Optimization Models

This master’s thesis analyzes the adoption and implementation of artificial intelligence in organizations in Bosnia and Herzegovina, through the lens of managerial perceptions, organizational capabilities, and sectoral context. The research applies a combined methodological framework—integrating quantitative analysis with qualitative data interpretation using thematic analysis—to capture both the structural and interpretative dimensions of AI-related decision-making. The findings demonstrate that managerial readiness for AI implementation is not a linear function of perceived usefulness or ease of use, but is instead deeply rooted in symbolic interpretations of threats, institutional culture, and perceived control. A particular contribution of the study lies in its conceptualization of singularity as a dynamic threshold at which AI is no longer merely a tool, but a means of redefining organizational decisions, values, and strategic directions. In this context, singularity emerges as the outcome of accumulated subjective beliefs, leadership determination, and institutional empowerment- rather than a self-evident result of digital maturity.

Smart manufacturing environments (digitalized production systems with integrated sensor networks and data analytics capabilities) require advanced predictive maintenance capabilities, yet implementation faces significant barriers due to data privacy concerns and proprietary knowledge protection requirements. Traditional machine learning approaches necessitate centralized data repositories, creating obstacles for collaborative maintenance optimization across organizational boundaries. This research develops and evaluates a federated learning framework that enables effective predictive maintenance while preserving data privacy in manufacturing networks. The study implemented a horizontal federated learning architecture with secure aggregation protocols and differential privacy techniques across multiple aerospace manufacturing facilities. System performance was evaluated through comparative analysis against centralized and standalone approaches across multiple predictive maintenance use cases. The federated approach achieved 93.7% of centralized model accuracy while eliminating cross-facility data sharing, with failure prediction lead times approaching centralized performance while substantially outperforming standalone models. Computational overhead increased modestly, but network data transfer requirements decreased by 94%. Privacy analysis confirmed that proprietary process parameters could not be reconstructed from shared model updates. This research advances smart manufacturing capabilities by providing a practical implementation framework for privacy-preserving predictive maintenance across organizational boundaries, enabling industry collaboration while maintaining intellectual property protection.

Machine Learning (ML) is already part of everyday life, powering applications such as Netflix recommendations, Google Maps navigation, and online fraud detection. This article surveys ten practical, real-world use cases of ML, explaining both the technical models and workflows behind them. Case studies include Netflix’s recommendation engine, Google Assistant’s speech models, and Tesla’s autonomous driving systems. By combining technical insights with practical examples, this study highlights how ML is shaping modern society while pointing to future directions such as emotional AI and real-time translation.both researchers and practitioners.

Developing models and algorithms to generate robust project schedules that are less sensitive to disturbances are essential in today's highly competitive uncertain project environments. This paper addresses robust scheduling in project environments; specifically, we address the discrete time/cost trade-off problem (DTCTP). We formulate the robust DTCTP with three alternative optimization models in which interval uncertainty is assumed for the unknown cost parameters. We develop exact and heuristic algorithms to solve these robust optimization models. Furthermore, we compare the schedules that have been generated with these models on the basis of schedule robustness.

This paper explores how machine learning (ML) is transforming software development by enhancing functionality, decision-making, and performance across diverse industries. It details methodologies for integrating ML models into software systems, including model selection, training, optimization, and real-time deployment. Case studies in fraud detection and embedded applications highlight challenges, optimization strategies, and the importance of continuous learning in maintaining model accuracy. With advancements such as machine learning operations (MLOps) and federated learning, ML is becoming more scalable, transparent, and accessible, even in resource-constrained environments.

Keywords: Machine Learning, Artificial Intelligence, MLOps, Federated Learning, Explainable AI, Predictive Analytics, Fraud Detection, Edge Computing, Software Engineering.
DOI: https://doi.org/10.58812/wsnt.v3i02.1964

Discovery of the optimal best possibility of location for facilities is the central problem associated in logistics management. The optimal places for the distribution centres (DCs) can be based on the selected attributes that are crucial to locate the best possible locations to increase the speed of the facility service and thus reduce the overall transport cost and time and to provide best service. The major task is to identifying and locating the required number of DCs and its optimum locations are considered as the important goals for the design of any logistics network. The number of DCs will clearly depends upon many factors like population, capacity of the facility, type of facility etc. but locating the optimum locations of DCs will reduce the overall cost. But, for solving such a wide problem space, the powerful tools are the soft computing based approaches and that are well suited and find a meaningful solution in finite time. In this work, we are going to find the optimum locations of DCs for logistics using various soft computing methods. .

Developing models and algorithms to generate robust project schedules that are less sensitive to disturbances are essential in today's highly competitive uncertain project environments. This paper addresses robust scheduling in project environments; specifically, we address the discrete time/cost trade-off problem (DTCTP). We formulate the robust DTCTP with three alternative optimization models in which interval uncertainty is assumed for the unknown cost parameters. We develop exact and heuristic algorithms to solve these robust optimization models. Furthermore, we compare the schedules that have been generated with these models on the basis of schedule robustness.

