Abstract India is the third-largest seafood-producing country in the world. Shrimp aquaculture relies heavily on precise water quality monitoring to ensure optimal shrimp health and productivity. Shrimp prices have sharply declined since April 2022. In April 2022, the average price was $10.21 per kilogram. By the latest months of 2024, it has dropped to $6.76 per kilogram. Considering all these factors the research has focused on brackish water aquaculture monitoring. To demonstrate the shortcomings of traditional monitoring methods in brackish water prawn aquaculture, a study was conducted. The investigation revealed that farmers often stay up all night to maintain optimal water quality for healthy shrimp production, as ponds are sensitive to environmental factors such as temperature, pH, and humidity. Implementing an intelligent forecasting approach can address the complexity of crop monitoring issues. A hybrid intelligence mechanism can efficiently forecast and handle large volumes of streaming data. Dissolved Oxygen (DO) levels are critical indicators affecting water salinity and pollutant levels that impacts shrimp health. This study introduces a novel deep learning hybrid algorithm for real-time prediction of dissolved oxygen and quality in shrimp ponds using an IoT-based DO measurement kit. We utilize a uniform DO threshold of ≥5 ppm during the course of the study to determine healthy water conditions for shrimp development. The main objective of the model is to forecast dissolved oxygen (DO) concentration in shrimp ponds, with other environmental factors as secondary objectives. The proposed model integrates CNN and Long Short-Term Memory networks to capture spatial and temporal features of DO data. Results demonstrate the effectiveness of our approach, achieving high prediction accuracy compared to traditional methods. Specifically, our CNLSTMN architecture achieved an MSE of 0.065, RMSE of 0.255, and R 2 of 0.85, significantly outperforming linear regression (MSE: 0.120, RMSE: 0.346, R 2 : 0.65), decision trees (MSE: 0.090, RMSE: 0.300, R 2 : 0.75), and random forests (MSE: 0.075, RMSE: 0.274, R 2 : 0.80). These results demonstrate the superiority of our approach. It provides immediate alerts when DO levels deviate, enhancing environmental monitoring and management practices in shrimp aquaculture. The approach achieves an accuracy of 92.5 %.

Abstract In this study, the thermal-hydraulic performance of four finned heat sink models, namely square, cylindrical, conical, and pyramidal fins, was numerically investigated. The main objective of this work is to evaluate and compare the thermal performance of these fin geometries under identical operating conditions. Numerical analysis was conducted using the ANSYS Fluent solver, where the governing energy and Navier–Stokes equations were solved in conjunction with the k-ɛ turbulence model. The simulations were performed for Reynolds numbers ranging from 5,000 to 25,000, under a uniform heat flux condition of 5000 W/m 2 . The ideal fin spacing and length for the square fins were tested and applied as reference parameters for all fin geometries. The results indicated that cylindrical fins exhibit the best thermal performance, achieving the lowest heated wall temperature and the highest Nusselt number compared with other models. Unlike previous studies focusing on single fin geometries, this study provides a comparative assessment of multiple fin shapes, offering clearer insights into thermal-hydraulic performance and guidance for future research.

Abstract The proposed cascade control design for both minimum and non-minimum phase delayed time processes is based on the keeping the open loop transfer function same. The controller parameters are interms of the tuning paramter and process parameters for both primary and secondary time delayed processes. The proposed controller parameters having tuning paramters and are tuned based on the desired robust stability factor Ms. Numurous different order minimum and non minimum processes are tested by the proposed method and four examples are presented to show the effectiveness of the proposed method. The proposed method have evaluated based on the integral error creteria such as ITSE, ISE, ITAE and IAE values with desired robustness. The proposed method have compared with the recent literature methods and gives improved performance both unperturbed and purturbed processes. The proposed technique was validated on the cascade experimental setup and outcomes demonstrated significant improvements in performance.

