Accurate Performance Prediction Research Articles

Improving Student Performance Prediction Through Feature Selection: Insights from ‘Offee’ Assessment Data during the Covid-19 Pandemic

Post-pandemic student performance prediction has gained considerable attention within the academic community. Educational institutions worldwide have adopted blended learning approaches for delivering education and assessments. In this study, we harnessed data from the ‘Offee’ assessment platform based in Mumbai, India, spanning from September 2020 to October 2022. This dataset allows us to analyze the real impact of the Covid-19 pandemic on examinations.To tackle the challenges posed by high-dimensional data, we employed feature selection techniques to retain only the most significant features for predicting student performance. Specifically, we assessed the effectiveness of two decision tree-based methods, J48 classifier and Random Forest, in conjunction with the ranker search method using wrapper subset evaluation. The resulting feature set was then utilized to train and test a Support Vector Machine model for predicting student performance.Our findings reveal that both J48 and Random Forest feature selection methods yield impressive precision accuracy rates for student performance prediction, achieving 93.14% and 86.24%, respectively. In contrast, using the dataset without feature selection resulted in lower precision accuracy, standing at 79.9%, and consumed more time during model training and development.In summary, this study underscores the significance of feature selection in enhancing the accuracy of student performance prediction compared to utilizing the dataset without such feature curation.

Maximizing blue energy: the role of ion partitioning in nanochannel systems.

This study describes a numerical analysis on blue energy generation using a charged nanochannel with an integrated pH-sensitive polyelectrolyte layer (PEL), considering ion partitioning effects due to permittivity differences. The mathematical model for ionic and fluidic transport is solved using the finite element method, and the model validation is performed against existing theoretical and experimental results. The study investigates the influence of electrolyte concentration, permittivity ratio, and salt types (KCl, BeCl2, AlCl3) on the energy conversion process. The findings illustrate the substantial role of ion partitioning in modulating ionic concentration and potential fields, thereby affecting current profiles and energy conversion efficiencies. Remarkably, overlooking ion partitioning leads to significant overestimations of power density, highlighting the necessity of this consideration for accurate device performance predictions. This work introduces a promising configuration that achieves higher power densities, paving the way for the next generation of efficient energy-harvesting devices. The findings offer valuable insights into the development of state-of-the-art blue energy harvesting nanofluidic devices, advancing sustainable energy production.

Research on Performance Prediction Method Based on Gaussian Process Regression

Utilizing performance prediction models to predict students' graduation outcomes can offer crucial data support for academic early warning systems in higher education institutions, potentially becoming an essential tool for enhancing teaching quality. Therefore, in order to build a more accurate performance prediction model, a performance prediction method based on Gaussian process regression is proposed, and a series number of random experiments are carried out on the real performance data of seven majors such as broadcast and television directing major in a university to verify the effectiveness and feasibility of the method. The experimental results show that the proposed method has higher prediction accuracy and stability, and can provide effective reference for improving teaching quality.

Quality-Driven Regularization for Deep Learning Networks and Its Application to Industrial Soft Sensors.

The growth of data collection in industrial processes has led to a renewed emphasis on the development of data-driven soft sensors. A key step in building an accurate, reliable soft sensor is feature representation. Deep networks have shown great ability to learn hierarchical data features using unsupervised pretraining and supervised fine-tuning. For typical deep networks like stacked auto-encoder (SAE), the pretraining stage is unsupervised, in which some important information related to quality variables may be discarded. In this article, a new quality-driven regularization (QR) is proposed for deep networks to learn quality-related features from industrial process data. Specifically, a QR-based SAE (QR-SAE) is developed, which changes the loss function to control the weights of the different input variables. By choosing an appropriate inductive bias for the weight matrix, the model provides quality-relevant information for predictive modeling. Finally, the proposed QR-SAE is used to predict the quality of a real industrial hydrocracking process. Comparative experiments show that QR-SAE can extract quality-related features and achieve accurate prediction performance.

