• Không có kết quả nào được tìm thấy

Matrix-Based pt router

N/A
N/A
Nguyễn Gia Hào

Academic year: 2023

Chia sẻ "Matrix-Based pt router"

Copied!
242
0
0

Loading.... (view fulltext now)

Văn bản

While traffic flow dynamics are simulated by MATSim's mobsim using a queuing model, this flow is not considered in the multimodal contribution. From a transportation planning perspective, the essential element of car sharing—the importance of its availability at precise points in time and space—does not fit traditional models that consider vehicle-per-hour flows.

Figure 21.1 shows the implementation’s basic concept—a multimodal contribution is added to each link object in the mobsim.
Figure 21.1 shows the implementation’s basic concept—a multimodal contribution is added to each link object in the mobsim.

Free-Floating Carsharing

Generalized Cost of Carsharing Travel

Carsharing Membership

Validation

Applications

Dynamic Transport Services

  • DVRP Contribution
  • DVRP Model
    • Schedule
    • Least-Cost Paths
  • DynAgent
    • Main Interfaces and Classes
    • Configuring and Running a Dynamic Simulation
    • RandomDynAgent Example
  • Agents in DVRP
    • Drivers
    • Passengers
  • Optimizer
  • Configuring and Running a DVRP Simulation
  • OneTaxi Example
  • Research with DVRP

The org.matsim.contrib.dvrp.run package contains VrpLauncherUtils and other utility classes that simplify certain steps of the above schema. Theorg.matsim.contrib.dvrp.examples.onetaxipackage contains a simple example of simulating online taxi transportation with the DVRP contribution.

SUBPART FIVE

Commercial Traffic

Freight Traffic

Basic Information

Carriers

WagonSim

Basic Information

Summary

Since Sebastian has left science for now, he allowed us to take his code and integrate it into the repository under the GPL (GNU General Public License). For now it will just "sit" here until someone tries to make it work.

Figure 25.1: WagonSim process chain.
Figure 25.1: WagonSim process chain.

SUBPART SIX

Additional Choice Dimensions

Destination Innovation

  • Basic Information
  • Introduction
  • Key Issues in Developing the Module
    • Error Terms
    • Quenched Randomness
    • Search Space Construction and Evaluation
    • Destination Choice Set Specification
    • Facility Load
  • Application of the Module
  • The Module in the MATSim Context
  • Lessons Learned
  • Further Reading

In a modular environment like MATSim, which is designed to plug in users' own modules externally - possibly drawing their own random numbers - the risk of destroying the damping is too high, so approach (a) is impractical. Some random number generators have problems in the initial phase of the draw, e.g. the first few random numbers are correlated or never cover the entire probability space.

Figure 27.1: Search space: The search algorithm must be able to handle correlated, as well as uncorrelated, error terms as given by the MNL model
Figure 27.1: Search space: The search algorithm must be able to handle correlated, as well as uncorrelated, error terms as given by the MNL model

Joint Decisions

  • Basic Information
  • Joint Decisions and Transport Systems .1 Motivation
    • The Joint Planning Problem
  • A Solution Algorithm for the Joint Planning Problem: A Generalization of the MATSim Process
    • Algorithm
    • Technical Considerations on the Implementation
  • Selected Results
  • Further Reading

The result of the different copies will take into account the impact of the joint plan to which it belongs. Dubernet and Axhausen (forthcoming) present a validation of the model for the domestic issue, using a Z¨urich scenario.

