This paper investigates the challenges of intelligent traffic light control at urban intersections in the context of the Internet of Things. Increasing vehicle density and mobility have exacerbated traffic congestion and resulted in inefficient use of road infrastructure, particularly at intersections. In addition, the dynamic nature of traffic flow, the unpredictability of human driving behavior, and the complexity of network topologies pose significant obstacles to efficient traffic control. To address these issues, an intelligent control algorithm is proposed that exploits communication between vehicles and infrastructure, as well as between infrastructures. The algorithm incorporates an enhanced version of Dijkstra's algorithm to optimize traffic light operation and minimize vehicle waiting times by dynamically computing shortest paths based on real-time traffic data. Simulation experiments on an urban road network show that the proposed method significantly reduces delays and improves travel efficiency compared to traditional fixed-time traffic control systems. Performance evaluation based on metrics such as average vehicle delay and total travel time confirms significant improvements. In addition, the system demonstrates robustness under varying traffic loads and dynamic road conditions. Future research will focus on extending the approach to highway scenarios and integrating emergency vehicle prioritization mechanisms.

Download article

• APA 7th style

Hambli, C., & Amad, M. (2026). A real-time adaptive traffic light control algorithm at urban intersections for smart cities. Applied Computer Science, 22(1), 20–34. https://doi.org/10.35784/acs_7569

• Chicago style

Hambli, Chahrazad, and Mourad Amad. ‘A Real-Time Adaptive Traffic Light Control Algorithm at Urban Intersections for Smart Cities’. Applied Computer Science 22, no. 1 (2026): 20–34. https://doi.org/10.35784/acs_7569.

• IEEE style

C. Hambli and M. Amad, ‘A real-time adaptive traffic light control algorithm at urban intersections for smart cities’, Applied Computer Science, vol. 22, no. 1, pp. 20–34, doi: 10.35784/acs_7569.

• Vancouver style

Hambli C, Amad M. A real-time adaptive traffic light control algorithm at urban intersections for smart cities. Applied Computer Science. 2026; 22(1):20–34.

As drone-based object detection technology continues to evolve, the demand is shifting from simply detecting objects to enabling users to accurately identify specific targets. For example, users can enter specific targets as prompts to accurately detect the desired objects. To address this need, an efficient text-guided object recognition model has been developed to improve the recognition of small objects. Specifically, an improved version of the existing YOLO-World model is presented. The proposed method replaces the C2f layer in the YOLOv8 backbone with a C3k2 layer, allowing for a more accurate representation of local features, especially for small objects or those with well-defined boundaries. In addition, the proposed architecture improves processing speed and efficiency by optimizing parallel processing, while contributing to a more lightweight model design. Comparative experiments on the VisDrone dataset show that the proposed model outperforms the original YOLO-World model, with precision increasing from 40.6% to 41.6%, recall from 30.8% to 31%, F1 score from 35% to 35.5%, and mAP@0.5 from 30.4% to 30.7%, confirming its improved accuracy. In addition, the model exhibits superior lightweight performance, with the number of parameters reduced from 4 million to 3.8 million and the FLOPs reduced from 15.7 billion to 15.2 billion. These results indicate that the proposed approach provides a practical and effective solution for accurate object detection in drone-based applications.

Download article

• APA 7th style

Jung, H.-K. (2026). A text-guided vision model for enhanced recognition of small instances. Applied Computer Science, 22(1), 35–46. https://doi.org/10.35784/acs_7850

• Chicago style

Jung, Hyun-Ki. ‘A Text-Guided Vision Model for Enhanced Recognition of Small Instances’. Applied Computer Science 22, no. 1 (2026): 35–46. https://doi.org/10.35784/acs_7850.

• IEEE style

H.-K. Jung, ‘A text-guided vision model for enhanced recognition of small instances’, Applied Computer Science, vol. 22, no. 1, pp. 35–46, doi: 10.35784/acs_7850.

• Vancouver style

Jung H-K. A text-guided vision model for enhanced recognition of small instances. Applied Computer Science. 2026; 22(1):35–46.