A Simplified Chemical Reactor Network Approach for Aeroengine Combustion Chamber Modeling and Preliminary Design

In a time when low emission solutions and technologies are of utmost importance regarding the sustainability of the aviation sector, this publication introduces a reduced-order physics-based model for combustion chambers of aeroengines, which is capable of reliably producing accurate pollutant emission and combustion efficiency estimations. The burner is subdivided into three volumes, with each represented by a single perfectly stirred reactor, thereby resulting in a simplified three-element serial chemical reactor network configuration, reducing complexity, and promoting the generality and ease of use of the model, without requiring the proprietary engine information needed by other such models. A tuning method is proposed to circumvent the limitations of its simplified configuration and the lack of detailed geometric data for combustors in literature. In contrast to most similar frameworks, this also provides the model with the ability to simultaneously predict the combustion efficiency and all pollutant emissions of interest (NOx, CO and unburnt hydrocarbons) more effectively by means of implementing a detailed chemical kinetics model. Validation against three correlation methods and actual aeroengine configurations demonstrates accurate performance and emission trend predictions. Integrated within two distinct combustion chamber low-emission preliminary design processes, the proposed model evaluates each new design, thereby displaying the ability to be employed in terms of optimizing a combustor’s overall performance given its sensitivity to geometric changes. Overall, the proposed model proves its worth as a reliable and valuable tool for use towards a greener future in aviation.

A Protocol for Predicting Lithium-Ion Battery Performances Reflecting the Three-Dimensional Structure of Electrodes

As social needs for higher capacity and longer life of lithium-ion batteries increase, shortening the period from research to development and commercialization of batteries has become an extremely important issue. To minimize the above period, a protocol for predicting the performances of (lithium-ion) batteries is highly urged for. Many reports have been published so far regarding the simulation technology of charge-discharge profiles of lithium-ion batteries.1,2 However, in most of simulations, the porous structure of lithium-ion battery electrodes is simplified into a 1D or 2D models. To achieve more accurate battery performance prediction, we adopted the actual 3D structure of the positive and negative electrodes of lithium-ion batteries in simulations.A plasma FIB SEM (Helious5, Thermofisher Scientific Inc.) was used to obtain the 3D structures of the positive and negative electrodes of the lithium-ion battery. The porosity and tortuosity of the electrodes were calculated by reconstructing a 3D image and performing image analysis. Based on this 3D structure, we also calculated the lithium ion conductivity in the electrode. The obtained 3D structural parameters were applied to the simulation of battery characteristics and compared with actual charge/discharge test results. We have also developed a performance prediction protocol that allows to reflect the measured physical properties of battery materials in the simulation.Figure 1 shows cross-sectional images of two different high-nickel-type positive electrodes prepared under different experimental conditions. It can be seen that the active materials are packed densely for both the electrodes, but the level of conductive carbon and binder dispersion is different. The particle size distribution of the active materials was also found to be largely different between the two samples. Structural parameters were obtained from the 3D structure by image analysis, and highly accurate performance prediction was performed. It can be predicted that the rate capability reflects the structural homogeneity. Also, the simulation revealed that the particle size distribution plays an important role for achieving favorable rate capability. Acknowledgement This work was financially supported by the NEXT Center of Innovation Program (COI-NEXT, Grant Number JPMJPF2016) of the Japan Science and Technology Agency. References (1) M Doyle , J. Newman, A. S. Gozdz, C . N. Schmts, and M. Tarascon, J. Electrochem. Soc. 1996, 143, 1890(2) G. Li and C. W. Monroe, Annu. Rev. Chem. Biomol. Eng. 2020, 11 277. Figure 1

A Novel Circuit-Based Model of Electrodialysis Avoiding the Use of Empirical Parameters