Edge AI represents a transformative shift in artificial intelligence deployment, moving computational intelligence from centralized cloud infrastructure to distributed edge devices and servers. This paradigm evolution addresses critical challenges in latency, bandwidth utilization, and real-time processing requirements that modern applications demand. By leveraging optimized models and specialized hardware

This article employs a hydro-economic optimization model to analyze the effects of the Grand Ethiopian Renaissance Dam on the distribution and magnitude of benefits in the Eastern Nile. Scenarios are considered based on plausible institutional arrangements that span varying levels of cooperation, as well as changes in hydrological conditions (water availability). The results show that the dam can increase Ethiopia's economic benefits by a factor of 5-6, without significantly affecting or compromising irrigation and hydropower production downstream. However, increasing GERD water storage during a drought could lead to high costs not only for Egypt and Sudan, but also for Ethiopia.

In recent years, reinforcement learning (RL) has garnered increasing attention for its applications in various domains, including finance, robotics, and healthcare. One critical area in healthcare where RL has shown potential is the optimization of insurance premium pricing. Traditional pricing models in healthcare insurance often rely on a combination of actuarial calculations and statistical methods. However, these methods can be limited by a lack of dynamic adaptability to the constantly changing nature of healthcare risks. This article explores the potential for using RL to optimize healthcare insurance premium pricing, offering a detailed review of existing models, the challenges in the domain, and the fundamentals of reinforcement learning. Additionally, it provides insights into how RL can be integrated into pricing strategies, discusses various methodologies, and highlights the potential improvements in accuracy, efficiency, and adaptability.

Our study aims to assess the impact of integrating environmental, social, and governance (ESG) criteria into the portfolio optimization process in Morocco, providing practical information for socially responsible (SR) investors through the use of two distinct models. Based on Markowitz’s mean-variance (MV) theory, the first model aims to define the optimal allocations for the entire set of companies comprising the Moroccan portfolio. The second model adopts a multicriteria approach by considering a constraint that sets a minimum ESG score that the investor is willing to accept. The construction of the optimized portfolio in these two models involves a preliminary filtering based on the Sharpe ratio applied to a set of 69 companies listed on the Casablanca stock exchange (CSE). A comprehensive comparison of the results obtained by each model reveals that SR portfolios in Morocco do not outperform their conventional counterparts. However, attractive opportunities emerge, allowing for the creation of portfolios that strike a balance between returns and risk while maintaining significant ESG scores. It is worth noting that stricter choices regarding ESG scores require investors to accept higher levels of risk for higher returns or to settle for lower returns for reduced risk. These results underscore that, in the Moroccan context, SR portfolio optimization remains a relatively new exploration, as previous studies have primarily focused on socially responsible investing (SRI) or SR funds. Our approach brings an original contribution to this field, encouraging further exploration by other researchers to include different types of assets or explore alternative optimization models.

Developing models and algorithms to generate robust project schedules that are less sensitive to disturbances are essential in today's highly competitive uncertain project environments. This paper addresses robust scheduling in project environments; specifically, we address the discrete time/cost trade-off problem (DTCTP). We formulate the robust DTCTP with three alternative optimization models in which interval uncertainty is assumed for the unknown cost parameters. We develop exact and heuristic algorithms to solve these robust optimization models. Furthermore, we compare the schedules that have been generated with these models on the basis of schedule robustness.

Batch size optimization for deep learning training is a critical challenge that greatly affects model performance and training efficiency as well as resource utilization. In this paper, we present a comprehensive analysis of the difficulties brought by both large and small batch sizes in deep learning, from both big and small batch sizes perspectives. With increasing complexity in modern neural networks and increasing dataset size, it is paramount both as a practitioner and a researcher to understand the implications of batch size selection. Different batch size configurations, as well as proposed solutions and mitigation strategies from current literature, are explored in this paper in terms of technical, computational, and theoretical challenges from these different batch sizes.

When applying optimizations, a number of decisions are made using fixed strategies, such as always applying an optimization if it is applicable, applying optimizations in a fixed order and assuming a fixed configuration for optimizations such as tile size and loop unrolling factor. While it is widely recognized that these fixed strategies may not be the most appropriate for producing high quality code, especially for embedded systems, there are no general and automatic strategies that do otherwise. In this paper, we present a framework that enables these decisions to be made based on predicting the impact of an optimization, taking into account resources and code context. The framework consists of optimization models, code models and resource models, which are integrated for predicting the impact of applying optimizations. Because data cache performance is important to embedded codes, we focus on cache performance and present an instance of the framework for cache performance in this paper. Since most opportunities for cache improvement come from loop optimizations, we describe code, optimization and cache models tailored to predict the impact of applying loop optimizations for data locality. Experimentally we demonstrate the need to selectively apply optimizations and show the performance benefit of our framework in predicting when to apply an optimization. We also show that our framework can be used to choose the most beneficial optimization when a number of optimizations can be applied to a loop nest. And lastly, we show that we can use the framework to combine optimizations on a loop nest.

