Introduction

ITG provided the successful completion of major specialty metals manufacturer project featuring the implementation of SORBA-ML, a state-of-the-art Machine Learning Platform, aimed at advanced analytics for predictive maintenance (PdM) and process optimization.

About SORBA-ML

SORBA-ML is an automation tool designed to create, train, optimize, and deploy machine learning algorithms, known as "SORBOT agents," for solving complex PdM and process optimization challenges. These agents were seamlessly integrated into the SORBA-SDC edge platform for real-time proactive health monitoring.

Pilot Program Features

This approach ensured the ability to test the system rigorously and quantify the number of PdM failures effectively. The customer was able to scale these solutions across the entire plant post-successful pilot completion, which had a maximum duration of six months.

Offered Capabilities:

• Unlimited training data sets
• Flexibility to add or subtract agents from the subscribed pack
• Full access to a range of ML algorithms

Project Phases

Phase 1: SORBA-SDC - Understanding PSB Failure Modes [1st–3rd Months]

• Engineering Discovery of PSB (P&ID, Controls, Sensors)
• Deployment and Configuration of SORBA-SDC and SORBA-ML Platforms as Virtual Machines (VMs) Onsite

• Server onboarding and I/O (Tags) Configuration for PSB
• SDC networking and workflow configuration
• Development of rules and thresholds for baseline analytics and failure modes
• Comprehensive Project Management and Training

Phase 2: SORBA-ML - Deploying Predictive Analytics [4th–6th Months]

• Creation and training of ML anomaly detection and classification algorithms
• Testing of SORBOT agents against simulated data
• Building and documenting configuration templates for the PSB
• Retraining and deploying ML algorithms
• Real-time PdM dashboards for the GSS project team

• Comprehensive training manuals and project documentation

Outcomes

Upon completion of the pilot program, the customer opted to scale the deployment of the initial 20 SORBOT agents across four additional assets. The project proved the efficiency and effectiveness of SORBA-ML in both predictive maintenance and process optimization applications.

For more information on how the SORBA-ML platform can transform your operations, please

Empowering Businesses with Data-Driven Predictive Maintenance and Process Optimization Solutions.

‍