Figure 28.1 presents the repartition of “car passenger” trips by purpose, in the simulation as well as the Swiss National Travel Survey
Figure 28.1 presents the repartition of “car passenger” trips by purpose, in the simulation as well as the Swiss National Travel Survey

Socnetgen

Basic Information

Summary

SUBPART SEVEN

Within-Day Replanning

  • Basic Information .1 Implementation Alternative 1
    • Implementation Alternative 2
  • Introduction
  • Simulation Approaches .1 Iterative Simulation Approaches
    • Within-Day Replanning Approach
    • Combined Approaches
  • Implementation .1 General Thoughts
    • Implementation Alternative 1: Plan-Based Implementation
    • Implementation Alternative 2: Replacing the Agent

In the current implementation, their random number generator is reinitialized for each rescheduled agent, using a deterministic value (eg, a combination of the agent's ID and the current time step). The rest consists of relatively simple bookkeeping methods like getId()—the agent needs to know its own identifier.

Figure 30.1: Exceptional event in a network.
Figure 30.1: Exceptional event in a network.

Making MATSim Agents Smarter with the Belief-Desire-Intention Framework

  • Basic Information
  • Introduction
  • Software Structure
  • Building an Application Using BDI Agents
    • The ClosetoDest Percept
    • The RequestLocation Query
    • The DriveTo BDI-Action
    • Discussion
  • Examples
    • Bushfire Example
    • Taxi Example
    • Discussion

Instructions and examples for the BDI application can be found in the integration repository (listed at the beginning of the chapter). All detection collection functions for the BDI system are defined in the AgentActivityEventHandler class. The BDI system is responsible for allowing only one active BDI action per agent.

Figure 31.1: Conceptual BDI-ABMS integration architecture.
Figure 31.1: Conceptual BDI-ABMS integration architecture.

SUBPART EIGHT

Automatic Calibration

CaDyTS: Calibration of Dynamic Traffic Simulations

  • Basic Information
  • Introduction
  • Adjusting Plans Utility
  • Hooking Cadyts into MATSim Hooking Cadyts into MATSim is based on the following operations
  • Applications

Initialization: When starting the calibration, it requires all available traffic counts and some further parameters. Iterations: The calibration is performed together with the simulation until (calibrated) stationary conditions are reached. Demand simulation: The calibration needs an entry point in the simulation to influence the plan selection.

Figure 32.1: Z¨urich case study results: mean relative error in link volumes.
Figure 32.1: Z¨urich case study results: mean relative error in link volumes.

SUBPART NINE

Visualizers

Senozon Via

  • Basic Information
  • Introduction
  • Simple Usage
  • Use Cases and Examples .1 Agent Visualization
    • Facility Analysis
    • Public Transport Analysis
    • Scenario Comparisons
    • Aggregating Data

A limited version is available for free and can be downloaded from the product website (senozon AG, 2015). To add a layer, the small plus icon at the bottom left of the window can be pressed, or by choosing Add Layer.. from the file menu. Such data can be either point data (such as activity locations, trip start locations, GPS points or any other spatial point data) or origin-destination data (such as trips with a start and end location , or linking an activity location to the home location of the agent performing the activity).

Figure 33.1: Via’s window with default layout and a network query being shown.
Figure 33.1: Via’s window with default layout and a network query being shown.

OTFVis: MATSim’s Open-Source Visualizer

  • Basic Information
  • Introduction
  • Using OTFVis
    • MVI Files
    • Starting OTFVis
    • Use Cases of OTFVis
    • Viewing an MVI File
    • General Interaction with the Main Screen
    • User Interaction in the Live Mobsim
    • Running a Query in OTFVis Real Time Data
  • Extending OTFVis
    • Design Principles of OTFVis
    • Readers and Writers
    • Visualization of the Data
    • Layers
    • Patching the Connections
    • Sending the Data
    • Performance Considerations
    • Sending Live Data

This is a clear superset of the information available in event files and in MVI files. In the following sections we examine how data is calculated within OTFVis and how this can be extended. The OTFLinkAgentsHandler class should be a good example of extracting, sending, receiving and displaying data in an OTFVis context.

Figure 34.1: OTFVis Start Dialog.
Figure 34.1: OTFVis Start Dialog.

