Federico Pierucci · HAL (Le Centre pour la Communication Scientifique Directe) · 2017
Training machine learning methods boils down to solving optimization problems whose objective functions often decomposes into two parts: a) the empirical risk, built upon the loss function, whose shape is determined by the performance metric and the noise assumptions; b) the regularization penalty, built upon a norm, or a gauge function, whose structure is determined by the prior information available for the problem at hand.Common loss functions, such as the hinge loss for binary classification, or more advanced loss functions, such as the one arising in classification with reject option, are non-smooth. Sparse regularization penalties such as the (vector) l1- penalty, or the (matrix) nuclear-norm penalty, are also non-smooth. However, basic non-smooth optimization algorithms, such as subgradient optimization or bundle-type methods, do not leverage the composite structure of the objective. The goal of this thesis is to study doubly non-smooth learning problems (with non-smooth loss functions and non-smooth regularization penalties) and first- order optimization algorithms that leverage composite structure of non-smooth objectives.In the first chapter, we introduce new regularization penalties, called the group Schatten norms, to generalize the standard Schatten norms to block- structured matrices. We establish the main properties of the group Schatten norms using tools from convex analysis and linear algebra; we retrieve in particular some convex envelope properties. We discuss several potential applications of the group nuclear-norm, in collaborative filtering, database compression, multi-label image tagging.In the second chapter, we present a survey of smoothing techniques that allow us to use first-order optimization algorithms designed for composite objectives decomposing into a smooth part and a non-smooth part. We also show how smoothing can be used on the loss function corresponding to the top-k accuracy, used for ranking and multi-class classification problems. We outline some first-order algorithms that can be used in combination with the smoothing technique: i) conditional gradient algorithms; ii) proximal gradient algorithms; iii) incremental gradient algorithms.In the third chapter, we study further conditional gradient algorithms for solving doubly non-smooth optimization problems. We show that an adaptive smoothing combined with the standard conditional gradient algorithm gives birth to new conditional gradient algorithms having the expected theoretical convergence guarantees. We present promising experimental results in collaborative filtering for movie recommendation and image categorization.
Cited by 0Open access
Qian Song, Yanmin Liu, Xiaoyan Zhang et al. · International Journal of Computational Intelligence Systems · 2024
In the simultaneous optimization of multiple objectives, how to balance convergence promotion and diversity preservation in the evolutionary process is a key and challenging problem. In this research, a hyperplane-assisted multi-objective particle swarm optimization with a twofold proportional assignment strategy (tpahaMOPSO) is suggested to ameliorate the optimization performance of MOPSO. First, the external archive is maintained in combination with hyperplane-based convergence evaluation and shift-based density estimation to retain high-quality candidate solutions. Second, a twofold proportional assignment scheme is designed to search the surrounding region of candidate solutions with better potential to emphasize convergence and diversity, respectively. Third, the domination relationship and convergence difference are combined to select a more reasonable individual historical best and reduce the risk of particle aggregation. Finally, the proposed tpahaMOPSO was compared with ten representative and advanced multi-objective optimization algorithms on 22 widely used test functions with different characteristics. The simulation results present that the developed tpahaMOPSO got the best result in 11 benchmark functions for both IGD and HV criteria. Concurrently, the Friedman test was applied for ranking analysis and the proposed algorithm also obtained excellent statistical analysis results. The promising performance and strong competitiveness of the proposed tpahaMOPSO have been verified by different experimental studies.
Cited by 1Open access
Jiangyu Chen, Lan Li · IET conference proceedings. · 2024
This paper aims to optimize the integration of renewable energy sources into smart grids using artificial intelligence (AI) and data analytics, addressing the challenges posed by the intermittency and variability of renewable energy. The research methodology involves designing an AI-based energy management system that incorporates data analytics, optimization techniques, and renewable energy technologies. The system architecture includes modules for energy generation, storage, and transmission, with key technologies such as smart control strategies and real-time demand response systems. The study employs AI optimization algorithms, including Genetic Algorithms (GA), Particle Swarm Optimization (PSO), and Deep Learning (DL), to dynamically adjust the system based on real-time data inputs. The results demonstrate that Deep Learning outperforms GA and PSO across all metrics, achieving the highest energy efficiency (95%), cost savings (80%), and CO2reduction (70%). The significance of this research lies in its contribution to the reliability and sustainability of energy systems by effectively optimizing energy efficiency, cost savings, and environmental impact. The innovation of this paper is the introduction of a multi-objective optimization framework that simultaneously considers energy efficiency, economic cost, and environmental impact, enabling real-time optimization of the energy management system based on intelligent algorithms. This research underscores the importance of advanced AI-based optimization techniques in the design and operation of smart grids, particularly as the share of renewable energy sources continues to increase.
