In 1999, the TRG has started to create a database of the books, journals and proceedings in its possession.. This work, made by Myriam Matthys, has succeded to take into account several thousand references. These are available as BibTex entries.
If you wish to consult directly the TRG library, please contactPierre Yves Bernard
The script's name is query. Enter query, followed by
the search terms. The number of these words must be less or equal to
5. query shows the results on the standard output. To get a
result file, say results, you have only to enter
query arguments > results
TRG developed ATES, Another Traffic Equilibrium Software, a traffic assignment and demand matrix estimation software based on the studied assessments.
ATES is a static model (in opposition to the dynamic models such as PACSIM) as it provides a averaged image of the vehicles flows for the time considered during the modelisation.
ATES mainly proposes two applications : the assignment of a demand matrix on a network and the update of a travelling matrix based on an a priori demand and on the studied traffic counts on routes modelised in the network.
The assignment phase allocates the travelling demand on the network, determining then routes allowing to connect the various origins to destinations, also precising the number of users on the various nodes of the network, the travelling time, etc.
To solve the assignment traffic problem a rule must be given according to which the users will choose an itinerary. Many rules are existing, determining thus various assignment methods (see hereafter). Each of these rules calls out the notion of arc cost.
Actually the use of a transportation network for its users is always done at a certain cost. This one can be estimated by means of the covered distance, of the travelling time or other criteria determined according to the final purpose of the modelisation. ATES modelises this notion of global cost linked to the travelling by the choice of a performance function which associates to each arcs of the network a travelling time according to the flow he supports (i.e. the number of vehicles). The travelling time of an arc is so relatively independent of traffic when this one is weak, but a more important traffic can slow down the user up to a complete blocking, because the capacity of the arc is reached.
ATES offers the user the possibility to choose among 5 assignment methods. These are the all or nothing (shortest paths), user equilibrium, system equilibrium, stochastic userequilibrium and probit assignment methods. We will not develop these methods here. We can just say that we use an iterative process if an equilibrium is wished.
The method used by ATES to update an initial demand according to the studied traffic counts is a growing factors method. The choice of the a priori matrix is not random since it is to the structure of this one that the growing factors which will correct this matrix will be applied in order to adjust the traffic counts.
We remark that a zero initial demand will always remain zero in a growing factors model, since the method uses multiplicative factors. To avoid this rigidity ATES offers the possibility to complete the starting matrix so that an initially zero demand between an origin i and a destination j can develops to a strictly positive demand in the updated matrix according to the studied traffic counts. We also note that the user keeps the possibility to maintain a zero demand for certain origin/destination pairs.
The final matrix is obtained by successive iterations : there is first an assignment phase of the initial matrix on the network, then growing factors are applied on the matrix so that the assigned flows on the arcs equipped with traffic counters correspond to the traffic counts which are made on it. The obtained matrix is then assigned on the network and the flows are again estimated and compared to the traffic counts. The iterations followed then one other until there is convergence or until the maximum number of iterations determined by the user is reached.
We remark that for the assignment phase of each iteration the user has the choice between the 5 methods proposed by ATES, hereabove mentioned.
A new behavioral traffic assignment modelWhen we desire to study the effects of exceptional events (e.g. accident) or the impact of informations systems on traffic, classicals models like equilibrium models are not sufficient. These models gives an average situation of vehicles flow.
To compensate these inconvenients, we must have recourse to a dynamic model. PACSIM is one of these. Developed by the TRG, partially in the EEC "DRIVE 1" program and partially in the SSTC "Transport et Mobilité" program, this model simulate urban traffic. A special attention is take to modelize crossroads, where differents traffic flows meet, and are then sources of conflicts and traffic jams.
By its dynamic nature, Pacsim allow to study temporal traffic flows evolutions on differents modelized axes in peak hours. This dynamical disposition is done by an "events driven" simulation. It signifies that events (accident, traffic informations, vehicles movements...) occurs and have for consequences that action are made in the model to simulate the traffic evolution according to these events effects (congestion, barred road, ...).
PACSIM also include a behavioral component which allow to simulate how the users choice transport, map out a route or even how they react to traffic informations. Note that PACSIM do not refer to the hypothesis of exact knowledge of the network. Actually, we introduce in PACSIM the perceived network concept: users knows network according to zones that they usually frequent.
In urban network, a non negligible part of traffic is made by public transport. Thus, PACSIM also incorporate a traffic multimodal aspect. Buses are, like cars, modelized and then interdependant events (traffic jam stops buses, a bus which leave its stop slacks traffic,...) occurs.
Finally, the informations systems also take place in the model. It allow us to measure the effect of sending an information in the network by the way, for example, of a variable message signal or RDS information.
Now, PACSIM is still developed to include a pollution model and a dynamic lights phase management.
The software HieLoW was developed by Michel Bierlaire and is sold by Stratec. It is a software allowing to calibrate hierarchical logit models and it runs under Windows.
The demonstration version of the software is available at the adress ftp://thales.math.fundp.ac.be/pub/hielow/demo/english/. Download the 12 files and launch install.exe. Once HieLoW is installed, execute the program hielow.exe. We recommand that you read the installation guide the and help system distributed with HieLoW.
The MEUSE method has been developed by Michel Bierlaire. It is a O/D (Origin/Destination) matrix estimation that can explicity exploit detailed informations on the structure of some columns.
For more details: MEUSE: An origin-destination matrix that exploits structure .
For more information, go there
|Ludovic Platbrood, Fabian Bastin - November 16th 2000 - FUNDP - Sciences - Math|