The application of reinforcement learning techniques to optimization problems has gained increasing attention due to their adaptability, generalization potential, and capacity to handle complex decision-making processes. This study explores the opportunities and limitations of Q-learning, in the context of the classical Assignment Problem, which plays an important role in transportation logistics and resource allocation scenarios. Four variants of the algorithm were developed and evaluated: a basic version, a version incorporating min-max normalization of cost values, a long-term profitability strategy, and a backward optimization approach. For each of the algorithms, the hyperparameters were optimized using the Optuna library and tests were performed on randomly generated cost matrices of varying dimensions (5, 10, 50, 100, and 200). The quality of the solutions was evaluated based on degradation relative to the optimal objective function value. The time to generate solutions was also measured. The results indicate significant differences in the capabilities of different algorithm variants. The basic Q-learning version is characterized by limited effectiveness and high variability, particularly for larger problem instances. Normalization improved computational efficiency and reduced variance, but did not lead to substantial improvements in solution quality for more complex cases. In contrast, the long-term profitability variant demonstrated notable improvements in both solution quality and stability, especially for smaller and medium-sized problems. The backward optimization variant yielded the highest overall solution quality.

Download article

• APA 7th style

Misztal, W., & Nazarewicz, S. (2026). Reinforcement learning for solving optimization problems: Opportunities and limitations on the example of the assignment problem. Applied Computer Science, 22(1), 47–62. https://doi.org/10.35784/acs_8031

• Chicago style

Misztal, Wojciech, and Sybilla Nazarewicz. ‘Reinforcement Learning for Solving Optimization Problems: Opportunities and Limitations on the Example of the Assignment Problem’. Applied Computer Science 22, no. 1 (2026): 47–62. https://doi.org/10.35784/acs_8031.

• IEEE style

W. Misztal and S. Nazarewicz, ‘Reinforcement learning for solving optimization problems: Opportunities and limitations on the example of the assignment problem’, Applied Computer Science, vol. 22, no. 1, pp. 47–62, doi: 10.35784/acs_8031.

• Vancouver style

Misztal W, Nazarewicz S. Reinforcement learning for solving optimization problems: Opportunities and limitations on the example of the assignment problem. Applied Computer Science. 2026; 22(1):47–62.

The increasing complexity of spiral steel pipe production necessitates the implementation of intelligent forecasting methods to predict potential failures. This, in turn, enables the development of reliable evaluation techniques aimed at minimizing unanticipated breakdowns and enhancing the efficacy of maintenance strategies. In the present study, a novel SCADA-integrated framework is proposed, which incorporates Fuzzy Comprehensive Evaluation (FCE) and Artificial Neural Networks (ANN) into mid-to-long-term reliability analysis and machine learning-based short-term fault prediction. The architecture performs dynamic analysis on the health of the equipment, welding, alignment, hydraulics, and motor systems using a synthetic SCADA dataset that includes more than 100,000 time-series data points. The generation of imprecise reliability grades is predicated on essential indicators, including mean time between failures (MTBF), mean time to repair (MTTR), the level of failure, and the difficulty of its detection. These indicators are subsequently modeled through artificial neural networks (ANNs) to enable real-time inference. The multi-week sensor window and alarm logs are used with tree-based classifiers and statistical models to predict faults up to four weeks in advance. The mean prediction accuracy is over 91%, and a cost-benefit analysis indicates that active maintenance planning can result in significant financial savings. The combined use of fuzzy logic and neural networks is particularly valuable in manufacturing environments because it integrates human-like reasoning with data-driven learning, enabling robust decision-making under uncertainty. The all-inclusive solution is a financially reasonable and scalable alternative for implementing predictive diagnostics in industrial steel pipe production settings.

