Network suffers from various types of security threats which are dynamic, and traditional static methods are difficult to effectively validate network security. A network security validation model based on adaptive control theory (NSVMAC) is proposed to validate network security in dynamic network environment. The Markov decision process is adopted to model the adaptive control of NSVMAC model, which is solved based on reinforcement learning method, and the adaptive optimization algorithm based on Q-Learning is proposed. The adaptability of NSVMAC model is analyzed in experiments, and confirms the effectiveness of the proposed method.