To deal with the challenge of intelligence decision-making for high-value target in large-scale clusters under battlefield resource constraints, a method based on the random forest (RF) algorithm was proposed in this paper. The critical characteristics of adversarial battlefield targets, including geographical location, strategic value, resource consumption and the interdependencies between them were constructed. To effectively resolve the decision-making problems arising from the excessive number of targets, the RF algorithm was introduced, which possessed strong capabilities in handling nonlinear relationships and high-dimensional data. The algorithm fully leveraged the target feature information to accurately determine which targets need to be destroyed. By digital simulation, it was demonstrated that this method enabled rapidly, scientifically decision-making in large-scale battlefield scenarios. This method could provide both a reliable theoretical basis and feasible technical approaches for practical application.