The integration of large language models (LLMs) like Llama 2 into cloud-based machine learning platforms such as Amazon SageMaker presents a significant opportunity for advancing conversational AI applications. This paper explores the best practices for deploying and optimizing Llama 2 Chat, an advanced language model, within the SageMaker environment. Through a comparative study, we analyze the performance, scalability, and costeffectiveness of different deployment strategies, focusing on the unique capabilities of SageMaker that can enhance Llama 2's functionalities. We investigate several key areas, including model training, inference optimization, resource management, and security considerations. By leveraging SageMaker's robust features such as automated model tuning, elastic infrastructure, and integrated security, we aim to provide insights into achieving optimal performance and efficiency when utilizing Llama 2 Chat. Our study includes practical experiments and benchmarks to illustrate the impact of various configurations on model latency, throughput, and cost. The findings offer valuable guidance for developers and organizations aiming to implement Llama 2 in real-world applications, ensuring a balance between computational efficiency and conversational quality. This paper contributes to the growing body of knowledge on LLM deployment in cloud environments, providing a comprehensive overview of strategies that align with industry standards and best practices.

In the ever-evolving era of Artificial Intelligence (AI), model performance has constituted a key metric driving innovation, leading to an exponential growth in model size and complexity. However, sustainability and energy efficiency have been critical requirements during deployment in contemporary industrial settings, necessitating the use of data-efficient approaches such as few-shot learning. In this paper, to alleviate the burden of lengthy model training and minimize energy consumption, a finetuning approach to adapt standard object detection models to downstream tasks is examined. Subsequently, a thorough case study and evaluation of the energy demands of the developed models, applied in object detection benchmark datasets from volatile industrial environments, is presented. Specifically, different finetuning strategies, as well as utilization of ancillary evaluation data during training, are examined, and the trade-off between performance and efficiency is highlighted in this low-data regime. Finally, this paper introduces a novel way to quantify this trade-off through a customized Efficiency Factor metric.

This article comprehensively explores Low-Rank Adaptation (LoRa), an innovative optimization technique for deep learning models. It delves into the theoretical foundations, implementation strategies, and real-world applications of LoRa across various domains, including natural language processing, computer vision, and speech recognition. The article discusses how LoRa leverages low-rank matrix factorization to create efficient, task-specific adaptations of large models, significantly reducing the number of trainable parameters while maintaining performance. It also examines the advantages and trade-offs of LoRa, its compatibility with other optimization methods, and emerging research directions such as dynamic rank selection, hybrid approaches, and hardware-aware implementations. By presenting case studies and discussing advanced topics, this article aims to provide researchers and practitioners with a thorough understanding of LoRa's potential to push the boundaries of deep learning optimization and enable more efficient and accessible AI systems.

This paper investigates the equilibrium growth dynamics of an economy whose production is based on natural resources and which seeks to maximize welfare to the local community. This involves determining the optimal trajectories of consumption in the local area and the use of the environmental resource. Economic dynamics are affected by negative environmental externalities which are explicitly included as unfavourable effects in a linear production function. The analysis shows the existence of local and global indeterminacy.

The process of finding a correct forecast equation for photovoltaic electricity production from renewable sources is an important matter, since knowing the factors affecting the increase in the proportion of renewable energy production and reducing the cost of the product has economic and scientific benefits. This paper proposes a mathematical model for forecasting energy production in photovoltaic (PV) panels based on a self-organizing feature map (SOFM) model. The proposed model is compared with other models, including the multi-layer perceptron (MLP) and support vector machine (SVM) models. Moreover, a mathematical model based on a polynomial function for fitting the desired output is proposed. Different practical measurement methods are used to validate the findings of the proposed neural and mathematical models such as mean square error (MSE), mean absolute error (MAE), correlation (R), and coefficient of determination (R 2). The proposed SOFM model achieved a final MSE of 0.0007 in the training phase and 0.0005 in the cross-validation phase. In contrast, the SVM model resulted in a small MSE value equal to 0.0058, while the MLP model achieved a final MSE of 0.026 with a correlation coefficient of 0.9989, which indicates a strong relationship between input and output variables. The proposed SOFM model closely fits the desired results based on the R 2 value, which is equal to 0.9555. Finally, the comparison results of MAE for the three models show that the SOFM model achieved a best result of 0.36156, whereas the SVM and MLP models yielded 4.53761 and 3.63927, respectively. A small MAE value indicates that the output of the SOFM model closely fits the actual results and predicts the desired output.

This paper presents an extensive mathematical analysis of cutting-edge artificial neural network architectures optimized explicitly for edge computing devices. Leveraging sophisticated mathematical formulations and innovative research, we unveil the intricacies of neural network design and optimization in the context of edge computing. Through rigorous mathematical derivations, empirical validations, and extensive literature review, we demonstrate the efficacy and potential of advanced neural networks for edge-centric applications. Performance Metrics: Accuracy: We measured the classification accuracy for image classification tasks and the F1-score for speech recognition and anomaly detection. Latency: We recorded the inference time required for processing a single data sample on edge devices. Energy Consumption: We measured the energy consumed during inference tasks using power monitoring tools.

Optimization techniques have demonstrated their capability to obtain economic and sustainable designs while meeting the performance and safety requirements in a number of engineering disciplines. In the bridge engineering field, the increasing main span length of superlong span bridges makes the wind-resistant design a top priority. However, it has been traditionally conducted following heuristic rules based on experimental analyses. Alternatively, the last advances in CFD and metamodeling techniques enables the development of optimization frameworks. The right conception of the aerostructural optimization problem is crucial to achieve noticeable reductions in the required amount of material and comprehensive safe designs. This paper reports the last advances in the authors' line of research that pursues the definition of a numerical methodology for the full aerostructural optimization of long-span bridges considering shape and size design variables. The capabilities of the method are presented, and the necessity of considering uncertainty in the process is discussed.

