Abstract: To solve the problem of evacuation path planning in building fires, a method of evacuation path planning for fire fighting and rescue based on isolated forest algorithm was proposed. The wireless sensor network arranged at the fire scene is used to collect fire environment information, build fire data samples, randomly divide and train fire data samples, create multiple isolated binomial trees to form isolated forests, identify abnormal fire data, obtain the location of ignition points and obstacles, and build a dynamic map of fire rescue environment based on grid method, by updating the equivalent distance of location nodes. The improved ant colony algorithm of pheromone concentration and pheromone volatilization factor is used to build a combined optimization model of rescue and evacuation paths and plan the best fire rescue and evacuation path. The experimental results show that this method can accurately detect the location of the ignition point in the fire, and can better avoid the ignition point and obstacles in the multi start and multi end rescue and evacuation path planning, the best fire fighting and rescue evacuation path can be quickly and reasonably planned.

Key words: isolated forest algorithm, rescue evacuation, path planning, location of ignition point, ant colony algorithm, pheromone