Artificial Neural Network Framework Research Articles

Kriging-based surrogate data-enriching artificial neural network prediction of strength and permeability of permeable cement-stabilized base

Limited test data hinder the accurate prediction of mechanical strength and permeability of permeable cement-stabilized base materials (PCBM). Here we show a kriging-based surrogate model assisted artificial neural network (KS-ANN) framework that integrates laboratory testing, mathematical modeling, and machine learning. A statistical distribution model was established from limited test data to enrich the dataset through the combination of markov chain monte carlo simulation and kriging-based surrogate modeling. Subsequently, an artificial neural network (ANN) model was trained using the enriched dataset. The results demonstrate that the well-trained KS-ANN model effectively captures the actual data distribution characteristics. The accurate prediction of the mechanical strength and permeability of PCBM under the constraint of limited data validates the effectiveness of the proposed framework. As compared to traditional ANN models, the KS-ANN model improves the prediction accuracy of PCBM’s mechanical strength by 21%. Based on the accurate prediction of PCBM’s mechanical strength and permeability by the KS-ANN model, an optimization function was developed to determine the optimal cement content and compaction force range of PCBM, enabling it to concurrently satisfy the requirements of mechanical strength and permeability. This study provides a cost-effective and rapid solution for evaluating the performance and optimizing the design of PCBM and similar materials.

Electrocardiography Classification with Leaky Integrate-and-Fire Neurons in an Artificial Neural Network-Inspired Spiking Neural Network Framework.

Monitoring heart conditions through electrocardiography (ECG) has been the cornerstone of identifying cardiac irregularities. Cardiologists often rely on a detailed analysis of ECG recordings to pinpoint deviations that are indicative of heart anomalies. This traditional method, while effective, demands significant expertise and is susceptible to inaccuracies due to its manual nature. In the realm of computational analysis, Artificial Neural Networks (ANNs) have gained prominence across various domains, which can be attributed to their superior analytical capabilities. Conversely, Spiking Neural Networks (SNNs), which mimic the neural activity of the brain more closely through impulse-based processing, have not seen widespread adoption. The challenge lies primarily in the complexity of their training methodologies. Despite this, SNNs offer a promising avenue for energy-efficient computational models capable of displaying a high-level performance. This paper introduces an innovative approach employing SNNs augmented with an attention mechanism to enhance feature recognition in ECG signals. By leveraging the inherent efficiency of SNNs, coupled with the precision of attention modules, this model aims to refine the analysis of cardiac signals. The novel aspect of our methodology involves adapting the learned parameters from ANNs to SNNs using leaky integrate-and-fire (LIF) neurons. This transfer learning strategy not only capitalizes on the strengths of both neural network models but also addresses the training challenges associated with SNNs. The proposed method is evaluated through extensive experiments on two publicly available benchmark ECG datasets. The results show that our model achieves an overall accuracy of 93.8% on the MIT-BIH Arrhythmia dataset and 85.8% on the 2017 PhysioNet Challenge dataset. This advancement underscores the potential of SNNs in the field of medical diagnostics, offering a path towards more accurate, efficient, and less resource-intensive analyses of heart diseases.

Optimization and predictive modeling of membrane based produced water treatment using machine learning models

Membrane technology has become a crucial solution for treating produced water, playing a vital role in resource recovery and environmental preservation. The use of advanced Artificial Intelligence (AI) technique such as Machine Learning (ML) has introduced an innovative approach to define and model these treatment procedures such as trans-membrane pressure (TMP) and total membrane resistance. The research focuses specifically on the utilization of Artificial Neural Network (ANN) and Extreme Gradient Boosting (XGBoost) model to improve the precision and effectiveness of produced water treatment using membrane technology. Also, the study presented explores how the Artificial Neural Network (ANN) framework excels in predictive insights and process modeling, optimizing operational parameters in membrane-based treatment setups. Furthermore, integrating the Extreme Gradient Boosting (XGBoost) model highlights its ability to enhance system performance, providing accurate predictions and better control in complex treatment processes. It presents comprehensive research of combining Artificial Neural Network (ANN) and Extreme Gradient Boosting (XGBoost) model with membrane technology for produced water treatment, emphasizing their potential to improve optimization techniques and modeling approaches, ultimately enhancing water treatment efficiency and sustainability. Both the Artificial Neural Network (ANN) and Extreme Gradient Boosting (XGBoost) models achieved an impressive R-squared (R2) scores of 0.754, and 0.97 for Transmembrane Pressure (TMP), and 0.95 and 0.81 for Total Resistance prediction respectively.

