Back Propagation Neural Network Model Research Articles

Machine learning-based prediction of CoCrFeNiMo0.2 high-entropy alloy weld bead dimensions in wire arc additive manufacturing

Wire arc additive manufacturing (WAAM) is a promising method to fabricate large-sized components. By adjusting WAAM process parameters, dimensions of WAAM-produced weld beads can be optimized. To this end, the current study adopted the cold metal transition (CMT) process to produce different-sized beads of CoCrFeNiMo0.2 high-entropy alloy (HEA), varying WAAM parameters and linking them with the formed bead dimensions and bead-substrate contact angles. Three machine learning algorithms, namely back propagation neural network (BPNN), support vector regression (SVR), and random forest regression (RFR), were used to predict bead dimensions and contact angles under various WAAM parameters. The BPNN model has great prediction performance in the height of beads. The SVR model has the highest accuracy in predicting the width and cross-sectional area of beads. The RFR outperforms the other two models in contact angles prediction. This work not only provides a reference for the WAAM of high-entropy alloys, but also provides new ideas for predicting bead size in WAAM.

Predicting the Dry Density of Clay Soil Improved by Adding Glass Powder Using a Back Propagation Neural Network Model

Clay soil has undesirable engineering properties, which can compromise structural stability. This study aims to enhance the compaction properties of high-plasticity clay soil by adding glass powder using artificial intelligence (AI), specifically through the Back propagation Neural Network (BPNET), to accurately predict dry density. The model used influential factors, such as wet soil weight (Wnet), glass powder ratio (Wglass), and water content (ω %) as inputs, with dry density (γ) as the output. The model demonstrated high accuracy, achieving a Mean Squared Error (MSE) of 0.0000117 and a Mean Absolute Error (MAE) of 0.002849, reflecting its effectiveness in improving clay soil properties and supporting its stability.

Medium and long-term regional water demand prediction using Harris hawks optimisation–backpropagation neural network model

Precise water demand prediction is essential for the efficient allocation and rational utilisation of regional water resources. This study addressed the challenge associated with medium and long-term water demand prediction by introducing a novel coupled model, HHO-BPNN (Harris Hawks Optimisation–Backpropagation Neural Network). Principal component analysis was employed to reduce the dimensionality of potential water demand factors. The performance of the forecasting models was compared through mean square error (MSE), mean absolute percentage error (MAPE), mean absolute error (MAE), and coefficient of determination (R2). The findings indicated that the HHO-BPNN outperformed traditional methods, such as BPNN, support vector machines, and grey prediction model. The study utilised the sliding window method to predict water demand for the next 1, 3, and 5 years for five prefecture-level cities in northern Jiangsu Province, China. High prediction accuracy was achieved across various categories of water demand (agricultural, industrial, domestic, and ecological), with the overall accuracy being impressive at 97%. Additionally, the forecasts aligned well with local developmental plans, suggesting practical applicability for urban planning. This study elucidates the key drivers impacting water demand, providing an effective tool for regional water demand forecasting, facilitating efficient and precise water management and decision-making in the future.

Reliability analysis of DDV and SPM in BOP-based Weibull model and expand fault dataset using deep learning