Abstract Bioremediation is a sustainable and promising technology for the remediation of different kinds of pollutants, such as heavy metals, dyes, antibiotics, microplastics, and other contaminants present in industrial wastewater. These emergent toxic pollutants have adverse effects on human health. However, different conventional technologies such as filtration, ion exchange, precipitation, etc., are used to remove contaminants such as heavy metals and organic and inorganic pollutants. In this direction, the application of green materials, nanoparticles, and their composites, phyco-remediation, mycoremediation, and different kinds of bio(nano) sorbents such as biochar, hydrochar, chitin, and chitosan, etc., are used for the removal of hazardous pollutants from industrial wastewater. Green bio(nano) sorbent materials are eco-friendly, sustainable in nature, and offer enhanced adsorption efficiency and selectivity as compared to the conventional mode of wastewater treatment. A techno-economic analysis and a circular economy analysis are required for assessing bioremediation. The future perspective and challenges are addressed for the implementation of bioremediation-based technology. Integration of other techniques, such as artificial intelligence, advanced machine learning, and the Internet of Things (IoT) is used by researchers for the bioremediation of industrial wastewater. Thus, the application of green (nano)materials for industrial wastewater treatment is a sustainable environmental solution.

Abstract The application of machine learning algorithms in forecasting oil and gas production rates is critical for optimizing resource extraction and improving operational efficiency. Accurate predictions of future production rates enable operators to maximize yields, minimize downtime, and optimize resource management. This study proposes a new evolutionary Subtraction-Average-Based Optimizer (SABO), which updates searchers’ positions in the search space using the subtraction average of searcher agents. The SABO algorithm is integrated into each machine learning model such as Decision Trees, Random Forest, Extra Trees, Adaptive Boosting, Bagging, and Categorical Boosting, for performance improvement, and the resultant hybrid models are compared and examined using the test cases of the benchmark functions and the real-world dataset. The best performance belonged to the highest R 2 of 0.9327 and lowest RMSE of 2.0785 and MAE of 0.4665 for SABO-Bagging in the training set, and the test dataset recorded the best performance by the R 2 of 0.9267 and lowest RMSE of 2.2157 and MAE of 0.4558. These findings indicate that the SABO-Bagging hybrid model is the best performing model for the prediction of oil and gas production rates, and it has a competitive advantage compared to other machine learning methods for operational optimization in the energy industry.

Abstract This study explores advanced methods to improve the accuracy of hydrocarbon property predictions in oil reservoir fluids, utilizing novel data from an Iranian oil reservoir. The objectives include calculating molecular weights, specific gravity, and molar ratios of hydrocarbons (C11–C50) while evaluating the performance of models for critical properties and phase equilibria. Key approaches involve the gamma distribution and single carbon number (SCN) techniques for hydrocarbon characterization, alongside flash calculations using cubic and non-cubic equations of state, including Soave-Redlich-Kwong (SRK), Peng-Robinson (PR), and Perturbed Chain-SAFT (PC-SAFT). Empirical data were generated using the Pederson regression method and compared with predictions from the gamma distribution function. Results indicate that the gamma distribution method reduces prediction errors by 50 %, with deviations of 5.19 % compared to 10.39 % for the SCN method. Among the equations of state, PC-SAFT achieved the highest accuracy, predicting vapor phase mole fractions with a deviation of 0.0172 % and gas and liquid densities with deviations of 22.125 % and 0.24 %, respectively. The novelty of this study lies in integrating unique field data from an Iranian reservoir with advanced modeling techniques, providing a reliable framework for reservoir fluid analysis. These findings contribute to optimizing oil recovery and improving hydrocarbon prediction accuracy.

Abstract Photovoltaic (PV)-based multi-product systems are the most effective strategies for optimizing energy consumption and enhancing energy production efficiency, which integrates multiple energy production processes within a single framework for maximizing the utilization of available energy. This research aims to provide an inclusive analysis of the PV-based multi-product system performance, which integrates electrolysis, and desalination units for simultaneous generation of power, hydrogen, and freshwater. Therefore, the optimized Extreme Gradient Boosting (XGBoost) model was carefully developed and trained to ensure high levels of accuracy and precision in predicting the power, fresh water, and hydrogen outputs of the system under varying operational conditions. This optimized XGBoost model is achieved through the hyperparameter tuning by the HenryGas Solubility Optimization Algorithm, which ensures high levels of accuracy and precision. The day-by-day analysis of power and hydrogen generation demonstrated that in winter, a minimum PV power of 6,200 kW and a hydrogen mass flow rate of 65 kg were witnessed due to low solar radiation and short daylight hours. In contrast, summer and spring registered the highest PV power levels of 8,000 kW and 7,500 kW, respectively, and hydrogen mass flow rates of 75 kg due to high solar radiation and prolonged daylight hours, hence enhancing system performance.