Decoding green food safety information dependency in the digital era: An intelligent validation using SEM-ANN framework

Dependence on mobile phone information in the Internet era is a crucial issue, and psychological perception and personalized and accurate pushing significantly impact this field. This introduction of rational and planned behaviour combined with the theoretical framework of Structural Equation Modelling (SEM) and Artificial Neural Networks (ANN) are intricacy analytical tools to explore the key factors affecting the information dependence on green food safety. The key factors affecting the reliance on green food safety information were investigated. The survey covered 630 participants aged 17 to 70 in Guangdong Province, China. The study showed that perceived risk, regulation, price volatility, innovativeness, and anticipatory confirmation significantly influence the perceived usefulness of green food information, affecting satisfaction with green food. The study also highlights the significant impact of the degree of anticipatory confirmation by users on satisfaction and perceived usefulness of green food. Model path analysis revealed that anticipatory confirmation and perceived usefulness explained 61% and 74% of the variance in financial sustainability, respectively. The application of deep ANN multilayer perceptron improved the prediction accuracy of perceived usefulness with 87.54% training accuracy and 90.34% testing accuracy, further deepening the understanding of the mechanism of relying on green food safety information. This study provides strong support for a deeper understanding of the relationship between consumer behaviour and green food safety information dependence, which helps promote enterprises and market camps to more effectively formulate product or service promotion strategies and provides a solid theoretical and empirical foundation for future green food safety and personalized and accurate push thus enhancing information dependence.

Sparse linear array synthesis using artificial neural network and nonconvex optimization

AbstractThe concept of sparse array synthesis has attracted considerable attention due to its potential to reduce hardware costs in array design and enhance the efficiency of array aperture utilization. In contrast to conventional artificial neural networks (ANNs), a more efficient encoder–decoding ANN framework is proposed in this letter for designing sparse linear arrays. The decoder is pretrained with a randomly generated training set, enabling its ability to rapidly predictiction of far‐field patterns. This process considers the actual coupling effect among the antenna elements. The output of the encoder is used as the input of the decoder to synthesize the sparse arrays by providing the desired pattern to the encoder. During the training process, a novel ‐norm minimization nonconvex optimization method is integrated into the proposed framework to reduce the number of activated elements. The effectiveness and efficiency of the method are demonstrated through numerical and full‐wave simulations for synthesizing various linear sparse arrays, including large‐scale arrays. In comparison to existing algorithms, our proposed method exhibits superior computational efficiency and predicted accuracy with fewer elements.

A data-driven model for a liquid desiccant regenerator equipped with an evacuated tube solar collector: Random forest regression, support vector regression and artificial neural network

The application of a solar-assisted liquid desiccant air-conditioning system equipped with evacuated tubes, focusing on the assessment and comparison of various artificial intelligence (AI) models is investigated. Specifically, Support Vector Regression (SVR), Multi-Layer Perceptron Artificial Neural Network (MLP-ANN), and Random Forest Regression (RFR) models are assessed for predicting key performance indicators: mass removal rate, efficiency, and effectiveness. Additionally, different optimizers within the Artificial Neural Network (ANN) framework—such as Adam, Stochastic Gradient Descent (SGD), and RMSprop—are systematically examined and tuned. The study encompasses the selection of the most suitable AI model for each target variable, considering parameters such as ambient temperature, solar radiation, timestamp, airflow rate, and initial solution concentration as influential factors in the modeling process. It is found that the best predictor model for effectiveness is SVR with RBF kernel. For MRR and efficiency, it is MLP-ANN with respectively AdamW and NAdam optimizers. The disparity of prediction of the MRR, efficiency and effectiveness target are respectively 0.72%, 1.07% and 0.5%, on average, indicating a precise prediction. Furthermore, including timestamps as model inputs significantly boosts accuracy, an aspect often neglected in prior research, leading to a noticeable minimum 5% rise in the R2 score.

