Artificial bee colony and adaptive large neighborhood search based approaches for the star-ring network design problem

This paper presents an improved artificial bee colony (I-ABC)-based offline optimization framework for tuning a fractional-order proportional–integral–derivative (FOPID) controller to regulate the speed of a permanent magnet synchronous motor (PMSM) in electric vehicle (EV) applications. The proposed approach addresses the practical challenge of achieving high dynamic performance and robustness while maintaining deployability in embedded EV motor control systems. The I-ABC algorithm incorporates adaptive neighborhood shrinking, elitist solution retention, and local refinement mechanisms to enhance global exploration and ensure reliable convergence in the five-dimensional FOPID parameter space. Unlike existing ABC- or GA-based PMSM control approaches that rely on online adaptation or hybrid co-simulation, the proposed I-ABC–FOPID framework performs complete offline tuning within MATLAB/Simulink, enabling direct deployment in embedded EV motor controllers with minimal computational burden. The novelty of this work lies in the integration of the I-ABC algorithm with offline FOPID tuning, which consistently outperforms conventional PID, GA–FOPID, and GA–RBL–FOPID controllers. Extensive simulation studies conducted at multiple speed setpoints (300, 600, and 900 rpm) demonstrate that the proposed controller achieves up to 75% reduction in peak overshoot, nearly 50% improvement in rise and settling times, and steady-state error below 0.3%. In addition, the controller maintains clean voltage and current waveforms, minimizes cumulative error indices (ISE, IAE, and ITAE), and exhibits strong robustness against load disturbances. These results confirm that the proposed I-ABC–tuned FOPID controller provides a high-performance, reliable, and energy-efficient solution suitable for real-time PMSM speed control in electric vehicle applications.

The age we are living in today is the age of digital advancement, and more needs to be done to tackle issues that have called for the creation of robust and efficient fraud recognition systems. This paper presents a hybrid model using the Proximal Policy Optimization (PPO) technique with improved artificial bee colony (ABC) optimization for detecting credit card fraud. Fraud detection belongs to one of the key difficulties in the accuracy of classification, particularly when it is required to deal with imbalanced classification—the majority class overwhelming the minority class leads quite often to its misclassification. Our architecture surmounts this challenge by enhancing the training of the reward mechanism of the PPO algorithm that increases the success rate of recognition of the minority class. It not only extends the definition of classification but also integrates it in an Artificial Neural Network (ANN) based architecture as a sequential decision-making task. More importantly, the rewards given for the right sampling with more emphasis on the minority class in turn enhance the capturing capabilities of the model concerning fraudulent transactions. Also, the ABC algorithm of cooperative communication features between the population is used since it can do better at the initial weight levels, which is important in solving problems that are highly complex and of high-dimensional space. The efficiency of this optimization algorithm is tested on credit card databases collected by Université Libre de Bruxelles. A comparison of the empirical results obtained against performance evaluation metrics shows that the algorithm is high in accuracy and, thus, effective in fraudulent transaction detection for e-commerce.

An Efficient Hybrid Artificial Bee Colony Algorithm for the Resource-Constrained Project Scheduling Problem

This study presents a hybrid, metaheuristic-driven optimization framework for power hyperparameter tuning in predictive modeling based on large-scale annual health examination data. Different from conventional grid and random search strategies, the proposed method directly incorporates particle swarm optimization, artificial bee colony, and gravitational search algorithm into the training pipeline of multiple machine learning models, enabling adaptive exploration of high-dimensional parameter spaces under clinical data constraints. The approach was evaluated on a comprehensive dataset comprising 93 clinical attributes and 1,000 patient records, with a specific focus on ischemic stroke risk prediction. Random Forest, decision tree, support vector machine, and logistic regression models were optimized using the proposed hybrid structure and benchmarked against baseline configurations. Experimental results demonstrate consistent and statistically significant reductions in mean squared error, mean absolute error, and root mean squared error, alongside improvements in R2 and classification accuracy exceeding 99% for optimized logistic regression models, while maintaining computational efficiency suitable for routine clinical deployment. Beyond performance gains, the study introduces a stacked ensemble architecture guided by metaheuristic-tuned base learners, enhancing model robustness and generalization across training and independent test sets. These findings demonstrate the practical novelty of integrating swarm and numerical optimization into clinical predictive pipelines, providing a scalable and domain-agnostic solution for high-accuracy risk decision support in preventive healthcare and other data-intensive applications.

