HIGHWAY SPACE INVENTORY CONTROL SYSTEM

摘要

In this paper, a new demand management concept - Highway Space Inventory Control System (HSICS) is proposed and modelled. The basic idea of the HSICS is that all road users have to make advance reservations to travel on the highway. Such a reservation system is flexible and could be applied during peak hours throughout the highway or only on highly congested sections of highways, or through out the day on highways depending on the requirements and considerations of road users, policy makers, and other relevant stakeholders. The proposed HSICS model consists of two modules - Highway Allocation System (HAS) and the Highway Reservation System (HRS). HAS is an offline module that allocates the highway sections to various vehicle types (single occupant cars, carpools, public transit, trucks) during different time periods with the goal of optimizing the objective function(s) value(s) subject to the existing supply and demand constraints. It develops the optimal highway allocations for different traffic "scenarios". The "traffic scenarios - optimal allocations" data obtained in this way enables the development of HRS. HRS is the on-line system that makes on-line decisions regarding the possibility to accept driver requests. The proposed model is illustrated using a numerical example of a highway section. The optimal allocation is obtained offline for a particular arrival pattern (the solution from the integer programming) and the performance of the proposed neural network can be checked against the optimal allocation. Many tests show that the outcome of the proposed neural network that is making real-time decisions to accept/reject the driver requests is nearly equal to the optimal solution. Further exploration of other objective functions will be necessary for specific application of the proposed procedure. On one hand, problem solving by optimization (or heuristic) technique and neural network training are time consuming procedures. On the other hand, the response time for the trained neural network is practically negligible. In other words, creation of the proposed Highway Space Inventory Control System might be time consuming, but once the system is created, it can be effectively used in real-time decision making.