rescue route and time simulation apparatus using reinforcement learing

摘要

The present invention relates to a structural path and rescue time simulation apparatus of a disaster structure through reinforcement learning, the structure of the disaster structure through reinforcement learning to perform a path search based on the behavior compensation value and integration compensation value calculated in each search node The present invention relates to a path and rescue time simulation apparatus. To this end, extract the outline of the structure based on the floor plan of the structure, create an environment setting unit that sets each property of the agent, doorway, and disaster that may occur in the structure, and create and arrange unit nodes based on the outline of the structure. A node unit for setting attribute values to each node, a reinforcement learning unit for predicting and determining an action direction of an agent to an adjacent node when searching for a structure path based on a unit node, and a structure for determining whether to proceed with the search for a structure path; The structure of the disaster structure through reinforcement learning includes setting a disaster scenario of the structure according to the path search, controlling to simulate the structure path according to the disaster scenario, and calculating a rescue time according to the structure path. A path and rescue time simulation apparatus is disclosed.