The unit commitment problem in power plant operation planning is addressed. For a real power system comprising coal-and gas-red thermal and pumped-storage hydro plants a large-scale mixed integer optimization model for unit commitment is developed. Then primal and dual approaches to solving the optimization problem are presented and results of test runs are reported.

This article employs a hydro-economic optimization model to analyze the effects of the Grand Ethiopian Renaissance Dam on the distribution and magnitude of benefits in the Eastern Nile. Scenarios are considered based on plausible institutional arrangements that span varying levels of cooperation, as well as changes in hydrological conditions (water availability). The results show that the dam can increase Ethiopia's economic benefits by a factor of 5-6, without significantly affecting or compromising irrigation and hydropower production downstream. However, increasing GERD water storage during a drought could lead to high costs not only for Egypt and Sudan, but also for Ethiopia.

Explaining the reason for model’s output as diabetes positive or negative is crucial for diabetes diagnosis. Because, reasoning the predictive outcome of model helps to understand why the model predicted an instance into diabetes positive or negative class. In recent years, highest predictive accuracy and promising result is achieved with simple linear model to complex deep neural network. However, the use of complex model such as ensemble and deep learning have trade-off between accuracy and interpretability. In response to the problem of interpretability, different approaches have been proposed to explain the predictive outcome of complex model. However, the relationship between the proposed approaches and the preferred approach for diabetes prediction is not clear. To address this problem, the authors aimed to implement and compare existing model interpretation approaches, local interpretable model agnostic explanation (LIME), shapely additive explanation (SHAP) and permutation f...

In this study, breast cancer prediction model is proposed with decision tree and adaptive boosting (Adboost). Furthermore, an extensive experimental evaluation of the predictive performance of the proposed model is conducted. The study is conducted on breast cancer dataset collected form the kaggle data repository. The dataset consists of 569 observations of which the 212 or 37.25% are benign or breast cancer negative and 62.74% are malignant or breast cancer positive. The class distribution shows that, the dataset is highly imbalanced and a learning algorithm such as decision tree is biased to the benign observation and results in poor performance on predicting the malignant observation. To improve the performance of the decision tree on the malignant observation, boosting algorithm namely, the adaptive boosting is employed. Finally, the predictive performance of the decision tree and adaptive boosting is analyzed. The analysis on predictive performance of the model on the kaggle b...

Breast cancer is the most common type of cancer occurring mostly in females. In recent years, many researchers have devoted to automate diagnosis of breast cancer by developing different machine learning model. However, the quality and quantity of feature in breast cancer diagnostic dataset have significant effect on the accuracy and efficiency of predictive model. Feature selection is effective method for reducing the dimensionality and improving the accuracy of predictive model. The use of feature selection is to determine feature required for training model and to remove irrelevant and duplicate feature. Duplicate feature is a feature that is highly correlated to another feature. The objective of this study is to conduct experimental research on three different feature selection methods for breast cancer prediction. Sequential, embedded and chi-square feature selection are implemented using breast cancer diagnostic dataset. The study compares the performance of sequential embedde...

90% of disponible data are created in recent years. Big Data term was know for the first time since 2005, and even before in Mesopotamia, in order to register the increased of their productions. But evolution erea of Big Data started at 20 century. Early data are from 1887, when Herman Hollerith created a computer that read wholes made on a card to organize registered data. Every our device is connected with internet of things (IoT), from which we can use and collect data. Collected data can help business understanding consumer model and behaviors. But big data is more than that. Big Data can help schientifics to face global problems, and business to face the right decision. The best example how big data had changed our live are social media. Use of big data collected from social media network help business to understand consummator behavior, audience groups and their dedication on studied situation. Our research focused in building an analysis informatic model, to analyse data collected from facebook pages.

Alborz multi-purpose reservoir dam is located in the Pashakola Babol basin in Mazandaran province. Optimal operation of the water volume stored in the dam reservoir is significant subject for managers and operators. In the rule curve prepared for Alborz dam, the purpose of flood control is not considered and rule curve was prepared only from the standpoint of properly operation for drinking and agricultural needs. According to the probable maximum flood of dam (1466.5 m3/s) and the outlets maximum capacity embedded in the dam (1183 m3/s) and lake of side spillway and fuse plug at the dam site, preparing the rule curve in flood conditions is necessary. In this study, in order to management and estimation of entering flood, a simulation-optimization mathematical model of flood was prepared which is capable of flood optimal routing for different return periods, with the goal of reducing flood damage. In case of equipping the dam basin to flood warning system in order to information of the flood time and volume, can be paid to reservoir optimal management in critical conditions of flood occurrence and reducing flood damage. The results were showed that for reducing 10000 years flood damage and PMF damage, respectively 10 and 35 hours before emptying of reservoir is needed until on the time of flood entering, the reservoir level is respectively at least 0.5 and 4 meters below the normal level.

