 Interdomain routing systems are critical components of the Internet and have been subject to numerous disruptions in recent years. Researchers have focused on understanding the attacker's behavior when targeting these systems and have identified the importance of selecting the most effective attack node group. However, existing research does not consider the cost of the attack, which can be significant. This paper proposes a new algorithm called partition multi-objective transformation, PMT, to optimize the selection of attack nodes by considering both the cost of the attack and the degree of non-linearity. PMT uses a network partition and node replacement strategy to initialize the algorithm and then uses a partition search method to find the best solution. Experimental results show that PMT outperforms other algorithms in terms of both objective function values and computational time. This article was authored by Wendian Zhao, Yu Wang, Liang Liang, and others.