Multilayer Model Research Articles

Fusion of infrared and visible images via multi-layer convolutional sparse representation

Infrared and visible image fusion is an effective solution for image quality enhancement. However, conventional fusion models require the decomposition of source images into image blocks, which disrupts the original structure of the images, leading to the loss of detail in the fused images and making the fusion results highly sensitive to matching errors. This paper employs Convolutional Sparse Representation (CSR) to perform global feature transformation on the source images, overcoming the drawbacks of traditional fusion models that rely on image decomposition. Inspired by neural networks, a multi-layer CSR model is proposed, which involves five layers in a forward-feeding manner: two CSR layers acquiring sparse coefficient maps, one fusion layer combining sparse maps, and two reconstruction layers for image recovery. The dataset used in this paper comprises infrared and visible images selected from public dataset, as well as registered images collected by an actual Unmanned Aerial Vehicle (UAV). The source images contain ground targets, marine targets, and natural landscapes. To validate the effectiveness of the proposed image fusion model in this paper, comparative analysis is conducted with state-of-the-art (SOTA) algorithms. Experimental results demonstrate that the proposed fusion model outperforms other state-of-the-art methods by at least 10% in SF, EN, MI and QAB/F fusion metrics in most image fusion cases, thereby affirming its favorable performance.

Compositional dependence of direct transition energies in SixGe1−x−ySny alloys lattice-matched to Ge/GaAs

SixGe1−x−ySny ternary alloys are a candidate material system for use in solar cells and other optoelectronic devices. We report on the direct transition energies and structural properties of Ge-rich SixGe1−x−ySny alloys with six different compositions (up to 10% Si and 3% Sn), lattice-matched to Ge or GaAs substrates. The direct interband transitions occurring at energies between 0.9 and 5.0 eV were investigated using spectroscopic ellipsometry, and the resulting data were used to obtain the dielectric functions of the SixGe1−x−ySny layer by fitting a multilayer model. Values for the E0, E1, Δ1, E0′, and E2 transition energies were then found by identifying critical points in the dielectric functions. Structurally, the composition of the samples was measured using energy-dispersive x-ray measurements. The lattice constants predicted from these compositions are in good agreement with reciprocal space maps obtained through x-ray diffraction. The results confirm that a 1 eV absorption edge due to direct interband transitions can be achieved using relatively low Si and Sn fractions (<10% and <3%, respectively), although the bandgap remains indirect and at lower energies. The higher-energy critical points show smaller shifts relative to Ge and match results previously observed or predicted in the literature.

Identification of historical maximum load on concrete beams based on crack and deformation response using knowledge-embedded machine learning algorithms

This study introduces an innovative method for identifying the historical maximum load on concrete beams, utilizing a multi-layer back propagation (BP) neural network model with physically interpretable features as inputs. The method establishes a mapping relationship among four-dimensional features (boundary constraints, system parameters, cracks, and deformations) and the maximum load in the load history, using incremental load action and structural response. Moreover, a hybrid optimization algorithm, combining Hunter Prey Optimization (HPO) and Particle Swarm Optimization (PSO), is introduced to optimize the model's parameters. The method is compared with commonly used machine learning models, demonstrating its accuracy and reliability. Additionally, the study performs a sensitivity analysis of the input features using the proposed intelligent identification model. The results reveal that the proposed intelligent identification model exhibits small prediction errors on both the training and testing sets, with average absolute percentage errors of 16.46 % and 19.56 %, and determination coefficients of 0.922 and 0.8185, respectively, showcasing its strong robustness and generalization capability. Compared to the non-optimized multi-layer BP neural network model, the proposed model achieves a reduction in average absolute percentage errors of 39.59 % and 48.33 % on the training and testing sets, respectively.