U radu se daje pregled literature koja prikazuje rezultate istraživanja o utjecaju pojedinih tipova organizacijske kulture na dijeljenje znanja u poduzećima. Istraživanje je obuhvatilo 17 znanstvenih radova iz citatnih baza Web of Science i Scopus, a koji su pružili raznolike dokaze o utjecaju pojedinih tipova kulture na različite dimenzije i vrste dijeljenja znanja. U analizi je naglasak stavljen na tipologiju utemeljenu na Modelu suparničkih vrijednosti, budući da ona predstavlja teorijsku podlogu za proučavanje utjecaja organizacijske kulture na dijeljenje znanja u većini dostupnih studija. Pokazalo se kako klanski tip organizacijske kulture najviše pogoduje dijeljenju znanja u poduzećima, zatim adhokratski tip, dok tržišni tip ima negativan utjecaj. Za hijerarhijski tip kulture rezultati su poprilično nekonzistentni i potrebna su dodatna istraživanja. Implikacije navedenih rezultata za praktičare ovim su jasne. Poduzeća koja žele poticati dijeljenje znanja među svojim zaposlenicima trebala bi njegovati vrijednosti kao što su timski rad, prijateljsko ozračje, međusobno povjerenje i odanost organizaciji, jer su to glavne karakteristike klanskog tipa kulture. U raspravi su, nakon analize rezultata, ponuđene neke smjernice za buduća istraživanja.

In recent years, machine learning is attaining higher precision and accuracy in clinical heart disease dataset classification. However, literature shows that the quality of heart disease feature used for the training model has a significant impact on the outcome of the predictive model. Thus, this study focuses on exploring the impact of the quality of heart disease features on the performance of the machine learning model on heart disease prediction by employing recursive feature elimination with cross-validation (RFECV). Furthermore, the study explores heart disease features with a significant effect on model output. The dataset for experimentation is obtained from the University of California Irvine (UCI) machine learning dataset. The experiment is implemented using a support vector machine (SVM), logistic regression (LR), decision tree (DT), and random forest (RF) are employed. The performance of the SVM, LR, DT, and RF models. The result appears to prove that the quality of the feature significantly affects the performance of the model. Overall, the experiment proves that RF outperforms as compared to other algorithms. In conclusion, the predictive accuracy of 99.7% is achieved with RF.

Considering the current state of Ukraine's energy sector in a dangerous and unstable environment, the operation of nuclear power plants is one of the most important sources of electricity supply for the state. The development of the theory of optimal control and the technical level of automated process control systems in the current state of affairs make it possible not to reduce the efficiency of controlling a nuclear power unit, not only by changing the parameters at any load, but also by changing the structure of automation equipment capable of implementing a certain static control program of power, which currently eliminates the influence of internal and external disturbances, without reducing safety indicators. The article considers the situation when the linear programming method is not applicable for solving the control problem. In this case, the decision at each stage does not depend on the decision at the previous one. This is where target programming comes in handythis is one of the methods of multi-criteria optimization, in the theory of which decision-making problems are solved simultaneously according to several criteria. The purpose of the work is to ensure the safe and efficient operation of a nuclear power unit in an energy system that is not constant in operational states. Using a mathematical model of a nuclear power unit, experiments were carried out to simulate the switching of control programs. The simulation model of the automated power control system, based on the bumpless switching of static programs according to the technological methods of operating power equipment, made it possible to form a change in the current model of the control program, as well as eliminate the static control error and get the response of process dependent parameters to program switching. Also, based on the results of the research, an objective function was proposed, which includes three normalized criteria with different objectives. In particular, the problem of optimizing the switching of static power control programs was solved by minimizing the objective function, which combines such characteristics as the efficiency and safety of the operation of a nuclear power unit.

In accordance with the principles of hierarchical management, a comprehensive two-level management system is presented for the development and manufacturing of products for the stages of pre-production (the upper level of the management hierarchy) and for the actual production stage (the lower level of the management hierarchy). At the stage of pre-production, the gear grinding operation design on the "MAAG" type machines was carried out. For this purpose, a technique for optimizing the gear grinding parameters for a two dish-wheel rolling scheme has been developed, a mathematical optimization model containing an objective function with restrictions imposed on it has been created. The objective function is the gear grinding machine time, which depends on the operation parameters (gear grinding stock allowance, cutting modes, grinding wheel specification, part material) and the design features of the gears being ground (module, diameter, number of teeth, radius of curvature of the involutes). The article shows that at the stage of pre-production, the gear grinding optimization is a method of operation design. At the stage of actual production, a closedloop automatic control system with feedback on the deviation of the adjustable value (gear grinding power) automatically supports the numerical power values that were found at the operation design stage, taking into account ensuring defect-free high-performance gear grinding (minimum number of working strokes and maximum longitudinal feeds). At this stage, i.e. when a robust longitudinal feed automatic control system is operating, the optimization carried out at the previous stage (pre-production) sets the functioning algorithm for the adaptive system with corresponding control algorithm. Thus, at the production stage (when the gear grinding machine is running), the operation optimization is a control method. Therefore, it is shown that with two-level control, the gear grinding operation optimization performs a dual function. On the one hand, it is a design method (at the pre-production stage), and on the othera management method (at the actual production stage). With this approach, i.e. with the integration of production and its preparation based on a single two-level management, the efficiency of a single integrated design and production automation system is significantly higher due to general (unified) optimization, rather than partial one.

