Integrated Deep Reinforcement Learning Technology for Platform Automation Operation and Fault Diagnosis Algorithm

The current mainstream protection solutions have encountered difficulties in application and performance degradation in the construction of security platforms due to the fragmentation and mismatch between security protection measures and business scenarios. Traditional control theory scheduling algorithms and heuristic based scheduling algorithms are difficult to achieve ideal results in dealing with complex and changing security scenarios. Therefore, this article explores a real-time optimization algorithm for Quality of Service (QoS) in software defined secure middleware based on the Deep Q-Network (DQN) algorithm. Firstly, in response to the problem of excessive resource waste caused by unreasonable scheduling algorithms in the current technology platform, this article introduces the use of deep integration of software and hardware to increase the intensity and depth of dynamic resource allocation, and expand the flexible calling of business and data to solve this problem. Subsequently, this study recommends the construction of deep Q-learning (DQN) algorithm technology for platform automation operation and fault diagnosis. Then, this article studies the algorithm reward function and training loss function, and establishes a fault recognition method in the behavior space by modeling the agent. The experimental results show that as time progresses, the success rate of system job scheduling remains at a high level of 80% to 95%, indicating the effectiveness of the scheduling mechanism. After the 40th hour, the success rate of the system gradually stabilized, indicating that the robustness of the scheduling algorithm has been effectively validated.