IIoT and data management enabling Artificial Intelligence in crystal growing 

Data management is a pillar for the application of Artificial Intelligence in industrial contexts. A significant case is addressed by SMART-Growth project, which aims at applying AI and CPS to control processes and equipment for laser crystal growing. 

Efficient data management represents the foundation for applying Artificial Intelligence (AI) in industrial environments. The success of AI systems strictly depends on the quality, accessibility, and structure of the data they have to process. In manufacturing sectors, data is generated continuously from sensors, devices, machines, equipment and digital platforms.  

Proper management ensures that this data is collected, cleaned, standardized, and securely stored, allowing AI models to detect patterns, make predictions (e.g., failures, health status, process variables), and optimize operations. Without solid data governance – covering integrity, integration, interoperability, and privacy – AI algorithms risk producing unreliable results. Therefore, robust data management transforms raw information into strategic and added value assets, enabling industries to achieve efficiency, innovation, and competitiveness through trustworthy and scalable AI solutions. 

A practical and significant case of IIoT data management enabling AI is addressed by SMART-Growth project (established within the I3 framework of the funding agency EISMEA), which aims at applying Artificial Intelligence and Cyber Physical System-based observation and control for laser crystal growing processes and equipment that provide multi-parametric process data sourcing from a sensor network around the process. In this context, Holonix Srl (IT provider) has designed, implemented and deployed a hardware and software infrastructure enabling integration, feeding and working of a new AI control which supports and adjusts the action of a traditional PID regulating the laser crystal growing at Filar Optomaterials Srl facitility.  

According to hw/sw requirements for supporting introduction and feeding of the AI in Filar Czochralski machine, performed activities have leveraged know-how in data management and have leaded to the definition and implementation of: 

• an “isolated” network for the machine; 

• proper interfaces with current legacy system; 

• a storage middleware lightweight solution aiming at collecting field data and returning AI control output.  

The designed solution consists of:  

• the existing pc hosting current control, that has been setted up to expose shared folders and data through Windows native functionalities; 

• new sensors for feeding new AI control, to be integrated with industrial PLC; 

• a new PC with specific sw monitoring the shared folders, interfacing PLC, getting/storing data and running AI control.  

A first testing phase has been positively executed, while an extensive experimental validation campaign on the whole solution (IIoT data management + AI) is currently ongoing at Filar premise.