Large-scale city emergency material distribution method based on deep reinforcement learning

Abstract

The invention relates to a large-scale city emergency material distribution method based on deep reinforcement learning, and the method combines the advantages of deep learning and reinforcement learning, carries out the training of a method through employing a DQN network, and achieves the emergency material distribution in a large-scale city disaster response scene. The method specifically comprises the following steps: step 1, modeling a disaster area, and constructing an emergency material distribution task model; 2, defining an emergency material distribution joint target optimization function for multiple disaster areas; 3, constructing a Markov decision process, and defining contents such as an environment state, an action space and a reward function; 4, training a deep reinforcement learning model; and 5, inputting the initial state of the disaster area, and generating an optimal emergency distribution strategy. According to the emergency material distribution method provided by the invention, on the premise of ensuring fair and effective emergency material distribution to relieve the pain of victims, the calculation efficiency is higher, the emergency response speed is improved, and the loss caused by disasters is further reduced.

Description

technical field [0001] The invention belongs to the field of emergency material distribution for emergencies, in particular to a large-scale urban emergency material distribution method based on deep reinforcement learning. Background technique [0002] In recent years, disasters have occurred frequently in the world. Large-scale urban disasters and emergencies often result in an extreme shortage of emergency supplies. Disasters will not only cause huge casualties, but the survivors in the affected areas will also face severe post-disaster environment. In the early stage of a disaster, as the pre-installed emergency materials and transportation channels such as roads in the affected areas may be destroyed by the disaster, emergency materials will be extremely scarce, resulting in an extreme imbalance in the supply and demand of emergency materials in the early stage of the disaster. At this stage, unreasonable distribution of materials will not only reduce the efficiency of...

AI Technical Summary

Problems solved by technology

[0004] For the distribution of emergency supplies for large-scale urban disasters and emergencies, most of the existing distribution methods cannot achieve a good balance between reasonable distribution and rapid generation strategies. Most of the distribution methods use traditional operational optimization algorithms to Solve the material distribution strategy, but the traditional method is difficult to solve the large-scale and complex emergency material distribution problem

Therefore, there is still no effective solution to the distribution of emergency supplies for large-scale urban disasters and emergencies.

Images