The hydraulic systems of deepwater Blowout Preventers (BOPs) are crucial for ensuring the reliability and safety of offshore oil and gas extraction operations. A functional failure can lead to uncontrolled blowouts, resulting in casualties and significant economic losses on the rig. The Direct Drive Valve (DDV) and the Subsea Plated Mounted (SPM) valve are key components that help maintain the proper functioning of the hydraulic system in deepwater BOPs. This study begins by utilizing the Weibull analysis method to assess the reliability of the DDV and SPM valves using limited fault data samples. To enhance the accuracy of predictions, Weibull parameters are estimated through various methods, including Maximum Likelihood Estimation (MLE), Least Squares Estimation (LSE), and a combination of Correlation Coefficient Optimization with Support Vector Regression (CCO + SVR).Given the challenges in gathering extensive fault data for DDV and SPM valves—due to complex subsea environments, cost constraints, time limitations, and other factors—this study proposes a method employing a Back Propagation Neural Network (BPNN) model to augment the limited fault data samples. To ensure the reliable operation of the DDV and SPM valves, preventive maintenance cycles are established at 2840 and 7550 operations, respectively. At the same reliability level, as the number of operational cycles increases, the remaining service life of the valves gradually decreases, leading to a higher probability of failure over a shorter timeframe. The Mean Remaining Life (MRL) of the DDV and SPM valves, corresponding to different operational times, is analyzed, providing essential reference points for their usage and maintenance. When the extended data sample is utilized for reliability evaluation, the reliability characteristics of the small fault data samples are effectively reflected, and the parameter prediction error remains low. This indicates that the extended data sample is more suitable for reliability evaluation.

A Case Study Comparing Methods for Coal Thickness Identification in Complex Geological Conditions

This study compares the effectiveness of different methods for coal thickness identification, aiming to identify the most accurate approach and provide a reference for intelligent coalmine development. Focused on the No. 2 coal seam in a mining area in Shanxi, China, the analysis employs well log-constrained impedance inversion and seismic multi-attribute techniques. The results show that the back propagation (BP) neural network model, as part of the seismic multi-attribute approach, delivers prediction accuracy comparable to the well log-constrained inversion method. Specifically, after applying proper static corrections, a four-layer BP neural network was constructed using four optimized sensitive attributes as the input layer, achieving an error range of 0.11% to 1.36%, compared to 0.03% to 6.59% for the logging-based method. The BP neural network demonstrated strong applicability in complex geological environments. Empirical analysis further validated the BP neural network’s geological reliability and practicality in systematic coal thickness determination.

A reliability calculation method based on ISSA-BP neural network

PurposeTo address the shortcomings of the traditional back propagation (BP) neural network agent model, such as insufficient fitting accuracy and low computational efficiency, an improved method is proposed.Design/methodology/approachIn this study, an improved sparrow search algorithm (ISSA) is developed to optimize the reliability calculation of the BP neural network (ISSA-BP) using an enhanced BP neural network model. The traditional sparrow search algorithm is enhanced by incorporating a golden sine strategy to improve its position-updating mechanism, thereby overcoming its tendency to converge prematurely to local optima. Additionally, an opposition-based learning strategy is integrated to explore the reverse solution around the optimal solution of the sparrow search algorithm, mitigating the risk of local optima.FindingsThe results of the test function demonstrate that the proposed method significantly enhances fitting accuracy while maintaining computational efficiency. Finally, by applying this approach to the metro bogie frame as a case study, the structural reliability of the bogie frame is evaluated using the Monte Carlo method, providing valuable insights for subsequent analysis and structural optimization.Originality/valueThe use of the surrogate model approach for structural reliability analysis significantly improves solution efficiency. Furthermore, the integration of ISSA with the BP neural network enhances both fitting accuracy and computational efficiency, demonstrating the superiority and practicality of the proposed method.

Data-driven pressure prediction for self-pressurization liquid hydrogen tank using transfer learning method

Prediction of pressure evolution in liquid hydrogen (LH2) tanks utilizing data-driven technology has gradually garnered significant attention. However, training an accurate data-driven model for pressure prediction in LH2 tanks is a challenge due to inadequate experimental data. To end this, this paper presents a Backpropagation Neural Network model for pressure prediction in LH2 tanks based on transfer learning on pre-trained models, using experimental data and results from thermodynamics models. Pre-trained models are firstly optimized by Bayesian optimization algorithms with large sample datasets from the thermal multi-zone model program and then performed the fine-tuning method with small sample datasets from experiments. The effectiveness of the proposed approach is validated by experimental data and numerical results. Compared with baseline models trained only with experimental data, the results demonstrate better model performance, with a maximum of 10.51% accuracy improvement. The proposed approach demonstrates considerable potential in facilitating the application of advanced machine learning algorithms for LH2 tank pressure prediction, significantly reducing the dependence on experimental data collection.

