Power internet of things terminal protocol vulnerability mining method based on multi-agent reinforcement learning

• Main Authors: ZUO XUEYING, HAN JIAJIA, ZHANG YUXIN, SUN XIN, SONG YUBO, QIN ZHONGYUAN, WANG WEN, SI XIAOKAI
• Format: Patent
• Language: Chinese; English
• Published: 24.11.2023
• Subjects: ELECTRIC COMMUNICATION TECHNIQUE; ELECTRICITY; TRANSMISSION OF DIGITAL INFORMATION, e.g. TELEGRAPHICCOMMUNICATION

The invention provides an electric power internet of things terminal protocol vulnerability mining method based on multi-agent reinforcement learning. The method comprises a target protocol full-state data pool construction step; a reinforcement learning fuzzy test multi-agent array construction step; a strategy network design step; a value network design step; a centralized training step; and a decentralized fuzzy test step. Compared with an existing method, the method can effectively improve the intelligent degree of a fuzzy test vulnerability mining process, the comprehensiveness of protocol testing and the efficiency of triggering abnormity and collapse; the method can effectively solve the problem that a static test case loses efficacy under the condition of environment fluctuation in a traditional protocol vulnerability mining method. The method has good expandability at the same time, can flexibly customize a framework for different protocols to be tested, and is high in universality. 本发明提供了一种基于多智能体强化学