The Sparrow Search Algorithm (SSA) is a novel optimization method inspired by sparrows’ foraging and anti-predator behavior. It mimics their exploration and exploitation strategies to find near-optimal solutions for various optimization problems. This paper presents the first application of SSA to the widely recognized Permutation Flowshop Scheduling Problem (PFSP) with the makespan criterion as the optimization target. Our study aims to assess the effectiveness and robustness of this cutting-edge metaheuristic through computational experiments and statistical analysis. The proposed SSA is a hybrid variant that incorporates the Variable Neighborhood Search (VNS) algorithm along with a Path Relinking Strategy. The effectiveness of the proposed method is evaluated through computational experiments on PFSP benchmark instances. The performance of the hybrid SSA is compared against several well-established swarm-intelligence metaheuristics, namely Grey Wolf Optimizer (GWO), Whale Optimization Algorithm (WOA), Tuna Swarm Optimization Algorithm (TSO), Particle Swarm Optimization Algorithm (PSO), Firefly Algorithm (FA), Bat Algorithm (BA), and the Artificial Bee Colony (ABC). To ensure fair comparison, all methods are implemented within the same computational framework as the hybrid SSA. The experimental results show that the proposed hybrid SSA achieves the lowest average mean error compared with the competing methods in solving the PFSP. The results were further validated through a comprehensive non-parametric statistical analysis using Friedman, Aligned Friedman, and Quade tests, followed by post-hoc analysis with p-adjusted values, as well as Kruskal–Wallis and Wilcoxon post-hoc tests.

Discrete artificial bee colony algorithm based on multiple neighborhood swaps for the hybrid flowshop scheduling problem with uncertain processing stages

A learning-driven artificial bee colony algorithm for mobile robot multi-objective path planning

Abstract Fault diagnosis in complex software systems remains a critical challenge in software engineering, particularly under multi-fault and reliability-critical settings. To address the limitations of traditional spectrum-based fault localization techniques, such as poor search efficiency and susceptibility to local optima, this paper proposes a hybrid artificial intelligence framework that integrates Artificial Bee Colony (ABC) optimization with a Chaotic Genetic Algorithm (CGA). Experimental results on the Defects4J benchmark demonstrate a 12.1% improvement in Top-5 accuracy and a 38% reduction in EXAM score compared to particle swarm optimization-based fault localization methods. The proposed framework exploits the global exploration capability of ABC and the intensified local refinement enabled by chaotic operators in CGA to improve fault ranking while avoiding premature convergence. A dynamically weighted fitness function combines information from spectrum-based fault localization metrics, ABC, and CGA to compute final suspicion scores. Statistical significance analysis using the Wilcoxon signed-rank test ( p = 0.007) confirms the effectiveness of the proposed approach, highlighting its potential to support automated debugging in large-scale and distributed software systems.

The integration of photovoltaic (PV) generation with bidirectional electric vehicle (EV) fast-charging systems offers a promising pathway toward sustainable transportation and grid support. However, the dynamic coupling between maximum power point tracking (MPPT) perturbations and grid-side power quality presents a fundamental challenge in such multi-converter architectures. This paper addresses this challenge through a coordinated design and optimization framework for a grid-connected, PV-assisted bidirectional off-board EV fast charger. The system integrates a 184.695 kW PV array via a DC-DC boost converter, a common DC link, a three-phase bidirectional active front-end rectifier with an LCL filter, and a four-phase interleaved bidirectional DC-DC converter for the EV battery interface. A comparative evaluation of three MPPT algorithms establishes the Fuzzy Logic Variable Step-Size Perturb & Observe (Fuzzy VSS-P&O) as the optimal strategy, achieving 99.7% tracking efficiency with 46s settling time. However, initial integration of this high-performance MPPT reveals system-level harmonic distortion, with grid current total harmonic distortion (THD) reaching 4.02% during charging. To resolve this coupling, an Artificial Bee Colony (ABC) metaheuristic algorithm performs coordinated optimization of all critical PI controller gains. The optimized system reduces grid current THD to 1.40% during charging, improves DC-link transient response by 43%, and enhances Phase-Locked Loop (PLL) synchronization accuracy. Comprehensive validation confirms robust bidirectional operation with seamless mode transitions and compliant power quality. The results demonstrate that system-wide intelligent optimization is essential for reconciling advanced energy harvesting with stringent grid requirements in next-generation EV fast-charging infrastructure.

