Recently, the industry has been driven to move industrial control systems to the cloud due to the significant advantages it offers in terms of storage and computing resources. However, this shift also brings forth significant security challenges. By moving control systems to the cloud, the potential for attackers to infiltrate the system and launch damaging attacks increases. These attacks can result in severe disruptions and potentially catastrophic consequences. Hence, attack detection and mitigation mechanisms are crucial for cloud control systems. In this paper, we present an approach that leverages the digital twins concept and virtual actuator method to detect and mitigate deception attacks on control signals within cloud control systems. By conducting experiments on a real testbed and subjecting it to a set of attacks, we validate the effectiveness of our solution. Our proposed method successfully detects attacks in a timely manner and keeps the plant stable, with a good performance during the attack.