Deep learning-based approach for COVID-19 spread prediction

AbstractSpread prediction models are vital tools to help health authorities and governments fight against infectious diseases such as COVID-19. The availability of historical daily COVID-19 cases, in conjunction with other datasets such as temperature and humidity (which are believed to play a key role in the spread of the disease), has opened a window for researchers to investigate the potential of different techniques to model and thereby expand our understanding of the factors (e.g., interaction or exposure resulting from mobility) that govern the underlying dynamics of the spread. Traditionally, infectious diseases are modeled using compartmental models such as the SIR model. However, this model shortcoming is that it does not account for mobility, and the resulting mixing or interactions, which we conjecture are a key factor in the dynamics of the spread. Statistical analysis and deep learning-based approaches such as autoregressive integrated moving average (ARIMA), gated recurrent units, variational autoencoder, long short-term memory (LSTM), convolution LSTM, stacked LSTM, and bidirectional LSTM have been tested with COVID-19 historical data to predict the disease spread mainly in medium- and high-income countries with good COVID-19 testing capabilities. However, few studies have focused on low-income countries with low access to COVID-19 testing and, hence, highly biased historical datasets. In addition to this, the arguable best model (BiLSTM) has not been tested with an arguably good set of features (people mobility data, temperature, and relative humidity). Therefore, in this study, the multi-layer BiLSTM model is tested with mobility trend data from Google, temperature, and relative humidity to predict daily COVID-19 cases in low-income countries. The performance of the proposed multi-layer BiLSTM is evaluated by comparing its RMSE with the one from multi-layer LSTM (with the same settings as BiLSTM) in four developing countries namely Mozambique, Rwanda, Nepal, and Myanmar. The proposed multi-layer BiLSTM outperformed the multi-layer LSTM in all four countries. The proposed multi-layer BiLSTM was also evaluated by comparing its root mean-squared error (RMSE) with multi-layer LSTM models, ARIMA- and stacked LSTM-based models in eight countries, namely Italy, Turkey, Australia, Brazil, Canada, Egypt, Japan, and the UK. Finally, the proposed multi-layer BiLSTM model was evaluated at the city level by comparing its average relative error with the other four models, namely the LSTM-based model considering multi-layer architecture, Google Cloud Forecasting, the LSTM-based model with mobility data only, and the LSTM-based model with mobility, temperature, and relative humidity data for 7 periods (of 28 days each) in six highly populated regions in Japan, namely Tokyo, Aichi, Osaka, Hyogo, Kyoto, and Fukuoka. The proposed multi-layer BiLSTM model outperformed the multi-layer LSTM model and other previous models by up to 1.6 and 0.6 times in terms of RMSE and ARE, respectively. Therefore, the proposed model enables more accurate forecasting of COVID-19 cases and can support governments and health authorities in their decisions, mainly in developing countries with limited resources.

Time in range prediction using the experimental mobile application in type 1 diabetes

BACKGROUND: Time in range (TIR) is a promising indicator of glycemic control used for evaluation of continuous glucose monitoring (CGM) for patients with diabetes mellitus (DM). The current problem is the assessment and prediction of TIR for patients who use self-monitoring of blood glucose (SМBG) corresponding low CGM availability for the majority of diabetic patients.AIM: To develop a predictive model of TIR for patients with T1DM based on data of the experimental mobile application.MATERIALS AND METHODS: An analysis of 1253 professional CGM profiles of patients with T1DM was performed. On the base of included records, TIR(CGM) was calculated and training models of 7-point SMBG profiles were generated. SMBG profiles’re loaded into the developed experimental mobile application that calculated standard glycemic control parameters. The dataset was divided into main and test samples (80 and 20%). For the main sample, the following methods’re used to develop predictive models: simple linear regression (SLR), multiple linear regression (MLR), artificial neural network (ANN). The effectiveness of the developed models was assessed on the test sample with the calculation of the mean absolute error (MAE), the root mean square error (RMSE).RESULTS: The 568 CGM profiles’re included in the study. TIR in the main group (n=454) — 45 [33; 65]%, in the test group (n=114) — 43 [33; 58]%. The most significant predictors of the regression models were the derived TIR (dTIR), p<0,001; derived time below range level 1 (dTBR1), p<0,001; standard deviation of blood glucose (SD), p=0,007. Determination coefficient for SLR (predictor: dTIR) — 0,844; for MLR (predictors: dTIR, dTBR1, SD) — 0,907. ANN multilayer perceptron models with two and one hidden layers’re developed, with the RMSE on the validation set 4,617 and 6,639%, respectively. The results of the forecast efficiency on the test sample were: dTIR: MAE — 6,82%, RMSE — 8,60%; SLR: MAE — 5,66%, RMSE — 7,34%; MLR: MAE — 4,18%, RMSE — 5,28%; ANN (2 layers): MAE — 4,14%, RMSE — 5,19%; ANN (1 layer): MAE — 4,44%, RMSE — 5,52%.CONCLUSION: ANN with two hidden layers and MLR demonstrated the best ability for TIR prediction. Further studies are required for clinical validation of developed prognostic models.