The digital transformation of oil and gas pipeline networks has generated substantial volumes of unstructured maintenance documentation from communication systems, creating an urgent need for automated summarization to improve operational efficiency. However, domain-specific text summarization for pipeline communication maintenance remains challenging due to scarce labeled data and the high computational cost of fine-tuning large pretrained models. Parameter-efficient fine-tuning alleviates this issue, but its effectiveness strongly depends on appropriate hyperparameter selection. This paper proposes a unified framework that combines weight-decomposed low-rank adaptation with an enhanced Artificial Bee Colony algorithm for automated hyperparameter optimization. The enhanced algorithm addresses two specific limitations of the standard Artificial Bee Colony algorithm: uninformed random initialization that ignores promising regions, and premature abandonment of stagnated solutions that discards partially useful search directions. These two components represent principled design choices, each targeting a distinct bottleneck in applying swarm intelligence search to high-dimensional mixed-type hyperparameter spaces. The method introduces a hybrid initialization strategy to exploit prior knowledge and a stagnation-guided local search mechanism to refine stagnated solutions instead of discarding them, achieving a better balance between exploration and exploitation. Experimental results on a public Chinese summarization benchmark and an industrial oil and gas pipeline communication maintenance corpus show that the proposed approach consistently outperforms full fine-tuning, manually tuned parameter-efficient methods, and several evolutionary optimization baselines in terms of ROUGE metrics. The automated search introduces modest additional computational overhead compared to manual tuning while eliminating expert-dependent hyperparameter configuration and achieving consistent performance gains across both datasets. Overall, the proposed framework provides an efficient and robust solution for adapting large language models to specialized summarization tasks in the context of pipeline communication system maintenance.

Abstract Software fault prediction (SFP) improves software reliability and reduces maintenance costs, but real-world datasets often suffer from class imbalance and redundant features, which limit model performance. We present a cascade framework for SFP that integrates (i) SMOTE for class rebalancing, (ii) feature selection via SHAP values and Artificial Bee Colony (ABC) optimization, and (iii) ensemble and non-ensemble classifiers arranged in cascades. Using the PROMISE repository’s KC1 class-level dataset, we binarize the NUMDEFECTS target (faulty = 1 if NUMDEFECTS ≥ 1; otherwise 0), apply SMOTE on the training split, and evaluate eight baseline classifiers and multiple cascades. With SHAP-selected features, SVM achieves 0.7941 accuracy along with strong F1 (0.7742) and MCC (0.6205); with ABC-selected features, Random Forest, AdaBoost, and Gradient Boosting each reach 0.7647 accuracy with F1 ≈ 0.77 and MCC ≈ 0.55. The best cascade (AdaBoost → Random Forest) achieves 0.7941 accuracy while delivering balanced precision–recall performance (F1 ≈ 0.7879, MCC ≈ 0.5893). By consolidating imbalance handling, interpretable feature attribution, and metaheuristic optimization into a reproducible pipeline, this study provides a practical template for class-level fault prediction that can be adapted to other projects and datasets.

We present a design-ready, multi-output GA-ANN framework for simultaneous prediction and design optimisation of blast impacts, rock fragmentation, ground vibration, and airblast at Debswana’s Jwaneng Mine. Using ten input parameters from 120 production blasts, the model both predicts these impacts and supports rapid, constraint-aware blast design that meets regulatory limits while improving productivity. We compare genetic algorithm-artificial neural network (GA-ANN) with particle swarm optimisation (PSO-ANN), artificial bee colony (ABC-ANN), and imperialist competitive algorithm (ICA-ANN), select GA-ANN on held-out accuracy, construct a multi-output solution surface (10-70-25-3 architecture via Monte Carlo), and apply gradient-descent inverse design. On the test set, GA-ANN attains R2 = 0.910 (fragmentation), 0.925 (ground vibration), and 0.967 (airblast); inverse design returns input settings that maximise fragmentation ($$\approx $$ 84%) while minimising vibration ($$\approx $$ 0.10 mm/s) and airblast ($$\approx $$ 41 dB). The learned solution surface reveals operational trade-offs, enables fast constraint-aware “what-if” queries, and provides a retrainable path to safer, more efficient, compliance-oriented blast design at scale.