Study on Strength Prediction and Material Scheme Optimization for Modified Red Mud Based on Artificial Neural Networks

Focusing on the complex nonlinear problems of strength prediction and the material scheme design of modified red mud for use as a road material in engineering applications, a strength prediction neural network is established and utilized to optimize the material scheme, including the compound-solidifying agent ratio, water content, and curing age, based on experimental data accumulated during years of engineering practice and an artificial neural network. In this study, a backpropagation (BP) neural network is adopted, and 114 sets of experimental data are used to train the parameters of the unconfined compressive strength prediction model. Then, using the BP strength prediction model, the material scheme optimization process is carried out, with the strength and material costs as the objectives. The results show that the BP neural network model has a high prediction accuracy, the relative prediction error is basically within 10%, the root-mean-squared error is less than 0.04, and the correlation coefficient is more than 0.99. According to the strength requirements of modified red mud in different road projects and the constraints of each property, an optimal material scheme with a lower cost and higher 7 d target strength is obtained using a mix of polymer agent–fly-ash–cement–speed-cement in a ratio of 0.02%:1.96%:4.78%:0%, with a 33.93% water content of raw red mud, so that the target strength and material cost are 2.987 MPa and 17.099 CNY/T. This study creates an optimal material scheme, incorporating the compound-solidifying agent ratio, curing age, and water content of the modified red mud road material according to the strength requirements of different projects, thereby promoting the popularization of the utilization of red mud with better engineering practicability and economy.

Design Method for Low-Ice-Class Propellers Based on Multi-Objective Optimization

The objective of this paper was to establish a comprehensive methodology for the optimized design of propellers for ice-class vessels, aiming to enhance hydrodynamic efficiency while ensuring structural integrity. This paper begins by introducing a novel approach for calculating blade stress, which takes into account both extreme ice loads and hydrodynamic loads, to be utilized in the propeller strength design process. Subsequently, a backpropagation (BP) neural network model was developed based on the data obtained from B-series propeller charts and integrated with a genetic algorithm to achieve a preliminary optimized design of the propeller’s hydrodynamic performance. To illustrate the application of this methodology, a case study of an ice-breaking tug propeller design is presented, detailing the optimization design process, including the preliminary, intermediate, and final design stages. The study also addresses key aspects such as geometric parameterization, the selection of optimization variables, the implementation of optimization algorithms, and the balance of multi-objective trade-offs. The proposed design approach can serve as a valuable reference for the practical engineering design of propellers for ice-class vessels, providing a systematic framework for achieving optimal performance in challenging operating conditions.

An Automated Approach for Discriminating Hole Cleaning Efficiency While Predicting Penetration Rate in Egyptian Western Desert Wells

AbstractHigher rate of penetration (ROP) indicates successful drilling operation but is not the only drilling success measure. However, Conventional ROP prediction methods focus on increasing ROP and neglect the hole cleaning state, which can be altered by ROP changes. Higher ROP in vertical and deviated wells may increase cutting concentration, leading to hole cleaning problems such as overpulling and stuck pipe. With this problem in mind, this paper utilized geological, rheological, and drilling data of 31 vertical wells across four oil fields located in the Egyptian Western Desert, developed intelligent ROP prediction model through back propagation neural network (BPNN), and compared the proposed BPNN results with an empirical model. Finally, the pattern recognition algorithms including discriminant analysis, support vector machines, and neural network pattern recognition (NNPR) were implemented to discriminate hole cleaning efficiency following the ROP prediction process. Recognition models were developed based on predicted ROP, bit wear rate, specific energy, and drilling fluid carrying capacity index to evaluate hole cleaning. The accuracy of the multi-strategy classifier was evaluated using area under curve, confusion matrix, and receiver operating characteristic. The BPNN model outperformed the empirical model in terms of linear correlation coefficient (R = 98.6%) and average absolute error (AAE = 5.5%). Additionally, the best classification performance was achieved using the NNPR algorithm with 91% accuracy and a cross-validation error equal to zero. For validity, the proposed approach predicted ROP and classified hole cleaning efficiency for new vertical well in adjacent oil field, resulting in a 6% improvement in ROP.