A multilevel model for organizational productivity management: an interpretive structural modeling approach

PurposeThis research highlights the ongoing concern about organizational productivity and the lack of focus on designing an optimal model. The authors aim to create a comprehensive model for managing organizational productivity, considering its impact on profitability, customer satisfaction, and employee morale. They use qualitative research methods, including Systematic Literature Review and Interpretive Structural Modeling (ISM).Design/methodology/approachIn this research using the qualitative research method of Systematic Literature Review, 57 variables affecting productivity were identified. These variables were placed in 16 layers by using the ISM method, which were classified analytically in four sections: INPUTS, OUTPUTS, OUTCOMES and IMPACTS. By determining the relationship between the sections, the research model was designed.FindingsThe potential model for organizational productivity management provides a comprehensive framework addressing critical factors like technology adoption, employee empowerment, organizational culture, and more. It identifies Linkage, Dependent, and independent variables. The lower layers consist of INPUTS such as Technological Tools, Organizational Values, and more. In the highest layer, impactful variables like Enhanced competitiveness, Improved decision-making, and Improved organizational culture are labeled as IMPACTS. Middle layer variables are categorized as OUTPUTS and OUTCOMES.Originality/valueIn this study, the concept of productivity management was redefined for the first time, and a multi-layered model for productivity management was creatively explicated using the structural equation modeling method.

Efficient nanoscale characterization of wafer surfaces using intelligent sampling and Bayesian optimization

The geometrical attributes of a wafer, such as thickness, uniformity, and local curvature, serve as eminent determinants of its quality. Therefore, measurements that are both rapid and accurate are paramount as multi-stage fabrication processes hinge on them to ensure high product reliability, proper process control, and maximum efficiency. The manual wafer profiling measurement approach in use today is, unfortunately, a time-consuming process. This creates a need for fast and prompt analysis of wafer attributes. This research thus proposes an iterative sampling method that boasts reducing the number of samplings down to a minimum without the need to discard a sufficient level of accuracy for the purpose of reliable attribute estimation. Subsequently, we suggest a series of methods to improve the monitoring and evaluation of the surface topography, all within the context of ultra-precision/nanoscale manufacturing processes, such as ultra-precision machining (UPM) and chemical–mechanical planarization (CMP). For the purpose of both real-time and exception-based detection of anomalies in the UPM process, we constructed a process-machine interaction (PMI) model with Bayesian learning, which amalgamates the received data across heterogeneous sensors such as force, vibration, and acoustic emission in situ. Similarly, the CMP process is quantitatively characterized by a multi-scaled multilayer nonlinear decomposition model with predictive capabilities. This model helps understand the physio-mechanical phenomena underlying both CMP and the UPM processes and describe and predict the nonlinear interactions, as evidenced by the experimentally collected vibration signal patterns.

A clustering machine learning approach for improving concrete compressive strength prediction

AbstractThis study investigates the application of clustering techniques to enhance the accuracy of hierarchical classification and regression (HCR) models for predicting concrete compressive strength (CCS). Following the hypothesis that integrating clustering at the initial levels of model hierarchy reduces classification errors and prevents their propagation to subsequent levels, HCR models were developed utilizing both the unweighted pair group method with arithmetic mean (UPGMA) and hard clustering (HC) methods. Findings demonstrate that models using UPGMA significantly outperform those based on HC. Furthermore, it was demonstrated that further hierarchical clustering allows for multilayered HCR models that improve predictive performance by further leveraging parent–child relationships within data clusters. Overall, this study demonstrates that the proposed methodology can be introduced in the model development pipeline to enhance the prediction accuracy of CCS models.