This paper employs ANN (Artificial Neural Network) models to estimate GHI (global horizontal irradiance) for three major cities in the UAE (United Arab Emirates), namely Abu Dhabi, Dubai and Al-Ain. City data are then used to develop a comprehensive global GHI model for other nearby locations in the UAE. The ANN models use MLP (Multi-Layer Perceptron) and RBF (Radial Basis Function) techniques with comprehensive training algorithms, architectures, and different combinations of inputs. The UAE models are tested and validated against individual city models and data available from the UAE Solar Atlas with good agreement as attested by the computed statistical error parameters. The optimal ANN model is MLP-based and requires four mean daily weather parameters; namely, maximum temperature, wind speed, sunshine hours, and relative humidity. The computed statistical error parameters for the optimal MLP-ANN model in relation to the measured three-cities mean data (referred to as UAE data) are MBE (mean bias error) ¼ À0.0003 kWh/m 2 , RMSE ¼ 0.179 kWh/m 2 , R 2 ¼ 99%, NSE (Nash-Sutcliffe model Efficiency coefficient) ¼ 99%, and t-statistic ¼ 0.005 at 5% significance level. Results prove the suitability of the ANN models for estimating the monthly mean daily GHI in different locations of the UAE.

There are many photovoltaic/thermal (PV/T) systems' designs that are used mainly to reduce the temperature of the PV cell by using a thermal medium to cool the photovoltaic module. In this study, a PV/T system uses nanophase change material (PCM) and nanofluid cooling system was adopted. Three cooling models were compared using nanofluid (SiC-water) and nano-PCM to improve the performance and productivity of the PV/T system. Three mathematical models were developed for linear prediction, and their results were compared with the predicted artificial neural network results, results were verified, and experimental results were appropriate. Three common evaluation criteria were adopted to compare that the results of proposed forecasting models with other models developed in many research studies are done, including the R 2 , mean square error (MSE), and root-mean-square error (RMSE). Besides, different experiments were implemented using varying number of hidden layers to ensure that the proposed neural network models achieved the best results. The best neural prediction models deployed in this study resulted in good R 2 score of 0.81 and MSE of 0.0361 and RMSE and RMSE rate is 0.371. Mathematical models have proven their high potential to easily determine the future outcomes with the preferable circumstances for any PV/T system in a precise way to reduce the error rate to the lowest level.

There are many photovoltaic/thermal (PV/T) systems' designs that are used mainly to reduce the temperature of the PV cell by using a thermal medium to cool the photovoltaic module. In this study, a PV/T system uses nanophase change material (PCM) and nanofluid cooling system was adopted. Three cooling models were compared using nanofluid (SiC-water) and nano-PCM to improve the performance and productivity of the PV/T system. Three mathematical models were developed for linear prediction, and their results were compared with the predicted artificial neural network results, results were verified, and experimental results were appropriate. Three common evaluation criteria were adopted to compare that the results of proposed forecasting models with other models developed in many research studies are done, including the R 2 , mean square error (MSE), and root-mean-square error (RMSE). Besides, different experiments were implemented using varying number of hidden layers to ensure that the proposed neural network models achieved the best results. The best neural prediction models deployed in this study resulted in good R 2 score of 0.81 and MSE of 0.0361 and RMSE and RMSE rate is 0.371. Mathematical models have proven their high potential to easily determine the future outcomes with the preferable circumstances for any PV/T system in a precise way to reduce the error rate to the lowest level.

This paper presents a novel technique that enhances the general precedence, mixed-integer programming approach for the optimal scheduling of process operations. It proves to effectively solve different types of industrial problems dealing with particular sequence-dependent issues, requiring less computational effort than other optimization models. The new formulation takes advantage of the general precedence modeling efficiency, overcoming one of its major limitations.

Mountainous watersheds have always been a challenge for modelers due to large variability and insufficient ground observations, which cause forcing data, model structure, and parameter uncertainty. This study employed Differential Evolution Adaptive Metropolis (DREAM) algorithm which utilizes Markov Chain Monte Carlo (MCMC) approach to account for forcing data uncertainty. A conceptual degree day snowmelt model, MIKE 11 NAM (Nedbor Afstromnings Model), was used to simulate snowmelt runoff from Ilgaz basin, with an area of 28.4 km2 area, located in the northern part of Turkey. The mean elevation is around 1700 m and the basin is covered with broadleaf forest and has mainly brown soil with a high water holding capacity. Precipitation and evapotranspiration (ET) values were optimized in combination with model parameters conditioned on observed discharges and corrected values of input data were utilized for calibration and validation. Results showed that the observed precipitation was o...

The purpose of the present paper is to highlight some features of global dynamics of the two-sector growth model with accumulation of human and physical capital analyzed by Brito, P. and Venditti, A. (2010). In particular, we explore two cases where the Brito-Venditti system admits two balanced growth paths each of them corresponding, after a change of variables, to an equilibrium point of a 3-dimensional system. In the former one, the two stationary states have, respectively, a 2-dimensional and a 1-dimensional stable manifold (i.e. they are, in the Brito-Venditti terminology, locally indeterminate of order 2 and determinate, respectively). In the latter case, instead, the stable manifolds of the two equilibria have, respectively, dimension two and three (i.e. they are locally indeterminate of order 2 and 3). In both cases we prove the possible existence of points P such that in any neighborhood of P lying on the plane corresponding to a fixed value of the state variable there exist points Q whose positive trajectories tend to either equilibrium point. Moreover we show examples where the 2-dimensional stable manifold of the order 2 locally indeterminate equilibrium, in the former case, and the basin of the attracting equilibrium, in the latter case, are proven to be both unbounded.