Daily and seasonal variations of soil respiration from maize field under different water treatments in North China

AbstractTo further evaluate the effect of water stress on soil respiration (RS), reveal the influencing factors of daily and seasonal RS, and systematically evaluate and compare the sensibility of different machine learning algorithms (multiple nonlinear regression [MNR], support vector machine regression [SVR], backpropagation artificial neural network [BPNN]) to estimate RS from a maize field under water stress condition, the field experiments were conducted within a maize field in Inner Mongolia, China, during the entire 2019 growing season. Various levels of deficit irrigation were conducted in the vegetative, reproductive, and mature stages. Our research indicated that soil CO2 fluxes from 100% evapotranspiration treatment (Tr1) were significantly greater than various deficit irrigation treatments (Tr2, Tr3, Tr4) during each growth stage of summer maize. The cumulative soil CO2 fluxes of Tr2, Tr3, and Tr4 decreased 24.8%, 30.3%, and 43.7% compared with Tr1, respectively. We determined that the drivers affecting the daily RS were soil temperature at 5 cm depth (TS,5) and soil surface temperature (TSF), followed by water‐filled porosity (WFPS) at 5 cm depth, but no significant correlations were observed at 25 cm depths. TS,5 and TSF also performed similar correlation with seasonal RS with R greater than 0.753 among all water treatments, followed by chlorophyll content with R greater than 0.726. During the whole growing season, the BPNN model exhibited the best predicting result, and could explain the 60%–80% and 87.8% of the variations of RS at the daily and seasonal scales, with root mean square error of 48.7–100.9 mg m−2 h−1 and 91.5 mg m−2 h−1, respectively. The SVR and MNR models could estimate the 47.9%–57% and 39.9%–52.1% of the daily RS and 81.4% and 78.6% of the seasonal RS, respectively. Overall, our study indicated the machine learning algorithms could be successfully applied to estimate RS at daily and seasonal scales from a maize field under water stress condition.

Predicting Work-in-Process in Semiconductor Packaging Using Neural Networks: Technical Evaluation and Future Applications

This review paper focuses on the application of neural networks in semiconductor packaging, particularly examining how the Back Propagation Neural Network (BPNN) model predicts the work-in-process (WIP) arrival rates at various stages of semiconductor packaging processes. Our study demonstrates that BPNN models effectively forecast WIP quantities at each processing step, aiding production planners in optimizing machine allocation and thus reducing product manufacturing cycles. This paper further explores the potential applications of neural networks in enhancing production efficiency, forecasting capabilities, and process optimization within the semiconductor industry. We discuss the integration of real-time data from manufacturing systems with neural network models to enable more accurate and dynamic production planning. Looking ahead, this paper outlines prospective advancements in neural network applications for semiconductor packaging, emphasizing their role in addressing the challenges of rapidly changing market demands and technological innovations. This review not only underscores the practical implementations of neural networks but also highlights future directions for leveraging these technologies to maintain competitiveness in the fast-evolving semiconductor industry.