WITHDRAWN: Investigating the Polymer Structure Effects on SBS-Modified Asphalt Binders under Short-Term Aging through the Artificial Neural Networks, Genetic Algorithms, and Fuzzy Logic

SBS-modified asphalt (SBSMA) exhibits complicated aging behavior in the weathering environment due to oxidation of the matrix asphalt, degradation of the SBS, and disruption of the SBS-asphalt relationship. The weathering process of SBS-modified asphalt has been studied extensively. However, most of this research has taken a broad, environmental approach. This research delved into the nuanced effects of aging on low polymer-modified asphalt (LPMA) compositions integrated with a rejuvenating agent (RA), leveraging advanced computational intelligence methodologies. The study uniquely employed fuzzy logic, Artificial Neural Networks (ANN), and genetic algorithms to forecast and analyze the performance dynamics of distinct asphalt mixtures. Five asphalt mixtures were examined: a PG 58-28 binder served as the foundational benchmark, while the subsequent four used modified PG 58-28 with different ratios of Ethylene Vinyl Acetate (EVA) and sesame oil (RA). Assessment methodologies comprised the Resilient Modulus Test, Wheel Track Test, Flow Time, Direct Tension, and Quarter-Circle Bend tests. The ANN framework was instrumental in predicting rutting and crack trajectories, while fuzzy logic facilitated understanding the intricate interplays between component ratios and resultant performance metrics. Genetic algorithms optimized the EVA and RA ratios for enhanced efficacy. Initial results demonstrated that the amalgamation of EVA and RA strengthened the overall lifespan and rutting resilience of the asphalt composites. Enhanced blends, optimized using the genetic algorithm, exhibited improved resistance to initial crack formations relative to the baseline. As the aging process advanced, the modified blend's resilience to recurrent stress-induced cracks experienced minor reductions but paralleled the foundational benchmark. A harmonized combination, pinpointed via the genetic algorithm containing 6.5% EVA and 6% RA, emerged as the most proficient in amplifying durability, resistance against deformations, and immediate crack resistance.

Assessment of metal extraction from e-waste using supported IL membrane with reliable comparison between RSM regression and ANN framework

Recently, efficient techniques to remove indium ions from e-waste have been described due to their critical application. This paper illustrates the recovery of indium ions from an aqueous solution using a liquid membrane. CyphosIL 104 described the excellent potential for the extraction of indium ions. Evaluation of the five process parameters, such as indium concentration (10–100 mg/L), carrier concentration (0.05–0.2 mol/L), feed phase acidity (0.01–3 mol/L), chloride ion concentration (0.5–4 mol/L) and the stripping agent concentration (0.1–5 mol/L) were conducted. The interactive impacts of the various parameters on the extraction efficiency were investigated. The response surface methodology (RSM) and artificial neural network (ANN) were employed to model and compare the FS-SLM process results. RSM model with a quadratic equation (R2 = 0.9589) was the most suitable model for describing the efficiency. ANN model with six neurons showed a prediction of extraction efficiency with R2 = 0.9860. The best-optimized data were: 73.92 mg/L, 0.157 mol/L, 1.386 mol/L, 2.99 mol/L, and 3.06 mol/L for indium concentration, carrier concentration, feed phase acidity, chloride ion concentration, and stripping agent concentration. The results achieved by RSM and ANN led to an experimentally determined extraction efficiency of 93.91%, and 94.85%, respectively. It was close to the experimental data in the optimization condition (95.77%). Also, the evaluation shows that the ANN model has a better prediction and fitting ability to reach outcomes than the RSM model.

