Abstract: To address the distributed denial of service （DDoS）network attack problem of cloud control system, a spatio-temporal feature detection model CNN-MHA-BiLSTM is proposed to realize the extraction of traffic features from temporal and spatial dimensions, respectively, and to improve the ability of the model to distinguish the key features by combining multi-headed attention mechanism. A single water tank cloud control system based on the IAP cloud platform is built to study the impact of DDoS attacks on the cloud control system. The simulation results show that the CNN-MHA-BiLSTM model outperforms the existing DDoS attack detection models and can effectively detect DDoS attacks of the cloud control system.