Research On Resource Assignment Problem Based On Deep Reinforcement Learning

Resource assignment is how to adjust the assignment relationship between resources and targets to achieve the maximum assignment effect.Resource assignment problem is widely used in civil and military scenes.Electric power resource assignment,firepower resource assignment,emergency material assignment,etc.,are typical applications of resource assignment problem.Reinforcement learning is a kind of machine learning method,which is one of the key technologies to promote the transition from weak artificial intelligence to strong artificial intelligence.The key of reinforcement learning is strategy.The agent interacts with the environment to learn the strategy.The agent first acquires the state of the environment,and chooses the action according to the strategy in the current state,so that the state of the environment changes.After the state update,the environment returns the new state and reward to the agent;The agent further updates the strategy based on the reward.Deep reinforcement learning uses deep neural network on the basis of reinforcement learning and further improves the perception ability of the model on the basis of intelligent decision making.Firstly,this thesis studies the resource assignment problem based on deep reinforcement learning algorithm.In the resource assignment problem,according to the complexity of environment state modeling,this thesis designs two Markov decision models,designs and implements several deep reinforcement learning algorithms.And then this thesis carries out experimental verification,so as to study the rationality of modeling and the feasibility of deep reinforcement learning algorithm in resource assignment problem,and finally trains a deep reinforcement learning algorithm model with good decision-making ability.On the basis of using deep reinforcement learning algorithm to study the resource assignment problem,this thesis further combines expert experience with deep reinforcement learning algorithm,through experiments to explore the influence of introducing expert data on the performance of resource assignment algorithm.Finally,the performance and effect of the methods used in this thesis are evaluated and compared comprehensively to analyze the performance of each method.In this thesis,the proposed solution of resource assignment problem is studied,and a large number of comparative experiments are done,which demonstrates that the proposed methods can effectively address the resource assignment problem and provides the basis for the future research.