A macro-micro FE and ANN framework to assess site-specific bone ingrowth around the porous beaded-coated implant: an example with BOX® tibial implant for total ankle replacement.

The use of mechanoregulatory schemes based on finite element (FE) analysis for the evaluation of bone ingrowth around porous surfaces is a viable approach but requires significant computational time and effort. The aim of this study is to develop a combined macro-micro FE and artificial neural network (ANN) framework for rapid and accurate prediction of the site-specific bone ingrowth around the porous beaded-coated tibial implant for total ankle replacement (TAR). A macroscale FE model of the implanted tibia was developed based on CT data. Subsequently, a microscale FE model of the implant-bone interface was created for performing bone ingrowth simulations using mechanoregulatory algorithms. An ANN was trained for rapid and accurate prediction of bone ingrowth. The results predicted by ANN are well comparable to FE-predicted results. Predicted site-specific bone ingrowth using ANN around the implant ranges from 43.04 to 98.24%, with a mean bone ingrowth of around 74.24%. Results suggested that the central region exhibited the highest bone ingrowth, which is also well corroborated with the recent explanted study on BOX®. The proposed methodology has the potential to simulate bone ingrowth rapidly and effectively at any given site over any implant surface.

Machine learning-based evaluation of punching shear resistance for steel/FRP-RC slabs

With the emergence of fiber-reinforced polymer (FRP) reinforcement as a substitute for conventional steel reinforcement, different design codes have been developed to account for the different mechanical properties of the FRP, specifically the elastic modulus. The ACI 440.11-22, CSA/S806-12, and JSCE-97 are well-known standards for FRP-reinforced concrete structures. In particular, these design standards show significant variations in estimating the punching shear resistance and accounting for the elastic modulus. This study provides a statistical and machine learning-based evaluation of the punching shear models in design standards. An Artificial neural network (ANN) framework is used to develop a generalized punching resistance model of flat slabs reinforced with steel and FRP bars utilizing a large experimental dataset with 539 tests. The study presents a parametric study to examine the effect of different factors on the punching shear strength. The parametric study provided a graphical presentation and comparison between the design models and the developed ANN model.

Control-relevant neural networks for feedforward control with preview: Applied to an industrial flatbed printer

The performance of feedforward control depends strongly on its ability to compensate for reproducible disturbances. The aim of this paper is to develop a systematic framework for artificial neural networks (ANN) for feedforward control. The method involves three aspects: a new criterion that emphasizes the closed-loop control objective, inclusion of preview to deal with delays and non-minimum phase dynamics, and enabling the use of an iterative learning algorithm to generate training data in view of addressing generalization errors. The approach is illustrated through simulations and experiments on an industrial flatbed printer.

A decreasing failure rate model with a novel approach to enhance the artificial neural network's structure for engineering and disease data analysis

The study focuses on key metrics used to examine the characteristics of a lifetime random variable distribution in reliability and survival theory research. In this analysis, metrics including the probability density function time, mean residual lifespan, mean time between failures, hazard rate, and reliability function are essential. The focus of the inquiry is these important parameters in relation to the Burr-Hatke exponential model specifically. The study focuses on key metrics used to examine the characteristics of a lifetime random variable distribution in reliability and survival theory research. In this analysis, metrics including the probability density function time, mean residual lifespan, mean time between failures, hazard rate, and reliability function are essential The focus of the inquiry is these important parameters in relation to the Burr-Hatke exponential model specifically. A key component of the research is a comparison of the outcomes from the artificial intelligence approach and those from conventional literature-based methodologies. This comparison study sheds light on how well the artificial neural network framework performs while evaluating the Burr-Hatke exponential model's technical features. The study allows a comprehensive analysis of the training and prediction capabilities of the growing neural network by calculating multiple performance measures. This comprehensive strategy improves our comprehension of the model's survival traits and reliability, offering significant contributions to the larger field of study. The network structure's mean square error was estimated to be 5.19E-04, and its coefficient of determination value was 0.99987 for the first neural network model. For the second neural network model, the coefficient of determination value was 0.99999 and the mean square error value was 4.58E-06. The outcomes amply revealed the neural network structure's extraordinarily high prediction accuracy and the degree to which the prediction outputs agree with those of the Maximum Likelihood Estimation technique.