Download article

• APA 7th style

Bakhtiyarov, B., Jabiyeva, A., & Khudaverdiyeva, M. (2026). SCADA-Driven big data framework for fault prediction in spiral steel pipe manufacturing using fuzzy and neural network models. Applied Computer Science, 22(1), 63–81. https://doi.org/10.35784/acs_8104

• Chicago style

Bakhtiyarov, Bakhshali, Aynur Jabiyeva, and Mahabbat Khudaverdiyeva. ‘SCADA-Driven Big Data Framework for Fault Prediction in Spiral Steel Pipe Manufacturing Using Fuzzy and Neural Network Models’. Applied Computer Science 22, no. 1 (2026): 63–81. https://doi.org/10.35784/acs_8104.

• IEEE style

B. Bakhtiyarov, A. Jabiyeva, and M. Khudaverdiyeva, ‘SCADA-Driven big data framework for fault prediction in spiral steel pipe manufacturing using fuzzy and neural network models’, Applied Computer Science, vol. 22, no. 1, pp. 63–81, doi: 10.35784/acs_8104.

• Vancouver style

Bakhtiyarov B, Jabiyeva A, Khudaverdiyeva M. SCADA-Driven big data framework for fault prediction in spiral steel pipe manufacturing using fuzzy and neural network models. Applied Computer Science. 2026; 22(1):63–81.

This study presents a robust decision-making framework for evaluating strategic artificial intelligence (AI) initiatives within DHL's smart supply network. The objective is to prioritize four AI alternatives-autonomous warehouse routing, predictive delivery optimization, AI-driven demand forecasting, and intelligent inventory rebalancing-based on eight strategic criteria, including cybersecurity, adaptability, and infrastructure readiness. A cross-functional panel of experts provided linguistic assessments, modeled using Interval-Valued Hesitant Fuzzy Linguistic Term Sets (IVHFLTS) to capture hesitation and uncertainty. These inputs were aggregated and processed by an extended ELECTRE III method incorporating fuzzy thresholds for indifference, preference, and veto. Sensitivity analysis confirmed the stability of the final ranking under ±10% threshold variation, while consensus evaluation revealed expert divergence, which was mitigated by dynamic reweighting. Predictive delivery optimization and intelligent inventory rebalancing emerged as the top-ranked initiatives, aligning with DHL's strategic goals of customer responsiveness and operational resilience. The methodology demonstrates high robustness, interpretability, and practical relevance for AI-driven logistics planning.

Download article

• APA 7th style

Tamtam, F., & Tourabi, A. (2026). Enhanced ELECTRE III method with interval-valued hesitant fuzzy linguistic sets for multi-criteria group decision-making in smart supply networks. Applied Computer Science, 22(1), 82–98. https://doi.org/10.35784/acs_7703

• Chicago style

Tamtam, Fadoua, and Amina Tourabi. ‘Enhanced ELECTRE III Method with Interval-Valued Hesitant Fuzzy Linguistic Sets for Multi-Criteria Group Decision-Making in Smart Supply Networks’. Applied Computer Science 22, no. 1 (2026): 82–98. https://doi.org/10.35784/acs_7703.

• IEEE style

F. Tamtam and A. Tourabi, ‘Enhanced ELECTRE III method with interval-valued hesitant fuzzy linguistic sets for multi-criteria group decision-making in smart supply networks’, Applied Computer Science, vol. 22, no. 1, pp. 82–98, doi: 10.35784/acs_7703.

• Vancouver style

Tamtam F, Tourabi A. Enhanced ELECTRE III method with interval-valued hesitant fuzzy linguistic sets for multi-criteria group decision-making in smart supply networks. Applied Computer Science. 2026; 22(1):82–98.

The paper is devoted to the problem of comprehensive quality assessment in offset printing. On the basis of the research conducted, the quality indicators of sheet-fed offset printing with dampening are determined, namely, the print color difference, the fine line width, the color combination accuracy, the “gray balance” and the dot gain. These indicators were divided into two groups: the first group reflects the color reproduction, while the second concerns the reproduction of fine image elements. Based on the principles of fuzzy logic, the evaluation terms “low”, “medium”, “high” are assigned to the print quality, and a fuzzy knowledge base of the print quality parameters with the fulfillment of the “if-then” condition is formed. Fuzzy logic equations for the calculation of print quality options are constructed, and the defuzzification operation carried out using the “center of gravity” method allows to obtain a quantitative print quality indicator as a result of observing the corresponding modes of the offset printing technological process. The values of the indicators of the parameters of the offset printing quality obtained according to the results of the control and the calculation of the integral indicator serve as data for the reporting of the process in the Industrial Internet of Things system.​

