Search

The light rail (LR) regarded as the ideal solution for a city that has grown too large for buses, yet cannot be supported by heavy rail with fully dedicated light rails and elevated guide-ways. To improve the efficiency of LR, it is important to develop appropriate dispatching system. The Light Rail Intelligent Dispatching System (LRIDS) proposed in this paper is established according to the status of light rail operating including vehicle location and number of passengers, making the best use of Reinforcement Learning (RL). The algorithm uses a team of RL agents, each of which is responsible for controlling one route. Finally, the developed algorithm is implemented with light rail network of Beijing City. The results demonstrate the power of RL on light rail dispatching problem.

In this paper, we describe a system offering a solution to the problem of public transportation guidance. It is based on the integration of geographic information system (GIS) and global system for mobile communication (GSM) technologies. A significant operation for the handling of public transportation guidance system is the routing of passengers to their destinations. The response time of a real-time system like ours to such queries is of vital significance. By using efficient data structures for the implementation of the graph representing the bus network, the time performance of the shortest-path algorithm can be enhanced. Incorporating the efficient algorithm within the GIS will increase our system's viability. The developed system is implemented with Guangzhou containing 471 routes and 1040 stops. The test result indicates that the description, model and algorithm proposed in this paper are effective.

Aiming at the need of work to protect the urban underground gas pipeline, this paper puts forward the model of the safety evaluation based on artificial neural network. To overcome the unilateralism that was brought by only depending on the experts' experience to select the learning samples of artificial neural network, this model applied clustering analysis method in the selection of learning samples. This model also generally simulates the relation between corrosion extend of the urban underground gas pipeline and the genes that influence them. Finally the developed model was implemented with the real urban underground gas pipeline data of Shenzhen City. The test result indicate that this model can truly evaluate the pipeline's corrosion extent and the identify ratio is more than 90%.