Intelligent Multi Objective Resource Allocation Method Based on D3QN

The multi-objective resource allocation problem refers to how to allocate limited resources reasonably to multiple objectives under given conditions and environment, in order to obtain the best returns. This is an important and challenging problem in the field of command and control, as well as a complex combinatorial optimization problem that involves comprehensive consideration of various factors and constraints such as resource performance, target characteristics, operating environment, combat objectives, and other parameters. It has a high degree of nonlinearity, dynamism, and uncertainty, making it difficult to solve using traditional mathematical methods. Based on this, this article proposes a method of using Deep Q-Network (DQN) to solve the Dynamic Weapon Target Allocation (DWTA) problem. On the basis of reinforcement learning, a deep neural network is used to predict the Q-value function, and the Dueling method is used to improve the DQN algorithm by using a Double network for prediction. In the final simulation experiment, compared with the traditional DQN method, faster convergence speed and more accurate Q-value prediction were achieved.