Within the scope of predictive and preventive maintenance in industrial environments, fault detection in three-phase induction motors is crucial to reduce and, in many cases, prevent downtime and operational losses. Therefore, this work focuses on intermittent stator inter-turn short circuit anomalies, which can remain undetected by conventional monitoring systems due to their sporadic occurrence and weak signatures. In this sense, the main objective is to evaluate the sensitivity of low-cost piezoelectric sensors in comparison with power quality analyzers (PQAs) for the identification and characterization of stator faults under different operating conditions. The experimental campaign consisted of subjecting a three-phase induction motor to controlled insertion of a purely resistive impedance into the stator windings, enabling repeatable fault emulation. Different fault severities were reproduced through frequency modulation to generate intermittent short circuit patterns. Data were acquired simultaneously using piezoelectric sensors and a PQA. After the measurements, the recorded signals are processed using time–frequency signal processing techniques to compare the effectiveness of electrical variables and acoustic emission features extracted from both sensing technologies. Finally, preliminary results indicate that the proposed processing strategy enables low-cost piezoelectric sensors to detect inter-turn short circuit faults with satisfactory performance when compared with established power quality analyzers, supporting their potential as a cost-effective alternative for industrial condition monitoring applications.

Three-phase induction motors are the most widely used electrical machines in industrial applications worldwide due to their robustness, low cost and reliability. Inter-turn short-circuit faults represent one of the most critical incipient failure modes in these machines and can lead to severe performance degradation if they are not detected at an early stage. Although vibration-based condition monitoring techniques have shown promising results, many recent approaches rely on complex time–frequency representations and deep learning models, which increase computational cost and implementation complexity. Therefore, a lightweight and interpretable fault diagnosis framework based on vibration signals is proposed, combining frequency-domain feature extraction and a multilayer perceptron classified. In this work, vibration signals measured by MEMS accelerometers were segmented and processed using the Fast Fourier Transform. From the resulting spectra, a compact set of spectral features, including energy, spectral centroid, bandwidth, kurtosis and skewness, was extracted and used as input to the lightweight multilayer perceptron network. Also, the method was evaluated under healthy operating conditions and multiple inter-turn short-circuit fault scenarios, considering different phases and fault severity levels. Model performance was assessed using accuracy, precision, recall, F1-score and confusion matrix analysis, along with an evaluation of preprocessing, training and inference times. The results demonstrate that the proposed FFT-based MLP framework achieves competitive classification performance while significantly reducing computational complexity when compared to deep learning approaches. These findings indicate that frequency-domain statistical features combined with shallow neural networks provide an effective and efficient solution for vibration-based inter-turn short-circuit fault diagnosis in three-phase induction motors.

The coordination of Unmanned Aerial Vehicle (UAV) swarms for environmental monitoring faces significant challenges due to the limitations of centralized control, including single points of failure and high communication latency in dynamic environments. This study addresses the need for robust, scalable, and adaptive coordination without relying on a central controller. Methods: We propose a decentralized swarm intelligence algorithm based on local interaction rules, including separation, alignment, and cohesion, to govern collective agent behavior. The model was implemented and validated using a custom Python simulation environment, focusing on formal stability metrics that link local agent rules to global swarm dynamics. Results: Performance analysis using Swarm Performance Indicators (SPIs) demonstrates that the decentralized approach ensures high swarm stability and resilient network connectivity. Quantitative evaluations show that the system maintains operational integrity even under partial agent failure, outperforming traditional centralized architectures in scalability and fault tolerance. Specifically, the algorithm optimizes the trade-off between tracking accuracy and communication link quality, maintaining stable coordination with linear computational complexity. Conclusions: The findings highlight the efficacy of decentralized algorithms for enhancing the autonomy and resilience of mechatronic systems. This research provides a scalable analytical framework for next-generation autonomous systems in complex monitoring tasks, directly contributing to the field of automation and machine design.

Continuous operation of static equipment is not possible without assessing the metal condition, and one method to implement the asset management strategy is ultrasonic (UT) analysis of existing imperfections. A focus is placed on the risk-based inspection (RBI) of a high-pressure amine absorber in service where a zone with permissible short, scattered indications is found before the commissioning. The vessel has been hydrotested and put into operation, and the corresponding risk category has been defined. It has been in service for ten years, with regular thickness measurements and maintenance performed. The subsequent scheduled inspection is set at every five years. Initially, RBI recommendations are followed based on the criticality of the asset and the severity evaluation of the imperfections. For an absorber made from P355NH with a height of roughly 17 m and a diameter more than a meter, shell metal UT zones of a total size of 2.115 m² have been evaluated. No crack-like faults have been detected, and the existing rounded imperfections with a diameter of less than 4 mm are being monitored for eventual growth. Since no change is observed after the strategy-defined five-year period and all the inspection protocols show consistent results, an RBI reevaluation is performed for the definition of the new risk category. As a result, there is a decrease in the calculated risk of the corresponding asset.