NSGA-II based short-term building energy management using optimal LSTM-MLP forecasts

To conduct analysis on the field of electricity management in buildings is crucial to contribute to the clean energy promotion, energy efficiency, and resilience against climate change. This manuscript proposes a methodology for modeling a predictive and calibrated energy management system (EMS) using a hybrid methodology that combines long short-term memory and multilayer perceptron models (LSTM-MLP) optimized by non-dominated sorting genetic algorithm II (NSGA-II). The proposed approach utilizes global forecast system (GFS) data to anticipate energy consumption fluctuations and optimize the use of distributed energy sources, such as photovoltaic (PV) production, based on knowledge of the electricity prices in the free market one day ahead. This energy trade-off in the building is conducted with NSGA-II, guaranteeing the exploration and the exploitation of energy in the building while minimizing energy costs and energy wastes. The research carried out demonstrates the effectiveness of the hybrid LSTM-MLP model and the advantages of NSGA-II in hyperparameter tuning and in energy balance for sustainable building practices. The methodology conducted is tested in an existing building, the Industrial Engineering School located on the Campus Lagoas-Marcosende of the Universidade de Vigo, Spain.

Internal-multiple-elimination with application to migration using two-way wave equation depth-extrapolation scheme

Internal multiple interference, affecting both seismic data processing and interpretation, has been observed for long time. Although great progress has been achieved in developing a variety of internal-multiple-elimination (IME) methods, how to increase accuracy and reduce cost of IME still poses a significant challenge. A new method is proposed to effectively and efficiently eliminate internal multiples, along with its application in internal-multiple-eliminated-migration (IMEM), addressing this issue. This method stems from two-way wave equation depth-extrapolation scheme and associated up/down wavefield separation, which can accomplish depth-extrapolation of both up-going and down-going wavefields simultaneously, and complete internal-multiple-elimination processing, adaptively and efficiently. The proposed method has several features: (1) input data is same as that for conventional migration: source signature (used for migration only), macro velocity model, and receiver data, without additional requirements for source/receiver sampling; (2) method is efficient, without need of iterative calculations (which are typically needed for most of IME algorithms); and (3) method is cost effective: IME is completed in the same depth-extrapolation scheme of IMEM, without need of a separate processing and additional cost. Several synthesized data models are used to test the proposed method: one-dimensional model, horizontal layered model, multi-layer model with one curved layer, and SEG/EAGE Salt model. Additionally, we perform a sensitivity analysis of velocity using smoothed models. This analysis reveals that although the accuracy of velocity measurements impacts our proposed method, it significantly reduces internal multiple false imaging compared to traditional RTM techniques. When applied to actual seismic data from a carbonate reservoir zone, our method demonstrates superior clarity in imaging results, even in the presence of high-velocity carbonate formations, outperforming conventional migration methods in deep strata.

Mode-Mixed Effects Based Intralayer-Dependent Impulsive Synchronization for Multiple Mismatched Multilayer Neural Networks.

This article focuses on the intralayer-dependent impulsive synchronization of multiple mismatched multilayer neural networks (NNs) with mode-mixed effects. Initially, a novel multilayer NN model that removes the one-to-one interlayer coupling constraint and introduces nonidentical model parameters is first established to meet diverse modeling requirements in complex applications. To help the multilayer target NNs with mismatched connection coefficients and time delays achieve synchronization, the hybrid controller is designed using intralayer-dependent impulsive control and switched feedback control approaches. Furthermore, the mode-mixed effects caused by the intralayer coupling delays and switched intralayer topologies are incorporated into the novel model and analysis method to ensure that the subsystems operating within the current switching interval can effectively use the topology information of the previous switching intervals. Then, a novel analysis framework including super-Laplacian matrix, augmented matrix, and mode-mixed methods is developed to derive the synchronization results. Finally, the main results are verified via the numerical simulation with secure communication.

