This paper focuses on enhancing DDoS attack prevention ca-pabilities through the combination of the Cumulative Sum (CUSUM) algorithm and the Backpropagation method, aiming to detect attack indicators early and accurately. The CUSUM algorithm is used to monitor and analyze network traffic over time, identifying unusual fluctuations in traffic without requir-ing prior knowledge of attack types. Meanwhile, the Back-propagation method is applied to optimize neural networks, enabling the system to learn from previous traffic data and distinguish clearly between legitimate traffic and attack traf-fic. Compared to previous research methods, this combined approach offers several significant advantages. First, CUSUM provides high-accuracy attack detection, allowing the system to respond promptly. Second, Backpropagation enables the system to improve automatically over time, reducing false alarm rates and enhancing prevention effectiveness. Finally, the feasibility and effectiveness of the solution are demon-strated through real-world experiments, showing improved detection rates and faster response times compared to tradi-tional methods.