Cited by 6
Jiaxin Han, Manman He, Ying Cuan · Journal of Physics Conference Series · 2019
Abstract Multi-objective optimization (MOP) is a hot and difficult problem in the field of industrial production and scientific research, involving simultaneous optimization of multiple objectives. In this paper, an improved decomposition-based multi-objective evolutionary algorithm is proposed to solve MOP. Firstly, the Latin hypercube sampling method is used to generate the initial population in order to improve the diversity of the initial population and the probability of convergence to the global optimal solution. Secondly, the adaptive differential crossover operator is used to adjust the parameter settings adaptively for different problems, including the differential proportionality factor and crossover probability, so as to improve the performance of the algorithm. Finally, the reference point is set to coordinate origin, which can pull the population back to the real front faster and accelerate the convergence speed of the population. In a group of benchmark problems, ZDT and DTLZ test problems, the algorithm is widely compared with four most advanced algorithms. The experimental results show that the algorithm has better ability to solve MOP problems.
Cited by 0Open access
Breno Miranda · 2023
The electrical motors are electro-mechanical energy converters mostly used in the industrial sector, although even with modest presence in the transportation sector. The need for reducing the energy consumption by industries, as well as the emerging market demand for vehicle electrification require the manufacturing of low-cost and high-efficiency electrical motors. The Switched Reluctance Motors (SRMs) are attractive relating to induction and permanent magnet motors, having characteristics as robustness, simpler construction, low-cost and higher allowable operating temperature due to the absence of magnets on the rotor. However, the SRMs present drawbacks as poor torque quality, mechanical vibration and acoustic noise, usually related to the salient teeth, switching components and control strategies imposed by the power converter. The multi-objective optimization techniques, in particular meta-heuristics, coupled to the finite element analysis form a robust tool for either searching the solution, or mitigating such problems. The objective of this thesis is to provide easy implementation of appropriate methods for designing SRMs and controlling some drive system’s parameters by numerical optimization in an efficient way. Two optimization problems are formulated: search for the geometric dimensions that maximize the magnetic flux linkage per magnetic material bulk relation; search for the commutation angles that minimize the torque ripple and maximize the mean torque value for a given operating point. The problems are solved by using the Differential Evolution (DE) and the Speed-constrained Multi-objective Particle Swarm Optimization (SMPSO) algorithms. The obtained results indicate that the optimization algorithms were capable to solve the problems satisfactorily, given the inherit limitations to the proposed methodology, improving the torque quality and production capability, however with motor efficiency and power factor reduction. Finally, an advanced speed and torque technique control, Torque Sharing Function (TSF), with higher improvement capability of the torque quality and efficiency, is proposed to be applied to the converter as future work.
Cited by 0Open access
Vedran Jurdana · Repository of the University of Rijeka Library · 2023
This dissertation addresses the challenges in the spectral representation of nonstationary signals using joint time-frequency distributions (TFDs). Heuristic time-frequency methods introduce additional interfering energy clusters, while the wrong choice of parameters in advanced methods may lead to the loss of useful components, thus restricting their practical use. Existing concentration and entropy measures, along with their optimization methods, are appropriate when the structure of the useful components is preserved. However, these methods lack local positional information on the useful components, thereby making it difficult to account for the possible undesired loss of such components. The absence of a suitable criterion and optimization procedure for advanced methods, such as the compressive sensing (CS) method, has led to an experimental selection of the method's parameters, which requires specialist knowledge and increases the unreliability of its application for a wide range of signals. To surmount these challenges, this dissertation proposes performance criteria based on the localized Rényi entropy (LRE) and the estimated instantaneous frequencies and group delays, formulated as objective functions of multi-objective meta-heuristic optimization algorithms. Two methods are also proposed to further improve the TFD reconstruction method: a locally adaptive sparse reconstruction algorithm based on the LRE and an improved selection of the CS area. The here-proposed methods are tested on synthetic and real-life signals, with a focus on their application in electroencephalogram signal analysis. The obtained results show improvements compared with the existing state-of-the-art concentration measures, optimization procedures, and TFD reconstruction algorithms.