Prediction of chronological age from resting-state EEG power in the first three years of life.

The infant brain undergoes rapid and significant developmental changes in the first three years of life. Understanding these changes through the prediction of chronological age using neuroimaging data can provide insights into typical and atypical brain development. We utilized longitudinal resting-state EEG data from 457 typically developing infants, comprising 938 recordings, to develop age prediction models. The multilayer perceptron model demonstrated the highest accuracy with an R2 of 0.82 and a mean absolute error of 92.4 days. Aperiodic offset and periodic theta, alpha, and beta power were identified as key predictors of age via Shapley values. Application of the model to EEG data from infants later diagnosed with autism spectrum disorder or Down syndrome revealed significant underestimations of chronological age. This study establishes the feasibility of using EEG to assess brain maturation in early childhood and supports its potential as a clinical tool for early identification of alterations in brain development.

Influence of urbanization on meteorological conditions and ozone pollution in the Central Plains Urban Agglomeration, China

Changes in aerodynamic and thermal conditions caused by urbanization can impact regional meteorological conditions, subsequently affecting air quality. Updated Moderate-resolution Imaging Spectroradiometer (MODIS) land use data and coupled with the urban canopy models (UCMs) in the Weather Research and Forecasting (WRF) model. This enabled the impact of urban land expansion on meteorological conditions and ozone (O3) concentrations to be evaluated. Urban expansion increased the temperature at 2 m (T2) and the probability of precipitation in urban expansion areas, and enhanced the surface urban heat island at night. As the expansion areas became progressively larger, the increase in T2 became more pronounced. The proportions of urban surfaces in June 2016, 2018, and 2020 compared to 2001 increased by 0.69%, 0.83%, and 1.04%, respectively, while T2 increased by 0.12, 0.19, and 0.20 °C in urban areas, respectively. With urban expansion, the O3 concentration increased by 1.12, 1.37, and 0.76 μg/m3 (three-year averages) in urban, suburban, and rural areas, respectively. After coupling a multi-layer urban canopy model (building effect parameterization, BEP), or a multi-layer urban canopy model with a building energy model including anthropogenic heat due to air conditioning (BEP + BEM, abbreviated as BEM simulation), the O3 concentration changed significantly in the urban expansion area, compared to the results of a single-layer urban canopy model (UCM). O3 concentrations decreased most in the BEP simulation (−0.77 μg/m3), while O3 concentrations increased most in the BEM simulation (+1.85 μg/m3). The average observed O3 concentration was 108.35 μg/m3 (three-year average), while the simulated value was 75.65–83.72 μg/m3 (R = 0.69−0.77). The validation results in the BEM and Global Optimal Scenario (GOS) simulations were relatively good, with the GOS simulation producing slightly better results than the BEM. The simulation of O3 in urban agglomerations could be improved by integrating the results of the UCMs.

Human Head Transcranial Magnetic Stimulation Using Finite Element Method

Transcranial magnetic stimulation (TMS) is a wearable neuromodulation technique. It is approved for several therapies for various neurological disorders, including major depressive disorder, traumatic brain injury, Parkinson’s disease, and post-traumatic stress disorder. This method became an alternative neuromodulation technique for such brain-related disorders. However, it has shown significant improvement in this alternative approach. Studies based on this technique have shown limited efficacy. They might be associated with current levels, poor coil locality, optimal coil size, and neuromodulator settings. It has been shown in this research that coil heating is related to higher levels of current. Thus, it is required to analyze the impact of the current levels on the induced magnetic distribution to define the optimal current range for the TMS coils. It is not feasible to investigate this research with experimental tests and analytic methods. Alternatively, using an advanced computational model of the coils and accounting for different human head anatomical layers, coil current capacity can be optimized based on finite element magnetic field distribution. This paper aims to investigate the impact of the coil current levels on the induced magnetic field distribution. The current capacity of the coils can be optimized based on the required magnetic field. In this way, the overheating may be reduced and may result in increased efficacy. As a proof-of-concept, a prototype coil and multi-layered geometrical human head models were generated using geometric shapes. The fundamental human head tissue layers were generated based on their average thickness. The model was simulated based on a finite element magnetic simulation using appropriate boundary conditions and neuromodulator settings. The various coil current levels were applied to analyze the outcome. The models were simulated, and the results were recorded based on these current levels. Results showed that there is a direct relation between applied current levels and induced magnetic flux density in the region of interest.

