Taking a central stage at IMARC 2024: Real-Life Examples of AI Sensor Fusion-Based Predictive Maintenance in Mining
November 19, 2024
Michael Zolotov, CTO and co-founder of Razor Labs, recently delivered a keynote at the IMARC 2024 conference that drew a large audience of mining maintenance professionals. Known for his pioneering work in predictive maintenance for the mining sector, Michael captivated attendees by showcasing how AI-powered sensor fusion technology can transform equipment monitoring and maintenance. From remote mines in South Africa to complex operations in Latin America, Razor Labs has been leading advancements in predictive maintenance, helping mining operations avoid costly downtime and extend the lifespan of critical assets. In this blog post, we’ll walk through real-life case studies from Michael’s keynote, highlighting how Razor Labs’ technology is redefining the standards of mining maintenance.
Predicting Ball Mill’s Motor Failure in South Africa’s Precious Metal Site
In one of Michael’s examples, Razor Labs deployed its DataMind AI predictive maintenance system at a precious metal mine in South Africa, where a motor failure would have gone unnoticed by traditional monitoring methods. The mill, a crucial piece of equipment, must operate continuously; even a short downtime results in considerable losses in throughput and recovery, amounting to millions of dollars.
When the mill restarted after a scheduled shutdown, DataMind AI system identified rapid deterioration by analyzing vibration frequencies that had spiked by a factor of 40 in just a week—a telltale sign of an impending failure. Unlike typical monitoring systems, DataMind AI wasn’t just a “black box”; it provided clear, actionable evidence of the issue. With this insight, the site scheduled a preventive maintenance shutdown to replace the motor bearings and avoid a potential catastrophe right before the Christmas break. This example illustrates the value of timely, data-driven decisions enabled by advanced AI.
Removing Operational Noise with AI Sensor Fusion
One of the critical elements of Razor Labs’ technology is AI sensor fusion, which reduces operational noise to make subtle issues more visible. In his keynote, Michael likened this process to driving a car uphill. As the load increases, the engine vibrates more, but this doesn’t indicate a problem—only strain from the additional load. By leveraging multiple data points like speed, load, and vibration frequencies, DataMind AI system isolates meaningful signals from operational noise, making it possible to identify real deterioration trends even when the machine is under varying operational conditions.
In the South African case, this sensor fusion approach allowed operators to distinguish normal variations in mill operation from early signs of failure. Without it, the signs of motor bearing deterioration might have been missed entirely.
Visual AI for Crusher Monitoring and Material Flow Control
The keynote continued with another application example: crusher monitoring. At another mining site, the mill faced stoppages every few weeks due to unexplained blockages. Razor Labs installed cameras that monitored the material as it passed through the crushers, analyzing the size of individual ore particles. Through deep learning and visual AI, DataMind AI was able to track Particle Size Distribution (PSD) in real time, a key metric for ensuring smooth mill operation.
This approach revealed an upstream issue—the crusher was releasing oversized ore that occasionally blocked the mill. Thanks to real-time visual data from DataMind AI, site operators could now monitor material flow and identify trends. The team proactively addressed issues before they escalated, allowing for timely liner replacements and crusher gap adjustments. This proactive approach saved maintenance time and prevented the costly stoppages that plagued the operation
Extending Predictive Maintenance to Mobile Fleet Assets
Razor Labs has developed predictive maintenance capabilities for haul trucks, excavators, and other mobile equipment, which are critical assets in mining. He shared the example of a Caterpillar 793D haul truck where traditional OEM alerts often misinterpret operational noise—such as sharp turns or overspeeding—as faults.
Using AI Sensor Fusion, DataMind AI filtered out these operational factors and isolated critical data under consistent conditions. This approach detected a drop in oil pressure weeks in advance, indicating a potential oil pump failure. With this insight, maintenance teams could intervene before a breakdown occurred, ensuring the truck continued operating smoothly without unexpected downtime. This method has been shown to reduce nuisance alarms, allowing maintenance teams to focus only on genuine issues.
A Holistic View of Equipment Health Through Comprehensive Data Analysis
In addition to equipment sensors, DataMind AI leverages fluid analysis, maintenance logs, and even tire information for a comprehensive view of equipment health. As Michael pointed out, these combined data sources provide a 360-degree view of asset health. For instance, tire tread depth and road segment information help correlate tire wear with driving patterns, allowing teams to optimize both vehicle operation and tire lifespan.
By combining fluid viscosity, fuel quality, and coolant status with real-time sensor data, Razor Labs can catch issues that would otherwise go undetected. This level of insight allows mining operators to manage maintenance proactively, making data-driven decisions that save time, reduce costs, and increase asset reliability.
Michael’s keynote at IMARC 2024 underscored how Razor Labs’ DataMind AI system, powered by cutting-edge AI Sensor Fusion technology, is setting new standards in mining maintenance. Through real-life examples, he demonstrated the practical benefits of predictive maintenance—from avoiding costly mill shutdowns to optimizing mobile fleet assets and extending the lifespan of the equipment. For mining professionals looking to improve operational efficiency and reduce unexpected failures, these solutions offer a way forward.
