Aquacultural Engineering最新文献

{"title":"Quantum-Inspired Hybrid Bald Eagle-Ukari Algorithm with Reinforcement Learning for cybersecurity optimization in smart aquaculture systems: A novel bio-inspired intrusion detection and prevention framework","authors":"Wael Said , Mohssen Elbagory , Faizah Amer Altihani , Ibrahim Mohamed , Mahmoud El-Sharkawy , Ashwaq M. Alnemari , Farahat S. Moghanm , Mostafa Elbaz","doi":"10.1016/j.aquaeng.2026.102703","DOIUrl":"10.1016/j.aquaeng.2026.102703","url":null,"abstract":"<div><div>This study introduces a novel Quantum-Inspired Hybrid Bald Eagle-Ukari Algorithm with Reinforcement Learning (QI-HBEUA-RL) for cybersecurity optimization in smart aquaculture systems, addressing critical vulnerabilities in IoT-enabled fish farming networks that threaten production efficiency, fish health, and food security. The proposed framework synergistically integrates quantum computing principles (superposition and entanglement), bio-inspired metaheuristics (Bald Eagle Search and Ukari Algorithm), and Q-learning-based reinforcement learning to achieve superior intrusion detection and prevention performance. A comprehensive multi-objective optimization formulation simultaneously maximizes detection accuracy while minimizing false positives, computational overhead, energy consumption, and response time—critical factors for resource-constrained aquaculture IoT devices. The algorithm was validated using the publicly available CICIoT2023 dataset from the Canadian Institute for Cybersecurity (<span><span>https://www.unb.ca/cic/datasets/iotdataset-2023.html</span><svg><path></path></svg></span>), containing 33 attack types across 7 categories (DDoS, DoS, Reconnaissance, Web-based, Brute Force, Spoofing, and Mirai attacks) captured from real IoT devices. Statistical validation through 30 independent runs demonstrated exceptional performance: detection accuracy of 99.87 % (+4.23 % improvement over state-of-the-art), false positive rate of 0.043 % (-68.5 % reduction), detection latency of 1.34 ms enabling real-time capability, F1-score of 99.84 %, and statistically significant superiority (p < 0.001, Wilcoxon test) over six competing algorithms (NSGA-II, MOEA/D, MOPSO, MOGWO, MOHHO, standard BES) with 7.42 % hypervolume improvement, 29.35 % reduction in inverted generational distance, and 19.49 % better solution spacing. The algorithm achieved 47.3 % computational complexity reduction compared to deep learning approaches, enabling edge deployment on low-power aquaculture sensors, while generalization validation across seven benchmark datasets (NSL-KDD, UNSW-NB15, Bot-IoT, ToN-IoT, Edge-IIoTSet, MQTT-IoT-IDS2020, IoT-23) confirmed robustness with 3.8–6.2 % average accuracy improvements. Three real-world case studies validate practical implementation: offshore fish farm in Norway with 2500 sensors achieving 99.92 % threat detection accuracy and preventing $2.3 million annual losses from cyber-physical attacks, recirculating aquaculture system in Singapore with 15 tanks reducing attack response time from 8.5 s to 1.2 s, and shrimp farming operation in Thailand with 450 underwater sensors achieving 94.7 % zero-day attack detection through adaptive learning mechanisms. Economic analysis demonstrates rapid payback periods (12–18 days) and annual cost savings of $185,000-$890,000 through prevention of production losses, equipment damage, and data breaches. This research makes ten groundbreaking contributions including first application of quantu","PeriodicalId":8120,"journal":{"name":"Aquacultural Engineering","volume":"113 ","pages":"Article 102703"},"PeriodicalIF":4.3,"publicationDate":"2026-03-30","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"146185017","PeriodicalName":null,"FirstCategoryId":null,"ListUrlMain":null,"RegionNum":2,"RegionCategory":"农林科学","ArticlePicture":[],"TitleCN":null,"AbstractTextCN":null,"PMCID":"","EPubDate":null,"PubModel":null,"JCR":null,"JCRName":null,"Score":null,"Total":0}

