The development of electromechanical linear actuators (EMLAs) aims at compactness, energy efficiency, and high reliability. Conventional design methods often rely on costly prototypes and individual considerations of mechanics, electromagnetics, and control dynamics. This leads to long development cycles, inadequate treatment of nonlinear effects, and suboptimal performance. To address these challenges, our paper introduces a novel hybrid design methodology, integrating Analytical Modeling, Finite Element Analysis (FEA), Genetic Algorithms (GAs), and targeted experiments. Analytical Modeling provides rapid sizing, FEA combined with a GA refines geometry, and targeted experiments quantify nonlinear effects (friction, wear, thermal variability, and dynamic resonances). Unlike conventional methods, the integration is performed within iterative loops, using empirical data to refine simulation assumptions. As a result, development time is reduced by 30% and nonlinear effects are precisely addressed. The method is demonstrated on an automotive-grade EMLA. Its design is based on a claw-pole Permanent Magnet Stepper Motor, a trapezoidal lead screw, and an open-loop control with Hall effect end-position detection. After applying the method, the EMLA delivers more than 40 N of push force and achieves 600,000 actuations under the required conditions, making it suitable for various applications.

Operation and maintenance (O&M) events resulting from environmental factors (e.g., precipitation, temperature, seasonality, and unexpected weather conditions) are among the primary sources of operating costs and downtime in run-of-river small hydropower plants (SHPs). This paper presents a data-driven methodology for predicting such long events using machine learning models trained on historical power production, weather radar, and forecast data. Case studies on two Slovenian SHPs with different structural designs and levels of automation demonstrate how environmental features—such as day of year, rain duration, cumulative amount of rain, and rolling precipitation sums—can be used to forecast long events or shutdowns. The proposed approach integrates probabilistic classification outputs with threshold-consistency smoothing to reduce noise and stabilize predictions. Several algorithms were tested—including Logistic Regression, Support Vector Machine (SVM), Random Forest, Gradient Boosting, and k-Nearest Neighbors (k-NN)—across varying feature combinations for O&M model development, with cross-validation ensuring robust evaluation. The models achieved an F1-score of up to 0.58 in SHP1 (k-NN), showing strong seasonality dependence, and up to 0.68 in SHP2 (Gradient Boosting). For SHP1, the best model (k-NN) correctly detected 36 long events, while 15 were misclassified as no events and 38 false alarms were produced. For SHP2, the best model (Gradient Boosting) correctly detected 69 long events, misclassified 23 as no events, and produced 42 false alarms. The findings highlight that probabilistic machine learning-based forecasting can effectively support predictive O&M planning, particularly for manually operated or service-operated SHPs.

This research analyses the influence of welding parameters in pulsed TIG welding on the electrical response of the welding power source, and consequently on heat input and weld geometry. In the experiment, we varied pulse frequency and the proportion of base current. By measuring welding current and arc voltage, we analysed the response of welding power source, determined heat input, and examined weld geometry. The results show that, particularly at higher pulse frequencies, the welding power source fails to follow the intended pulse shape and fails to achieve set values of welding current. This directly affects heat input and weld geometry. Increasing pulse frequency up to 5 kHz results in higher weld penetration depths and weld width, mainly due to higher heat input.

Reliable traffic sign recognition is a key component of intelligent transportation systems and advanced driver assistance systems, where real-time performance and robustness under varying environmental conditions are essential. This paper presents a deep learning-based approach for traffic sign detection using the YOLOv8 object detection model. The proposed methodology covers dataset selection, model configuration, training, and evaluation using a dataset of European traffic signs annotated with bounding boxes across 55 classes. The YOLOv8s architecture was trained using high-resolution input images to improve the detection of small traffic signs commonly present in roadscene imagery. Training was performed with carefully selected hyperparameters and conservative data augmentation to preserve the visual characteristics of traffic signs. Model performance was evaluated using standard detection metrics, including precision, recall, mean Average Precision, and F1 score. In addition to quantitative evaluation on a validation dataset, qualitative validation was conducted on video recordings captured under real-world conditions, including different lighting and weather scenarios. The results demonstrate stable convergence, strong detection performance, and good robustness under favorable conditions, while highlighting performance degradation in challenging environments such as nighttime and adverse weather. The presented approach confirms the suitability of YOLOv8 for traffic sign detection in real-world applications and provides a foundation for further improvements targeting robustness under difficult conditions.