With the rapid development of the Internet and digital technology, network security issues are increasingly being taken seriously, and the frequency and complexity of attacks are also increasing. This article focuses on the innovative application of Bayesian algorithm in the field of network security risk assessment, aiming to construct a highly adaptable dynamic risk assessment framework to cope with the complex and uncertain network environment. This article delves into the unique advantages of Bayesian networks in integrating diverse heterogeneous data, including network traffic, system logs, vulnerability intelligence, etc., with the aim of achieving comprehensive and in-depth risk assessment. This study achieved early warning and precise analysis of potential network threats by finely constructing a Bayesian network model and applying its powerful reasoning ability. During the experimental phase, we carefully planned a series of realistic network environment simulation scenarios to fully validate the effectiveness, robustness, and generalization ability of the proposed model. The model scores for data points 1 to 3 are all between 0.55 and 0.75, and the adaptability scores are also relatively high (0.60 to 0.80), indicating that on platform A, the model's judgment of normal and abnormal data points is relatively accurate. In summary, the risk assessment model based on Bayesian algorithm proposed in this study has laid a solid technical foundation for building a stable and reliable network security ecosystem.