Abstract Fuel cell energy systems have a potential as a clean and efficient power generation method, but suffer a high cost of production and engineering of fuel cells, multi-faceted operating principles, loss of performance over time, and vulnerability under the changing operation. Therefore, accurate predictions of energy output from fuel cell systems are critical for performance optimization, ensuring stable integration into the energy grids, and improving overall economic viability. This work presents advanced research on predictive modeling for fuel cell system energy forecasting, with a focus on increasing the accuracy and reliability of energy forecasting. The main innovation of this work is the use of the Bootstrap Aggregation (Bagging) ensemble farmwork that aggregate the outputs of multiple models such as Support Vector Regression, Long Short-Term Memory, Recurrent Neural Networks, and AdaBoost to mitigate individual model biases, reduce variance, and enhance generalization performance. The findings indicates that the Bagging algorithm performed excellently in improving the accuracy of energy output from fuel cell predictions by exploiting the benefits of multiple learners. It shows the best R 2 of 0.984352, with lowest MAE and RMSE equal to 17.00 and 23.65 on training set meaning that it perfectly fitted the predicted and actual values. Notably, Bagging continues with good generalization in the test set having the lowest MAE and RMSE equal to 44.60 and 60.77 and the highest R 2 value of 0.888772.

Abstract There is a massive, accelerating restructuring of the global energy sector as it attempts to meet the pivotal need to address climate change and earn a sustainable future. Electrolysis, using renewable energy to create green hydrogen, is critical as we move to sustainability. However, reliable prediction of hydrogen production based on the solar sources remains problematic. This paper recommends a breakthrough hybrid framework based on machine learning that combines Adaptive Boosting (AdaBoost) and the Black Widow Optimization Algorithm (BWOA) for forecasting Hydrogen Production Rate (HPR) and Photovoltaic Cell Power (PPV) in the proposed solar-powered system, considering meteorological data from Abu Dhabi. The model was designed and evaluated using standardized data in an 80:20 train-test ratio as preparation and evaluation. The R2 of the AdaBoost-BWOA model was outstanding at 0.998 for HPR and 0.998 for PPV, and RMSE value was as low as 11.282 and 2,114.667, respectively. The proposed best model (AdaBoost-BWOA) outperformed other hybrid approaches, such as MLP-BWOA, LGB-BWOA, DT-BWOA, PSO-AdaBoost, GA-AdaBoost, and DE-AdaBoost by providing significant improvements in both RMSE and R2 values. The refined system also estimated a maximum hydrogen yield of 542.91 kg per hour in Summer. This performance validates the reliable predictive capabilities of the model that are crucial for maximizing hydrogen production in solar systems and for practical purposes in the technologies of sustainable energy.

Abstract The temperature profile is a critical parameter in industrial batch cooling crystallization processes which directly impacts the crystal size distribution (CSD) and subsequent down streaming operations. Conventional cooling approaches, including natural and linear cooling, often fail to achieve optimal results. Linear cooling may trigger excessive primary nucleation by crossing metastable limits, while natural cooling leads to inconsistent nucleation and growth, resulting in smaller and non-uniform crystals. Hence, optimized cooling profiles are used to achieve the desired target CSD. This study employs reinforcement learning (RL) to optimize the temperature profile for batch cooling crystallization for two systems: paracetamol in water and potassium dihydrogen phosphate (KDP) in aqueous solution. RL dynamically adjusts the cooling trajectory by iteratively interacting with a process model, aiming to maximize mean crystal size while minimizing the coefficient of variation. A comprehensive mathematical model incorporating population balance equations, mass balance, and energy balance is developed to simulate the crystallization process. For the paracetamol system, the RL-optimization strategy resulted in a 107.5 % increase in mean crystal size compared to the natural cooling profile. In the case of the KDP system, a 12 % increase in mean crystal length is achieved relative to a linear cooling profile, along with a significant reduction in the coefficient of variation, indicating improved crystal size uniformity. The optimization results obtained using RL are also compared with that from a genetic algorithm for both cases, and RL demonstrated superior performance. This work underscores the potential of RL in advancing broader applications in chemical process optimization.