SUBPART TEN

Analysis

Accessibility

  • Basic Information
  • Introduction
  • The Measure of Potential Accessibility
  • Accessibility Computation Integrated with Transport Simulation As mentioned above, accessibility computations are often based on travel times (Bundesinstitut
  • Econometric Interpretation
  • Spatial Resolution, Data, and Computational Aspects
  • Conclusion

If the coordinate-based (=grid-based=grid-based=cell-based) version of the MATSim accessibility calculation is chosen, its results can be interpreted as an accessibility field, i.e. as a measure that is continuous in space vary. In the concrete case of the MATSim reachability calculation, exploring the entire network through the lowest cost path tree is a computationally expensive task. First, transport system dynamics are represented through the accessibility calculation integration with the MATSim dynamic traffic simulation.

Figure 35.1: Accessibility of work places in Nelson Mandela Bay Municipality calculated by the grid-based MATSim accessibility computation
Figure 35.1: Accessibility of work places in Nelson Mandela Bay Municipality calculated by the grid-based MATSim accessibility computation

Emission Modeling

  • Basic Information
  • Introduction
  • Integrated Approaches for Modeling Transport and Emissions
  • Emission Calculation
  • Software Structure

At present, these are not considered in the emissions modeling tool because they contribute little to the overall level of emissions. Parking duration refers to the time a vehicle is not moving before cold start emissions are produced. The current implementation uses the MATSim vehicle interface as the vehicle container for storing data relevant to the vehicle type.4 The last two mandatory lookup tables (avgHbefaWarmTableandavgHbefaColdTable) provide the warm and cold emission factors respectively.

Figure 36.1: Software structure of the emission modeling tool.
Figure 36.1: Software structure of the emission modeling tool.

Interactive Analysis and Decision Support with MATSim

  • Basic Information
  • Introduction
  • Requirements of a Decision Support Interface to MATSim
    • Users
    • Functional Requirements
  • General Framework for Decision Support
    • Entity Relationship Diagram (ERD) for General Purpose Analysis
    • Interactive Analysis Using Business Analytics Software
  • Diaries from Events

To this end, we started to compile requirements specifications for potential audiences and their use case scenarios, to come up with a common interactive analysis framework and decision support to meet most requirements. This chapter provides a decision support tool aimed at decision makers and researchers in the fields of transport planning and operations, spatial planning and spatial economics and geography. The decision support framework should facilitate classical transport assessment methods, such as cost/benefit analysis and evaluation of transport infrastructure's spatial impact and policy measures.

Figure 37.1 shows the general framework as we envision it: data from various sources feeds into a spatially-enabled database, with all geodata transformed to use the same spatial reference system (ideally, using the same projection used for MATSim coordina
Figure 37.1 shows the general framework as we envision it: data from various sources feeds into a spatially-enabled database, with all geodata transformed to use the same spatial reference system (ideally, using the same projection used for MATSim coordina

The “Analysis” Contribution

Basic Information

Summary

SUBPART ELEVEN

Computational Performance Improvements

Multi-Modeling in MATSim: PSim

  • Basic Information
  • Introduction
  • Basic Idea
  • Performance
    • Distributed Computing

Initial tests on the Z¨urich scenario (described in Chapter 56) have shown a dramatic decrease in computation time, compared to the default QSim-only approach; performance improves linearly with an increasing number of compute cores. Rescheduling is fully multi-threaded, with no inter-core synchronization required, so its performance increases linearly as the number of cores increases; times improve more dramatically with the new approach than with the standard one. Slave nodes perform rescheduling operations and evaluate plans in a predetermined number of PSim iterations per cycle.

Figure 39.1: Operation of a MATSim run implementing pseudo-simulation.
Figure 39.1: Operation of a MATSim run implementing pseudo-simulation.