Cited by 0Open access
Fei, Zesong, Li, Bin, Yang, Shaoshi et al. · arXiv (Cornell University) · 2016
Wireless sensor networks (WSNs) have attracted substantial research interest, especially in the context of performing monitoring and surveillance tasks. However, it is challenging to strike compelling trade-offs amongst the various conflicting optimization criteria, such as the network's energy dissipation, packet-loss rate, coverage and lifetime. This paper provides a tutorial and survey of recent research and development efforts addressing this issue by using the technique of multi-objective optimization (MOO). First, we provide an overview of the main optimization objectives used in WSNs. Then, we elaborate on various prevalent approaches conceived for MOO, such as the family of mathematical programming based scalarization methods, the family of heuristics/metaheuristics based optimization algorithms, and a variety of other advanced optimization techniques. Furthermore, we summarize a range of recent studies of MOO in the context of WSNs, which are intended to provide useful guidelines for researchers to understand the referenced literature. Finally, we discuss a range of open problems to be tackled by future research.
Cited by 0Open access
Seyyedeh Newsha Ghoreishi, Anders Clausen, Bo Noerregaard Joergensen · 2017
Over the last decades, evolutionary algorithms have been extensively used to solve multi-objective optimization problems. However, the number of required function evaluations is not determined by nature of these algorithms which is often seen as a drawback. Therefore, a robust and reliable termination criterion is needed to stop the algorithm. There is a huge amount of knowledge encapsulated in the studies targeting termination criteria in evolutionary algorithms, but an updated integrated overview of this knowledge is missing. For this reason, we aim to conduct a systematic research through a comprehensive literature study. We extended the basic categorization of termination criteria to a more advanced one that takes the most common used termination criteria into consideration based on their specifications and the way they have been evolved over time. The survey is concluded by suggesting a road-map for future research directions.
Cited by 26Open access
S.S. Yun, Xiuyan Yang, Bailing Wang et al. · IEEE Transactions on Components Packaging and Manufacturing Technology · 2024
Surface mount technology (SMT) is pivotal in electronic manufacturing, with 60%-70% of quality issues linked to solder paste printing (SPP). This paper presents a real-time optimization strategy using an improved particle swarm optimization (PSO) algorithm. Enhancements in the particle velocity update and a multi-population parallel computing PSO algorithm with ring migration topology improve global search capabilities and optimization speed. A SPP quality prediction model, based on XGBoost, defines the optimization objective. The improved PSO algorithm optimizes five key process parameters, employing a data-driven approach to systematically determine weights. In defect cases, predefined printing parameters undergo adjustments based on defect categories, integrating expert experience and theoretical calculations. The efficacy of the enhanced PSO algorithm is rigorously validated through multiple experiments. Additionally, experiments using RNN and RBF neural network models offer insights into its effectiveness. This comprehensive approach not only addresses real-time quality concerns but also contributes to advancing the understanding and application of advanced optimization algorithms in the context of SMT.