{"title":"Numerical and experimental studies on the flow field of a semi-submersible aquaculture platform","authors":"Kai Wang ,&nbsp;Yi-Han Wen ,&nbsp;Jia-Ming Liu ,&nbsp;Cheng Liang ,&nbsp;Shuo Huang","doi":"10.1016/j.aquaeng.2025.102665","DOIUrl":"10.1016/j.aquaeng.2025.102665","url":null,"abstract":"<div><div>To enhance the hydrodynamic design of semi-submersible aquaculture platforms and mitigate potential risks, this study focuses on quantifying the nonlinear effects of structural design and nets on the surrounding flow field by carrying out both scaled model experiments and numerical simulations. The porous media method integrated with the Shear-Stress Transport k-ω turbulence model is employed to characterize fluid behavior, with results validated against experimental measurements. Investigations are conducted under inflow angles of 0° and 90°, at velocities of 0.2–0.4 m/s, and for configurations of net-free and net-equipped platforms. Good agreement is achieved, with a maximum drag coefficient error of 15 % for the net-free platform and a velocity reduction percentage error of 10 % for the net-equipped case. The platform structure induces abrupt velocity variations in the surrounding near-field, with flow significantly altered both in proximity to the structure and within the enclosed internal regions due to structural interference. The influence of nets on flow attenuation intensifies with increasing inflow velocity, with maximum velocity reduction reaching 18.4 % at 0° and 9.7 % at 90°, while dissipating coherent vortices and resulting in homogenized vorticity distribution throughout the internal regions enclosed within the platform structure. Structural details are significant: chamfered cross-sections produce a wake width 0.2 times that of non-chamfered ones. Furthermore, traditional empirical formulas for velocity attenuation overpredict values by approximately 10 % at 0°. At 90°, these formulas fail to capture the flow behavior, as local flow acceleration may occur due to structural interference. These results provide critical insights for safety assessments and design refinements of semi-submersible aquaculture platforms.</div></div>","PeriodicalId":8120,"journal":{"name":"Aquacultural Engineering","volume":"113 ","pages":"Article 102665"},"PeriodicalIF":4.3,"publicationDate":"2026-03-30","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"145610556","PeriodicalName":null,"FirstCategoryId":null,"ListUrlMain":null,"RegionNum":2,"RegionCategory":"农林科学","ArticlePicture":[],"TitleCN":null,"AbstractTextCN":null,"PMCID":"","EPubDate":null,"PubModel":null,"JCR":null,"JCRName":null,"Score":null,"Total":0}

{"title":"Evaluating a dual-energy X-ray absorptiometry protocol to determine the body composition of Atlantic Salmon (Salmo salar)","authors":"Michael J. Salini ,&nbsp;Rehan Mohammed ,&nbsp;Kristy DiGiacomo ,&nbsp;Jayde Kirkham ,&nbsp;David S. Francis ,&nbsp;Chethana Minoli Tissera ,&nbsp;Aaron Spence","doi":"10.1016/j.aquaeng.2025.102676","DOIUrl":"10.1016/j.aquaeng.2025.102676","url":null,"abstract":"<div><div>Knowledge of fish proximate composition is vital for farmers and researchers to make proactive management decisions and informed research interpretations. Currently, this is a cumbersome process requiring destructive sampling and specialised laboratory methods. This study strengthens the decision-making process for farmers and researchers by evaluating a rapid method for understanding fish proximate composition. We hypothesise that dual-energy X-ray absorptiometry (DXA) technology, designed and intended for human and livestock subjects, may also prove applicable for aquatic life. If proven, it enables a rapid, non-invasive and harmless method for analysing fish proximate composition and prediction of overall condition. In this study, we compare a DXA protocol with gold-standard “comparative slaughter techniques” (CST) of laboratory analysis for the examination of whole-body composition in Atlantic salmon (<em>Salmo salar</em>). The DXA results demonstrated a high degree of correlation (&gt;0.90) with CST analyses for total mass, fat mass and lean mass analysis. However, there was less agreement between the methods for bone mineral content analysis. Furthermore, regression equations generated from the dataset can predict CST equivalents for fat, lean and total mass with high precision. We conclude that DXA is a feasible substitute over traditional laboratory analysis methods and warrants further study.</div></div>","PeriodicalId":8120,"journal":{"name":"Aquacultural Engineering","volume":"113 ","pages":"Article 102676"},"PeriodicalIF":4.3,"publicationDate":"2026-03-30","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"145692425","PeriodicalName":null,"FirstCategoryId":null,"ListUrlMain":null,"RegionNum":2,"RegionCategory":"农林科学","ArticlePicture":[],"TitleCN":null,"AbstractTextCN":null,"PMCID":"","EPubDate":null,"PubModel":null,"JCR":null,"JCRName":null,"Score":null,"Total":0}

