Weapon-to-target assignment (WTA), which minimizes the damage to our forces by launching interceptor missiles (weapons) against ballistic missiles of the enemy (targets), is a critical decision-making problem of ballistic missile defense missions. The primary objective is to launch an interceptor with a high hit probability for each target. The existing research on WTA assumes that the hit probability is known before an engagement regardless of whether the probability varies during the engagement. However, the hit probability in actual engagement situations is a time-dependent variable that changes in accordance with the flight states of the target and interceptor that are unknown in advance. Therefore, a rolling horizon-based decision approach is necessary. In this research, we propose an adaptive WTA (AWTA) model that makes WTA decisions at each radar scanning time based on the hit probability predicted using radar information about the engagement situation-for each target, an interceptor with a hit probability higher than a threshold is launched, thereby maximizing the total hit result. A machine learning model is suggested to learn the probabilistic relationship between the flight states and hit results, and this model is embedded in the solution procedure of the AWTA model. The performance of the AWTA model is evaluated via a simulation-based experiment, and the results confirm that the proposed AWTA model is appropriate for real-time engagement situations.
All Science Journal Classification (ASJC) codes
- Computer Science(all)
- Materials Science(all)