Cited by 4
Anil Kumar D. B., Raghu N · Journal of Robotics and Control (JRC) · 2026
Cloud computing has rapidly gained prominence due to its fault tolerance, flexibility, scalability, and pay-as-youuse policy, making it ideal for hosting diverse applications. It has become a dominant paradigm for executing workflow applications with optimized resource utilization. Among its critical components, workflow scheduling plays a pivotal role inmanaging complex constraints and processing large volumes of data. However, while existing multi-objective workflow scheduling algorithms assign tasks effectively, they oftenstruggle to minimize both cost and execution time simultaneously under dynamic conditions. To address these limitations, this article proposes an advanced optimizationbased multi-objective scheduling algorithm named Chronological One-to-One Based Optimization Work schedule (COOBO Work schedule). The main objective of this study is todevelop an optimization-based multi-objective scheduling approach that achieves efficient task allocation while reducing make span, energy consumption, and memory usage, andimproving resource utilization. The approach begins with the simulation of a cloud computing environment incorporating dynamic voltage and frequency scaling. Task scheduling isguided by multiple fitness parameters, including energy consumption predicted using a Deep Feedforward Network. Subsequently, COOBO is employed to optimize the schedulingprocess. Experimental results demonstrate that COOBO-Work schedule outperforms existing methods in terms of make span of 0.016, resource utilization of 0.936, energy consumption of65.877J, memory usage of 23.765 MB, and computational overhead of 0.357sec, thereby validating its effectiveness in cloud workflow management.
Cited by 0Open access
Ravi Ucheniya, Amit Saraswat, Shahbaz Ahmed Siddiqui et al. · IEEE Access · 2024
This paper proposes a new multi-objective stochastic market settlement model for a coupled energy and reactive power ancillary service, named as MOSMS-CERP model, for a prospective grid integration of wind energy. It is a comprehensive two-stage scenario-based stochastic optimization model to incorporate the system uncertainties associated with wind power generations along with load demands. In this multi-objective framework, two competing objective functions as the expected total payment functions for energy and reactive power ancillary services are simultaneously minimized while satisfying all the system operating constraints including technical as well as market driven constraints. An efficient hybrid payment strategy is designed for energy market as well as reactive power ancillary service market for creating a competitive and fair market auction. A pay-as-bid payment mechanism is adopted for all thermal generating units whereas a pay-at-MCP (Market Clearing Price) is adopted for all hydro-powered generating units and wind farms for the energy market settlement. However, the ancillary service market of reactive power is settled using uniform price payment mechanism. It is a unique market settlement model, which combines three significant complexities such as stochastic nature, multi-objective constrained optimization, and meta-heuristics. The proposed MOSMS-CERP model is examined by applying five well-known, recent, and advanced meta-heuristic methods such as Non-dominated Sorting Genetic Algorithms (NSGA-II & I-NSGA-III), Multi-Objective Keshtel Algorithm (MOKA), Hybrid Fuzzy Multi-Objective Evolutionary Algorithm (HFMOEA), and Multi-Objective Particle Swarm Optimization (MOPSO) for producing the sets of Pareto-optimal solutions. A fair statistical comparison is carryout for these five meta-heuristic methods based on No Free Launch Theorem. All simulations are conducted on a standard IEEE 24-bus RTS with a doubly fed induction generator (DFIG) based wind farm on a DIgSILENT Power Factory software platform. The obtained simulation results ensure the robustness and effectiveness of the proposed MOSMS-CERP model.
Cited by 7Open access
Baokang Peng, Gang Cheng, Jingyi Qiu et al. · 2025
This paper proposes MOIL, an efficient multi-objective optimization framework for SRAM design that combines neural network (NN) surrogate modeling with NSGA-II evolutionary algorithms. The framework features: 1) An NN surrogate model employing a two-phase training strategy — initially trained on Latin Hypercube Sampling data and progressively refined through incremental learning cycles — achieving 99.83% prediction accuracy on critical SRAM metrics (read/write delay, SNM, leakage power) while eliminating iterative SPICE simulations; 2) An adaptive NSGA-II optimizer incorporating dynamic crowding distance calculation to effectively explore 5-dimensional device parameters (Fin width/height, Lg, PHIG) and generate Pareto-optimal solutions. Experimental results demonstrate MOIL's superior efficiency with 20.8× faster decision-making and 13.7× speedup over Bayesian optimization method, while maintaining solution diversity (hypervolume ratio of 3.52). The framework establishes a robust methodology for rapid SRAM evaluation and optimization in advanced technology nodes.
