This paper proposes a hybrid BW-SOA optimization strategy to increase the sensing layer’s reaction time, as well as an enhanced Hidden Markov Model-based method of network security posture evaluation. The efficient use of situational awareness for smart power IoT security monitoring is made possible by the use of edge computing based on deep learning. Experiments have shown that the augmented HMM model improves assessment performance by more successfully identifying security flaws in the system. The state transfer probability of the improved HMM model is more reasonable compared to the original HMM model, S3→S1 is improved from 0.15 to 0.6635, and the concentration of the observation distribution is significantly improved, and the probability of observing V5 in S5 is increased from 0.61 to 0.9962.The converged DL neural network-based edge computation method has an average reward of -39.73, which is 77.17% higher compared to the traditional DDPG.