{"title":"Recovery of phosphate from actual aquaculture wastewater using calcined eggshell under continuous flow condition and its phytotoxicity","authors":"Jun Jie Lim ,&nbsp;Sumathi Sethupathi ,&nbsp;Sin Ying Tan ,&nbsp;Nor Ismaliza Mohd Ismail","doi":"10.1016/j.aquaeng.2026.102689","DOIUrl":"10.1016/j.aquaeng.2026.102689","url":null,"abstract":"<div><div>Oversupplying one of the non-renewable essential macronutrients, phosphorus, in the aquaculture industry has given rise to water pollution issues. Hence, recovering phosphorus from aquaculture wastewater is essential to prevent eutrophication in water bodies and to make effective use of the nutrient. Thus far, no one has reported on the utilization of calcined eggshell as an adsorbent to recover phosphate from actual aquaculture wastewater. Therefore, this study focuses on a sustainable approach in recovering phosphorus through the adsorption process using calcined eggshell (CES) as adsorbent. The effect of influent flow rate (10–50 mL) and dosage of adsorbent (0.2–1.8 g) was assessed via breakthrough curves. The breakthrough curve behaviour was determined through linear and non-linear models of Adam-Bohart, Thomas, and Yoon-Nelson model. The phytotoxicity of the spent adsorbents was assessed using the germination index. The optimized adsorption capacity was recorded as 21.2 mg/g of phosphate under a wastewater flow rate of 28 mL/min and using 1.17 g of CES. The ANOVA analysis (R² = 0.9766) proved that the experimental and predicted values significantly agree with each other. The non-linear Yoon-Nelson model was found to be the best model to demonstrate the column behaviour in this study with the significant value of the correlation coefficient. Only mild phytotoxicity effect was observed using spent CES. This study provides a sustainable solution for recovering phosphorus from aquaculture wastewater and the possible reutilization of the spent CES as a soil conditioner.</div></div>","PeriodicalId":8120,"journal":{"name":"Aquacultural Engineering","volume":"113 ","pages":"Article 102689"},"PeriodicalIF":4.3,"publicationDate":"2026-03-30","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"145973042","PeriodicalName":null,"FirstCategoryId":null,"ListUrlMain":null,"RegionNum":2,"RegionCategory":"农林科学","ArticlePicture":[],"TitleCN":null,"AbstractTextCN":null,"PMCID":"","EPubDate":null,"PubModel":null,"JCR":null,"JCRName":null,"Score":null,"Total":0}

{"title":"MFSD-YOLO: A multi-scenario fish small target detection method in aquaculture","authors":"Yupeng Mei ,&nbsp;Yingyi Chen ,&nbsp;Yi Liu ,&nbsp;HuiHui Yu ,&nbsp;Ling Yang ,&nbsp;Daoliang Li","doi":"10.1016/j.aquaeng.2025.102677","DOIUrl":"10.1016/j.aquaeng.2025.102677","url":null,"abstract":"<div><div>Fish target detection is of great significance in aquaculture. Through target detection, the growth and health status of fish can be monitored, which is the key to realize intelligent breeding. However, in the intensive aquaculture model, fish images have problems of blur, occlusion, deformation and large-scale changes. Here, we propose a multi-scenario fish small target detection method based on YOLOv8 (MFSD-YOLO). Firstly, we propose a fish image enhancement method based on generative adversarial network (FIE-GAN) for the situation of blur, noise and low contrast. Among them, we design a noise reduction network optimization generator to solve the problem of fish image noise. In addition, the occlusion between fish targets due to high-density aquaculture and the deformation due to fish movement can affect the accuracy of fish target detection. For this reason, we introduce a bidirectional feature pyramid network (BiFPN), which extracts contextual information by fusing the up and down sampled feature maps to improve the detection accuracy. Meanwhile, to address the problem of large-scale changes caused by fish movement, we add a small target detection layer to YOLOv8 to detect fish small target information, which enhances the combination of deep semantic information and shallow semantic information. The experimental results show that the precision, recall rate, [email protected] and [email protected]:0.95 of the proposed algorithm MFSD-YOLO on the FD-dataset are 92.2 %, 92.4 %, 96.7 % and 59.5 %, respectively, which are 0.9 %, 1.2 %, 1.4 % and 3.5 % higher than those of YOLOv8. In summary, our proposed algorithm has achieved good results and can better meet the needs of target detection in the field of aquaculture.