Aerzen Digital Systems Builds Scalable AI Solutions with MATLAB Production Server
“Through the use of MATLAB, we were able to create an industrial machine learning solution that can be adapted to any environment, including simulation, cloud, and edge devices.”
Key Outcomes
- Early detection of issues and proactive maintenance using real-time anomaly detection enabled
- Efficiency improved through AI-based machine state classification to provide actionable suggestions to customers
- Models rapidly deployed, monitored, and automatically retrained at scale using premade MATLAB components in a DevOps pipeline
Aerzen Digital Systems provides software and automation services for the industrial sector. Recently, the company created a cloud-based, scalable AI solution for critical industrial complexes, such as wastewater treatment plants, using MATLAB®.
The core service is a module created with MATLAB Production Server™ that runs in Kubernetes® on Azure®. The module runs machine learning models that process incoming data from programmable logic controllers, customer databases, or other sources—providing data visualization and dashboard functionality for monitoring and maintenance. The service is structured in layers that automatically detect whether the underlying model is up-to-date or needs to be retrained with new data.
The Aerzen team also designed their service to be customizable to the needs of each customer. In addition to the cloud, their machine learning pipeline supports direct deployment on edge devices.
The team has already used their system for several applications, including machine state classification and anomaly detection, to improve efficiency and reliability of industrial plants. In the future, Aerzen plans to improve the ecosystem by adding automatic model selection and digital twin creation to further accelerate monitoring and optimization solutions for the process industry.