Electrodialysis (ED) is an emerging membrane-based electroseparation technology for desalination, water treatment, and ionic pollutant removal. Compared to conventional separation methods such as reverse osmosis, ED offers several advantages including lower energy consumption, higher selectivity, and lower fouling propensity. Further, coupling ED with renewable electricity sources enables it to be far more sustainable than competing technologies (reverse osmosis and evaporation) across a wide range of scales. However, the design and operation of ED systems are still challenging due to the complex interactions between the membrane, the electrolytes, and the applied electric field. To address this issue, advanced process models are required to provide accurate predictions of ED performance and to guide system optimisation. Existing ED models tend to be highly dependent on empirical parameters and thus are only applicable to a narrow range of process conditions and require several data sets to validate. Therefore, in this work we aimed to develop a glabal model of ED without reliance on experimental training data.Herein, we present a novel circuit-based model for ED. The model considers the membrane stack as a series of resistors, where the membranes and electrolytes are represented as separate resistive elements. Crucially, the membrane resistance is calculated directly from the electrostatic interaction between ions and fixed charge groups, rather than from manufacturer data or empirical sub-model. This allows for consideration of the ion identity and electrolyte concentration when determining the membrane resistance. Ohm's law is used to relate the applied voltage and stack electrical resistance to a current density, which is then converted to a flux using Faraday's law and current efficiency model. Historically, current efficiencies are calculated from transport numbers provided from membrane manufactures and assumed constant. A novel model for the current efficiency has been developed by us in which the current efficiency is a function of the trans-membrane concentration difference. A space-wise material balance is employed to determine the concentration profile along the stack, while a time-wise material balance tracks the changes in reservoir concentration for recirculating batch experiments.To validate the model, desalination experiments were conducted on a PC BED 1-4 recirculating batch system and compared the model predictions with experimental data. The model demonstrated excellent matching on a range of variables across a range of conditions, including current density, current efficiency, and ion concentration.The proposed model has various applications in ED process modelling, optimisation, and economic analysis. It can be used to evaluate the impact of different design parameters such as membrane thickness, membrane charge density, and flow rate on system performance as well as to optimise these.In summary, the circuit-based model presented in this work offers a robust and versatile tool for ED process simulation and optimisation, which can be used for effective and efficiency desalination and water treatment. Two major advancements presented include models for the membrane electrical resistance and current efficiency, which have historically been considered constant. Figure 1

A holistic framework for improving the prediction of reverse osmosis membrane performance using machine learning

Accurate prediction and modeling of RO membranes performance is crucial in desalination processes for proper process control and operation. Existing models do not consider all process parameters, leading to less understanding of the parameter's importance. In this study, 5 non-ensemble and 7 ensemble machine learning models were employed to predict the performance of RO membrane. Data from a modern RO desalination plant in the UAE was utilized for the models' building. Thirteen input parameters, including operational parameters, water characteristic parameters, and time-dependent parameters, were used to predict salt rejection. The results suggested that ensemble techniques are more capable of predicting the performance of RO membranes. Among ensemble methods, the XGBoost model was found to outperform other models. Recursive feature elimination was integrated with Shapley additive explanation analysis to gain insights into the most influential predictors and confirm the model's ability to comprehend the RO membrane mechanism. The findings highlighted that five parameters are critical for predicting RO membrane performance and could be prioritized for future monitoring to provide timely membrane performance warnings: the membrane's age, feed water temperature, pressure, feed water flow, and chloride concentration. It also indicated that maintaining lower temperatures, increasing feed water pressure, and increasing feed flow can improve process efficiency. The optimal XGBoost model was found to have an outstanding predictive performance with a high R2 (94.75) and a low RMSE (0.181). Ultimately, the framework proposed by this study can serve as a tool to simplify and understand complex membrane processes.

Experimental and CFD modeling of a progressive cavity pump (PCP) using overset unstructured mesh part 2: Two-phase flow.

