1,720,972 research outputs found
Short-term speed predictions exploiting big data on large urban road networks
Full text linkBig data from floating cars supply a frequent, ubiquitous sampling of traffic conditions on the road network and provide great opportunities for enhanced short-term traffic predictions based on real-time information on the whole network. Two network-based machine learning models, a Bayesian network and a neural network, are formulated with a double star framework that reflects time and space correlation among traffic variables and because of its modular structure is suitable for an automatic implementation on large road networks. Among different mono-dimensional time-series models, a seasonal autoregressive moving average model (SARMA) is selected for comparison. The time-series model is also used in a hybrid modeling framework to provide the Bayesian network with an a priori estimation of the predicted speed, which is then corrected exploiting the information collected on other links. A large floating car data set on a sub-area of the road network of Rome is used for validation. To account for the variable accuracy of the speed estimated from floating car data, a new error indicator is introduced that relates accuracy of prediction to accuracy of measure. Validation results highlighted that the spatial architecture of the Bayesian network is advantageous in standard conditions, where a priori knowledge is more significant, while mono-dimensional time series revealed to be more valuable in the few cases of non-recurrent congestion conditions observed in the data set. The results obtained suggested introducing a supervisor framework that selects the most suitable prediction depending on the detected traffic regimes
Modeling Car following with Feed-Forward and Long-Short Term Memory Neural Networks
Full text linkThe paper investigates the capability of modeling the car following behavior by training shallow and deep recurrent neural networks to reproduce observed driving profiles, collected in several experiments with pairs of GPS-equipped vehicles running in typical urban traffic conditions. The input variables are relative speed, spacing, and vehicle speed. In the model, we assume that the reaction is not instantaneous. However, it may occur during a time interval of the order of a few tenth of seconds because of both the psychophysical driver’s reaction process and the mechanical activation of braking or dispensing the traction power to the wheels. Experimental results confirm the reliability of this assumption and highlight that the deep recurrent neural network outperforms the simpler feed-forward neural network
'Big Data' opportunities for new transport models and services
No full textLa diffusione di dispositivi mobili di localizzazione e comunicazione produce con continuità un'immensa quantità di dati – i cosiddetti Big Data– che possono essere sfruttati per migliorare la conoscenza dello stato del sistema di trasporto e per mettere in atto opportune politiche di mobilità e appropriate azioni di regolazione. Peraltro, gli attuali sistemi di informazione presentano tuttora notevoli lacune e sono quindi suscettibili di significativi miglioramenti. La presente memoria esamina i principali problemi connessi all’uso dei Big Data e tratta dell’opportunità offerta da questi di realizzare modelli di mobilità individuale basati su un aggiornamento continuo dell’informazione. La prima parte, di carattere discorsivo, è completata da una sintetica rassegna delle ricerche eseguite dagli autori e finalizzate ad analizzare i Big Data sotto diversi aspetti: per ottenere una caratterizzazione quantitativa dei fenomeni di mobilità, per comprendere i comportamenti e le preferenze degli utenti, per migliorare i modelli di previsione e studiare nuovi servizi di mobilità.The vast diffusion of mobile location and communication devices continuously produce a huge amount of data ('Big Data') that can be exploited to improve the knowledge about the state of the transport system and perform appropriate regulatory actions and execute suitable policy actions. However, current information systems still have noteworthy drawbacks and are in the meantime open to significant improvements. The paper discusses the main problems related to the use of the Big Data and examines the opportunities offered by Big Data to introduce new transport models based on continuous updating of microscopic models. The first part of the paper, of conversational nature, is complemented by a concise overview of the research conducted by the authors with multiple objectives: derive a quantitative characterization of the mobility phenomena, understand users' behaviour and preferences, develop enhanced prediction models, and study new mobility services
Optimization and Simulation Approach of Containers handling operations at Intermodal Terminals
No full textThe paper deals with the problem of minimizing reshuffling of containers in an inland intermodal terminal. The problem is tackled according to a simulation-optimization approach. A simulation model computes the operational costs of containers, related to storage and pick-up operations in an inland yard. The optimization is carried out by a double genetic algorithm that applies two genetic algorithms in series. The first optimizes the locations of the containers to store in the yard and identifies the blocking containers that have to be reshuffled. The second genetic algorithm takes the solution of the first and optimizes the reshuffling of blocking containers together with the unloaded ones. The proposed optimization method has been tested on a theoretical example of a realistic size. Results highlighted that the double genetic algorithm reduces the total operational costs by 5% with respect to the single genetic algorithm
Comparative analysis of implicit models for real-time short-term traffic predictions