In-depth study of adsorption mechanisms and interactions in the removal of pharmaceutical contaminants via activated carbon: a physicochemical analysis.

This study presents a theoretical analysis of the adsorption process of pharmaceutical pollutants, specifically acetaminophen (ATP) and diclofenac (DFC), onto activated carbon (AC) derived from avocado biomass waste. The adsorption isotherms of ATP and DFC were analyzed using a multilayer model, which revealed the formation of two to four adsorption layers depending on the temperature of the aqueous solution. The saturation adsorption capacities for ATP and DFC were 52.71 and 116.53mg/g, respectively. A steric analysis suggested that the adsorption mechanisms of ATP and DFC involved a multi-molecular process. The calculated adsorption energies (ΔE1 and ΔE2) varied between 12.86 and 22.58kJ/mol, with the highest values observed for DFC removal. Therefore, the adsorption of these organic molecules was associated with physisorption interactions: van der Waals forces and hydrogen bonds. These findings enhance the understanding of the depollution processes of pharmaceutical compounds using carbon-based adsorbents and highlight the potential of utilizing waste biomass for environmental remediation.

A multi-grain multi-layer astrochemical model with variable desorption energy for surface species

Interstellar surface chemistry is a complex process that occurs in icy layers that have accumulated onto grains of different sizes. The efficiency of the surface processes often depends on the immediate environment of the adsorbed molecules. We investigated how gas-grain chemistry changes when the surface molecule binding energy is modified, depending on the properties of the surface. In a gas-grain astrochemical model, molecular binding energy gradually changes for three different environments -- (1) the bare grain surface, (2) polar water-dominated ices, and (3) weakly polar carbon monoxide-dominated ices. In addition to diffusion, evaporation, and chemical desorption, photodesorption was also made binding energy-dependent, in line with experimental results. These phenomena occur in a collapsing prestellar core model that considers five grain sizes with ices arranged into four layers. Variable desorption energy moderately affects gas-grain chemistry. Bare-grain effects slow down ice accumulation, while easier diffusion of molecules on weakly polar ices promotes the production of carbon dioxide. Efficient chemical desorption from bare grains significantly delays the appearance of the first ice monolayer. The combination of multiple aspects of grain surface chemistry creates a gas-ice balance that is different from simpler models. The composition of the interstellar ices is regulated by several binding-energy dependent desorption mechanisms. Their actions overlap in time and space, explaining the similar proportions of major ice components (water and carbon oxides) observed in all directions.

A multi-layer model for double-skin Pd-based membranes: Layer-by-layer parameter fitting

The Pd-based double-skin (DS) membrane has high potential to be implemented in various industrial processes. However, its multi-layer structure makes it difficult to model and understand the limiting transport mechanisms through each of these layers. In this work an extensive set of experiments on several membranes with different configurations was performed with the purpose of deriving a detailed model for the double-skin Pd-based membrane. This is important for the general understanding of the mass-transport taking place within each of the membrane layers, namely the support, the selective and the protective layer. Characterization, permeation tests and data-fitting led to the construction and validation of a multi-layer model for the DS-membrane. The experimental results also pointed out that the protective layer of HT or YSZ/Al2O3 has only a minor or negligible effect on the overall transport resistance of the membrane. The influence of the support was found to be small but not negligible compared to the dominating resistance of the dense selective PdAg layer. Experiments and simulations of the membrane permeation in multi-component mixtures also highlighted the large importance of concentration polarization and H2 depletion, which were successfully described in the model using a film-layer model together with a Sherwood correlation.

Forecasting fish mortality from water and air quality data using deep learning models.

