<p>Security threats in cloud environments have gathered significant attention due to denial-of-service attacks. These issues stem from challenges like multi-tenancy, loss of data control, and trust issues. This paper presents a novel Levy Flight Secretary Bird Optimized Attention Deep Convolutional Generative Network approach aimed at enhancing data security in cloud orchestration systems. Initially, data collection is carried out using different datasets. Then these data are pre-processed using preprocessing techniques such as normalization, inaccurate data detection, anonymization, and minimization to enhance data quality and privacy. The significant features are selected by integrating the ANOVA, chi-square, and Principal Component Analysis. Furthermore, this paper proposes an advanced attack prediction phase by leveraging Deep Convolutional Generative Adversarial Network, Attention Mechanism, and Secretary Bird Optimization with Levy Flight Strategy to improve the accuracy and reliability of attack predictions in cloud infrastructures. Thus the proposed framework significantly improves the data security of cloud orchestration systems by effectively predicting potential attacks. Overall, the proposed method achieves high-performance metrics with a detection rate of 99.0%, accuracy of 99.1%, recall of 98.3%, true negative rate of 98.4%, precision of 97.7%, true positive rate of 97.4%, and F1-score of 97.9%.</p>