Traditional security strategies and methods are difficult to adapt to constantly changing patterns of network attacks and threats, so a method is needed to improve the security and defense capabilities of the network. This study proposed a network information security system based on reinforcement learning using the Q-Learning algorithm. The system autonomously adjusts security policies and behaviors through interactive learning with the environment to improve the detection and response capabilities to threats. The characteristics of the Q-Learning algorithm enable the system to learn and optimize based on observed network activity and feedback signals, in order to adapt to constantly changing network attacks and security threats. Through experiments and evaluations, the system resource utilization rate based on the Q-Learning algorithm was between 75 and 88%, and the system can more accurately identify and classify threats, reduce false alarm rates, and improve its adaptability to unknown threats.