Neural network-based liquidity risk prediction in Indian private banks

Liquidity risk stands as a pivotal financial concern within the Indian banking landscape, particularly in the context of Basel Recommendation. This recommendation underscores the vital role of liquidity risk management, urging banks to establish robust information systems for the assessment, anticipation, and control of liquidity risks. Recognizing the nuanced nature of this risk, banks across India employ diverse tools and methodologies in managing liquidity risk. This study delves into the efficacy of artificial neural networks (ANNs) in forecasting liquidity risk within the private banking sector in India. Drawing from prior research and leveraging accounting data, a unique structure and architecture for a multilayer perceptron neural network was developed. Subsequently, the study harnessed this ANN framework, along with Matlab software, to forecast liquidity risk in private Indian banks over the period spanning 2009 to 2019. The research outcomes shed light on the promising potential of artificial neural networks as a valuable tool for predicting liquidity risk in private Indian banks. This discovery holds significance in enhancing the resilience and stability of the Indian banking sector by providing more precise and timely insights into liquidity risk management.

Geometric optimization of magnetorheological damper for prosthetic ankles using artificial neural networks

In this work, the primary design constraints of a magnetorheological (MR) actuator and its stroke dimension have been found based on biomechanical requirements and anthropometric constraints of the ankle in a transtibial prosthesis. Based on the inverted slider-crank mechanism models, the force controller parameters of the MR damper are identified. Parameters of the MR dampers are evaluated through optimization that minimises the error between the prosthetic ankle moment and the desired ankle moment for normal level ground walking from experimental data. Furthermore, an artificial neural network (ANN) framework for the MR valve is developed where a three-layered ANN model has been utilised to forecast the magnetic flux density (MFD) across different regions of the MR valve. The data have been generated from an ANSYS-APDL software package using finite element magnetostatic analysis (FEMS). The ANN model outcomes match the FEMS results reasonably well. Finally, the ANN model is employed to find MFD and is used to optimize the MR valve. Optimal solutions are obtained that satisfy the goal function of maximising the damper force and minimising the energy consumption and weight of the MR damper. Subsequently, the optimized MR damper has been fabricated and tested experimentally and it has been found to produce enough force to act as an actuator in a prosthetic ankle.

Artificial Neural Networks and Time Series of Counts: A Class of Nonlinear INGARCH Models

Abstract Time series of counts are frequently analyzed using generalized integer-valued autoregressive models with conditional heteroskedasticity (INGARCH). These models employ response functions to map a vector of past observations and past conditional expectations to the conditional expectation of the present observation. In this paper, it is shown how INGARCH models can be combined with artificial neural network (ANN) response functions to obtain a class of nonlinear INGARCH models. The ANN framework allows for the interpretation of many existing INGARCH models as a degenerate version of a corresponding neural model. Details on maximum likelihood estimation, marginal effects and confidence intervals are given. The empirical analysis of time series of bounded and unbounded counts reveals that the neural INGARCH models are able to outperform reasonable degenerate competitor models in terms of the information loss.

Artificial neural network-augmented stabilized finite element method

An artificial neural network-augmented Streamline Upwind/Petrov-Galerkin finite element scheme (SPDE-NetII) is proposed for solving singularly perturbed partial differential equations. In particular, an artificial neural network framework is proposed to predict optimal values for the stabilization parameter to be used in Streamline upwind/Petrov-Galerkin stabilization schemes. The neural network is trained by minimizing a physics-informed cost function, where the equation's mesh and physical parameters are used as input features. Further, the predicted stabilization parameter is normalized with the gradient of the solution to treat the boundary/interior layer region adequately. The proposed approach suppresses the undershoots and overshoots in the stabilized finite element solution and outperforms the existing neural network-based partial differential equation solvers such as Physics-Informed Neural Networks and Variational Neural Networks.

