An exhaustive examination and evaluation of hierarchy network security measuring and optimal active defense in the context of cloud computing is provided in this paper. After collecting all the cloud platform’s security-related data obtained via evaluation, the pertinent security data is summarized and analyzed to create the cloud platform’s safety record index. This index serves as a benchmark for the cloud platform administrators to evaluate the security vulnerabilities associated with their cloud platform. It offers cloud platform administrators a standard against which to assess the security hazards of cloud platforms. We primarily investigate the building of the cloud platform, the building process of the security situational awareness system, and the estimation of the safety circumstance value and use via the cloud platform’s security circumstance understanding system, thereby significantly enhancing the cloud platform’s stability, security, and dependability. By implementing the approach, the limitations of conventional network security management are circumvented, as the latter is predicated solely on historical data and is incapable of perceiving real-time alterations in the security condition of the system. Concurrently, the anticipated outcomes are incorporated into the imprecise decision-making system’s input, enhancing the evaluation’s precision. The approach improves the efficiency and efficacy of predicting the network safety situation in real time, augments the algorithm’s rate of convergence and prediction precision, and prevents overfitting from occurring. Simulation experiments conducted using the internet network safety posture dataset demonstrate that this research approach exhibits superior learning efficiency and lower prediction error compared to conventional machine learning and other deep learning techniques.