A Deep Learning Approach for Enhanced Real-Time Prediction of Winter Road Surface Temperatures in High-Altitude Mountain Areas

Low temperatures and icing in winter are significant factors that severely affect highway safety and traffic mobility. To enhance the precision and reliability of real-time winter road surface temperature (RST) prediction, a short-term prediction model is developed that harnesses both feature selection and deep learning. Leveraging meteorological data from a mountain highway in Yunnan, China, the key environmental variables affecting road surface temperature were first extracted using a random forest (RF) model for feature selection. These features were then combined with RST data to construct multiple groups of input variable combinations for the prediction model. A short-term prediction model with a 10-minute update frequency was built using a long short-term memory neural network (LSTM), namely RF-LSTM. The best input variable combination and preset parameters for the prediction model were determined through comparative testing with prevalent machine learning models, and the transferability of the prediction model was verified. The results showed that the best input variable combination for the RF-LSTM prediction model was road surface temperature and air temperature. The model recognised that the short-term RST was affected by long and short-term memory characteristics within a two-hour timeframe. When compared to the RF model, backpropagation (BP) neural network model and the standard LSTM model, the proposed model reduces prediction errors by 59.15%, 31.10% and 20.26%, respectively, while the prediction accuracy is 99.13% within an error margin of ±0.5℃. On the verification dataset, the proposed model maintains its time transferability with an average prediction absolute error of 0.0478. In all, the proposed model not only achieves a higher level of precision in real-time RST predictions but also ensures a more consistent and reliable performance under the challenging conditions of high altitude and mountainous terrain, offering enhanced support for traffic safety and road maintenance decision-making.

Nondestructive freshness prediction of large yellow croaker (Pseudosciaena crocea) using computer vision and machine learning techniques based on pupil color.

Conventional methods for evaluating of fish freshness based on physiological and biochemical methods are often destructive, complicated, and costly. This study aimed to predict the freshness of large yellow croaker which was sampled every second day in 9 consecutive days at 4°C, using computer vision technology combined with pupil color parameters and different machine learning algorithms (back propagation neural network, BPNN; radial basis function neural network; support vector regression; and random forest regression, RFR). In the process of model building, the RFR model provided the most accurate prediction for the value of total volatile basic nitrogen (TVB-N), with the R-square of the test set ( ) of 0.993. The BPNN model exhibited the best fit for predicting the value of thiobarbituric acid (TBA), with of 0.959. Additionally, the RFR model was the most effective in forecasting total viable count (TVC), with of 0.935. After validation, the root mean square error values of the RFR model for predicting TVB-N value, TBA value, and TVC value were the lowest, which were 0.764, 0.067, and 0.219, respectively. It demonstrated the applicability and good predictive performance of the RFR model for predicting biochemical and microbiological indicators. These findings also demonstrated that monitoring the changes in pupil color could successfully predict the freshness of chilled fish. PRACTICAL APPLICATION: Application Scenario: Quality inspectors detect changes in thefreshness of large yellow croaker in real time from the beginning of distribution to the selling site.

Spectral Variations of Reclamation Vegetation in Rare Earth Mining Areas Using Continuous–Discrete Wavelets and Their Impact on Chlorophyll Estimation

Ion-adsorption rare earth mining areas are primarily situated in the hilly regions of southern China. However, mining activities have led to extensive deforestation of the original vegetation. The reclamation vegetation planted for ecological restoration faces significant challenges in surviving under environmental stresses, including heavy metal pollution, ammonia nitrogen contamination, and soil drought. To rapidly and accurately monitor the growth of reclamation vegetation, this study investigates the spectral variations and their impact on the accuracy of chlorophyll estimation, utilizing hyperspectral data and relative chlorophyll content (SPAD). Specifically, continuous–discrete wavelet transforms were applied, along with the original spectra and first derivative spectra, to enhance spectral anomalies in the reclamation vegetation and identify chlorophyll-sensitive spectral features. Additionally, multiple linear stepwise regression and backpropagation neural network models were employed to estimate chlorophyll content. The results revealed the following: (1) the d5 and d6 scales of the discrete wavelet effectively highlighted spectral anomalies in the reclamation vegetation; (2) Salix japonica (Salix fragilis L.), among typical reclamation species, exhibited poor adaptability to the environmental conditions of the rare earth mining area; (3) the backpropagation neural network model demonstrated superior performance in chlorophyll estimation, with the spectral features Fir, Fir_d4, Fir_d5, and Fir_d6 significantly enhancing the accuracy of the model, achieving an R2 of 0.93 for Photinia glabra (Photinia glabra (Thunb.) Maxim.). The application of continuous–discrete wavelet transforms to hyperspectral data significantly improves the precision of chlorophyll estimation, underscoring the potential of this method for the rapid monitoring of reclamation vegetation growth.