Prediction of entire thermal degradation process of polymethyl methacrylate infiltrated with kerosene by a modified artificial neural network

AbstractPredicting the entire thermal degradation process of solid combustibles infiltrated with flammable liquids is a challenge at present. In the current study, a novel artificial neural network (ANN) framework containing data preprocessing, data normalization and data transformation is proposed to predict the entire thermal degradation process of polymethyl methacrylate infiltrated with kerosene at three scenarios: (1) fixed kerosene mass fraction with various heating rates, (2) fixed heating rate with various kerosene mass fractions, and (3) various kerosene mass fractions with various heating rates. The entire thermal degradation process of scenario (1) can be accurately predicted using the ANN with 2‐4‐2‐1 topology. Using the data transformation formula exp(x1(1 + x2) to generate a new input variable based on temperature and kerosene mass fraction, a new ANN with 3‐4‐2‐1 topology can accurately predict the entire thermal degradation process of scenario (2). Two new input variables are generated using the data transformation formula 1/(1 + log((1 + x1)(1 + x2))) based on two data sets: (1) kerosene mass fraction and temperature, and (2) heating rate and temperature. The new ANN with 5‐4‐2‐1 topology can accurately predict the entire thermal degradation process at all three scenarios. The new ANN with Levenberg–Marquardt training function and Tanh activation function possesses the best prediction performance.Highlights Data preprocessing can significantly improve the prediction accuracy of ANN. ANN with a suitable hidden layer structure has high prediction accuracy. The new ANN can predict the entire pyrolysis process in various scenarios. ANN with LM and Tanh has the highest prediction accuracy.

Healthcare 3.0: How to Transform Machine Learning Prototypes into Functional Healthcare Applications for Diagnostic Assistance?

AbstractBackgroundAs the number of elderly people is rapidly increasing, we are faced with the challenge that diagnostic services are demanded more frequently. At the same time, the number of medical centers and available experts remains almost constant. Tools for diagnostic assistance are urgently needed to improve the effectiveness of healthcare. In four externally funded projects, we will investigate aspects and strategies of how machine learning prototype systems can be translated into functional healthcare applications.MethodIn the ongoing project “ExplAInation” funded by the German research foundation (DFG), we are developing an artificial neural network framework to generate visual and textual explanations to improve the comprehensibility and interpretability of deep learning models. This effort includes participatory research with clinical users. In the complementary project “TESIComp” funded by the Federal Ministry of Education and Research (BMBF), we will investigate ethical and social aspects of the emerging field of computational psychiatry. Patients, caregivers, and medical doctors will be interviewed, which will be qualitatively analyzed. Within the German Medical Informatics Initiative, we will contribute to the project “Open Medical Inference” (OMI), which will develop a network of distributed machine learning evaluation services. We also participate in the international “Clinical AI‐based Diagnostics” (CAIDX) consortium, which receives funding from the European Interreg Baltic Sea Region program.ResultWe developed a deep learning application for the detection of dementia atrophy patterns in brain MRI scans. Derived relevance maps highlight diagnostically important brain areas for further evaluation by the radiologists. Interviews with clinicians will provide information on expectations, key requirements and the functional utility of machine learning‐based assistance. Interviews with patients and caregivers will elucidate future changes and challenges in the doctor’s role and responsibilities. The OMI network will allow hospitals to use distributed machine learning evaluation services remotely, without the need of operating all the tools locally. In CAIDX, we will develop best‐practice guidelines for the overarching process of integrating machine learning prototypes and commercial tools into the hospital.ConclusionIn “Healthcare 3.0”, the digital transformation will change current diagnostic procedures and roles. Our activities focus on the involved stakeholders, regulatory aspects and implementation strategies to better steer this process.

