Large amounts of sensitive and important data are at danger of cyberattacks, which are typically preceded by network reconnaissance, due to the Internet’s fast expansion. HAM, a moving target defense strategy, aids in combating network reconnaissance. Nonetheless, HAM continues to face a number of serious issues: (1) Existing methods are unable to adjust to hostile tactics on their own. (2) Because each host chooses whether to change its IP address, the network state varies over time. (3) The majority of approaches disregard the viability of current connections in favor of improving security. In this paper, The aforementioned issues are addressed by proposing an Intelligence-Driven Host Address Mutation (ID-HAM) scheme. To clarify the mutation process, we first build a MDP and create a smooth mutation mechanism. Second, to remove impractical actions from the MDP action space, we formulate address-to-host assignments as a constrained satisfaction issue. Thirdly, in order to learn from scanning behaviors, we create an advantage actor-critic algorithm for HAM. Finally, thorough simulations and security analysis show how effective ID-HAM is. When compared to state-of-the-art systems, ID-HAM can cut scanning hit times by up to \(25\%\) with only a slight impact on communication. In order to test various scanning technologies, we also put in place a proof-of-concept prototype system.