Download article

• APA 7th style

Repeta, V., Ryvak, P., & Krykhovets, O. (2026). Models for calculating the integral quality indicator of the offset printing process for the IIOT-system. Applied Computer Science, 22(1), 99–109. https://doi.org/10.35784/acs_7953

• Chicago style

Repeta, Vyacheslav, Pavlo Ryvak, and Oleksandra Krykhovets. ‘Models for Calculating the Integral Quality Indicator of the Offset Printing Process for the IIOT-System’. Applied Computer Science 22, no. 1 (2026): 99–109. https://doi.org/10.35784/acs_7953.

• IEEE style

V. Repeta, P. Ryvak, and O. Krykhovets, ‘Models for calculating the integral quality indicator of the offset printing process for the IIOT-system’, Applied Computer Science, vol. 22, no. 1, pp. 99–109, doi: 10.35784/acs_7953.

• Vancouver style

Repeta V, Ryvak P, Krykhovets O. Models for calculating the integral quality indicator of the offset printing process for the IIOT-system. Applied Computer Science. 2026; 22(1):99–109.

Due to the increasing frequency of forest fires and their rapid spread, early detection is critical for effective containment and mitigation. This paper proposes a cost-effective edge-based forest fire detection system that receives images from multiple sources (terrestrial cameras and UAVs) to predict and alert authorities of potential forest fires. The model of the system is built using transfer learning with MobileNetv2 on a realistic and diverse dataset, resulting in a lightweight CNN model that is further optimized by using quantization to reduce its size and improve the inference speed. The proposed model is deployed on an 8-node Raspberry Pi cluster, using Slurm and MPI to manage cluster task scheduling and parallel processing. The proposed system achieves 99.21% accuracy, precision, recall, and F1 score on a realistic test dataset containing vague real-world scenarios such as fog and sunset conditions, with an inference speed of 69 frames per second. These results, along with the system's autonomous and offline operation, cost effectiveness, power efficiency, and scalability, make it ideal for real-time forest fire monitoring at the edge, even in off-grid and remote areas.​

Download article

• APA 7th style

Belferd, A. N. E., Bensenane, H., & Rahmoun, A. (2026). A scalable and cost-effective forest fire detection approach using deep transfer learning on a Raspberry Pi cluster. Applied Computer Science, 22(1), 110–122. https://doi.org/10.35784/acs_8125

• Chicago style

Belferd, Achraf Nasser Eddine, Hamdan Bensenane, and Abdellatif Rahmoun. ‘A Scalable and Cost-Effective Forest Fire Detection Approach Using Deep Transfer Learning on a Raspberry Pi Cluster’. Applied Computer Science 22, no. 1 (2026): 110–122. https://doi.org/10.35784/acs_8125.

• IEEE style

A. N. E. Belferd, H. Bensenane, and A. Rahmoun, ‘A scalable and cost-effective forest fire detection approach using deep transfer learning on a Raspberry Pi cluster’, Applied Computer Science, vol. 22, no. 1, pp. 110–122, doi: 10.35784/acs_8125.

• Vancouver style

Belferd ANE, Bensenane H, Rahmoun A. A scalable and cost-effective forest fire detection approach using deep transfer learning on a Raspberry Pi cluster. Applied Computer Science. 2026; 22(1):110–122.

