Reverse Deduction Model of Forest Fire Spread Based on Cellular Automata

Nowadays, forest fires occur more and more frequently due to global warming, posing a huge threat to the economy and people’s security. In order to prevent forest fires and carry out timely rescue, the research about reverse deduction model which can trace the source of fires is of great significance. At present, there are many mature spread models of forest fires that can accurately simulate the spread of forest fires, but most of them are forward simulation models, and there are few studies on reverse simulation. Even the existing reverse simulation models have complex and single structure, and the input data are difficult to obtain. Therefore, this paper comes up with ideas about reverse deduction model of forest fire spread based on Cellular Automata (CA) without considering human intervention, which needs parameters easily accessible and can reversely simulate the fire process. Firstly, the basic component of Cellular Automata, principles of Discrete Event System Specification (DEVS) and principles and common methods of Shortest Path Algorithms are analyzed, which provides the basis for building CA models and methods for searching for ignition points. Then a suitable Cellular Automata model is designed, including cell neighborhood, combustion state, and state transition rules, and a CA-based reverse deduction model of forest fire spread is constructed based on the DEVS framework, thus being used to simulate reversely the spread of forest fires and estimate the fire point based on minimum travel time (MTT), which reflects the possibility of the ignition point by the MTT standard deviation. After being tested by experimental data collected in the fire happened in Hulunbuir in China, it shows that the model provides good simulation effect and accuracy. The areas of fire at different time of the reverse simulation model are close to that of ordinary forward simulation model. Also, the location of the fire point can be accurately estimated, within limited error and being close to the real fire point. In conclusion, this model provides a more convenient forest fire simulation method, which can simulate the burning process of the fire and predict the location of the fire point when the fire point is unknown. It can be used for the prediction of forest fire burning trend and fire point, so as to provide help for fire extinguishing and disaster relief, and can also trace the cause of the fire after the disaster to prevent the occurrence of fire. Therefore, this model is of great significance in disaster prevention.