Other Experiences with Computational Performance Improvements

A parallel version of JDEQSim (Waraich et al., 2015) never made it into the main MATSim repository. PSim (Chapter 39) approaches the problem from a different angle: Instead of speeding up QSim itself, it tries to take advantage of the fact that (1) adding or removing a small number of synthetic passengers does not change congestion patterns very much, and thus alternative plans can be evaluated in parallel, and (2) the congestion patterns generated by mobsim do not vary that much from one iteration to the next, so mobsim does not need to be restarted every time a few synthetic commuters switch to different alternatives. In summary: (1) the behavioral parts of MATSim are easily parallelized; the main challenge is mobsim.

Table 40.1: Computing times of the mobsim for the Gauteng scenario (see Chapter 69) with 523 000 links for different computers, different population sizes, and different numbers of threads
Table 40.1: Computing times of the mobsim for the Gauteng scenario (see Chapter 69) with 523 000 links for different computers, different population sizes, and different numbers of threads

SUBPART TWELVE

Other Modules

Evacuation Planning: An Integrated Approach

  • Basic Information
  • Related Work
  • Download MATSim and Evacuation
  • The Fifteen-Minute Tour
  • Input Data (any Place and any Size)
  • Scenario Manager
    • Scenario Configuration
    • Evacuation Area
    • Evacuation Demand
    • Road Closures
    • Bus Stop Editor
    • Running the Scenario
    • Analysis
  • Conclusion

The path to the network file covering the evacuation area: OSM XML files (*.osm) are currently supported. The path to an ESRI shapefile describing the extent of the evacuation area, depicted by a simple polygon. The ScenarioManager integrates modules for the definition of the evacuation area and the distribution of the affected population.

Figure 41.1: Illustration of a configuration opened in the ScenarioManager and as XML file.
Figure 41.1: Illustration of a configuration opened in the ScenarioManager and as XML file.

MATSim4UrbanSim

Basic Information

Discontinued Modules

DEQSim

Planomat

All the best answer modules here face the challenge of not being able to run full mobsim (=network loading=synthetic reality) every time they need such information. As a result, all best response modules are forced to build an internal synthetic reality model. After that, maintaining Planomat's best response capability would have been a permanent labor-intensive commitment.

PlanomatX

Any strategy module that generates the best response plans must be able to compare plans and choose a better one, at least according to the considered choice dimensions. Planomat always tended to return the same solution: understandable from a best-answer module, but it becomes a problem when what the module thinks is a best answer starts to differ from what the MATSim core thinks. However, it should be noted that the improved software architecture does not solve the general conceptual problem; best answer modules must somehow follow core system development.

SUBPART THIRTEEN

Development Process & Own Modules

Organization: Development Process, Code Structure and Contributing to MATSim

  • MATSim’s Team, Core Developers Group, and Community The MATSim team currently consists of three research groups and a spin-off company
  • Roles in the MATSim Community The MATSim community includes the following roles
  • Code Base
    • Main Distribution
    • Core
    • Contributions
    • Playgrounds
    • Contributions and Extensions
    • Releases, Nightly Builds and Code HEAD
  • Drivers, Organization and Tools of Development
  • Documentation, Dissemination and Support
  • Your Contribution to MATSim

A small group of the MATSim team defines the group of core MATSim developers, maintaining the core of MATSim as defined below in section 44.3.2. Support is provided by the MATSim team via these mailing lists and via http://matsim.org/faq, both on a best-effort basis. Information on such publications can be obtained, for example, from http://matsim.org/publications and from this book.

Figure 44.1: MATSim events and community.
Figure 44.1: MATSim events and community.

How to Write Your Own Extensions and Possibly Contribute Them to MATSim

Introduction

The scoring uses events to track each agent's success in executing its plan and calculates the scoring value based on this. As a result, it is possible to add additional executable code to the execution flow of QSim byMobsimListeners in the same way as is possible with the ControllerListeners mentioned above. The router, on the other hand, is most importantly configured by replacing the definition of the generalized travel cost.

