Large cities around the world increasingly face problems like traffic congestion and pollution in commutes, which impacts the population’s quality of life. Researches have been developed to try to minimize these effects, challenging the rising car ownership. A portion of them addresses the improvement of public transport planning by using computational techniques such as machine learning and genetic algorithms. This research deals with the subject by proposing a methodology for the simulation and optimization of vehicles stop times of public transport networks, aiming to reduce users’ travel times. The methodology was applied to São Paulo’s night bus network. First, a multi-agent model of the network was built with the aid of the MATSim (Multi-Agent Transport Simulation) software. It simulates the schedules of the vehicles and the commuting population. For that, procedures were developed to obtain, treat and convert input data such as travel demand, road network and GTFS files. Afterwards, the simulation was run to obtain vehicles times at stops and routes taken by the users. Output files were created, including the one containing all the actions in the simulation (events.xml). Finally, the reduction in travel times was sought by optimizing the vehicles times. The genetic algorithm heuristic was applied to perform the public transport network optimization, adding delay values to each vehicle time. Also, routines were developed to convert the simulation events into adequate data structures for the algorithm. The optimization resulted in a reduction of the total travel time of the commuting population, mainly by reducing transfer times.
Le grandi città del mondo affrontano in modo crescente problemi come la congestione del traffico e l'inquinamento nei spostamenti, che influisce sulla qualità della vita della popolazione. Ricerche sono state realizzate per cercare di minimizzare questi effetti, sfidando il crescente uso dell'automobile. Una parte si occupa del miglioramento della pianificazione del trasporto pubblico utilizzando tecniche computazionali come l'apprendimento automatico e algoritmi genetici. Questa ricerca affronta l'argomento proponendo una metodologia per la simulazione e l'ottimizzazione degli orari alle fermate dei veicoli di una rete di trasporto pubblico, con l'obiettivo di ridurre i tempi di viaggio della popolazione. La metodologia è stata applicata alla rete di autobus notturna di São Paulo. Innanzitutto, è stato creato un modello multiagente della rete con l'uso del software MATSim (Multi-Agent Transport Simulation). Il software simula gli orari dei veicoli e gli spostamenti della popolazione. Per questo, sono state sviluppate procedure per ottenere, trattare e convertire dati di input come domanda di trasporto, rete viaria e file GTFS. Successivamente, la simulazione è stata eseguita per ottenere gli orari dei veicoli alle fermate e i percorsi effettuati da persone. Sono stati creati file di output, incluso quello che contiene tutte le azioni della simulazione (events.xml). Finalmente, la riduzione dei tempi di viaggio è stata ricercata ottimizzando gli orari dei veicoli. Un algoritmo genetico è stato usato per eseguire l'ottimizzazione della rete di trasporto pubblico a partire dalla inclusione di ritardi sugli orari dei veicoli. Inoltre, metodi sono stati sviluppati per convertire gli eventi della simulazione in strutture di dati adeguate per l'algoritmo. L'ottimizzazione ha portato alla riduzione del tempo di viaggio totale della popolazione, principalmente riducendo i tempi di cambiamento.
A methodology to simulate and optimize vehicles stop times of public transport networks
BRUNI, LUIZ RAPHAEL
2018/2019
Abstract
Large cities around the world increasingly face problems like traffic congestion and pollution in commutes, which impacts the population’s quality of life. Researches have been developed to try to minimize these effects, challenging the rising car ownership. A portion of them addresses the improvement of public transport planning by using computational techniques such as machine learning and genetic algorithms. This research deals with the subject by proposing a methodology for the simulation and optimization of vehicles stop times of public transport networks, aiming to reduce users’ travel times. The methodology was applied to São Paulo’s night bus network. First, a multi-agent model of the network was built with the aid of the MATSim (Multi-Agent Transport Simulation) software. It simulates the schedules of the vehicles and the commuting population. For that, procedures were developed to obtain, treat and convert input data such as travel demand, road network and GTFS files. Afterwards, the simulation was run to obtain vehicles times at stops and routes taken by the users. Output files were created, including the one containing all the actions in the simulation (events.xml). Finally, the reduction in travel times was sought by optimizing the vehicles times. The genetic algorithm heuristic was applied to perform the public transport network optimization, adding delay values to each vehicle time. Also, routines were developed to convert the simulation events into adequate data structures for the algorithm. The optimization resulted in a reduction of the total travel time of the commuting population, mainly by reducing transfer times.File | Dimensione | Formato | |
---|---|---|---|
2019_07_Bruni.pdf
non accessibile
Descrizione: Thesis text
Dimensione
3.63 MB
Formato
Adobe PDF
|
3.63 MB | Adobe PDF | Visualizza/Apri |
I documenti in POLITesi sono protetti da copyright e tutti i diritti sono riservati, salvo diversa indicazione.
https://hdl.handle.net/10589/149382