Full text linkPredicting future traffic conditions in real-time is a crucial issue for applications of intelligent transportation systems devoted to traffic management and traveller information. The increasing number of connected vehicles equipped with locating technologies provides a ubiquitous updated source of information on the whole network. This offers great opportunities for developing data-driven models that extrapolate short-term future trend directly from data without modelling traffic phenomenon explicitly. Among several different approaches to implicit modelling, machinelearning models based on a network structure are expected to be more suitable to catch traffic phenomenon because of their capability to account for spatial correlations existing between traffic measures taken on different elements of the road network. The study analyses and applies different implicit models for short-term prediction on a large road network: namely, time-dependent artificial neural networks and Bayesian networks. These models are validated and compared by exploiting a large database of link speeds recorded on the metropolitan area of Rome during seven months. © The Institution of Engineering and Technology 2016
Hybrid Metaheuristic Approach to Solve the Problem of Containers Reshuffling in an Inland Terminal
Full text linkThe paper deals with the problem of minimizing the reshuffling of containers in an inland intermodal terminal. The problem is tackled according to a hybrid approach that combines a preliminary selection of heuristics and a genetic algorithm. The heuristics are used to determine the initial population for the genetic algorithm, which aims to optimize the locations of the containers to store in the yard in order to minimize the operational costs. A simulation model computes the costs related to storage and pick-up operations in the yard bay. The proposed optimization method has been calibrated by selecting the optimal parameters of the genetic algorithm in a toy case and has been tested on a theoretical example of realistic size. Results highlighted that the use of a suitable heuristic to generate the initial population outperforms the genetic algorithm, initialized with a random solution, by 20%
Investigation and modeling on drivers’ route and departure time choices from a big data set of floating car data
No full textIn this paper, a general analysis methodology aimed at processing a large set of Floating Car Data (FCD) reconstructing the routes followed by the drivers and then clustering them to achieve suitable choice sets- is applied to a broad set of FCD collected in the metropolitan city of Rome over six months. Through the observation of about 10,000 trips, an analysis of Wardrop's principle is carried out focused on the morning peak period: the results show that about 75% of the routes chosen by the users have travel times that exceed the minimum value by less than 35%, a value having the same magnitude of the average coefficient of variation of the observed link travel times, that is 24%. The possibility of modeling drivers' route choice behavior among a set of similar routes is investigated, and different utility functional forms are defined and calibrated. The values of rho(2) obtained are low, as expected considering that the drivers mostly perceive the routes that were actually chosen as equivalent alternatives. Nevertheless, the coefficients' values are statistically significant: results confirmed that length, travel time, and traffic lights represent three attributes that affect the path choice mechanism with a probability of 95%. Finally, the users' process to improve their choice is also investigated, and the day-to-day route and departure time choice processes are analyzed to verify the possible existence of a correlation between observed changes and possible delays experienced by the users in the days before the change: for travel time increases or reductions between 5 and 20 minutes, a correlation has been identified with the number of route changes
Short-term traffic predictions on large urban traffic networks: applications of network-based machine learning models and dynamic traffic assignment models
Full text linkThe paper discusses the issues to face in applications of short-term traffic predictions on urban road networks and the opportunities provided by explicit and implicit models. Different specifications of Bayesian Networks and Artificial Neural Networks are applied for prediction of road link speed and are tested on a large floating car data set. Moreover, two traffic assignment models of different complexity are applied
on a sub-area of the road network of Rome and validated on the same floating car data set
Statistical and Clustering-Based Assessment of Variable Speed Limits Effects on Motorway Performance from Real-World Observations
Full text linkVariable Speed Limit (VSL) systems aimed at reducing congestion and improving safety performance have been implemented around the world in previous years. However, field studies have shown controversial results regarding traffic performance improvement. This study integrates statistical testing methods and clustering techniques for assessing the effect of a non-mandatory VSL system on traffic flow performances on a 14-km portion of the Padua–Mestre motorway in Italy. Statistical analysis is conducted on the observed speeds, collected for almost one year, to identify any significant differences provided by VSL activation. The changes in global motorway performances induced by the VSL in typical traffic patterns under recurring congestion are assessed using both statistical tests and two specific clustering algorithms, namely K-means and DBSCAN. The results indicate that the VSL system effectively affects the observed speeds and alleviates congested conditions: the observed reduction in mean travel time ranges is around 4% with the VSL system active across various lanes; the standard deviation of vehicular speeds witnessed a decrease of 12% to 20% in the most congested segments, while no notable distinction is observed in traffic flows
- …