Extension Points

  • Config Group
  • ObjectAttributes and Customizable
  • Scenario Element
  • ControlerListener: Handling Controler Events
  • Events
  • Mobsim Listener
  • TripRouter
  • Mobsim
  • PlanStrategy
  • Scoring

Please check the ObjectAttributes documentation (seehttp://matsim.org/javadoc→main distribution) for more details and example instructions. Please check the PlanStrategy documentation (see http://matsim.org/javadoc → main distribution) for more details and example instructions. Please check the eScoringFunction documentation (seehttp://matsim.org/javadoc→main distribution) for more details and example instructions.

Understanding MATSim

Some History of MATSim

Scientific Sources of MATSim

In this sense, agent-based travel demand modeling had been developed in Germany since the 1970s (see references in Axhausen and Herz, 1989), as well as in English-speaking countries, as described in Jones et al. Complex adaptive systems/co- evolutionary algorithms Nash equilibrium-like approaches have been developed in transport assignments since the formative Wardrop (1952) paper. These approaches translated Nash equilibrium logic into coevolutionary search schemes, which efficiently identified the optima of each agent's daily schedule.

Stages of Development .1 Kai Nagel’s Perspective

  • Kay W. Axhausen’s Perspective

In the MATSim context, the competition for objects was taken from Horni et al. Departure time, mode, and route choice are the heart of the transportation modeling enterprise and have been addressed in MATSim almost from the beginning (Raney and Nagel, 2004; Balmer et al., 2005b; Rieser et al., 2009). With these approaches, it is impossible to directly model destination choice, as the best response destination would be the closest possible destination (Horni et al., 2009).

Agent-Based Traffic Assignment

  • Introduction
  • From Route Swapping to Agent Plan Choice
    • Static Traffic Assignment
    • Dynamic Traffic Assignment
    • Individual Travelers
    • Stochastic Network Loading
    • Extending the Route Assignment Loop to Other Choice Dimensions
  • Agent-Based Simulation
    • Agent-Based UE; One Plan per Traveler
    • Agent-Based SUE; Multiple Plans per Traveler
  • Conclusion

Defining the network load as more on the “physical” side of the system, the behaviorally relevant steps are the generation of choice sets and choices (Bowman and Ben-Akiva, 1998). The simulation of an agent-based UE is possible through the following implementation of the behavioral elements. Repetitions: Repeat the following many times. a) Choice/choice: Select one of the plans for each agent.

MATSim as a Monte-Carlo Engine

  • Introduction
  • Relaxation as a Stochastic Process .1 Probabilistic Model Components
    • Markov Chain Perspective
  • Existence and Uniqueness of MATSim Solutions
  • Analyzing Simulation Outputs

This means that Algorithm 48.1 can be a valid solution method for the model system shown in Figure 48.1. Intuitively, the removal of the stage index means that equation (48.5) is now valid for any stage in the long run. To obtain the Markov chain representation of Algorithm 48.1, it is necessary to determine (i) which variables in MATSim represent the states of this chain and (ii) which distribution of transitions is the basis for the MATSim simulation logic.

Figure 48.2: Example of (a)periodicity
Figure 48.2: Example of (a)periodicity

Hình ảnh

Figure 21.1 shows the implementation’s basic concept—a multimodal contribution is added to each link object in the mobsim.
Figure 27.1: Search space: The search algorithm must be able to handle correlated, as well as uncorrelated, error terms as given by the MNL model
Figure 27.2: Error term runs for the Z¨urich scenario.
Figure 27.3: Daily traffic volumes for 123 links compared to traffic counts. Per link k the relative error is used, i.e., (vol simulated,k − vol counted,k )/vol counted,k .
+7

Tài liệu tham khảo

Tài liệu liên quan

Căn cứ các báo cáo khảo sát địa chất đã được thu thập và lưu trữ trong quá trình nghiên cứu, có thể lựa chọn một số mặt cắt đại diện cho các đơn hành chính của thành phố