Cited by 0
Armah, Abraham, Abankwa, Benjamin, Minnoh, Peter Kenneth · Zenodo (CERN European Organization for Nuclear Research) · 2025
The rapid development of renewable energy systems has created an unprecedented demand for critical minerals, which places the United States at a crossroads of energy security, environmental sustainability and social responsibility. Traditionally, mining has focused on production efficiency, often leading to extensive environmental damage, displacement of communities and worker safety concerns, thereby undermining long-term sustainability goals. This paper introduces a multi-objective optimization model that combines Artificial Intelligence with ethical considerations and sustainable development goals, which offers a rigorous approach to responsible critical mineral mining across the United States. The paper conducts a systematic literature review to examine five key areas: 1) the vulnerabilities of the critical mineral supply chain and its geopolitical implications; 2) the environmental and social impacts of conventional mining; 3) the role of Artificial Intelligence in the mining industry; 4) ethical principles guiding responsible AI management; and 5) multi-objective optimization of decision-support systems. Synthesis of recent empirical research shows that AI technologies improve ore-grade prediction accuracy by about 30 percent, however, geopolitical risks significantly influence mineral price volatility and supply stability. The analysis reveals that 54% of global mining operations are located on Indigenous land, often without permission and that by 2035, automation could displace 30-45% of the mining workforce. The proposed framework addresses the complex trade-offs among production efficiency, environmental protection, social equity and economic viability, using advanced optimization algorithms. These algorithms incorporate environmental monitoring, community impact considerations and regulatory compliance, which ensures comprehensive decision-making.
Cited by 0Open access
Armah, Abraham, Abankwa, Benjamin, Minnoh, Peter Kenneth · Zenodo (CERN European Organization for Nuclear Research) · 2025
The rapid development of renewable energy systems has created an unprecedented demand for critical minerals, which places the United States at a crossroads of energy security, environmental sustainability and social responsibility. Traditionally, mining has focused on production efficiency, often leading to extensive environmental damage, displacement of communities and worker safety concerns, thereby undermining long-term sustainability goals. This paper introduces a multi-objective optimization model that combines Artificial Intelligence with ethical considerations and sustainable development goals, which offers a rigorous approach to responsible critical mineral mining across the United States. The paper conducts a systematic literature review to examine five key areas: 1) the vulnerabilities of the critical mineral supply chain and its geopolitical implications; 2) the environmental and social impacts of conventional mining; 3) the role of Artificial Intelligence in the mining industry; 4) ethical principles guiding responsible AI management; and 5) multi-objective optimization of decision-support systems. Synthesis of recent empirical research shows that AI technologies improve ore-grade prediction accuracy by about 30 percent, however, geopolitical risks significantly influence mineral price volatility and supply stability. The analysis reveals that 54% of global mining operations are located on Indigenous land, often without permission and that by 2035, automation could displace 30-45% of the mining workforce. The proposed framework addresses the complex trade-offs among production efficiency, environmental protection, social equity and economic viability, using advanced optimization algorithms. These algorithms incorporate environmental monitoring, community impact considerations and regulatory compliance, which ensures comprehensive decision-making.
Cited by 0Open access
V - Dharshini, Anburaman Seetharaman -, S - Srilekha · International Journal For Multidisciplinary Research · 2024
In today's dynamic global economy, supply chain management has emerged as a critical component for businesses seeking to gain a competitive edge. However, the complexity of modern supply chains presents numerous challenges, ranging from cost minimization and environmental impact reduction to ensuring product availability. This research paper proposes a comprehensive approach to address these challenges through the integration of data analytics and machine learning techniques. By leveraging advanced algorithms such as random forest classifier, logistic regression, decision tree algorithm, clustering algorithms, and genetic algorithm, the proposed Multi-Objective Supply Chain Optimization platform aims to revolutionize supply chain management. The platform offers decision-makers actionable insights for informed and sustainable choices, empowering them to achieve a balance between financial efficiency and ecological responsibility. Through a detailed examination of costs, sustainability metrics, and product availability, this paper demonstrates the potential of data-driven approaches to transform supply chain management. Furthermore, the paper explores the implementation of the Multi-Objective Supply Chain Optimization web application, highlighting its key functionalities and modules. By providing real-time tracking, dynamic route optimization, and expanded environmental impact analysis, the platform offers a holistic solution to the multifaceted challenges faced by modern supply chains. Ultimately, this research paper underscores the critical role of data analytics and machine learning in driving supply chain efficiency, reducing costs, and promoting environmental sustainability, thereby positioning businesses for long-term success in an increasingly interconnected world.
