Empowering Mining Operations: A Cutting-Edge Weakly Supervised Learning Approach for Iron Ore Feed Load Estimation

Accurate load estimation is crucial for optimizing mining operations, ensuring efficient resource allocation, and enhancing productivity. Traditional supervised learning methods require precise data labeling, which can be laborious and costly. In this study, we present a cutting-edge weakly supervised learning approach for real-time iron ore feed load estimation in mining operations. By leveraging readily available process data and weak labels, our framework enables accurate load estimation without exhaustive data labeling efforts. We integrate Convolutional Neural Networks (CNNs) and Recurrent Neural Networks (RNNs) to analyze real-time sensor data and historical process logs. The results demonstrate the effectiveness of weakly supervised learning, outperforming traditional supervised learning methods and empowering mining operations with data-driven decision-making capabilities.