ABSTRACT To enhance the accuracy of data acquisition in low-voltage distribution station areas and achieve high-precision detection of three-phase unbalanced load line loss, a method based on IoT sensing and swarm intelligence is investigated. This approach designs an IoT sensing terminal consisting of hardware such as a core board and a serial communication board. Using these terminal, real-time operating data of the low-voltage distribution station area under three-phase unbalance conditions are collected. Based on the local outlier factor algorithm, the information entropy of the real-time operating data is calculated to determine whether it exhibits outlier characteristics. The local reachable density among operating data points is computed using weighted distance. By comparing with a set outlier factor threshold, outlier attributes of the operating data are identified, thereby obtaining abnormal data points of three-phase unbalanced load line loss in the low-voltage distribution station area. The load line loss rate corresponding to the line loss detection results in the low-voltage distribution station area is calculated. Detection results with a line loss rate below 10% are eliminated, and the weighted distance parameter is optimised using the artificial bee colony algorithm from swarm intelligence to accomplish three-phase unbalanced load line loss detection in the low-voltage distribution station area. Experimental results demonstrate that the proposed method achieves an accuracy of 96.8% in abnormal line loss detection, with an F1-score of 0.954, representing the highest accuracy among comparative methods. It also accurately acquires voltage data under three-phase unbalance conditions in the low-voltage distribution station area, demonstrating favourable application performance.

There has been huge positive changes in smart infrastructure management due to the creation of systems that perform real time environmental tracking, and process cyber-physical data. These changes are apparent due to the combination of Wireless Sensor Networks (WSNs) with Internet of Things (IoT). Large WSN deployments face obstacles in scheduling, due to restricted energy supplies and operational environments being too hostile or inaccessible. Genetic Algorithms like Simulated Annealing and Artificial Bee Colony tends to perform in suboptimal standards as these algorithms fail to adapt well to environments that have energy depletion along with changing topologies. Hence, a new method for WSN scheduling known as RL-HAPSO, that utilizes Ant Colony Optimization (ACO) and the Particle Swarm Optimization (PSO) algorithms along with the adaptive capability of Q-learning Reinforcement Learning has been addressed in this paper. An energy-efficient node selection by ACO operates during the first phase, followed by PSO optimization, which improves coverage and minimizes redundancy before execution of real-time reinforcement learning algorithm that selects activation schedules based on network states. The model runs multiple simulations, and does performance validation by assessing its execution time and convergence cost along with energy utilization, which is compared to each algorithm independently and the also the hybrid model without RL implementation. Results indicate execution in microseconds interval by each algorithm, yet RL-HAPSO stands out, as it achieves better optimization costs through enhanced fault tolerance, coverage and minimal energy usage. During performance alterations the system automatically adjusts its operations leading to consistent robust behaviour, even in case of node failure and environmental variations. The obtained results indicate that this proposed methodology functions as a viable approach for future-generation IoT applications that support resource-aware and smart WSN scheduling.

Accurate prediction of oil–nitrogen interfacial tension (IFT) is critical for designing efficient enhanced oil recovery (EOR) strategies. Traditional empirical correlations often lack generalizability and demand detailed compositional data, motivating the need for robust machine learning frameworks. In this study, Gradient Boosting Machine (GBM) models were developed and optimized using four metaheuristic algorithms including Artificial Bee Colony (ABC), Particle Swarm Optimization (PSO), Coupled Simulated Annealing (CSA), and Whale Optimization Algorithm (WOA) to predict equilibrium IFT under varying pressures, temperatures, and API gravities. A curated dataset of 148 experimental measurements was validated through outlier detection and evaluated using five-fold cross-validation to ensure generalization. Model performance was assessed using R2, mean squared error (MSE), and average absolute relative error percentage (AARE%). Comparative results demonstrate that the ABC-optimized GBM achieved the highest test R2 and competitive error metrics, outperforming other optimization strategies in predictive reliability. SHAP analysis further confirmed pressure and temperature as the dominant factors influencing IFT, with API gravity exerting a secondary effect. The findings not only establish the ABC-GBM framework as a powerful predictive tool but also reinforce the physical plausibility of the results, offering practical guidance for process optimization in nitrogen-based EOR applications.

Maritime transportation plays a crucial role in inter-island logistics distribution in Indonesia, one of which is through the Sea Toll Road program. Despite its economic benefits, the shipping sector also contributes to environmental pollution due to the high consumption of fossil fuels and the resulting pollutant emissions. The implementation of a carbon tax is one effort to reduce fuel consumption by increasing the burden of ship operational costs. As the Sea Toll route expands every year, this study was conducted to evaluate and optimize shipping routes to be more technically efficient and environmentally friendly. The research focused on 7 Sea Toll routes departing from Tanjung Perak Port, Surabaya, by solving the Capacitated Vehicle Routing Problem (CVRP) using the Artificial Bee Colony (ABC) algorithm, an optimization algorithm inspired by the behavior of honeybees in foraging. The results showed a decrease in total distance traveled from 16,708 NM to 13,816 NM, an increase in average load capacity from 67.86% to 95%, and a decrease in carbon tax costs by 17.3%. The research results prove that the ABC algorithm is effective in designing optimal shipping routes in terms of route efficiency, load distribution, operational costs, and environmental sustainability.

