Introduction

ITG is proud to announce the successful completion of our predictive analytics project for the compactor system using SORBA Machine Learning services. The project aimed to employ SORBA as a proactive analytical tool, allowing for early identification of machine faults to prevent operational downtime. Below, you will find an outline of how the project was executed, from auditing to live monitoring.

Summary

ITG provided a predictive analytics solution integrating SORBA Predictive Analytics and Machine Learning software with SORBA Cloud services. Our team went beyond software implementation to analyze results and provide insightful reviews.

Services and Deliverables

1) Audit, Analyze & Train

Deliverables:

• Validation of Data Sets: Imported, audited, and defined off-line states for sensor history.
• Sensor Groups: Excluded bad sensors and modeled asset & sensor-type sub-components.

• Population Rule Analysis: Completed.
• Failure Identification: Identified and audited for hidden failures.
• Failure Codes: Reviewed and edited.
• Training: Trained Failure and Anomaly Agents.
• Rule Creation: Created Rules Agents & Planned Maintenance Agents.

Services:

• Validated actual time-series data and defined appropriate offline states.
• Grouped sensors representing particular assets or asset sub-components.
• Executed Population Rule Analysis.
• Identified past failures and unreported failures for agent training.

• Created and tuned agents for monitoring failures and anomalies.

2) Monitor Live Equipment

Deliverables:

• Agent Creation: Created Failure and Anomaly Agents.
• Agent Deployment: Deployed Anomaly Agents and Failure Agents.

Services:

• Monitored live equipment with specifically trained agents.
• Allowed SORBA to adapt and learn new failure patterns and operating conditions.
• Sent out alert notifications upon anomaly or failure detection.

3) Visualize & Manage Asset Health

Deliverables:

• SORBA Dashboards: Checked operation.
• User-Defined Views: Checked.
• Alert Management: Validated.
• Agent Retraining: Verified.

• System Health: Reviewed.

Services:

• Established and validated operational dashboards and views for asset health monitoring.
• Validated alert management workflows.
• Checked the retraining of agents upon anomaly events.

• Updated agents across equipment when new conditions were detected.

SORBA Machine Learning Annual Subscription

The project also included an annual subscription to SORBA Machine Learning, which offers:

• A wide range of machine learning algorithms including SPC calculations.

• Supervised and unsupervised learning options.
• A license for 20 agents per asset, per data collector.
• Email, PLC, and SCADA alert notifications.
• Variable output prediction and regression.
• Parameter optimization.

• PLC machine learning agent creation.

We are excited about the successful implementation and management of this predictive analytics solution for the compactor system. We look forward to continued collaboration on future projects.

For further inquiries, please Contact Us.

Delivering Excellence in Predictive Analytics, ITG.

‍