In oil well operations, accurate modeling of two-phase liquid-gas flow in Progressive Cavity Pumps (PCPs) is vital. This study aimed to comprehensively analyze such flow in a metal-metal PCP using computational fluid dynamics (CFD) simulations validated by experimental data. The methodology applied to the High-Resolution Interface Capturing (HRIC) scheme is to compute the convective transport of phase fractions, yielding insights into the intricate dynamics of these flows. Validation was achieved using laboratory-acquired data and comparisons with extant publications. The findings confirmed the CFD simulations' alignment with field data for two-phase liquid-gas flows. Gas compression and effective viscosity gradient were crucial for accurate pump performance prediction. An investigation into the “gas seal” phenomenon using the Eulerian Multiphase Model elucidated the distinct separation of gas and liquid phases. The influence of the gas volume fraction on pump metrics like torque, power, and efficiency was explored, with CFD predictions exhibiting less than 8% error. The study underscores the pivotal role of CFD tools in predicting pump behavior under diverse conditions. It advances our understanding of two-phase flows in oil wells, especially regarding the gas's impact on pump components. This research offers the oil industry a refined modeling tool, enhancing the comprehension of oil well pump dynamics.

Thermal-Economic Analysis of an Organic Rankine Cycle System with Direct Evaporative Condenser

The organic Rankine cycle (ORC) system for power generation has proven to be an effective technology for low-temperature waste heat utilization. Accurate prediction and comprehensive comparison of system performance under different conditions are necessary for the development and application of suitable ORC configurations. This paper proposed an organic Rankine cycle (ORC) system using a direct evaporative condenser to realize performance enhancement and analyzed its dynamic performance based on the actual climatic condition, which is beneficial for the performance optimization of this system. This study begins with an introduction to the thermal economics model of the proposed system and evaluates the performance of the system based on the 3E (energy, exergy, economy) analysis method. Secondly, four candidate working fluids were compared and analyzed, leading to the selection of R142b as the best working fluid for the proposed system. Finally, the dynamic performance of the proposed system using the working fluid of R142b was analyzed based on the hourly environment temperature. The result showed that the net thermos-electric conversion efficiency of the system was negatively correlated with the ambient wet-bulb temperature. The annual average exergy efficiency of the system is about 65.79%, and the average exergy loss of the heat absorption unit, evaporative condenser, pump, and expander account for 61.07%, 6.92%, 2.99%, and 29.01% of the exergy loss of the system respectively. In the case 8760 h of operation per year, the payback period of the proposed ORC system using direct evaporative condenser is about 2.14 years.

A computer code for optimizing the neutronics model parameters based on results of reactor physics experiments

The paper describes in brief the functional capabilities of a computer code for optimizing the neutronics model parameters (neutron data, technological parameters, and their covariance matrices) based on results of reactor physics experiments using conditional nonlinear multi-parameter optimization algorithms. The code’s application scope includes adjustment of neutron constants, technological parameters and their covariance matrices based on integral measurement results, formulation of requiremen117198ts with respect to the neutron data uncertainties for achieving the target accuracies in calculation of the reactor functionals, and estimation of the reactor performance prediction accuracy, as well as the informativity and similarity metrics of reactor physics experiments with respect to each other and in relation to the target reactor system. The paper also considers some examples of using the code to refine the neutronics models of nuclear reactor and fuel cycle systems based on results of reactor physics experiments.

A novel generalized source-free domain adaptation approach for cross-domain industrial fault diagnosis

Domain adaptation has been widely applied in data-driven fault diagnosis tasks to address the domain shift problem between source and target data. However, conventional domain adaptation methods require both domains to be known, which is not always feasible due to privacy concerns and big data transmission. To overcome this limitation, a dedicated method called source-free domain adaptation (SFDA) has been developed to ensure reliable performance without relying on source data during target model adaptation. SFDA can achieve accurate classification tasks under domain shift problems and source data-free scenarios. We propose a generalized source model with manifold Mixup data augmentation and label smoothing techniques to avoid overfitting during the source model training. Based on this model, a novel self-training framework is proposed to implement the domain adaptation task and achieve accurate prediction performance. The experimental results from three real-world datasets demonstrate the effectiveness of the proposed approach.