The high rate of aquatic mortality incidents recorded in Taiwan and worldwide is creating an urgent demand for more accurate fish mortality prediction. Present study innovatively integrated air and water quality data to measure water quality degradation, and utilized deep learning methods to predict accidental fish mortality from the data. Keras library was used to build multilayer perceptron and long short-term memory models for training purposes, and the models' accuracies in fish mortality prediction were compared with that of the naïve Bayesian classifier. Environmental data from the 5 days before a fish mortality event proved to be the most important data for effective model training. Multilayer perceptron model reached an accuracy of 93.4%, with a loss function of 0.01, when meteorological and water quality data were jointly considered. It was found that meteorological conditions were not the sole contributors to fish mortality. Predicted fish mortality rate of 4.7% closely corresponded to the true number of fish mortality events during the study period, that is, four. A significant surge in fish mortality, from 20% to 50%, was noted when the river pollution index increased from 5.36 to 6.5. Moreover, the probability of fish mortality increased when the concentration of dissolved oxygen dropped below 2mg/L. To mitigate fish mortality, ammonia nitrogen concentrations should be capped at 5mg/L. Dissolved oxygen concentration was found to be the paramount factor influencing fish mortality, followed by the river pollution index and meteorological data. Results of the present study are expected to aid progress toward achieving the Sustainable Development Goals and to increase the profitability of water resources.

A Soft Sensor for Flow Estimation and Uncertainty Analysis Based on Artificial Intelligence: A Case Study of Water Supply Systems

The fourth industrial revolution has transformed the industry, with information technology playing a crucial role in this shift. The increasing digitization of industrial systems demands efficient sensing and control methods, giving rise to soft sensors that have the potential to replace traditional physical sensors in order to reduce costs and enhance efficiency. This study explores the implementation of an artificial neural network (ANN) based soft sensor model in a water supply system to predict flow rates within the system. The soft sensor is centered on a Long Short-Term Memory (LSTM) artificial neural network model using Monte Carlo dropout to reduce uncertainty and improve estimation performance. Based on the results of this work, it is concluded that the proposed soft sensor (with Monte Carlo dropout) can predict flow rates more precisely, contributing to the reduction in water losses, as well as cost savings. This approach offers a valuable solution for minimizing water losses and ensuring the efficient use of this vital resource. Regarding the use of soft sensors based on LSTM neural networks with a careful choice of Monte Carlo dropout parameters, when compared to the multilayer perceptron model, the LSTM model with Monte Carlo dropout showed better mean absolute error, root mean square error, and coefficient of determination: 0.2450, 0.3121, and 0.996437 versus 0.2556, 0.3522, and 0.9954. Furthermore, this choice of Monte Carlo dropout parameters allowed us to achieve an LSTM network model capable of reducing uncertainty to 1.8290, keeping the error metrics also at low levels.

An Investigation into the Interaction of Smart City Governance and Green Economy

Cities can be effectively transformed into smart cities (SCs) to support sustainability, green economy (GE) and sustainable development (SD). Urbanization policies are important for SCs’s effectiveness in GE and sustainability. The study aims to propose suggestion on preliminary pillars of the multi-layered conceptual model towards contribution of smart governance (SG) in SCs to GE and sustainability. Therefore, the study reviews the literature in different search engines (e.g., Web-of-Science, Scopus, Sciencedirect) with keyword combinations of SC, GE, SC governance and sustainability. The study examines SC examples and SG strategies to support GE. The findings highlight the contribution of SC and effective SG to the GE and the multi-layered nature of SC that changes strategically through a multi-actor and multi-sectoral approach. The results emphasize that sustainability interacts with the SC, creating a closed loop that supports SC governance to develop GE. Furthermore, the results provide preliminary suggestion on main pillars of the conceptual model as five aspects (technological, governance, environmental, social, economic) and five keys (ICT-led urban-growth, collaboration, integration, transparency & accountability, green performance) that can support GE because of the interactions of SG and SD. The study can be beneficial for municipalities, urban planners, and researchers in the relevant field.