seleccionar la classificació de referència a partir de la qual s’uneixen la resta de classificacions. Aquestes tres tècniques són: agafant la primera classificació, triant la que presenta una major inèrcia (relació variança entre-classes i intra-classes) i triant la que aporta una major informació (mitjançant el càlcul d’Informació Mútua de Shannon).
Finalment es van provar les tècniques d’inèrcia i informació mútua amb dades ambientals reals preses d’una depuradora d’aigües residuals, per tal de comprovar l’efectivitat dels resultats respecte al mètode tradicional.
Totes les implementacions i proves es van dur a terme sobre el Sistema
Intel·ligent d’Anàlisi de Dades GESCONDA, el qual es descriurà en el pròxim apartat.
L’estudi finalitza amb una breu discussió dels resultats obtinguts i unes conclusions sobre el treball realitzat.On geometrical properties of preconditioners in IPMs for classes of block-angular problems
http://hdl.handle.net/2117/84257
On geometrical properties of preconditioners in IPMs for classes of block-angular problems
Castro Pérez, Jordi; Nasini, Stefano
One of the most efficient interior-point methods for some classes of block-angular structured problems
solves the normal equations by a combination of Cholesky factorizations and preconditioned conjugate gradient for,
respectively, the block and linking constraints. In this work we show that the choice of a good preconditioner depends on geometrical properties of the constraints structure. In particular, it is seen that the principal angles between the subspaces generated by the diagonal blocks and the linking constraints can be used to estimate ex-ante the efficiency of the preconditioner. Numerical validation is provided with some generated optimization problems. An application to the solution of multicommodity network flow problems with nodal capacities and equal flows of up to 127 million of
variables and up to 7.5 million of constraints is also presented
J. Castro, S. Nasini, On geometrical properties of preconditioners in IPMs for classes of block-angular problems, Research Report DR 2016/03, Dept. of Statistics and Operations Research, Universitat Politècnica de Catalunya, 2016.
2016-03-11T18:48:48ZCastro Pérez, JordiNasini, StefanoOne of the most efficient interior-point methods for some classes of block-angular structured problems
solves the normal equations by a combination of Cholesky factorizations and preconditioned conjugate gradient for,
respectively, the block and linking constraints. In this work we show that the choice of a good preconditioner depends on geometrical properties of the constraints structure. In particular, it is seen that the principal angles between the subspaces generated by the diagonal blocks and the linking constraints can be used to estimate ex-ante the efficiency of the preconditioner. Numerical validation is provided with some generated optimization problems. An application to the solution of multicommodity network flow problems with nodal capacities and equal flows of up to 127 million of
variables and up to 7.5 million of constraints is also presentedModelización y análisis de la red de autobuses de la ciudad de Vigo
http://hdl.handle.net/2117/82875
Modelización y análisis de la red de autobuses de la ciudad de Vigo
Montero Mercadé, Lídia
Informe tècnic EIO desenvolupament conveni C253330) Estudio Plan Director de Tráfico y Transporte de la ciudad de Vigo. Proyecto vinculado al Dept. de Ingeniería Mecánica de ETSEIB de la UPC (Responsable Juli García Ramón)
2016-02-11T17:36:36ZMontero Mercadé, LídiaInforme tècnic EIO desenvolupament conveni C253330) Estudio Plan Director de Tráfico y Transporte de la ciudad de Vigo. Proyecto vinculado al Dept. de Ingeniería Mecánica de ETSEIB de la UPC (Responsable Juli García Ramón)A cutting-plane approach for large-scale capacitated multi-period facility location using a specialized interior-point method
http://hdl.handle.net/2117/80887
A cutting-plane approach for large-scale capacitated multi-period facility location using a specialized interior-point method
Castro Pérez, Jordi; Nasini, Stefano; Saldanha da Gama, Francisco
We propose a cutting-plane approach (namely, Benders decomposition) for a class of capacitated multi-period facility location problems. The novelty of this approach lies on the use of a specialized interior-point method for solving the Benders subproblems. The primal block-angular structure of the resulting linear optimization problems is exploited by the interior-point method, allowing the (either exact or inexact) efficient solution of large instances. The effect of different modeling conditions and problem specifications on the computational performance are also investigated both theoretically and empirically, providing a deeper understanding of the significant factors influencing the overall efficiency of the cutting-plane method. This approach
allowed the solution of instances of up to 200 potential locations, one million customers and three periods, resulting in mixed integer linear optimization problems of up to 600 binary and 600 millions of continuous variables. Those problems were solved by the specialized approach in less than one hour, outperforming other stateof-
the-art methods, which exhausted the (144 Gigabytes of) available memory in the largest instances.