Metaheuristic learning algorithms for accurate prediction of hydraulic performance of porous embankment weirs

A porous weir is an environmentally friendly structure with minimal negative environmental impact. Due to the complex flow mechanism around porous weirs, it is difficult to provide a general empirical relationship to estimate the free and submerged discharge coefficients and threshold submergence or modular limit. Hence, the current study attempts to introduce an efficient alternative approach leveraging on multilayer perceptron (MLP) coupled with generalized normal distribution optimization (GNDO), Runge-Kutta optimization algorithm (RUN), slime mould algorithm (SMA), whale optimization algorithm (WOA), and grey wolf optimizer (GWO). These models' robustness and accuracy are investigated to solve the free discharge coefficient (FDC), modular limit index (MLI), and submerge discharge reduction factor (SDRF) of the porous embankment weirs (PEWs) modeling problem where 6514 datasets are studied experimentally for free, modular limit, and submerged flow conditions. The performance of developed metaheuristic-MLP models is compared with a gradient algorithm-based MLP, namely Levenberg-Marquardt (LM) model. The obtained results from analysis of different performance indicators demonstrate that the GNDO-MLP outperforms other developed MLP models. The results demonstrate that the metaheuristic-based MLP models are more reliable in forecasting FDC, MLI, and SDRF of PEWs than the ML-MLP model. To measure how much the MLP model improved when combined with meta-heuristic algorithms, an aggregated statistical index called objective function was applied. The results of this index showed that the MLP model with GNDO performed better than the MLP models with WOA, SMA, GWO, RUN, and LM by 82%, 66%, 69%, 69%, and 96% for FDC, by 41%, 13%, 9%, 8%, and 48% for MLI, and by 64%, 20%, 8%, 4%, and 69% for SDRF condition, respectively. Furthermore, the results of the uncertainty analysis, using the P-factor and R-factor, demonstrated that GNDO-MLP had lower uncertainty compared to other models. The robustness of the GNDO-MLP model indicates that it can be used for further research on the probabilistic analysis of the PEWs.

Modeling strain-induced martensitic transformation in austenitic stainless steels subjected to cryogenic temperatures using incremental mean-field homogenization schemes

Austenitic stainless steels are commonly used as structural materials in high-field superconducting magnet systems because they retain high strength, ductility, and toughness at very low temperatures, and they are paramagnetic or antiferromagnetic under the Néel temperature in their fully austenitic state. However, they are susceptible to strain-induced martensitic transformation, especially at cryogenic temperatures, which modifies the material properties, induces volume changes and additional strain hardening, and leads to ferromagnetic behavior. Thus, accurate predictions of the structural performance of these materials at very low temperatures are of great interest for the conception and design of these cryo-magnetic systems. In this paper, we propose an adequate constitutive model for the evolving bi-phase material—austenite and martensite—based on a Hill-type incremental formulation. Two different versions of the model are proposed based on the linear mean-field homogenization scheme: Mori-Tanaka and Self-Consistent. Moreover, a rate-independent nonlinear mixed kinematic-isotropic hardening law is used for each phase, and the martensitic transformation is described by the nonlinear kinetic law proposed by Olson and Cohen (1975). The constitutive model is implemented in ABAQUS/Standard through a UMAT user subroutine, for which a return mapping algorithm based on the implicit backward Euler integration scheme is used and a closed-form expression of the consistent Jacobian tensor is provided. The Mori-Tanaka and Self-Consistent approaches are evaluated in terms of their ability to describe the mechanical behavior of the bi-phase aggregate by comparing the predictions of the homogenization schemes with unit-cell finite element calculations with an explicit description of the martensite inclusions and the austenite matrix. The comparison is carried out for different stress states with controlled triaxiality and Lode parameter under monotonic and cycling loading, paying special attention to the evolution of the mechanical fields in each phase. The unit-cell calculations are performed for both constant and evolving martensite volume fractions. In addition, numerical simulations of tensile tests on samples subjected to different initial temperatures are carried out for the transforming bi-phase material and the results are compared to experimental data for AISI 304L and AISI 316LN steels.

