Research Of Theory And Method For Dynamic Data-Driven Simulation Base On Forest Fire Spread

Computational model, measurement infrastructure, and information technology arc currently used to analyze and predict the characteristics and behaviors of complex systems.Most of the computational models used to date, however, only allow data inputs that are fixed when the simulations are launched. These simulation and measurement approaches are serialized and static but not synchronized and cooperative. The lack of capability to simultaneously inject measured data into simulation models limits the dynamic requirements for simulations in response to the real-time changing conditions and therefore is unable to catch the instantaneous reactions and occurrences in nature. The lack of capability to simultaneously inject measured data into simulation models limits the dynamic requirements for simulations in response to the real-time changing conditions and therefore is unable to catch the instantaneous reactions and occurrences in nature.Integrating simulation system with actual system,dynamic data driven application simulations offer the promise of more accurate analysis,more accurate predictions,more precise controls,an more reliable outcomes. Dynamic data-driven simulation has become a hot spot of the system simulation in current. According to dynamic data driven application systems conceptn the theory and methods apply to simulation wildfire proagation. Studied the theory and implementation of a dynamic data driven application system in this paper.Forest fire spreading have significant impact on both ecosystems and human society. In order to make decision and prevent wildfires spread. people have studied the fire spread mechanism of forest,and accurated forecasting and simulatling on the development of forest fire spread. However, as a complex system. The accuracy of wildfire spread simulations depends on many factors, including GIS data, fuel data, weather data, and high-fidelity wildfire behavior models. Unfortunately, due to the dynamic and complex nature of wildfire, it is impractical to obtain all these data with no error.The establishment, development and application of the theory of dynamic data driven application system for forest fire spread prediction accuracy improved to provide new ideas and theoretical support. First of all.based on the research and analysis of the basic concepts and the principle of realization of dynamic data driven application system, the technical architecture is established of ynamic data driven application system,the key components of the dyanmic data driven system is the system model,the measurement model.and the data assimilation system, consequently the dynamic data driven application system for forest fire spread simulation is established. By analyzing the function widely used the system model of forest fire spread simulation, the applicable conditions and the impact factor, the system model base is established and the object-oriented data-driven system model class base is constructed,and the forest fire spread based on discrete event simulation model DHVS-FIRE is constructed according to the principle of dynamic data-driven.As the source of dynamic data. not only data collecting、transporting and utilizing,the measurement model,which is used to couple the application model and real time data,should be develop,thus to compare the application model’s output with the real data,and further estimate the real state of the system to improve the simulation results. Intending to estimate the evolving fire front,which represents the most important information in a forest fire spread simulation,and collecting the real time temperatures is deployed to map the system state to assimilate sensor data from real forest fire,in paper,it is studied that acquires fire temperature and implements measurement model.In dynamic data driven system, the data assimilation is to use observation information to improve state estimation of system under study. It tries to find the solutions by minimizing the errors between the real system and the models and to correct the models.It is an analysis techniquc,in which the observed data is accumulated into the model to relate approaches are designed to estimate a system state from the observation data.In this paple.the Sequential Monte Carlo methods SMC is emploied to carry out data assimilation in forest fire spread simulation using a discrete event simulation model.The SMC methods used in this paper implement the sequential importance sampling with resampling principle and their associated algorithms.for assimilationg data in DEVS-FIRE simulations.Finally,experiments are designed by chosing the incorrect wind conditions for the dynamic data driven application system of forest fire spread.including wind speed and wind direction. Experiments show the methods with assimilation-enhance simulation give more accurate simulation results by assimilation observation data from the real fire than the methods of static data.In the other hand, the density of deployed sensors affects the accuracy of the fire spread simulation through the different deployment strategies for sensor.Finally.the thesis analyzed the effectiveness of SMC on convergence,degeneracy and sample impoverishment.Based on the structure of DDDAS, this dissetation presents the dynamic data driven application system simulation and simulation results in forest fire spread simulation, from the experimental results.it can be known that by applying DDDAS,the simulation results could be improved.and it can provide a reference for complex system simulation and new idea for forest fire spread simulation.