|
A Hybrid Model Combining Advanced Mathematical Modeling and Reinforcement Learning to Enhance Command Effectiveness and Decision-Making on the Battlefield
|
| |
|
|
|
Abstract: (5 Views) |
Nowadays, the use of mathematical modeling and artificial intelligence methods in the analysis and simulation of battlefield scenarios plays an important role in improving decision-making processes and evaluating different strategies. One of the important challenges in this field is improving the decision-making and effectiveness of a team that is in a weaker position than its opponent in terms of combat capabilities and resources. In this study, a hybrid model based on mathematical modeling using game theory and reinforcement learning is proposed to improve the command and decision-making process of the weaker team. In the proposed model, the competitive interactions between two teams and their offensive and defensive decisions are simulated using a mathematical model based on game theory. Q-Learning and Deep Q-Network (DQN) algorithms are then employed to train an intelligent coach to select appropriate actions for the weaker team. The performance of the proposed model is evaluated in a conventional battlefield scenario over 1,000 simulation episodes based on the total accumulated reward and the remaining health of the team members. The simulation results show a positive and increasing trend in the total reward of the weaker team from approximately episode 400 onward. This improvement coincides with a reduction in the exploration rate (ε) to below 0.2 and increased exploitation of the learned policy, indicating the gradual improvement and stabilization of the intelligent coach's decision-making process.
|
|
| Keywords: Mathematical Modeling, Reinforcement Learning, Battlefield, Command and Control. |
|
|
Full-Text [PDF 900 kb]
(9 Downloads)
|
Type of Study: Research |
Subject:
Command
Received: 2025/02/16 | Accepted: 2025/11/18 | Published: 2026/10/19
|
|
|
|
|
|
|
| Add your comments about this article |
|
|
|
