Tomra Mining launched CONTAIN, a deep learning technology designed to improve the recovery of inclusion-type ores that traditional sorting methods struggle to detect. CONTAIN integrates seamlessly into TOMRA Mining’s ecosystem, showcasing the latest advancements in its AI-driven sorting platform.
Tomra’s software engineers and mining experts developed CONTAIN entirely in-house. The technology utilizes convolutional neural networks to analyze X-ray imagery in real time, categorizing rocks based on their likelihood of subsurface ore mineral inclusions. It excels in identifying complex mineralizations such as tungsten, nickel, and tin ores, which traditional sorting systems often misclassify or exclude.
Through its proprietary deep learning algorithms, CONTAIN transforms pattern recognition into operational value, analyzing the structure of each rock to detect subtle mineralogical patterns. The system assigns probability scores to rocks based on the presence of valuable metals like tungsten, nickel, or tin, enabling precise sorting decisions. Mining operations adjust strategies in real time, maximizing concentrate grade, minimizing material loss, or optimizing processing costs.
CONTAIN delivers industrial-scale performance by maintaining high accuracy even in dense, fast-paced input streams. It proves particularly effective in high-volume processing plants where speed, consistency, and recovery rates significantly impact profitability.
The technology successfully handles a wide range of ore grades, including low-grade, inclusion-rich rocks. It enables economically viable recovery of such ores by accurately evaluating the value of rocks and configuring sorting thresholds. Tomra trained the system on tens of thousands of ore samples, ensuring high effectiveness in classifying ores containing tungsten, nickel, and tin.
Field trials at Wolfram Bergbau in Mittersill, Austria, demonstrated the transformative impact of CONTAIN. The trials increased total plant throughput by eight percent, reduced ore mineral losses by 33%, and achieved the lowest tails grade to date. The technology identified tungsten-bearing inclusions undetectable by traditional sorting systems, improving concentrate quality and reducing ore mineral loss.
CONTAIN complements Tomra’s sensor-based sorting ecosystem, synchronizing with COM XRT and OBTAIN technologies for a comprehensive approach to ore processing. These technologies provide mining operations with unparalleled flexibility and precision in sorting inclusion-type ores, streamlining system scalability and protecting infrastructure investments.
More information is posted on www.Tomra.com/mining.
Comments