Evaluation of supervised machine learning regression models for CFD-based surrogate modelling in indoor airflow field reconstruction

Fast and reliable prediction of indoor airflow distribution is critical for indoor environment control. While neural networks (NN), often interchangeably referred to as Back Propagation Neural Networks (BPNNs), are popular for airflow predictions, optimising these models is challenging due to their ”black box” nature and complex network structures. This study explores alternative robust regression models, including decision-tree-based models (e.g., XGBoost, LightGBM, Random Forest) and Support Vector Regression (SVR), for predicting indoor airflow. Two BPNN structures were initially developed to evaluate feasibility of NN models. BPNN A was trained using airflow velocities from two inlets as input neurons to directly predict the airflow velocity distribution within the domain. BPNN B was trained additionally with spatial information, including space samples and boundary wall data. Higher-dimensional training structures of BPNN B were applied to decision tree-based models and SVR to assess their capability in predicting non-linear airflow patterns. Results indicated that BPNN A achieved the highest accuracy, while the inclusion of higher-dimensional data in BPNN B led to decreased accuracy. Among all decision-tree-based models, XGBoost demonstrated the greatest potential, achieving an R2 above 99.5% and predictive errors below 10%. XGBoost also outperformed both BPNN models in speed, being 15.78 times faster than BPNN A and 252 times faster than BPNN B. The interpretability of XGBoost was further explored by analysing feature importance, which helps identify the most influential input variables while predicting the airflow velocity. This analysis is expected to offer an enhanced understanding of boundary conditions leading to optimised indoor environment strategy.

Study on Compression Bearing Capacity of Tapered Concrete-Filled Double-Skin Steel Tubular Members Based on Heuristic-Algorithm-Optimized Backpropagation Neural Network Model

A tapered concrete-filled double-skin steel tubular (TCFDST) structure has been used as the main framework in transmission towers, offshore facility platforms, and turbine towers owing to its excellent mechanical properties. In order to solve the difficulties of calculating the axial compressive capacity of TCFDST members due to the variations in cross-section, this paper applied heuristic optimization algorithms such as Genetic Algorithms (GAs), Particle Swarm Optimization (PSO), Simulated Annealing (SA), and Ant Colony Optimization (ACO) to enhance a Backpropagation Neural Network (BPNN) model. A predictive model incorporating both global and local optimization strategies for the axial compressive capacity of a TCFDST structure is proposed. A comprehensive axial database for TCFDST members, comprising 1327 sets of experimental and finite element analysis results, was established, with ten types of component dimensions and material parameters selected as input variables and compressive bearing capacity as the output variable. This study developed and assessed four BPNN models, each optimized by a different heuristic algorithm, against various machine learning algorithms and standards. The heuristic-algorithm-optimized BPNN models demonstrated superior accuracy in predicting the axial compressive capacity of TCFDST members. Through parametric analysis, this study identified the relationship between the model’s bearing capacity predictions and each input parameter, confirming the model’s broad applicability. The optimized BPNN model, refined with heuristic algorithms, provides a significant reference for addressing the computational challenges associated with the load-bearing capacity of TCFDST structures and facilitating their application.

