The cloud model serves as a powerful tool for representing uncertainty in cloud-based artificial intelligence and data mining applications, which are critical in modern cloud networking and communication systems. Among its key techniques, the inverse cloud transformation algorithm enables the conversion of quantitative network data into qualitative concepts for improved decision-making under uncertainty. This paper presents a dynamic incremental approach to the inverse cloud transformation, addressing limitations in existing methods related to expectation estimation based on first-order central moments. By integrating offset points from the forward cloud transformation and dynamically generating new cloud drops using normal random variables, the proposed method incrementally updates sample sets to more accurately estimate cloud model parameters. Two dynamic incremental inverse cloud transformation algorithms are introduced and evaluated. Experimental results demonstrate that the proposed algorithms achieve lower estimation errors, enhanced stability, and faster convergence. They also show strong robustness against variations in network data, making them suitable for real-time cloud communication environments where data uncertainty and dynamics are prevalent. Finally, the proposed approach is applied to modelling and assessing the performance of a real-time system—in this case, shooting accuracy in a networked environment—demonstrating its practical utility in cloud-based communication and control systems.