<p>Privacy protection is a key issue in data mining, facing security risks. The existing differential privacy protection algorithms have certain limitations in real-time data mining, such as inflexible privacy protection mechanisms and large errors. Therefore, a real-time aggregated data protection method has been proposed, which optimizes differential privacy to improve data security and accuracy. Firstly, privacy quality analysis is used to evaluate data errors and the length of the privacy protection window and then adaptively adjust the length of the privacy protection window. In addition, the perturbation algorithm of intelligent grouping is combined with budget allocation to reduce the errors caused by perturbations introduced in optimizing differential privacy algorithms. Compared with the infinite flow <i>ω</i>-event sequence data privacy protection algorithm, the optimized differential privacy protection method improves accuracy and privacy quality by 21% and 1.4 times, respectively. Compared with the optimized differential privacy protection algorithm, the fixed sliding window algorithm reduces the average absolute error by 73% and improves privacy quality by about 1.3 times. Compared with the optimized differential privacy protection algorithm, the privacy protection data aggregation algorithm of fog computing reduces privacy quality by 36%. By adaptively adjusting the length of the privacy protection window and optimizing the perturbation algorithm, research has achieved a reduction in the risk of privacy leakage while improving data accuracy, providing strong technical support for information security.</p>