This work examines the impact that economic growth can have on biodiversity and on the ecological dynamics that would naturally emerge in the absence of human activity. The loss of biodiversity may induce policy makers to implement defensive actions that prevent single species from extinction. These defensive actions, however, may deeply alter the natural dynamics of interaction between species, leading to an ecological equilibrium that is completely different from the one that would exist in the absence of human intervention. This suggests that there might exist a conflict between preserving biodiversity (through stabilization of the ecological system) and preserving the intrinsic features of the ecological dynamics.

In this paper we use global analysis techniques to analyze an economic growth model with environmental negative externalities, giving rise to a three-dimensional dynamic system (the framework is the one introduced by Wirl (1997) [53]). The dynamics of our model admits a locally attracting stationary state P * 1 , which is, in fact, a poverty trap, coexisting with another stationary state P * 2 possessing saddle-point stability. Global dynamical analysis shows that, under some conditions on the parameters, if the initial values of the state variables are close enough to the coordinates of P * 1 , then there exists a continuum of equilibrium trajectories approaching P * 1 and one trajectory approaching P * 2. Therefore, our model exhibits global indeterminacy, since either P * 1 or P * 2 can be selected according to agent expectations. Moreover, we prove that conditions guaranteeing the attractivity of P * 1 also imply the saddle-point stability of P * 2. However, when P * 1 is not attractive, numerical simulations show the possible existence of one or two limit cycles: an attractive one surrounding P * 1 and one endowed with a two-dimensional stable manifold surrounding P * 2 .

As an emerging technology, photovoltaic/thermal (PV/T) systems have been gaining attention from manufacturers and experts because they increase the efficiency of photovoltaic units while producing thermal energy for a variety of uses. Likewise, electric cars are gaining ground as opposed to cars powered by fossil fuels. Electrical vehicles (EVs) are thought to increase environmental pollution by consuming fossil electricity. By combining the two technologies, the environment can be improved, and their popularity can be increased. For the first time, according to authors knowledge, this paper provides a comprehensive review of the applications of PV/T systems for EVs. The paper begins by discussing the need for sustainable and renewable energy sources for EVs, and the advantages of PV/T systems in this context. The review then covers various PV/T system designs, including air-based, water-based, nanofluid, phase change material, hybrid systems, and their integration with EVs. The paper also discusses PV/T systems efficiency and performance in different climates, and the impact of various parameters, such as operating conditions and system design, on their performance. Additionally, the review discusses the challenges and future opportunities for PV/T systems in EV applications, such as the need for cost-effective solutions and the potential for integrating PV/T systems with energy storage and smart charging technologies. The paper concludes that PV/T have significant potential to provide a sustainable and efficient energy source for EVs, and further research and development in this area can lead to innovative and practical solutions for sustainable transportation.

One of the most important issues in the field of water resource management is the optimal utilization of dam reservoirs. In the current study, the optimal utilization of the Aydoghmoush Dam Reservoir is examined based on a hybrid of the bat algorithm (BA) and particle swarm optimization algorithm (PSOA) by increasing the convergence rate of the new hybrid algorithm (HA) without being trapped in the local optima. The main goal of the study was to reduce irrigation deficiencies downstream of this reservoir. The results showed that the HA reduced the computational time and increased the convergence rate. The average downstream irrigation demand over a 10-year period (1991–2000) was 25.12 × 106 m3, while the amount of water release based on the HA was 24.48 × 106 m3. Therefore, the HA was able to meet the irrigation demands better than some other evolutionary algorithms. Moreover, lower indices of root mean square error (RMSE) and mean absolute error (MAE) were obtained for the HA. In a...

The aim of this study is to design and implement soft computing techniques called Support Vector Machine (SVM) and Multilayer Perceptron (MLP) for great management of energy generation based on experimental work. Solar energy could be utilized through thermal systems or Photovoltaics (PV) and it is renewable energy source, environmental friendly and proven globally for a long time. The SVM and MLP models are consist of two inputs layers and one layer output. The inputs of SVM network are solar radiation and time, while the output is the PV current. The inputs of MLP network are solar radiation and ambient temperature, while the output is the PV current. The practical implementation of the proposed SVM model is achieved a final MSE of (0.026378744) in training phase and (0.035615759) in cross validation phase. Besides, MLP is achieved a final MSE of (0.005804253) in the training phase and it is achieved (0.010523501) in cross validation phase. The final MSE of cross validation with standard deviation is (0.000527668). The experiments achieved in the predicting model a value of determination factor (R² = 0.9844388787) for SVM and (R² = 0.9701310549) for MLP which indicates the predicting model is very close to the regression line and a well data fitting to the statistical model. Besides, the proposed model achieved less MSE in comparison with other related work.

