2 May 2022 in Environmental Pollution
1 November 2020 in Water Research
15 December 2023 in Chemosphere
9 December 2024 in Water Research
1 January 2022 in SSRN Electronic Journal
The successful preparation and identification of Keggin-structure Fe13 clusters in recent years further enriched the potential application scenarios of ferric coagulants. Comparing the coagulation efficiencies and mechanisms of Fe13 in the removal of nano/microplastics with conventional polymeric Al13 and monomeric Al/Fe, this work aimed to elucidate the coagulation behaviour of Fe13 compared with the traditional mono ferric coagulant, which has the coagulation applied bottleneck of quick and violet hydrolysis. The results showed that Fe13 has a similar electrostatic neutralization potential to Al13, which could keep a positively charged species, especially in acid conditions. The Fe13 species has a selective removal potential toward the microplastics with a polar functional group like ester. Moreover, Fe13 could hydrolyze to form active sol-gel hydroxides in neutral and alkalinity conditions, which is like the behaviour of traditional monomeric Fe coagulants but seldom restabilization. The electrostatic neutralization of Fe13 could enhance the removal of nano plastic from -25% to 75% compared with monomeric Fe at pH 4. The higher floc density as a monomeric Fe coagulant and better electrostatic neutralization potential of Keggin Fe13 posed a good prospect for Fe13 to replace the monomeric Fe coagulants in conventional coagulation.
20 February 2026 in SSRN Electronic Journal
High-accuracy leakage detection in urban water-supply networks is critical for water infrastructure safety. Although ground-penetrating radar (GPR) enables non-destructive inspection, automated interpretation remains challenging due to background interference and the high cost of labeled data. Motivated by practical need for convenient on-site deployment of GPR-based pipeline leakage detection and real-time capability of YOLOv8m, this study proposes an intelligent, low-cost leakage detection framework based on four-channel fused YOLOv8m model. Firstly, when evaluated on full test dataset, YOLOv8m demonstrates strong performance with a test mAP@0.5 of 0.972 and real-time inference speed of 111 FPS, highlighting its practical viability for on-site applications. Subsequently, grayscale B-scans are fused with instantaneous-amplitude RGB images to jointly exploit geometric contours and energy-anomaly cues, enhancing model sensitivity and cross-scenario generalization. To address label scarcity, a few-shot transfer learning strategy is implemented, pre-training the model on abundant simulated data followed by fine-tuning on limited real samples. Experimental results demonstrate that in cross-domain evaluations, four-channel strategy improves mAP@0.5 from 0.16 to 0.615 compared to single-channel models, while few-shot fine-tuning with only 15% real samples further raises metric to 0.859. Proposed framework offers a practical and field-deployable solution that reduces dependence on extensive labeled datasets and supports more sustainable water‑resource management.