Optimising Neural Networks for Enhanced Fracture Density Prediction in Surrounding Rock of Coalbed Methane Reservoir

ABSTRACTFractures influence the mechanical strength of coal roof and floor, constraining the design of hydraulic fracturing for coalbed methane production. Currently, the predominant approach involves the integration of petrophysical logging with machine learning for fracture prediction. Nevertheless, challenges exist regarding the model's accuracy. In this study, we present a novel approach to predict fracture density. Our method optimises a back‐propagation (BP) neural network and utilises principal component analysis for feature extraction. We employ logging parameters (density, compensated neutron and acoustic time difference) obtained from Shouyang Block well SY‐1 and fracture density data from electrical imaging logging to construct the FVDC model's dataset. The BP neural network model is optimised using the Sparrow Search algorithm and Tent Chaotic Mapping. The results demonstrate a substantial enhancement over the BP neural network model, with reductions of 80.102% in mean absolute error, 94.182% in mean square error, 75.879% in root mean square error and 79.764% in mean absolute percentage error. When considering accuracy, the optimised model (97.098%) surpasses the support vector regression model (96.478%), the random forest model (94.404%) and the BP neural network model (85.657%). Scalability testing for the optimised model was conducted using data from well SY‐2, yielding a remarkable prediction accuracy of 96.775%. This performance exceeds that of the BP neural network (with an accuracy of 85.102%), as well as the random forest and support vector regression models (with accuracies of 91.234% and 90.384%, respectively). These results underscore the potential of well logging and machine learning in FVDC prediction.

Forecasting Orange Juice Futures: LSTM, ConvLSTM, and Traditional Models Across Trading Horizons

This study evaluated the forecasting accuracy of various models over 5-day and 10-day trading horizons to predict the prices of orange juice futures (OJ = F). The analysis included traditional models like Autoregressive Integrated Moving Average (ARIMA) and advanced neural network models such as Long Short-Term Memory (LSTM), Recurrent Neural Network (RNN), Backpropagation Neural Network (BPNN), Support Vector Regression (SVR), and Convolutional Long Short-Term Memory (ConvLSTM), incorporating factors like the Commodities Index and the S&P500 Index. We employed loss function metrics and various tests to assess model performance. The results indicated that for the 5-day horizon, the LSTM and ConvLSTM consistently outperformed the other models. LSTM achieved the lowest error rates and demonstrated superior capability in capturing temporal dependencies, especially in single-factor and S&P500 Index predictions. ConvLSTM also performed strongly, effectively modeling spatial and temporal data patterns. In the 10-day horizon, similar trends were observed. LSTM and ConvLSTM models had significantly lower errors and better alignment with actual values. The BPNN model performed well when all factors were included, and the SVR model maintained consistent accuracy, particularly for single-factor predictions. The Diebold–Mariano (DM) test indicated significant differences in forecasting accuracy, favoring advanced neural network models. In addition, incorporating multiple influencing factors further improved predictive performance, enhancing investment outcomes and reducing risk.

Derivation of marine water quality criteria for copper based on artificial neural network model

The water chemical effects of copper have been a focus in the study of water quality criteria (WQC). Currently, multiple regression models are commonly used to quantitatively describe the impact of environmental factors on Cu toxicity in WQC studies. However, the influence of species-specific effects may consequently lead to poor prediction results of the regression models in practical application. For this issue, a backpropagation neural network (BPNN) model optimized using a genetic algorithm was developed in this study. The results showed when pooled data of given taxonomic groups were used, the BPNN mixed models had higher Adj.R2 for five out of seven groups in the predicted toxicity values compared to the MNLR mixed models. When using species-specific models, the BPNN model still showed higher predictive performance. Further comparison of the two models for the species M. galloprovincialis revealed that, in addition to the good predictive performance of the BPNN models, the pre-set species codes of different species in the taxonomic group for the BPNN mixed model also reduced the impact of species-specific effects among species. Finally, the WQCs under different water quality parameter ranges were obtained using predicted toxicity values from mixed BPNN and MNLR models. The short-term WQC range for common water quality parameters (salinity: 25–30 ppt, DOC: 0.5–2.5 mg/L) obtained from the BPNN mixed model in natural marine environments was 1.6–4.41 μg/L, which aligns with guidance values provided by major global institutions, demonstrating the feasibility of applying the BPNN mixed model to WQC derivation. This study aims to provide valuable references for future research on WQC.