Stationarity is a fundamental assumption in time series modeling that underlies reliable statistical inference and forecasting. Time series data can be found in many domains, including industry, engineering, finance, economics, epidemiology, and health care analysis. This study addresses stochastic non-stationarity arising from unit root processes. It explores the efficacy of fractional differentiation as a means of achieving stationarity, especially in the context of limited-sample time series data, and attempts to confirm it statistically through experiments. To this end, 24 series of malaria and typhoid incidence were used, from the Adamawa region of Cameroon, collected weekly from January 2021 to December 2023, 14 of which were non-stationary. Four models were tested: Autoregressive Integrated Moving Average (ARIMA), Fractional ARIMA (ARFIMA), Long Short-Term Memory (LSTM), and a hybrid Fractionally-Differentiated-LSTM (FD-LSTM) proposed in this paper. The accuracy of the prediction models was assessed by the Root Mean Square Error (RMSE), Mean Absolute Error (MAE), and Coefficient of Determination (R²) values. The results show that the Pearson correlations between the original time series and the integer-differentiated series are weak, mainly between 0.2 and 0.4, while they are between 0.75 and 0.98 for the fractional-differentiated series. Moreover, ARFIMA outperforms ARIMA by 93% in training and 100% in testing, while FD-LSTM achieves a 100% improvement over the standard LSTM model. These results contribute to the methodological toolkit for time series forecasting in data analytics and highlight the statistical and practical advantages of fractional differencing in small sample preprocessing.​​

Download article

• APA 7th style

Batoure Bamana, A., Sokdou Bila Lamou, Y., Fotsa-Mbogne, D. J., & Shafiee Kamalabad, M. (2026). Addressing non-stationarity with stochastic trend in the context of limited time series data: An experimental survey in healthcare analytics. Applied Computer Science, 22(1), 123–139. https://doi.org/10.35784/acs_7862

• Chicago style

Batoure Bamana, Apollinaire, Yannick Sokdou Bila Lamou, David Jaures Fotsa-Mbogne, and Mahdi Shafiee Kamalabad. ‘Addressing Non-Stationarity with Stochastic Trend in the Context of Limited Time Series Data: An Experimental Survey in Healthcare Analytics’. Applied Computer Science 22, no. 1 (2026): 123–139. https://doi.org/10.35784/acs_7862.

• IEEE style

A. Batoure Bamana, Y. Sokdou Bila Lamou, D. J. Fotsa-Mbogne, and M. Shafiee Kamalabad, ‘Addressing non-stationarity with stochastic trend in the context of limited time series data: An experimental survey in healthcare analytics’, Applied Computer Science, vol. 22, no. 1, pp. 123–139, doi: 10.35784/acs_7862.

• Vancouver style

Batoure Bamana A, Sokdou Bila Lamou Y, Fotsa-Mbogne DJ, Shafiee Kamalabad M. Addressing non-stationarity with stochastic trend in the context of limited time series data: An experimental survey in healthcare analytics. Applied Computer Science. 2026; 22(1):123–139.

Collaborative environmental exploration by a fleet of mobile robots is of growing interest, especially in the context of unknown environments. Exploration algorithms find diverse and critical applications, such as search and rescue, underwater surveillance, and space observation. However, despite significant advances in the field, a persistent gap between research results and their translation into real-world applications is a major obstacle to the deployment of effective solutions. This paper proposes a hybrid approach, called IACO-RL, which combines inverse ant colony optimization (IACO) with reinforcement learning (RL) to improve exploration efficiency. This method aims to maximize space coverage and minimize exploration time, with the additional goal of accurately locating mines hidden in the environment. The IACO algorithm directs robots to scarce or unexplored areas by reversing the classical pheromone deposition mechanism, thus promoting efficient spatial dispersal. For its part, the RL module allows each agent to learn autonomously from its interactions with the environment, thus enhancing its adaptability and local decision-making capacity. Experimental results, obtained through simulations in different environmental scenarios, show that the IACO-RL approach outperforms single methods in terms of coverage, speed and mine detection capacity. These performances confirm the relevance of this hybridization and highlight that effective mine detection results directly from the efficiency of the exploration performed by the multi-robot system.​

Download article

• APA 7th style

Rahmoune, N., & Rahmoune, A. (2026). Efficient multi-robot exploration of unknown environments using inverted ant colony optimization and reinforcement learning. Applied Computer Science, 22(1), 140–153. https://doi.org/10.35784/acs_7891

• Chicago style

Rahmoune, Nabila, and Adel Rahmoune. ‘Efficient Multi-Robot Exploration of Unknown Environments Using Inverted Ant Colony Optimization and Reinforcement Learning’. Applied Computer Science 22, no. 1 (2026): 140–153. https://doi.org/10.35784/acs_7891.