The notable developments in renewable energy facilities and resources help reduce the cost of production and increase production capacity. Therefore, developers in renewable energy evaluate the overall performance of the various equipment, methods, and structure and then determine the optimal variables for the design of energy production systems. Variables include equipment characteristics and quality, geographical location, and climatic variables such as solar irradiance, temperature, humidity, dust, etc. This paper investigated and reviewed the current big data methods and tools in solar energy production. It discusses the comprehensive two-stage design and evaluation for examining the optimal structure for renewable energy systems. In the design stage, technical and economic aspects are discussed based on a robust analysis of all input/output variables for determining the highest performance. Next, assess and evaluate the effectiveness of each method under different circumstances...

This article evaluates a 1.4-kW building integrated grid-connected photovoltaic plant. The PV plant was installed in the Faculty of Engineering solar energy lab, Sohar University, Oman, and the system data have been collected for a year from July 2017 to June 2018. The grid-connected system was evaluated in terms of power, energy, specific yield, capacity factor, and cost of energy, and payback period. The measured diffuse and global solar irradiations are 3289 and 6182 Wh/m 2 , respectively. Four predictive models (TLRN, FRNN-1, FRNN-2, and FRNN-3) using deep learning approach based on RNN and TLRN were proposed to predict the PV current performance through data input of temperature (T) and solar irradiance (G). The experiment results found that the highest energy production, array, reference, and final yields are 245.8 kWh, 3.43 to 5.65 kWh/kWp-day, 4.61 to 7.33 kWh/kWp-day, and 3.24 to 4.82 kWh/kWp-day, respectively. Meanwhile, CF, CoE, and PBP were found to be 21.7%, 0.045 USD/kWh and 11.17 years, respectively. The highest performance for prediction models were found for FRNN-2 and FRNN-3 due to they exhibit lower MSE which means being tightly fitted to experiments.

The aim of this study is to design and implement soft computing techniques called Support Vector Machine (SVM) and Multilayer Perceptron (MLP) for great management of energy generation based on experimental work. Solar energy could be utilized through thermal systems or Photovoltaics (PV) and it is renewable energy source, environmental friendly and proven globally for a long time. The SVM and MLP models are consist of two inputs layers and one layer output. The inputs of SVM network are solar radiation and time, while the output is the PV current. The inputs of MLP network are solar radiation and ambient temperature, while the output is the PV current. The practical implementation of the proposed SVM model is achieved a final MSE of (0.026378744) in training phase and (0.035615759) in cross validation phase. Besides, MLP is achieved a final MSE of (0.005804253) in the training phase and it is achieved (0.010523501) in cross validation phase. The final MSE of cross validation with standard deviation is (0.000527668). The experiments achieved in the predicting model a value of determination factor (R² = 0.9844388787) for SVM and (R² = 0.9701310549) for MLP which indicates the predicting model is very close to the regression line and a well data fitting to the statistical model. Besides, the proposed model achieved less MSE in comparison with other related work.

This work examines the impact that economic growth can have on biodiversity and on the ecological dynamics that would naturally emerge in the absence of human activity. The loss of biodiversity may induce policy makers to implement defensive actions that prevent single species from extinction. These defensive actions, however, may deeply alter the natural dynamics of interaction between species, leading to an ecological equilibrium that is completely different from the one that would exist in the absence of human intervention. This suggests that there might exist a conflict between preserving biodiversity (through stabilization of the ecological system) and preserving the intrinsic features of the ecological dynamics.

Federated learning (FL) is a novel methodology aiming at training machine learning (ML) and deep learning (DL) models in a decentralized manner in order to solve three main problems seen in the artificial intelligence (AI) sector, namely, (a) model optimization, (b) data security and privacy, and (c) resource optimization. FL has been established as the "status quo" in today's AI applications especially in the industrial and critical infrastructure (CI) domain, as the three aforementioned pillars are invaluable in assuring their integrity. CIs include important facilities such as industrial infrastructures (smart grids, manufacturing, powerlines, etc.), medical facilities, agriculture, supply chains, and more. Deploying AI applications in these infrastructures is an arduous task that can compromise the CI's security and production procedures, requiring meticulous integration and testing. Even a slight mistake leading to the disruption of operations in these infrastructures can have dire consequences, economical, functional, and even loss of life. FL offers the needed functionalities to galvanize the integration and optimization of

The purpose of the present paper is to highlight some features of global dynamics of the two-sector growth model with accumulation of human and physical capital analyzed by Brito, P. and Venditti, A. (2010). In particular, we explore two cases where the Brito-Venditti system admits two balanced growth paths each of them corresponding, after a change of variables, to an equilibrium point of a 3-dimensional system. In the former one, the two stationary states have, respectively, a 2-dimensional and a 1-dimensional stable manifold (i.e. they are, in the Brito-Venditti terminology, locally indeterminate of order 2 and determinate, respectively). In the latter case, instead, the stable manifolds of the two equilibria have, respectively, dimension two and three (i.e. they are locally indeterminate of order 2 and 3). In both cases we prove the possible existence of points P such that in any neighborhood of P lying on the plane corresponding to a fixed value of the state variable there exist points Q whose positive trajectories tend to either equilibrium point. Moreover we show examples where the 2-dimensional stable manifold of the order 2 locally indeterminate equilibrium, in the former case, and the basin of the attracting equilibrium, in the latter case, are proven to be both unbounded.