Failures in critical industrial components (bearings, compressors, and conveyor belts) often lead to unplanned downtime, high costs, and safety concerns. Traditional diagnostic approaches underperform in noisy or changing environments due to heavy reliance on manual feature engineering and rule-based systems. In response, advanced machine learning, deep learning, and sophisticated signal processing techniques have emerged as transformative solutions for fault detection and predictive maintenance. To address the complexity of these advancements and their practical implications, this review combines analyses from large language models with expert validation to categorize key methodologies—spanning classical machine learning models, deep neural networks, and hybrid physics–data approaches. It also explores essential signal processing tools (e.g., Fast Fourier Transform (FFT), wavelets, and Complete Ensemble Empirical Mode Decomposition with Adaptive Noise (CEEMDAN)) and methods for estimating Remaining Useful Life (RUL) while highlighting major challenges such as the scarcity of labeled data, the need for model explainability, and adaptation to evolving operational conditions. By synthesizing these insights, this article offers a path forward for the adoption of new technologies (deep learning, IoT/Industry 4.0, etc.) in complex industrial contexts, anticipating the collaborative and sustainable paradigms of Industry 5.0, where human–machine collaboration and sustainability play central roles.

Chalcopyrite, the main source of copper worldwide, faces challenges in its flotation due to the complexity of its surface interactions and its coexistence with other minerals. Given the need for papers that show the current state of knowledge and new trends, this article reviews the developments of chalcopyrite flotation, with a focus on molecular design. A comprehensive bibliography search was conducted using keywords and specific queries in the Scopus database, applying inclusion and exclusion criteria to select the most relevant articles. The results were structured in three research periods, according to temporal and thematic criteria. The first period approaches the fundamentals of the process, considering variables as reagent dosage, surface chemistry and the influence of metal ions on recovery and selectivity. The second period explores the analysis and measurement techniques for the development of more selective and sustainable reagents. The third period analyzes the integration of advanced tools, such as molecular dynamic simulations and machine learning, into the understanding of adsorption mechanisms and custom reagent design. It is expected that this work will become a theoretical reference in future research and for mining companies that intend to innovate in their copper flotation and recovery processes.

Conveyor belts are critical equipment in mining operations, where continuous and reliable material transport is essential for production efficiency. This systematic review aims to analyze computer vision-based techniques applied to conveyor belt condition monitoring. Following PRISMA guidelines, a search was conducted in the Scopus and Web of Science databases, and 80 studies were selected after applying predefined eligibility criteria. These studies were synthesized through quantitative bibliometric methods and structured qualitative thematic categorization. The findings reveal a significant increase in scientific output after 2020, as well as its geographic distribution and potentially the most influential contributions. The main research lines focus on damage detection, deviation detection, and foreign object detection. A clear transition is also observed from traditional image processing methods—such as filtering, segmentation, and geometric analysis—toward deep learning models, including YOLO, CenterNet, and hybrid architectures, with improvements in precision, speed, and stability. Nevertheless, challenges remain related to datasets representativeness, the heterogeneity of evaluation protocols, and variability in operational conditions. Finally, opportunities for advancement are identified through multimodal datasets, adaptive models, and lightweight solutions that facilitate integration into asset management systems and support scalable industrial adoption.

The increasing scarcity of freshwater in mining regions of Chile has promoted the use of low-quality water as an alternative in flotation processes, significantly modifying their operating conditions. In particular, high salt concentrations and the presence of dissolved ionic species may interfere with the adsorption of collectors on chalcopyrite, thereby reducing its hydrophobicity. In this context, the present study analyzes the adsorption characteristics and the mechanistic role of selected representative ionic species on the chalcopyrite surface. To this end, simulations based on density functional theory (DFT) were employed to describe the interaction between the chalcopyrite (112) surface and Na⁺, Ca²⁺, Mg²⁺, and OH⁻ ions. After geometric convergence of the optimized structures was achieved, adsorption energies, charge redistribution based on Mulliken population analysis, and the final structural configurations were evaluated for each case. The results revealed clearly differentiated behaviors among the ionic species considered. The OH⁻ ion exhibited a localized and specific interaction with metal-centered sites. By contrast, Ca²⁺ and Mg²⁺ show stable adsorption near sulfur atoms, indicating a higher affinity that may lead to the occupation or blocking of active surface sites. Meanwhile, Na⁺ displays a weak interaction without inducing significant structural modifications. Overall, these findings provide an atomistic-level interpretation of how ionic species present in low-quality water can influence the surface reactivity of chalcopyrite under flotation operating conditions.

This fully synthetic study tests a structure-preserving surrogate on NRTL-generated phase equilibrium systems; it contains no experimental molecular validation. A symmetric neural potential represents $G^E/\left(nRT\right)$, from which activity coefficients, thermal derivatives, binodals, flashes, and tangent-plane-distance diagnostics are derived. Across ten binary training seeds, its median $\ln \gamma$ MAE was $1.16\times {10}^{-3}$, better than matched direct regression ($2.40\times {10}^{-3}$) but not Redlich–Kister ($9.65\times {10}^{-4}$) or correctly specified refitted NRTL ($2.36\times {10}^{-4}$). Gibbs–Duhem residuals decayed as $O\left(h^2\right)$ under grid refinement, confirming diagnostic truncation around an analytic identity. A 41,600-point binary hull gave endpoint MAE $2.31\times {10}^{-4}$. On 100 ternary type-I feeds, phase-count agreement with numerical reference labels was 98.0% (exact 95% interval, 93.0–99.8%), with two false negatives after recovery of one infeasible-floor solver failure. A 20-member ensemble showed pooled error ranking ($\rho =0.894$) but weak in-domain ranking ($\rho =0.164$); a temperature–distance heuristic was stronger ($\rho =0.914$), and nominal 90% conformal coverage fell from 88.0% in-domain to 0.0% under shift. Controlled flashes were 2.77× slower than NRTL. A hard thermodynamic structure is therefore valuable within neural modeling, but classical dominance, shift sensitivity, and absent molecular data bound the claim.