• IEEE style

N. Rahmoune and A. Rahmoune, ‘Efficient multi-robot exploration of unknown environments using inverted ant colony optimization and reinforcement learning’, Applied Computer Science, vol. 22, no. 1, pp. 140–153, doi: 10.35784/acs_7891.

• Vancouver style

Rahmoune N, Rahmoune A. Efficient multi-robot exploration of unknown environments using inverted ant colony optimization and reinforcement learning. Applied Computer Science. 2026; 22(1):140–153.

Path planning and optimization are essential topics in robotics because they directly affect the effectiveness and safety of robot navigation. The application of metaheuristic methods and algorithms in the field of robot motion planning has attracted the attention of researchers in the field of robotics, given the ease of use and efficiency of the methods in coordinating agents. Metaheuristic algorithms have attracted much attention in recent years due to their efficiency in solving complex optimization problems. This paper summarizes the mobile robot path planning with metaheuristic algorithms, along with their strengths and weaknesses. In this paper, we will focus on a few meta-algorithms: Ant Colony Optimization (ACO), Particle Swarm Optimization (PSO), Genetic Algorithms (GA), Artificial Fish Swarm Algorithm (AFSA), Grey Wolf Optimizer (GWO), Bat Algorithm (BA), Firefly Algorithm (FA), and Cuckoo Algorithm (CA). In addition, this study reviews the status of path planning research and its major difficulties to be solved, along with the future trends of path planning.​

Download article

• APA 7th style

Sadiq, S., Abrahim, A., & Sadeeq, H. (2026). A comprehensive review of metaheuristic algorithms for mobile robot path planning. Applied Computer Science, 22(1), 154–170. https://doi.org/10.35784/acs_8050

• Chicago style

Sadiq, Sheren, Araz Abrahim, and Haval Sadeeq. ‘A Comprehensive Review of Metaheuristic Algorithms for Mobile Robot Path Planning’. Applied Computer Science 22, no. 1 (2026): 154–170. https://doi.org/10.35784/acs_8050.

• IEEE style

S. Sadiq, A. Abrahim, and H. Sadeeq, ‘A comprehensive review of metaheuristic algorithms for mobile robot path planning’, Applied Computer Science, vol. 22, no. 1, pp. 154–170, doi: 10.35784/acs_8050.

• Vancouver style

Sadiq S, Abrahim A, Sadeeq H. A comprehensive review of metaheuristic algorithms for mobile robot path planning. Applied Computer Science. 2026; 22(1):154–170.

Autolube systems have been widely adopted in the mining industry to improve equipment reliability, but most still operate at fixed time intervals without adapting to real conditions in the field, and monitoring systems use LED lights, making it difficult to diagnose failures due to the minimal system interface. To overcome these issues, this study developed Smart Autolube based on Artificial Intelligence of Things (AIoT), which integrates sensor-based monitoring with machine learning models for adaptive lubrication pressure prediction. With industry support from PT. Multindo Technology Utama, the system was tested under mining simulation conditions using pressure, temperature, and stress sensor data. After preprocessing, which includes winsorization, feature engineering (lag, rolling statistics, and trends), two ensemble algorithms, Random Forest Regressor (RFR) and Gradient Boosting Regressor (GBR), are used to build a prediction model. The base model showed low accuracy (R² < 0.1), but after feature engineering and extreme hyperparameter tuning, the performance improved significantly with an R² of 0.9816 for GBR and 0.9711 for RFR. Explanability analysis using SHAP (SHapley Additive exPlanations) shows that engineering features such as trends, lag_2, and rolling_mean_3 contribute the most to the predictions compared to native features such as temperature and voltage. This study proves that Smart Autolube can provide accurate and explainable lubrication pressure predictions. Further research is suggested to expand the scope of the data, add other mechanical parameters, and test the generalization of the model in different industrial environments.​

