The use of smart sensing technologies in production systems has grown in popularity in recent years. These systems use sensors to gather data and interpret it in real time, making industrial processes more automated and efficient. However, strong security measures are more important than ever to guard against possible cyber attacks and vulnerabilities due to the increasing complexity and interconnection of these smart sensing production systems. These difficulties include the possibility of unwanted access to private information, falsification of sensor readings, and interference with device-to-device communication. Hence, creating a security architecture that can successfully counteract these new dangers and guarantee the availability, integrity, and confidentiality of the smart sensing production system is central to the issue description. Conventional security systems often depend on intrusion detection systems, firewalls, and encryption methods to secure networks and information. These steps might not be enough, nevertheless, to handle the unique difficulties presented by smart sensing manufacturing systems. Additionally, subtle and sophisticated assaults targeting the networked sensors and communication channels are difficult for traditional systems to detect. Therefore, a more intelligent and adaptable security solution is required, one that can recognize the special traits of smart sensing settings and take proactive measures to counter new threats. Furthermore, any security breech in contemporary industrial systems can have dire repercussions, including possible safety risks, production interruptions, and data leaks. The industrial sector is rapidly embracing Industry 4.0 principles, which emphasize the need for enhanced security measures due to the dependence on networked equipment and data-driven decision-making. Because deep neural networks (DNNs) are excellent at processing complex and high-dimensional data, this research presents a promising method for securing smart sensing production systems. DNNs are particularly well-suited for analyzing the various streams of information generated by sensors in a production environment. DNN models may identify abnormalities suggestive of security vulnerabilities and understand patterns of typical behavior by utilizing artificial intelligence and machine learning. In the context of networked and data-driven production settings, these models offer a more intelligent and adaptable approach to security, providing a better degree of protection against emerging cyber threats.