Cited by 0Open access
Akvile Giedraityte, Sigitas Rimkevičius, Mantas Marčiukaitis et al. · Applied Sciences · 2025
The growing need for sustainable energy solutions has propelled the development of Hybrid Renewable Energy Systems (HRESs), which integrate diverse renewable sources like solar, wind, biomass, geothermal, hydropower and tidal. This review paper focuses on balancing economic, environmental, social and technical criteria to enhance system performance and resilience. Using comprehensive methodologies, the review examines state-of-the-art algorithms such as Multi-Objective Particle Swarm Optimization (MOPSO) and Non-Dominated Sorting Genetic Algorithm II (NSGA-II), alongside Crow Search Algorithm (CSA), Grey Wolf Optimizer (GWO), Levy Flight-Salp Swarm Algorithm (LF-SSA), Mixed-Integer Linear Programming (MILP) and tools like HOMER Pro 3.12–3.16 and MATLAB 9.1–9.13, which have been instrumental in optimizing HRESs. Key findings highlight the growing role of advanced, multi-energy storage technologies in stabilizing HRESs and addressing the intermittency of renewable sources. Moreover, the integration of metaheuristic algorithms with machine learning has enabled dynamic adaptability and predictive optimization, paving the way for real-time energy management. HRES configurations for cost-effectiveness, environmental sustainability, and operational reliability while also emphasizing the transformative potential of emerging technologies such as quantum computing are underscored. This review provides critical insights into the evolving landscape of HRES optimization, offering actionable recommendations for future research and practical applications in achieving global energy sustainability goals.
Cited by 60Open access
Pengchao Zhang, Xiang Liu, Zebang Yi et al. · Sustainability · 2024
Big data and artificial intelligence have promoted mining innovation and sustainable development, and the transportation used in open-pit mining has increasingly incorporated unmanned driving, real-time information sharing, and intelligent algorithm applications. However, the traditional manual scheduling used for mining transportation often prioritizes output over efficiency and quality, resulting in high operational expenses, traffic jams, and long lines. In this study, a novel scheduling model with multi-objective optimization was created to overcome these problems. Production, demand, ore grade, and vehicle count were the model’s constraints. The optimization goals were to minimize the shipping cost, total waiting time, and ore grade deviation. An enhanced multi-objective beluga whale optimization (IMOBWO) algorithm was implemented in the model. The algorithm’s superior performance was demonstrated in ten test functions, as well as the IEEE 30-bus system. It was enhanced by optimizing the population initialization, improving the adaptive factor, and adding dynamic domain perturbation. The case analysis showed that, in comparison to the other three conventional multi-objective algorithms, IMOBWO reduced the shipping cost from 7.65 to 0.84%, the total waiting time from 35.7 to 7.54%, and the ore grade deviation from 14.8 to 3.73%. The implementation of this algorithm for truck scheduling in open-pit mines increased operational efficiency, decreased operating costs, and advanced intelligent mine construction and transportation systems. These factors play a significant role in the safety, profitability, and sustainability of open-pit mines.
