Detecting Anomalies with Embedded AI Technology

By Ebele Maduekwe

Category:
Industry Trends

Artificial Intelligence (AI) is a major buzzword in the industrial world at the moment.  ARC Advisory has been monitoring the AI topic for a while now as we are convinced that AI will play a vital role in supporting cost reduction and developing flexible production environments in manufacturing.  For more background on AI, please refer on our blogs on policy, observations and products.

Omron has released one of the first embedded AI solution for manufacturing in a controller that collects, analyzes and utilizes data on edge devices.  The hardware is based on the company’s Sysmac NY5 IPC and the NX7 CPU and analyzes data locally, or on the edge, without the need to connect to additional on-premise or off-premise infrastructure.

embedded AI embeddedai.JPG

Detecting Anomalies with Embedded AI Technology

The AI function of the controller uses a machine learning solution called Isolation Forest technique to analyze machine information and detect anomalies.  This way, the AI controller monitors machine status, like voltage and temperature, and alerts maintenance when this data shows anomalies that could lead to machine failure and unplanned downtime.

The controller comes preloaded with machine learning models, including the AI Predictive Library, a library block for detecting operational anomalies for air cylinders, ball screws and conveyor belts.  The library block combines Timeseries database functions with the isolation forest algorithm (Machine learning-based AI algorithm) to match patterns in machine data and to extract anomaly points from the equipment.  Using this anomaly data extracted from the machine, the AI algorithm then repeatedly “learns” about the system.  It then uses the information learned to predict the best timing for maintenance services.

Advantages of Embedded AI Technology

Omron argues that the new AI controller brings real-time analysis from the machine to the user’s fingertips with a return value for automated data-driven decisions based on anomaly detection.  Companies can significantly reduce their dependency on recovery services and periodic maintenance, thus avoiding losses from unplanned machine downtime.  This new wave of advanced controllers could save companies time and money on R&D as well.

Conclusion

ARC continues to cover all topics regarding the use of Artificial Intelligence in Manufacturing.  At the time when automation is getting “automated”, products like Omron’s AI controller show how AI could be applied by manufacturers who want to achieve Industry 4.0 goals at little costs.

Engage with ARC Advisory Group