Download article

• APA 7th style

Khumaidi, A., Risanto, D., Aditya, L., Hamka, W. H., & Al. Jihad, H. (2026). Smart Autolube: Optimized machine learning-based pressure prediction for AIoT lubrication systems. Applied Computer Science, 22(1), 171–183. https://doi.org/10.35784/acs_7947

• Chicago style

Khumaidi, Ali, Risanto Darmawan, Lukman Aditya, Wardhana Halking Hamka, and Hudzaifah Al Jihad. ‘Smart Autolube: Optimized Machine Learning-Based Pressure Prediction for AIoT Lubrication Systems’. Applied Computer Science 22, no. 1 (2026): 171–183. https://doi.org/10.35784/acs_7947.

• IEEE style

A. Khumaidi, R. Darmawan, L. Aditya, W. H. Hamka, and H. Al Jihad, ‘Smart Autolube: Optimized machine learning-based pressure prediction for AIoT lubrication systems’, Applied Computer Science, vol. 22, no. 1, pp. 171–183, doi: 10.35784/acs_7947.

• Vancouver style

Khumaidi A, Darmawan R, Aditya L, Hamka WH, Al Jihad H. Smart Autolube: Optimized machine learning-based pressure prediction for AIoT lubrication systems. Applied Computer Science. 2026; 22(1):171–183.

The study employed applied computer modelling to identify the optimal process parameters for resistance projection welding using the original procedure. The influence of technological parameters (welding power, welding time, electrode pressure) on the quality of 184 welded joints produced by resistance projection welding of steel nuts and S235JR steel plates was examined using computer modelling methods, specifically a combination of machine learning and an evolutionary algorithm. A tree-based model was used to identify relationships between signals, and a genetic algorithm for multi-criteria optimisation. The prepared joints were then examined to determine the impact of the welding parameters on the microstructure, Vickers hardness, and strength of the welded joints (as assessed by pull-off testing). The superior strength of the projection welding joints was achieved through short welding times and high power. Additionally, limited welding time effectively restricted the heat-affected zone, reducing weld hardness and improving the joint's plasticity. The original modelling process enables energy consumption (welding current) to be minimised while maximising joint strength, which was the main aim of the work. Finally, the set of optimised welding parameters selected by AI was verified through sample welding and strength testing, and this was confirmed through final strength testing experiments. ​

Download article

• APA 7th style

Karski, S., Awtoniuk, M., & Szala, M. (2026). Application of artificial intelligence methods to determine the optimal process parameters in resistance projection welding of steel nuts. Applied Computer Science, 22(1), 184–198. https://doi.org/10.35784/acs_8247

• Chicago style

Karski, Szymon, Michał Awtoniuk, and Mirosław Szala. ‘Application of Artificial Intelligence Methods to Determine the Optimal Process Parameters in Resistance Projection Welding of Steel Nuts’. Applied Computer Science 22, no. 1 (2026): 184–198. https://doi.org/10.35784/acs_8247.

• IEEE style

S. Karski, M. Awtoniuk, and M. Szala, ‘Application of artificial intelligence methods to determine the optimal process parameters in resistance projection welding of steel nuts’, Applied Computer Science, vol. 22, no. 1, pp. 184–198, doi: 10.35784/acs_8247.

• Vancouver style

Karski S, Awtoniuk M, Szala M. Application of artificial intelligence methods to determine the optimal process parameters in resistance projection welding of steel nuts. Applied Computer Science. 2026; 22(1):184–198.