Healthcare 3.0: How to Transform Machine Learning Prototypes into Functional Healthcare Applications for Diagnostic Assistance?

AbstractBackgroundAs the number of elderly people is rapidly increasing, we are faced with the challenge that diagnostic services are demanded more frequently. At the same time, the number of medical centers and available experts remains almost constant. Tools for diagnostic assistance are urgently needed to improve the effectiveness of healthcare. In four externally funded projects, we will investigate aspects and strategies of how machine learning prototype systems can be translated into functional healthcare applications.MethodIn the ongoing project “ExplAInation” funded by the German research foundation (DFG), we are developing an artificial neural network framework to generate visual and textual explanations to improve the comprehensibility and interpretability of deep learning models. This effort includes participatory research with clinical users. In the complementary project “TESIComp” funded by the Federal Ministry of Education and Research (BMBF), we will investigate ethical and social aspects of the emerging field of computational psychiatry. Patients, caregivers, and medical doctors will be interviewed, which will be qualitatively analyzed. Within the German Medical Informatics Initiative, we will contribute to the project “Open Medical Inference” (OMI), which will develop a network of distributed machine learning evaluation services. We also participate in the international “Clinical AI‐based Diagnostics” (CAIDX) consortium, which receives funding from the European Interreg Baltic Sea Region program.ResultWe developed a deep learning application for the detection of dementia atrophy patterns in brain MRI scans. Derived relevance maps highlight diagnostically important brain areas for further evaluation by the radiologists. Interviews with clinicians will provide information on expectations, key requirements and the functional utility of machine learning‐based assistance. Interviews with patients and caregivers will elucidate future changes and challenges in the doctor’s role and responsibilities. The OMI network will allow hospitals to use distributed machine learning evaluation services remotely, without the need of operating all the tools locally. In CAIDX, we will develop best‐practice guidelines for the overarching process of integrating machine learning prototypes and commercial tools into the hospital.ConclusionIn “Healthcare 3.0”, the digital transformation will change current diagnostic procedures and roles. Our activities focus on the involved stakeholders, regulatory aspects and implementation strategies to better steer this process.

Forecasting meteorological impacts on the environmental sustainability of a large-scale solar plant via artificial intelligence-based life cycle assessment

Although large-scale solar (LSS) is a promising renewable energy technology, it causes adverse impacts on the ecosystem, human health, and resource depletion throughout its upstream (i.e., raw material extraction to solar panel production) and downstream (i.e., plant demolition and waste management) processes. The LSS operational performance also fluctuates due to meteorological conditions, leading to uncertainty in electricity generation and raising concerns about its overall environmental performance. Hitherto, there has been no evidence-backed study that evaluates the ecological sustainability of LSS with the consideration of meteorological uncertainties. In this study, a novel integrated Life Cycle Assessment (LCA) and Artificial Neural Network (ANN) framework is developed to forecast the meteorological impacts on LSS's electricity generation and its life cycle environmental sustainability. For LCA, 18 impact categories and three damage categories are characterised and assessed by ReCiPe 2016 via SimaPro v. 9.1. For ANN, a feedforward neural network is applied via Neural Designer 5.9.3. Taking an LSS plant in Malaysia as a case study, the photovoltaic panel production stage contributes the highest environmental impact in LSS (30 % of human health, 30 % of ecosystem quality, and 34 % of resource scarcity). Aluminium recycling reduces by 10 % for human health, 10 % for ecosystem quality, and 9 % for resource scarcity. The emissions avoided by the forecasted LSS-generated electricity offset the environmental burden for human health, ecosystem quality, and resource scarcity 12–68 times, 13–73 times, and 18–98 times, respectively. The developed ANN-LCA framework can provide LSS stakeholders with data-backed insights to effectively design an environmentally conscious LSS facility, considering meteorological influences.