</div></div>","PeriodicalId":8120,"journal":{"name":"Aquacultural Engineering","volume":"113 ","pages":"Article 102677"},"PeriodicalIF":4.3,"publicationDate":"2026-03-30","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"145748926","PeriodicalName":null,"FirstCategoryId":null,"ListUrlMain":null,"RegionNum":2,"RegionCategory":"农林科学","ArticlePicture":[],"TitleCN":null,"AbstractTextCN":null,"PMCID":"","EPubDate":null,"PubModel":null,"JCR":null,"JCRName":null,"Score":null,"Total":0}

{"title":"Assessing enhanced geosmin depuration in Atlantic salmon (Salmo salar) using TiO₂/UV + H₂O₂ advanced oxidation under variable recirculating aquaculture system conditions","authors":"Matthew Stromberg ,&nbsp;Eric Sibanda ,&nbsp;Adiba Hossen ,&nbsp;John Stubblefield ,&nbsp;Upal Ghosh ,&nbsp;Yonathan Zohar","doi":"10.1016/j.aquaeng.2026.102694","DOIUrl":"10.1016/j.aquaeng.2026.102694","url":null,"abstract":"<div><div>Off-flavor compounds such as geosmin remain a major bottleneck in recirculating aquaculture systems (RAS), prolonging depuration and increasing energy and water demand. This study evaluated a TiO₂/UV + H₂O₂ advanced oxidation process (AOP) for enhancing geosmin removal from both RAS water and Atlantic salmon (<em>Salmo salar</em>) tissue under variable system conditions. In freshwater trials, the AOP + H₂O₂ combination achieved 89 % geosmin reduction from water within 8 h, whereas peroxide or ozone alone were far less effective. In fish tissue, AOP + H₂O₂ accelerated geosmin depuration by 4.3 × relative to controls, with a time to 50 % removal, t₅₀ = 23 h vs 101 h) and by 3.1 × relative to H₂O₂ alone. Biphasic modeling revealed balanced fast/slow pool kinetics (58 %/42 %) compared to predominantly slow-pool behavior in controls (35 %/65 %). Treatment effects were significant after 72 h (p &lt; 0.05). Although ozone achieved partial removal, unsafe oxidative reductive potential (ORP) levels (&gt; 800 mV) limit its practical application. All treatments showed minimal effectiveness in saltwater. Collectively, these results demonstrate that TiO₂/UV + H₂O₂ AOP can substantially accelerate off-flavor removal in Atlantic salmon RAS, potentially shortening commercial depuration by 2–3 days and reducing operational energy and water use.</div></div>","PeriodicalId":8120,"journal":{"name":"Aquacultural Engineering","volume":"113 ","pages":"Article 102694"},"PeriodicalIF":4.3,"publicationDate":"2026-03-30","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"146185015","PeriodicalName":null,"FirstCategoryId":null,"ListUrlMain":null,"RegionNum":2,"RegionCategory":"农林科学","ArticlePicture":[],"TitleCN":null,"AbstractTextCN":null,"PMCID":"","EPubDate":null,"PubModel":null,"JCR":null,"JCRName":null,"Score":null,"Total":0}

{"title":"Biofouling community succession and cleaning strategies for aquaculture net cages in the western Yellow Sea","authors":"Chun-Wei Bi ,&nbsp;Ming-Shan Sun ,&nbsp;Liu-Yi Huang ,&nbsp;Xin-Xing You ,&nbsp;Shu-Yue He","doi":"10.1016/j.aquaeng.2026.102690","DOIUrl":"10.1016/j.aquaeng.2026.102690","url":null,"abstract":"<div><div>Marine biofouling threatens the sustainable development of cage mariculture by reducing water exchange, degrading water quality, and increasing health risks for cultured fish. Although summer biofouling is severe in the western Yellow Sea, research on fouling communities and control methods in this area remains limited. Here, we present the first seasonal dataset of biofouling succession for aquaculture net cages in the western Yellow Sea. Based on weekly in situ net-panel deployments from March to June at depths of 0.5, 2.5, and 4.5 m. A total of 27 taxa were identified. In March–April, Amphipoda became dominant at 0.5 and 2.5 m, a pattern linked to flow velocity. In June, Bryozoa and Hydrozoa became dominant at 4.5 m, correlating with elevated nitrogen and phosphorus concentrations. After 4 months of immersion, the average biofouling