Research On Theory And Application Of Micro-simulation Model Of Pedestrian Based On Cellular Automata

Microscopic simulation models of pedestrian dynamic have been an effective toolto evaluate and optimize the program of construction and operation of modern facilitiesfor walking. Pedestrian micro-simulation model has become an important researchdirection of the computer science and traffic engineering. For the deficiencies ofexisting pedestrian simulation modes, this dissertation makes a deep research inpedestrian micro-simulation models based on cellular automata. The main work of thisdissertation are summarized as follows.(1) A safety interspace model based on psychosocial distance is proposed. Thismodel is designed to simulate the effect of psychological exclusion in normal pedestrianmovement. The velocity equations of the model indicate the positive correlationbetween the safe distance and instantaneous velocity, as well as normal randomdistribution of perceived distance. Verified by real data from controlled experiments, itis found that SIM can accurately reproduce the fundamental diagrams of empirical data.Compared with walk-following model, the distribution of individual space requirementsand individual velocity of SIM is closer to the real data.(2) A slow reaction model based on pedestrian movement delay is proposed. Thismodel is designed to simulate the reaction delay of single row pedestrian flow in thelow-speed movement. The states of pedestrian movement are classified and each stateapplies different velocity update formula. In addition, a slow reaction probability isintroduced to describe differences of pedestrian reactive agility. Simulation experimentsverify the validity of the model by compering with real data. It is found that SRM cansimulate and reproduce the structures of space-time evolution and the self-organizationphenomena of real pedestrian flow better than LG model.(3) A cellular automata model with a mechanism of dynamic route choice isproposed. This model is designed to simulate pedestrian commute between theresidential and commercial areas of a city. After allocation of OD informationfor pedestrian activities, a mechanism of dynamic route choice is introduced.Through simulation experiments, this dissertation investigates the relationshipof average velocity and flow rate with the size of the city and the businessdistrict area. The dissertation studies the impact of the business district layouts and urban development patterns on average velocity and flow. In addition, thedissertation makes finite-size scaling analysis and investigation of probabilitydistribution of arrival time, target arrival rate and convergence time.(4) A decentralized cooperation model for multi-directional pedestrian flow isproposed. This model is designed to simulate cooperative behavior of multi-directionalmixed pedestrian movement in weaving sections. The model consists of twocooperation mechanism: stepping aside and alternative route mechanism. Simulationexperiments quantitatively study the effect of two cooperative mechanism on averagevelocity, average flow rate and average travel time. It is found that compared with thepoint-to-point random walk model without cooperation mechanism, the average velocity,flow rate and travel time are improved under the combined effect of the two cooperationmechanism.(5) A practical prototype (WalkEva) of pedestrian micro-simulation software isresearched and developed. The system architecture of prototype is divided into fourlevels of models, algorithms, control and interface. Main function modules are realized,such as scene modeling, pedestrian property configuration, program control, pathplanning, data saving and output, evaluation and statistics of results,2D and3Danimation, etc. The prototype has almost complete workflow and demo-friendlyinterface. Through several tests of typical application scenarios, the validity of themodel and the practicality of the software prototype have been verified.