Conveyor belts are the most effective way to transport ore in a mining complex. The ore that comes from the mining areas can be heterogeneous in size and type. As the ore has to pass through several processing stages, online information about the ore type and degree of fragmentation can help improve mineral processing for both safety and efficiency. Current instrumentation systems are expensive and require frequent calibration and maintenance. This article presents a novel intelligent instrument for online recognition of type and degree of fragmentation. A 2-D light detection and ranging (LiDAR) sensor along with machine learning (ML) techniques was used to estimate the characteristics of iron ore particles on conveyor belts. An experiment was conducted using several types of ore and granulometry. Five ML models were compared by means of statistical methods, including average accuracy and normality and hypothesis tests. Among them, the random forest (RF) models achieved the highest rate of accuracy, 93.81% for ore type and 85.52% for degree of fragmentation. These models were improved by a voting mechanism that resulted in a reduction of classification errors of 93.3% for ore type and 99.2% for degree of fragmentation. These findings demonstrate that the system has the potential for improving mineral processing controls and heightening operational safety within the mining sector.
29 October 2024 in IEEE Latin America Transactions
Flooding is one of the most impactful natural disasters, causing significant losses and prompting extensive research into monitoring water levels in urban streams. Current technologies rely on pressure and ultrasonic sensors, which, while accurate, can be susceptible to damage from floods and are often costly. As an alternative, ground camera approaches offer a low-cost solution; however, most of these methods use raw images from the water stream and are sensitive to environmental factors. We address this gap with a dataset comprising a visual marker with black bars indicating the water level, which we refer to as "barcode panel". We employed various deep learning algorithms to predict the water level and compared their performance. The proposed approach was evaluated using classic classification and error metrics. The models demonstrated accuracy in detecting the water level. These promising results provide important insights for practical applications and future studies.
In this paper, we present a cost-effective system for monitoring and controlling alkaline electrolyzers, intending to improve hydrogen gas production on a laboratory scale. Our work includes two main innovations. Firstly, we suggest an approach to calibrate a standard air flow meter to accurately measure the flow of hydrogen-rich gas from electrolyzers, improving measurement accuracy while keeping costs low. Secondly, we introduce a unique cascade control method to manage hydrogen-rich gas production in the electrolyzer, ensuring precise control over gas flow rates. By combining affordable, energy-efficient devices with a PI control system, we achieve efficient gas production through electrolysis, replacing manual control approaches. Experimental results confirm the effectiveness of our cascade control method, demonstrating stable operation with minimal errors. These results provide a foundation for further research into control strategies to enhance the performance of electrolytic cells.
9 January 2026 in Engineering Applications of Artificial Intelligence
In the mineral sector, many processes use water for ore beneficiation processes. A lack of sensing or control of water content can lead to operational problems in various mineral processing operations, especially in ore transport. Current instrumentation systems are either slow or inaccurate. Therefore, a novel bench system was developed to address this gap by achieving a fast response time and improved accuracy. The developed instrument measures the ore moisture by using the real-dual-frequency method (RDFM) to assess the ore’s electrical conductivity and relative permittivity. Additionally, it takes into account the bulk density, the bench chamber level, and the compress torque. All these variables are used to create a tiny machine-learning (TinyML) model that evaluates the ore’s moisture with a low time response. This process is done while the ore sample is compressed to reduce air bubbles inside the samples and improve measurement. Experiments were performed using the bench system in a mining company’s physical analysis laboratory. The instrument was utilized to measure the moisture content in the ore, leading to the development of a dataset used to train and validate various tree-based tinyML models. The results indicate that ore compression enhances accuracy and that decision trees are effective for estimating moisture with a quicker response time.
The S11D mining complex in Brazil, situated in Pará state, extracts 20 million tons of iron each quarter. Connecting via a standard 802.11b/g/n wireless network is crucial for mine operations across vast distances. A local team employs a network monitoring tool called the Ekahau Site Survey to guarantee the proper functioning of the network. However, due to the harsh terrain and the dangerous nature of S11D operations, this tool fails to gather data from all points of interest, resulting in interpolated maps that may not accurately represent the network’s overall quality. In this work, we propose a platform that can be attached to mobile machines during operations to automatically collect network parameters, such as channelization, RSSI, latency, packet loss, and bandwidth, without requiring human intervention. Using these network data, we generate an RSSI map using Kriging, which the local team can use. Comparison tests conducted in the laboratory and the field demonstrate that the platform performs similarly to Ekahau in capturing network parameters, ensuring its use in day-to-day operations for mapping.