biomass reached ∼10 kg·m⁻², with net occlusion peaking at ∼ 90 %. Biofouling peaked in June, coinciding with a 50 % reduction in dissolved oxygen. Elevated nitrogen and phosphorus levels promoted Bryozoa growth, worsening water quality and amplifying environmental degradation through nutrient-fouling synergy. Based on these findings and offshore biofouling data, summer cleaning protocols are recommended: Monthly low-pressure jet cleaning during March–April, biweekly combined brush–jet cleaning from May, and weekly cleaning in June. Post-cleaning monitoring of high-risk regions (shallow inshore and deeper offshore layers) is recommended. These results provide region-specific guidance for biofouling management and support more sustainable mariculture practices in temperate coastal systems.</div></div>","PeriodicalId":8120,"journal":{"name":"Aquacultural Engineering","volume":"113 ","pages":"Article 102690"},"PeriodicalIF":4.3,"publicationDate":"2026-03-30","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"145973041","PeriodicalName":null,"FirstCategoryId":null,"ListUrlMain":null,"RegionNum":2,"RegionCategory":"农林科学","ArticlePicture":[],"TitleCN":null,"AbstractTextCN":null,"PMCID":"","EPubDate":null,"PubModel":null,"JCR":null,"JCRName":null,"Score":null,"Total":0}

{"title":"Evaluation of modified inoculum media for qualitative floc production and its effect on growth, immune response, digestive physiology, and antioxidant capacity of Anabas testudineus fry reared in biofloc system at varied stocking densities","authors":"Sushree Sangeeta Dey ,&nbsp;Asanaru Majeedkutty BabithaRani ,&nbsp;Naseemashahul Shamna ,&nbsp;Upasana Sahoo ,&nbsp;Gouranga Biswas","doi":"10.1016/j.aquaeng.2026.102687","DOIUrl":"10.1016/j.aquaeng.2026.102687","url":null,"abstract":"<div><div>Biofloc inoculum plays a crucial role during startup period in biofloc technology (BFT). So, an indoor trial was conducted to investigate the effect of inoculum media i.e., conventional (C) and novel (N) media, on microbial floc quality and <em>Anabas testudineus</em> fry reared in the biofloc system at three stocking densities i.e., 150 (T1), 250 (T2), and 350 (T3) fish/m³, following 2 × 3 factorial design. The analysis of the startup period revealed that the biofloc development phase reduced from 21 days using conventional media-based inoculum to 7 days using novel inoculum. After ten culture weeks, the novel media-based biofloc embodied considerably higher (<em>p</em> &lt; 0.05) nutritional content and bioactive compounds (total carotenoids, polyhydroxybutyrate, fatty acids) than the conventional one. The interactive effect of biofloc type and stocking density revealed significantly higher (<em>p</em> &lt; 0.05) weight gain, specific growth rate, biomass yield, and feed utilisation efficiency of fry in NT3 treatment. The N treatments remarkably enhanced (<em>p</em> &lt; 0.05) fish carcass <em>p</em>rotein and lipid contents. The improved immune response (serum protein, IgM, respiratory burst, and myeloperoxidase activity) and concurrently lower SOD, CAT, glucose, cortisol, triglyceride, and cholesterol levels in all N treatments substantiate reduced stress in fish compared to C treatments. NT3 treatment exhibited significantly increased (<em>p</em> &lt; 0.05) activities of amylase, <em>p</em>rotease, lipase, AST, and ALT. Conclusively, the novel media can be utilized in BFT for curtailing long startup constraints and enhancing biofloc quality and welfare of <em>A. testudineus</em> fry, whose density can be recommended at 350 fish/m³ for achieving maximum yield.</div></div>","PeriodicalId":8120,"journal":{"name":"Aquacultural Engineering","volume":"113 ","pages":"Article 102687"},"PeriodicalIF":4.3,"publicationDate":"2026-03-30","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"145973044","PeriodicalName":null,"FirstCategoryId":null,"ListUrlMain":null,"RegionNum":2,"RegionCategory":"农林科学","ArticlePicture":[],"TitleCN":null,"AbstractTextCN":null,"PMCID":"","EPubDate":null,"PubModel":null,"JCR":null,"JCRName":null,"Score":null,"Total":0}