Cited by 4Open access
Cong Luo, Xinyu Li, Wenyin Gong et al. · IEEE Transactions on Evolutionary Computation · 2025
The flexible job shop scheduling, as the most typical production mode in industrial manufacturing, aims to improve production efficiency. However, the proposal of energy-saving and emission-reduction policy implies that it is impossible to increase the processing speed to improve productivity, and energy consumption is also becoming another important optimization objective. For the multi-objective flexible job shop scheduling problem, the optimization process tends to converge faster in some regions. This is because different scheduling sequences obtain the same objective values, i.e. there is a multimodal characteristic, which is still hardly investigated. Therefore, optimizing the decision space and the objective space simultaneously has become an urgent challenge that needs to be solved. To overcome the above challenges, we model the multimodal multi-objective flexible job shop scheduling problem with variable speed (MMFJSP-S) and propose an affinity propagation hierarchical memetic algorithm (APHMA) to minimize makespan and total energy consumption. Firstly, four problem-specific neighborhood structures are employed to enhance the convergence; Then, an affinity propagation clustering combined with the random forests strategy is proposed to classify the global and local Pareto sets; Finally, a hierarchical environmental selection strategy is designed to ensure the convergence and diversity in the decision and objective spaces. Evaluations against seven advanced algorithms on MK and DP benchmarks demonstrate the competitive performance of APHMA in solving MMFJSP-S.
Cited by 27
Yuanyuan Sun · Engineering Research Express · 2025
Abstract Large-scale construction projects demand advanced methods to optimize scheduling, resources, and costs under dynamic conditions. However, traditional approaches like the Critical Path Method (CPM) struggle with multi-objective trade-offs and real-time adaptability, limiting their effectiveness in complex environments. This research introduces a Multi-Objective Ant Colony Optimization (MO-ACO) framework, designed to simultaneously optimize project duration, cost, and risk. Unlike conventional methods, MO-ACO leverages bio-inspired algorithms to dynamically balance competing objectives, offering a more adaptive and scalable solution for modern construction challenges. Real-world infrastructure project data, including task durations, labor availability, equipment utilization, material costs, and site constraints, were preprocessed using min-max normalization and outlier removal (IQR) to ensure consistency. Principal Component Analysis (PCA) was applied for feature extraction, and the construction workflow was modeled as a weighted directed graph. In this model, artificial ants explore execution paths probabilistically, guided by pheromone trails, enabling real-time, data-driven decision-making. Simulation results demonstrate that MO-ACO outperforms traditional scheduling methods, reducing project delays, improving resource utilization, and achieving an 87.85% scheduling quality benchmark. Beyond performance gains, the framework supports seamless integration with Building Information Modelling (BIM) and predictive analytics, enhancing its practical utility for construction managers. Research bridges the gap between theoretical optimization and real-world application, providing a scalable, adaptive, and integrative tool for modern project management. By prioritizing multi-objective optimization and real-time adaptability, MO-ACO sets a new standard for construction project efficiency.
Cited by 0
Mei Li, Rui Ge, Zian Zhu et al. · Case Studies in Thermal Engineering · 2025
This study conducts a comprehensive thermodynamic investigation and multi-objective optimization of the cryogenic system for the Hefei Advanced Light Facility (HALF), a critical subsystem responsible for maintaining operational stability and performance. Through systematic application of energy and exergy analysis methodologies, a thermodynamic framework is established to quantify system efficiency and pinpoint primary inefficiency sources. The optimization strategy employs five advanced multi-objective evolutionary algorithms (NSGA-III, MOPSO, SPEA2, MOGOA, MOAHA) for Pareto-frontier exploration, addressing the inherent trade-offs between thermodynamic performance and resource consumption. The dynamic analyses are performed within EcosimPro version 5.4, extended by a dedicated HALF cryogenic library based on CRYOLIB 3.2. Post-optimization results demonstrate a 3.66% relative improvement in exergy efficiency (from 55.79% to 59.45%), accompanied by 5.82% reduction in total power consumption and 31.7% decrease in liquid nitrogen usage. A novel life cycle cost analysis (LCCA) model integrates capital expenditure with operational parameters, revealing 6.7% in lifecycle cost reduction potential over the 30-year service period through optimal configuration selection. The dynamic simulation successfully captures the integrated thermo-fluid and control responses of the HALF cryogenic system, demonstrating robust performance from compression through liquefaction to cryomodule cooling. The proposed methodology establishes new benchmarks for cryogenic system design in particle accelerator facilities, offering both theoretical guidance for thermodynamic optimization and practical strategies for sustainable operation of large-scale scientific infrastructures. The engineering design of the HALF cryogenic system, developed by the RF Superconducting and Cryogenics Group of IHEP, successfully passed the design review on June 15, 2024.
Cited by 0Open access