Key Takeaways
- The frontier in ore sorting sensors is not a single sensor. It is sensor fusion: combining XRT, optical or machine vision, laser, and induction or NIR signals on one platform.
- Steinert’s KSS is built as a combined sensor system. It pairs XRT, XRF or NIR with colour, 3D laser and induction sensing, depending on ore type.
- The right combination depends on what physically distinguishes your ore from waste. Density and atomic number differences favour XRT, surface colour favours optical, and conductivity favours induction.
- AI classification software turns several simultaneous sensor streams into one accept or reject decision per particle. That is where vendors now differ most.
TL;DR
Choosing ore sorting sensors starts with the physical property that separates your ore from waste. Next, you rely on the vendor’s AI classification software to fuse the signals. Steinert’s combined KSS platform is built for multi-sensor fusion, while TOMRA offers XRT, NIR and laser as configurable options.
What Are Ore Sorting Sensors?
Ore sorting sensors are the instruments inside a sorting machine that look at each rock as it passes. Each one measures a different physical property, such as density, colour, surface mineralogy, shape or electrical conductivity. The sorter uses those readings to decide which rocks to keep and which to eject.
No single sensor sees everything. For example, a rock can look like waste on the surface but carry dense sulphide inside. That is why modern machines often combine two or more sensors. The sensor choice sets the ceiling on sorting performance, and the software decides how close you get to it.
How Do I Apply AI to Ore Sorting Technologies?
The process has three steps: match the sensor to your ore, combine sensors where needed, then train the fusion model.
Matching ore sorting sensors to your ore
First, figure out which property actually separates your ore from waste. X-ray transmission (XRT) detects density and effective atomic number differences. It is the workhorse for base-metal sulphide ores, including copper, where the valuable mineral is denser than the host rock. Optical or machine-vision sensing reads surface colour and texture. It helps when ore and waste look different but have similar density. Near-infrared (NIR) picks up surface mineralogy, which suits clay and alteration distinctions. Finally, laser and induction sensors add 3D shape and conductivity information.
Combining sensors on one platform
Steinert’s KSS is designed around combining sensors. It pairs one of three main sensors (XRT, XRF or NIR) with any of three additional sensors (induction, 3D laser, colour). For mining, the typical configuration is XRT combined with colour, laser and induction. This flexibility exists because real ore bodies rarely separate on one property. For instance, a copper porphyry might need XRT for bulk density plus a second sensor for the edge cases. Spectral sensing can add mineralogy too, as our page on SWIR mineral analysis explains.
Training the fusion model
The AI layer makes multi-sensor fusion work in real time. Hand-coded rules struggle to weight several readings per particle. Instead, modern platforms use trained classification models that learn the fused decision boundary from labeled test data. That is also why a proper sortability trial matters more than spec sheets. Both TOMRA and Steinert run test centers for these trials.
How Ore Sorting Sensors Fit the Wider Flowsheet
Sensors only help if the sorter sits in the right place. Most sorters handle coarse crushed rock, so they follow crushing and screening and come before grinding. The overall case for AI ore sorting rests on rejecting waste before it consumes grinding energy and water.
Sensor data also has value beyond the sorter. Particle-level readings describe the ore feeding the plant. As a result, they can support blending and stockpile decisions downstream. However, expect the sensor mix to need revisiting as mining moves into new ore types.
FAQs
Which ore sorting sensors are most common in mining?
X-ray transmission (XRT) is the workhorse sensor for mineral sorting, especially for sulphide ores. Optical colour sensors, NIR, 3D laser and induction sensors are common additions. Many platforms combine XRT with one or more of these.
Why combine several ore sorting sensors?
Real ore rarely differs from waste in only one property. A second sensor catches particles that fool the first, such as rocks with misleading surface colour. AI classification then weighs all the readings together.
How do I choose ore sorting sensors for my deposit?
Start with the mineralogy and the physical contrast between ore and waste. Then run representative samples through a vendor test center. The trial results, not spec sheets, should drive the final sensor choice.
What role does AI play in sensor-based sorting?
AI turns raw sensor signals into a keep or reject decision for each particle. It learns the decision boundary from labeled test data rather than fixed thresholds. That matters most when several sensors must be read at once.
Try It With Geocluster
Matching sensor technology to your ore’s physical and mineralogical properties is a research question worth doing properly. Do not simply default to whatever a vendor pitches first. Geocluster can help you work through the deposit characterization needed to make that call with evidence. You can also read how the Geocluster Research Harness works.