Summary

A major class 1 railroad had sought innovative products and services to bolster its Mechanical Engineering team's efforts in reducing maintenance and repair costs. With a reduced workforce, the need for a distributed system for data collection, analysis, and actionable insights was more important than ever. The project with ITG aimed at achieving an estimated maintenance cost reduction of 20% to 40% through various benefits:

• Early warning of equipment failure
• Enhanced efficiency of the locomotive maintenance staff
• Replacement and repair conducted only when essential
• Minimal disruption to regular locomotive service
• Cost control within acceptable and predictable levels

‍

Solution and Deliverables

ITG utilized historical data sets provided by the customer to demonstrate the effectiveness of machine learning via the SORBA platform. The aim was to uncover the root causes of the top 5 mechanical and sensor failures. Using SORBA's machine learning algorithms, specific failure signatures were created for each detected anomaly. The customer received comprehensive reports detailing the prediction confidence levels for each anomaly, sensor tag rankings, lead times to failure values, and financial ROI calculations. The customer was responsible for validating these findings, thereby paving the way for broader implementation of the SORBA IIoT platform within their Mechanical Engineering department.

Services

What The Customer Provided:

1. Identified DASH-3 locomotives that had experienced the top 5 mechanical and sensor issues. The specifics of these failures were discovered during the data set analysis.

2. Supported ITG in obtaining the required datasets, with a minimum of 3 months' worth of data from selected DASH-3 locomotives.

3. Supplied financial information related to the failures, allowing ITG to calculate ROI based on the findings.

4. Assembled key stakeholders for a meeting to discuss ITG's report findings.

What ITG Provided:

1. Deployed an Engineer and Data Scientist to retrieve and transform the locomotive datasets supplied by customer.

2. Imported these datasets into the SORBA platform for machine learning analysis.

3. Collated the results into a comprehensible report for presentation to customer.

4. Worked closely with customer to validate the findings.

5. Refined the report to include all validated anomalies and depicted failures, root causes based on tag rankings, and a financial model for each anomaly, based on fleet-wide deployment of Machine Learning Agents. ITG created up to 20 SORBOTS (Machine Learning agents) for each validated failure.

Conclusion

The Class 1 Freight Railroad AI-Predictive Maintenance was a milestone in the adoption of advanced analytics and machine learning for optimizing maintenance and repair operations. The pilot successfully demonstrated the significant cost-saving potential and increased operational efficiency achievable through the SORBA platform. With validated findings and ROI calculations, customer is now better positioned to expand the use of machine learning and IIoT technologies in its Mechanical Engineering department.

For further details on this retrospective or to discuss future collaborations, please

We remain committed to driving operational efficiencies through innovation and data analytics.

‍