2015-12-17T18:28:33ZCastro Pérez, JordiNasini, StefanoSaldanha da Gama, FranciscoWe propose a cutting-plane approach (namely, Benders decomposition) for a class of capacitated multi-period facility location problems. The novelty of this approach lies on the use of a specialized interior-point method for solving the Benders subproblems. The primal block-angular structure of the resulting linear optimization problems is exploited by the interior-point method, allowing the (either exact or inexact) efficient solution of large instances. The effect of different modeling conditions and problem specifications on the computational performance are also investigated both theoretically and empirically, providing a deeper understanding of the significant factors influencing the overall efficiency of the cutting-plane method. This approach
allowed the solution of instances of up to 200 potential locations, one million customers and three periods, resulting in mixed integer linear optimization problems of up to 600 binary and 600 millions of continuous variables. Those problems were solved by the specialized approach in less than one hour, outperforming other stateof-
the-art methods, which exhausted the (144 Gigabytes of) available memory in the largest instances.Computational framework for the estimation of dynamic OD trip matrices
http://hdl.handle.net/2117/80834
Computational framework for the estimation of dynamic OD trip matrices
Barceló Bugeda, Jaime; Montero Mercadé, Lídia
Origin-Destination (OD) trip matrices describe traffic behavior patterns across the network and play a key role as primary data input to many traffic models. OD matrices are a critical requirement, in traffic assignment models, static or dynamic. However, OD matrices are not yet directly observable; thus, the current practice consists of adjusting an initial a priori matrix from link flow counts, speeds, travel times and other aggregate demand data, supplied by a layout of traffic counting stations. The availability of new traffic measurements from ICT applications offers the possibility to formulate and develop more efficient algorithms, especially suited for real-time applications; whose efficiency depends, among other factors, on the quality of the seed matrix. This paper proposes an integrated computational framework in which an off-line procedure generates the time-sliced OD matrices, which are the input to an on-line estimator, whose sensitivity with respect to the available traffic measurements is analyzed.
Origin-Destination (OD) trip matrices describe traffic behavior patterns across the network and play a key role as primary data input to many traffic models. OD matrices are a critical requirement, in traffic assignment models, static or dynamic. However, OD matrices are not yet directly observable; thus, the current practice consists of adjusting an initial a priori matrix from link flow counts, speeds, travel times and other aggregate demand data, supplied by a layout of traffic counting stations. The availability of new traffic measurements from ICT applications offers the possibility to formulate and develop more efficient algorithms, especially suited for real-time applications. This work proposes an integrated computational framework in which an off-line procedure generates the time-sliced OD matrices, which are the input to an on-line estimator, whose sensitivity with respect to the available traffic measurements is analyzed.
2015-12-16T18:15:58ZBarceló Bugeda, JaimeMontero Mercadé, LídiaOrigin-Destination (OD) trip matrices describe traffic behavior patterns across the network and play a key role as primary data input to many traffic models. OD matrices are a critical requirement, in traffic assignment models, static or dynamic. However, OD matrices are not yet directly observable; thus, the current practice consists of adjusting an initial a priori matrix from link flow counts, speeds, travel times and other aggregate demand data, supplied by a layout of traffic counting stations. The availability of new traffic measurements from ICT applications offers the possibility to formulate and develop more efficient algorithms, especially suited for real-time applications; whose efficiency depends, among other factors, on the quality of the seed matrix. This paper proposes an integrated computational framework in which an off-line procedure generates the time-sliced OD matrices, which are the input to an on-line estimator, whose sensitivity with respect to the available traffic measurements is analyzed.