| The urban rail transit (URT) operation and management department has been required to gradually improve the transport technology and service quality, and implement the service concept of people-oriented in order to satisfy the safety and effectiveness of the huge volume of passenger ridership under the background that China has entered the era of network operation of urban rail transit. However, the distribution imbalance between the passenger demand and capacity supply often cause unpredictable congestion state in the network. And because of the dynamic evolution of the passenger flow, a variety of sudden and outburst safety transport problem with huge impact will be generated, which brings great challenge to the safe operation. As the complexity degree of the URT network increasing, the operation departments will take the management and control measures based on the existing experience at present when faced with the severe passenger security issue, which lacks of the accurate grasp and analysis of the passenger distribution under congestion state, lacks of the accurate understanding and evaluation of the passenger flow evolution, lacks of the effective grooming and control strategy to network grooming. These problems are also highly valued with the operation department, domestic and foreign researchers and the whole society. Therefore we need to bring out reasonable explanation from theory, scientific modeling from the technical method and effective verification from practical application. To solve the above problems will help to improve the large passenger flow evaluation and analysis technology, the congestion state analysis and management technology, the crowd and congestion grooming and control technology in decision making under network operation, and provide important theoretical foundation and technical means for URT emergency response and crisis response. Based on the above problems, it will cover the following contents:(1) Analyze the architecture and build the physical model of the URT network. The time-dependent characteristic of the schedule-based network is studied in the foundation of analyzing the train transportation organization. Integrated with the network pattern and the transportation organization diagram, the schedule-expanded network of URT network with the characteristic of temporal variation is proposed. Based on the schedule-expanded network, the author put forward the K-shortest path searching algorithm to solve the route searching problem in this network.(2) The author analyze the main behavior impact factor and characteristic of passenger flow based on the passenger travel behavior survey. According to the passengers’decision-making process, the travel impact factor of the passenger is divided into based decision-making factors and additional factors in the light of the factors’ deterministic and dynamics. According to the random utility theory, the author propose the path choice model. Considering the information-dependence, the rationality degree and the rational travel excitation threshold of the passenger, the author construct self-adaptive learning model based on studying on the self-adaptive learning behavior of the passenger.(3) The author establish the passenger flow dynamic assignment model in URT under congestion network. The passenger flow assignment problem is studied by considering the strict capacity constraint and overload delay impact factor. The stochastic user equilibrium (SUE) relation is demonstrated under capacity constraint and the passenger overload cost, and the effective algorithm to solve this problem is put forward. According to the assignment results of the SUE model proposed in the section, the sensitivity of the parameters is to be studied. The establishing model is help to improve the passenger assignment theory in URT network. It is possible to realize the error about 11.4% of passenger assignment results calculation according to the assignment model in the paper, with the calculation of the passengers’overload delay time in station under network congestion.(4) Based on the multi-agent simulation modeling technology and method, using the bottom-up modeling method according to the passengers’ travel behavior, which including path choice and self-adaptive learning, the author propose a path choice model under the situation of path overlapping and a train choice model based on serial selection process. The author also put forward a day-to-day passenger self-adaptive learning model. By constructing a day-to-day dynamic evolution simulation model, the model can simulate and produce the whole travel decision making under day-to-day conditions of the passengers considering the information-dependence, the rationality degree and the rational travel excitation threshold of the passenger. And the passenger flow evolution process can be analyzed and explained via the simulation experiments. In the foundation of above, the author discusses the evaluation method of the flow evolution and overload delay with establishment of the station-section related linear probability matrix and overload delay timing index. The description of the spread process is studied from the perspective of spatial diffusion and time expansion of passenger flow. The flow evolution mechanism of the network is explained by numerical examples, and the method of evaluating the network congestion is presented.(5) Based on the passenger flow distribution and network evolution, the author discusses the optimization of network supply and passenger demand control. And the flow control balancing model and transport capacity allocation model are set up. The author puts forward the passenger congestion release theory aiming at maximizing the transport capacity and network efficiency. The empirical study demonstrates that these models have high theoretical value and application significance on capacity allocation, passenger control and congestion release.The main innovation points of this paper are as follows:1)The schedule expanded network is proposed from the perspective of the passenger flow assignment with the K-shortest path searching algorithm. The algorithm contains depth-first branch edge-deletion algorithm in double-layer subgraph and schedule extension traversal algorithm. The searching algorithm reduces both the complexity in comparison of the existing algorithm, and has better searching efficiency. Meanwhile, the paper demonstrates the SUE condition of the flow assignment and the flow overload delay. And the paper constructs a SUE assignment model with capacity constraint that takes overload delay as decision variables, which provides a new modeling idea for schedule-based assignment model.2) The paper proposes a joint choice model of passengers’travel decision making. The model is built with path choice and train choice as a decision-making tree, which provide a new method to study the passengers’ traveling in URT network.3) The paper proposes a day-to-day dynamic evolution simulation model to simulate passenger self-adaptive learning process through their travel every day, by considering the information-dependence, the rationality degree and the rational travel excitation threshold of the passenger. Based on the model, the complicate day-to-day flow evolution process can be discussed and studied.4) The paper establishes the passenger flow balancing control model based on maximum of the inbound flow rate and section capacity utilization. With the model, the paper proposes a detailed flow balancing control method between space and time granularity. |
PDF Full Text Request