Stock photo.

Successes at SAICA with predictive and AI based monitoring

Authors:


A reliable predictive monitoring system is key when it comes to machine availability and maintenance planning. Voith’s predictive monitoring solution OnCare.Health Paper is an online monitoring, analysis and diagnostic solution, combining Voith’s expert know-how in papermaking, analytics and artificial intelligence.

In 2022, SAICA decided to upgrade its existing condition monitoring systems of three machines in Spain with the Voith OnCare.Health predictive machine and technology monitoring solution, following with two more machines in 2023. By continuously collecting and analyzing data on the machine condition, SAICA can now automatically detect deviations and anomalies. Timely response to malfunctions can be triggered to prevent unplanned machine downtimes.

Nevertheless, at diverse positions within the paper mill, it is not possible or costly to install sensors. In addition, experts must still investigate and analyze the root causes of issues at physically monitored components. Therefore, SAICA and Voith began a joint development project to enable overall machine monitoring and analytics based on existing machine data. This approach can facilitate both condition and behavior monitoring of different component and processes.

  • Self-trained AI models for mechanical and electrical components were set up based on SAICA's historical process data.
  • The running of AI models is designed to identify anomalous behaviors across diverse mechanical components.
  • SAICA, with consulting from Voith, is independently executing incremental learning of the AI models and fine-tuning its models to different machine modes and grades.
  • Automatically generated insights about the most common causes of the anomalies are enabling SAICA to take immediate actions.
  • In 2024, the joint development project will be enhanced with the new intelligent pattern recognition function, allowing SAICA to generate automatic recommendations for action on errors identified by the AI.

The presentation will showcase Voith’s data-driven monitoring approach and provide insights into the AI and associated technical solution. In addition, the results achieved at SAICA will be shown and the possibilities and limitations of AI-based predictive monitoring in paper production discussed.