Accurate classification of bone fracture severity is critical in orthopedic evaluation. Radiation-based methods such as X-ray, CT, and MRI provide anatomical detail but lack the ability to classify fracture severity. This study presents a non-invasive, vibration-based approach to assessing fracture severity by analyzing dynamic response characteristics. Five goat (metacarpal) bone specimens were examined, including one unfractured bone that served as a reference bone and four with induced fractures in the lateral, longitudinal, and two oblique orientations. Controlled impact excitation was applied using an impact hammer, and acceleration responses were measured using acceleration sensors. Frequency response function (FRF), coherence, and phase shift were calculated using Fast Fourier Transform (FFT) algorithms. Resonance frequency and FRF magnitude served as primary indicators of stiffness loss and damping changes caused by fractures. The reference bone had a resonance frequency of 376 Hz and an FRF magnitude of 12.96 g/N, which was considered the reference parameter. The lateral fracture showed the most severe response, with a 17.98% increase in resonant frequency and a 491% increase in FRF magnitude, indicating significant stiffness redistribution and low damping. Longitudinal and oblique fractures resulted in large resonant frequency reductions of up to 94.4%. The experimental results were obtained using Fast Fourier Transform (FFT) algorithms and Euler-Bernoulli ray theory. These results suggest that vibration analysis is a reliable, quantitative, and non-destructive tool for classifying the severity of bone fractures.​​

Download article

• APA 7th style

Jani, J., & Rachchh, N. (2026). Development of non-destructive vibration method for classification of bone fracture severity. Applied Computer Science, 22(1), 199–213. https://doi.org/10.35784/acs_8034

• Chicago style

Jani, Jignesh, and Nikunj Rachchh. ‘Development of Non-Destructive Vibration Method for Classification of Bone Fracture Severity’. Applied Computer Science 22, no. 1 (2026): 199–213. https://doi.org/10.35784/acs_8034.

• IEEE style

J. Jani and N. Rachchh, ‘Development of non-destructive vibration method for classification of bone fracture severity’, Applied Computer Science, vol. 22, no. 1, pp. 199–213, doi: 10.35784/acs_8034.

• Vancouver style

Jani J, Rachchh N. Development of non-destructive vibration method for classification of bone fracture severity. Applied Computer Science. 2026; 22(1):199–213.

Accurate pain assessment remains a cornerstone of effective clinical care as it significantly influences diagnosis, treatment planning, and evaluation of therapeutic outcomes. Traditional pain assessment tools such as the Visual Analog Scale (VAS), Numerical Rating Scale (NRS), and Verbal Rating Scale (VRS) rely heavily on the patient's ability to self-report their level of discomfort. However, these conventional approaches are inadequate for patient populations with impaired communication abilities, including individuals with neurological disorders, dementia, or those in postoperative recovery. To overcome these challenges, this study presents a novel, automated pain assessment framework that uses artificial intelligence (AI) and facial expression analysis to objectively quantify pain levels. The proposed system incorporates transfer learning and deep neural network models to improve the accuracy of pain detection using facial cues. Using the UNBC-McMaster Shoulder Pain Expression Archive Database, a widely recognized benchmark in pain research, the model was trained to identify and classify facial expressions associated with different levels of pain. A key innovation of this research is the development of an enhanced multilevel pain scale, which extends the traditional ten-point scale to sixteen different levels, allowing for more precise and granular assessment. Despite the inherent problem of class imbalance within the dataset, the model achieved a commendable classification accuracy of 91%. The results highlight the viability of AI-based tools as reliable, non-invasive alternatives to traditional self-report methods, particularly for non-communicative patients. This advancement promises to improve patient care by supporting clinicians with objective, data-driven pain assessment techniques.

Download article

• APA 7th style

Alshiha, A. A. M. (2026). Quantifying pain: An AI-driven approach to detecting pain levels via facial expressions. Applied Computer Science, 22(1), 214–227. https://doi.org/10.35784/acs_7747

• Chicago style

Alshiha, Abeer A. Mohamad. ‘Quantifying Pain: An AI-Driven Approach to Detecting Pain Levels via Facial Expressions’. Applied Computer Science 22, no. 1 (2026): 214–227. https://doi.org/10.35784/acs_7747.

• IEEE style

A. A. M. Alshiha, ‘Quantifying pain: An AI-driven approach to detecting pain levels via facial expressions’, Applied Computer Science, vol. 22, no. 1, pp. 214–227, doi: 10.35784/acs_7747.

• Vancouver style

Alshiha AAM. Quantifying pain: An AI-driven approach to detecting pain levels via facial expressions. Applied Computer Science. 2026; 22(1):214–227.