Forecasting Monthly Fish Landing in East Coast Peninsular Malaysia Using SARIMA and Artificial Neural Network

Marine resources management has become vital nowadays due to the increases of awareness of these resources becoming limited. With regards to fish landing, it has been a crucial for the managers to make better decision making particularly the amount of fish stock need to be kept for future food security since fish is one of the main source of protein supply to the country. Hence, the objective of this study is to predict the Monthly Fish Landing in East Coast Peninsular Malaysia using Box–Jenkins Autoregressive Seasonal Autoregressive Integrated Moving Average (SARIMA) and Artificial Neural Networks (ANN). The data obtained were split into two sets with ratio of 80% and 20%, training data was set from January 2012 until December 2019 while testing data from January 2020 to December 2021. The SAS and MATLAB were used to analyse and obtaining the best model of prediction. Based on Box-Jenkin approach, the best model of SARIMA was observed in SARIMA (1,1,0)(0,1,1)12 with AIC and SBC reach the minimum value at AIC=20.67209 and SBC=20.77894. By taking lag variables of SARIMA model as the input for ANN, we then developed SARIMA-ANN model. The findings showed that the integration of SARIMA–ANN model proven to be a feasible option for more accurate prediction of monthly fish landing performance in East Coast Peninsular Malaysia.

On falling film evaporator – A review of mechanisms and critical assessment of correlation on a horizontal tube bundle with updated development

Optimization of the energy efficiency and reduction of the greenhouse gas emission for the heat pump system is imperative to meet the net-zero goals at 2050. Chillers with falling film evaporator design not only possess better system performance but also contain less refrigerant inventory. Hence, accurate prediction of the evaporator performance is pivotal especially when charged with low-GWP refrigerant. The present study reviews the correlations for falling film evaporator with a horizontal tube bundle configuration. The major efforts of this study include four tasks: (a) literature review of the experimental studies and available empirical correlations; (b) comprehensive discussion of the falling film evaporation heat transfer mechanism; (c) development of a new rationally based correlation based on available literature; and (d) comparison of different correlations based on the existing data. The collected data includes 4114 data points from 8 sources, 6 refrigerants (R-600a, R-290, R-245fa, R-134a, R-1234ze(E), R-123), 5 types of the tubes (Plain, Turbo-GII-HP, GEWA-B, Low-fin, High-Flux), liquid Weber number from 2.2 × 10−6 to 0.7, imposed vapor Weber number from 0 to 37.2, heat flux from 2.5 to 151.5 kW/m2, and film Reynolds number from 1 to 3159.8. The new correlation gives a MAD of 28.3%, and an R2 of 0.86. Yet, the developed correlation considers various heat transfer mechanisms, including the transition point from falling film evaporation to the nucleate boiling, local evaporation, dry-out, mist flow, imposed flow, and enhanced tube effects.

Spatial variability of nitrate leaching and risk assessment of nitrate contamination in the Ghiss-Nekor alluvial aquifer system (Northeastern Morocco) through Disjunctive Kriging