With rapid economic development in China’s coastal regions, more oil stations are being built on soft soil foundations, facing risks such as foundation settlement and pipeline failures. Mechanical vibrations of oil pumps can induce resonance in pipelines, leading to rupture, leakage, and fire or explosion, threatening both safety and sustainable operation. Traditional monitoring methods, relying on physical models or data-driven approaches alone, are limited in capturing these coupled risks. This study proposes an ABC-XGBoost hybrid risk warning model, where the artificial bee colony algorithm optimizes XGBoost hyperparameters (iteration number, tree depth, learning rate) to improve predictive accuracy. By using multidimensional data—such as internal pressure, vibration amplitude, and ground settlement—the model evaluates stress and resonance risks in real time, supporting sustainable safety management. Validation with real station data shows an accuracy of 95.22%, 2.61% higher than the unoptimized model, demonstrating effective early warning and contribution to sustainable pipeline operation.

Purpose Capsule endoscopy offers a minimally invasive approach to gastrointestinal (GI) diagnostics but is limited by passive propulsion, which restricts controllability and risks suboptimal visualisation in complex anatomical regions. This study aims to develop and validate advanced control strategies for the CAPSUBOT, a vibro-impact robotic active capsule endoscope, enabling precise navigation and sustained motion within the GI tract. Methods A two-degree-of-freedom (2-DOF) navigation system is implemented to control the relative motion of the inner mass with respect to the outer mass of the capsule. The capsule dynamics are mathematically modelled and formulated in the state-space domain. Linear Quadratic Regulator (LQR) and various Sliding Mode Controllers (SMC), including first-order SMC, second-order SMC, Integral SMC (ISMC), and Terminal SMC (TSMC), are designed and simulated using MATLAB/Simulink. Frictional forces and interaction effects are incorporated to emulate realistic operational conditions. The controllers’ performance in tracking desired trajectories is systematically evaluated, and the potential for optimisation of LQR weighting matrices using computational techniques such as artificial bee colony, genetic algorithm, particle swarm optimisation, and reinforcement learning is discussed. Results Simulation results demonstrate that both LQR and SMC controllers can effectively guide the CAPSUBOT along commanded trajectories. First-order SMC, ISMC, and TSMC exhibit particularly strong performance in achieving steady-state tracking of the inner mass position and velocity, whereas LQR offers improved trajectory tracking under nominal conditions. The SMC controllers show robustness against disturbances and system uncertainties, supporting their use for sustained in vivo motion. Some time delays are observed in the SMC responses, highlighting areas for further improvement. Conclusion This work establishes a framework for actively controlling a vibro-impact robotic capsule endoscope using LQR and SMC strategies, providing unified principles for the interaction of inner and outer capsule masses under feedback-controlled actuation. While the controllers demonstrate feasibility and robustness, limitations remain: SMC time delays and computational complexity in real-time implementation warrant further investigation. Future work will focus on integrating artificial intelligence-based reinforcement learning, neural networks, and fuzzy logic to develop faster, adaptive controllers. Overall, the study contributes both practical guidance for CAPSUBOT navigation and generalized control principles that may inform the design of other active capsule endoscopy systems.

The Artificial Bee Colony (ABC) algorithm is a simple and effective population-based optimization method, but it may exhibit unstable convergence and weak exploitation capability in discrete and highly constrained problems. This study proposes an improved ABC framework that integrates a probabilistic Uniform crossover operator and a gene-level lock mechanism to enhance convergence stability and local refinement. The framework is applied to an integrated multi-resource allocation problem in liquid transportation, which has not previously been addressed within the ABC literature. The problem requires the simultaneous assignment of drivers, trucks, trailers, and ISO tanks under operational and regulatory constraints. Comparative analysis of different ABC configurations shows that integrating only Uniform crossover reduced the mean cost to 17.78, adding only the lock mechanism reduced it to 29.78, and combining both further decreased it to 14.94, indicating a complementary effect between the two mechanisms. The proposed configuration consistently achieved the lowest mean costs across small, medium, and large datasets. Compared with established metaheuristic algorithms and expert manual planning (34.72), the method produced lower-cost and feasible solutions, demonstrating both algorithmic robustness and practical relevance.