{"title":"Numerical modeling of dynamic responses of aquaculture net cages in waves: Snap load prediction and the effect of nonlinear added mass","authors":"Kexin Zhang ,&nbsp;Baoyu Liu ,&nbsp;Hua-Dong Yao ,&nbsp;Hao Chen ,&nbsp;Jikang Chen","doi":"10.1016/j.aquaeng.2026.102701","DOIUrl":"10.1016/j.aquaeng.2026.102701","url":null,"abstract":"<div><div>This study proposes an alternative time-domain model to simulate an aquaculture net cage equipped with a floating collar of circular cross-section and flexible nets. The hydrodynamic loads on the floating collar are calculated using a partially nonlinear strip theory, which incorporates the draft-dependent impulsive added mass of the cross sections along the floating collar to capture structural vibrations caused by water entry and exit. The structural responses are obtained in the time domain based on modal superposition, with the wave-radiation memory effects accounted for through Cummins-type equations. Results for regular waves show that the nonlinear added-mass loads excite high-frequency responses near the natural frequencies of the higher-order modes of the floater. An inextensible truss model is applied to simulate large-amplitude deformation of the nets. It is found that the phase difference in motion between the floating collar and the sinkers leads to transient snap loads. While the impulse of these loads has a limited impact on the motion of the floating collar due to the short duration, their peak tension poses a severe threat to the structural safety of the net structures. This study also clearly identifies the inherent limitations of the inextensible truss model in predicting the peak values of such snap-induced tensions. To address this limitation, we propose a straightforward extension of the original inextensible truss model. The enhanced extensible truss formulation accounts for the transient elastic response of the net, allowing it to replicate the snap-load phenomenon and thereby provide more reasonable and accurate predictions of the critical snap loads.</div></div>","PeriodicalId":8120,"journal":{"name":"Aquacultural Engineering","volume":"113 ","pages":"Article 102701"},"PeriodicalIF":4.3,"publicationDate":"2026-03-30","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"146184828","PeriodicalName":null,"FirstCategoryId":null,"ListUrlMain":null,"RegionNum":2,"RegionCategory":"农林科学","ArticlePicture":[],"TitleCN":null,"AbstractTextCN":null,"PMCID":"","EPubDate":null,"PubModel":null,"JCR":null,"JCRName":null,"Score":null,"Total":0}

{"title":"Integrating adsorptive pre-treatment and deep learning modeling for nitrogen management in marine recirculating aquaculture systems","authors":"Hyunsoo Choi ,&nbsp;Subha M. Roy ,&nbsp;Taeho Kim","doi":"10.1016/j.aquaeng.2025.102683","DOIUrl":"10.1016/j.aquaeng.2025.102683","url":null,"abstract":"<div><div>Excess total ammonia nitrogen (TAN) and its oxidized forms, nitrite and nitrate, pose significant risks to marine aquaculture. This study evaluated the feasibility of adsorption-based technologies for nitrogen removal under seawater aquaculture conditions. Three synthesized adsorbents—chitosan hydrogel (Cht-HG), <em>p</em>-hydroxybenzoic acid-modified Cht-HG (M-Cht-HG), and alginate–hydrogel–metal–zeolite (Al-H-M-Z)—were benchmarked against a moving bed biofilm reactor (MBBR) and commercial zeolite. In pilot-scale fish tanks, Al-H-M-Z achieved rapid physicochemical TAN removal, with 75.6 % efficiency within 24 h, reducing concentrations from 0.023 mg L^–1 to 0.006 mg L^–1. Furthermore, it maintained TAN and nitrite levels below 0.20 mg L^–1 for a 7-day period, demonstrating stable performance. In contrast, M-Cht-HG was effective for only 2 days, after which TAN levels exceeded 1.62 mg L^–1. At an initial concentration of 300 mg L^–1, a long short-term memory (LSTM) deep learning model predicted that Al-H-M-Z would remove 80.1 mg L^–1 TAN (26.7 %) and 86.5 mg L^–1 nitrite (28.8 %) within 24 h. This study provides a foundation for integrating advanced adsorbents with predictive modeling to improve nitrogen treatment strategies for sustainable marine aquaculture.</div></div>","PeriodicalId":8120,"journal":{"name":"Aquacultural Engineering","volume":"113 ","pages":"Article 102683"},"PeriodicalIF":4.3,"publicationDate":"2026-03-30","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"145836929","PeriodicalName":null,"FirstCategoryId":null,"ListUrlMain":null,"RegionNum":2,"RegionCategory":"农林科学","ArticlePicture":[],"TitleCN":null,"AbstractTextCN":null,"PMCID":"","EPubDate":null,"PubModel":null,"JCR":null,"JCRName":null,"Score":null,"Total":0}