In this study, groundwater nitrate monitoring data from water supply and water management organizations were processed in a GIS environment to evaluate spatial patterns of nitrate occurrence and explore its correlation with the corresponding land uses in Ghiss-Nekor alluvial groundwater (northeastern Morocco). Land-use/Land Cover within the Ghiss-Nekor plain incorporated rangeland, built area, crop fields, water bodies, dense vegetation, and bare ground categories, encompassing respectively about 39.06 %, 36.69 %, 17.57 %, 4.63 %, 1.24 %, and 0.81 % of the total area. According to the World Health Organization (WHO), the selected thresholds are an acceptable concentration of nitrate for potable water (50 mg/L) and a human affected value (13 mg/L). Simple Kriging and Disjunctive Kriging (D.K.) geostatistical methods were compared with a deterministic interpolation method (GPI) to identify the convenient approach to predict nitrate concentrations at unsampled sites. D.K. provided more accurate predictive performance and highlighted that nitrate contamination is higher in wells in the southern half of the study area, where nitrate concentrations exceeded 50 mg/L in 29.2 % (33,3 km²) of the study area. Areas occupying 21.28 % of the total surface, with nitrate contents exceeding the human-affected value, are located near small industrial sites, residential septic tanks, and agricultural areas. No significant correlation was found between the nitrate concentrations and the depth of the wells. The probability of nitrate content exceeding the thresholds of 13 and 50 mg/L is respectively less than 0.5 in 61.35 % (69.94 km²) and 77.63 % (88.4 km²) of the total study area, whereas the maximum likelihood of nitrate contamination is found along uncontrolled urban discharges located within highly permeable deposits. The assessment also demonstrated that areas showing high predicted probabilities of nitrate contamination diminish with increasing threshold value. The present study shed light on the necessity of depicting the extent of polluted areas, particularly in vulnerable alluvial aquifers, where exposure to higher levels of drinking water contaminants threatens human health.

Temporal data-driven short-term traffic prediction: Application and analysis of LSTM model

With the development of the economy, the number of private cars has increased, and traffic congestion is becoming increasingly common. In order to prevent traffic congestion, short-term prediction of traffic flow is urgent. This article is based on the data of a Hangzhou elevated bridge from October 4th to 18th, 2015, with a timestamp of 5 minutes and a total of over 4000 pieces of data, to have the Long Short-Term Memory (LSTM) model to be trained and the following days traffic flow to be predicted. The outcomes show that the model has high predictive capability and can reliably forecast future short-term transportation passenger flow, accurately reflecting the trend of data changes. Compared with the Auto-Regressive Moving Average (ARIMA) method with a Mean-Square Error (MSE) of 36, the LSTM model has a MSE of 22, which indicates a smaller MSE, indicating a more accurate prediction performance of LSTM. The LSTM models prediction is beneficial for alleviating traffic congestion, providing great convenience for peoples daily travel, and greatly reducing their travel time.

Fast prediction method of propeller hydrodynamic performance based on SPF-LSTM

In order to realize fast and accurate prediction of hydrodynamic performance of UUV propeller, a hydrodynamic performance prediction model based on SPF-LSTM neural network was proposed in this paper, which combined with real propeller flow test and CFD numerical simulation technology. The model comprehensively considered the test data of real paddle and the simulation data based on CFD numerical simulation, and adopted the Sample penalty factor (SPF) to optimize the initial weight of the model, adjust the sensitivity of the model to different samples, and further improve the prediction accuracy. The experimental results show that the prediction results of this model are in good agreement with the test data of real propeller, and the calculation period is very short and the efficiency is high, which meets the requirement of real-time and accurate prediction of the open water performance of propeller.

Impact of nanoplastics on membrane scaling and fouling in reverse osmosis desalination process

Nanoplastics (NPs) are a prevalent type of emerging pollutant in marine environment. However, their fouling behavior and impact on reverse osmosis (RO) membrane performance remain unexplored. We investigated the relationship between polystyrene (PS), one of the most abundant NPs, with silica scaling and humic acid (HA) fouling in RO. The results demonstrated that the surface potential of NPs played an important role in the combined scaling and fouling process. Compared with the negatively charged NPs (original PS and carboxyl group modified PS, PS−COOH), the amino-functionalized PS (PS−NH2) with positive surface charge significantly accelerated membrane scaling/fouling and induced a synergistic water flux decline, due to the strong electrostatic attraction between PS−NH2, foulants, and the membrane surface. The amino groups acted as binding sites, which promoted the heterogeneous nucleation of silica and adsorption of HA, then formed stable composite pollutants. Thermodynamic analysis via isothermal titration calorimetry (ITC) further confirmed the spontaneous formation of stable complexes between PS−NH2 and silicates/HA. Our study provides new insights into the combined NPs fouling with other scalants or foulants, and offers guidance for the accurate prediction of RO performance in the presence of NPs.