Funding: The author acknowledges support from project BG16RFPR002-1.014-0005.

Industrial and infrastructure systems increasingly operate in dynamic environments characterised by fluctuating loads, changing operating modes, and varying environmental conditions. These non-stationary behaviours challenge conventional condition monitoring and fault diagnosis approaches, which typically rely on static models, fixed thresholds, or pre-defined fault signatures. As operating conditions evolve, such approaches often suffer from reduced diagnostic accuracy and increased false alarm rates. Self-learning condition monitoring systems offer a promising pathway toward adaptive and resilient fault diagnosis in such dynamic operating environments. This study proposes a self-learning condition monitoring framework that continuously adapts to changing system behaviour through online and incremental learning mechanisms. The framework integrates multi-sensor data streams with adaptive machine learning techniques capable of updating feature representations, decision boundaries, and fault models in real time. By combining unsupervised anomaly detection with semi-supervised learning, the system can identify emerging degradation patterns and previously unseen fault modes while maintaining robustness to normal operational variability. A key contribution of the proposed approach is its ability to differentiate between benign operational changes and true fault-related anomalies. Context-aware feature extraction and adaptive thresholding are employed to reduce false alarms under varying load and environmental conditions. Feedback loops are incorporated to refine diagnostic confidence and improve model performance as new data become available, enabling the system to learn continuously from operational experience. Illustrative case studies demonstrate that the self-learning framework achieves improved fault detection accuracy, enhanced adaptability, and lower false alarm rates compared to traditional static monitoring methods. Overall, this work highlights the potential of self-learning condition monitoring systems as a foundation for intelligent, autonomous fault diagnosis, supporting reliable and efficient operation of complex assets in highly dynamic operating environments.

Critical assets such as power generation equipment, industrial machinery, transportation infrastructure, and manufacturing systems are essential to the reliability and safety of modern socio-technical environments. Unexpected failures in these assets can result in costly downtime, safety risks, and service disruptions. Conventional condition monitoring and fault diagnosis approaches—often based on periodic inspections, fixed thresholds, or isolated data analysis—are increasingly inadequate for complex assets operating under variable and uncertain conditions. In this context, Digital Twin-enabled condition monitoring and predictive fault diagnosis offers a powerful, data-driven paradigm for proactive asset health management.

This study proposes a Digital Twin-based framework that integrates real-time sensor data, physical system models, and advanced analytics to enable continuous monitoring and predictive fault diagnosis of critical assets. The Digital Twin acts as a dynamic virtual replica of the physical asset, continuously updated through operational data streams such as vibration, temperature, electrical, and process signals. Machine learning and statistical inference techniques are employed to detect anomalies, identify fault signatures, and capture degradation trends, while physics-informed constraints ensure consistency with underlying system behaviour. A key feature of the proposed approach is its predictive capability. By embedding remaining useful life estimation and fault progression modelling within the Digital Twin, the framework enables early warning of impending failures and supports proactive maintenance planning. Asset operators can evaluate “what-if” scenarios, assess the impact of operating conditions on asset health, and optimise maintenance strategies to minimise downtime and lifecycle costs. The effectiveness of the framework is demonstrated through illustrative use cases involving representative critical assets, showing improved fault detection accuracy and earlier diagnosis compared to conventional monitoring methods. Overall, this work highlights Digital Twin–enabled condition monitoring as a foundation for intelligent, predictive asset management, supporting enhanced reliability, safety, and operational efficiency in critical infrastructure and industrial systems.

Solar drying systems provide an energy-efficient and environmentally sustainable solution for processing agricultural and medicinal products; however, their performance is strongly influenced by nonlinear process dynamics, fluctuating solar irradiation, and continuously changing ambient conditions. These uncertainties often lead to unstable temperature and humidity regimes, increased energy consumption, and the degradation of product quality when conventional fixed-parameter control strategies are applied. Therefore, the development of intelligent and adaptive control approaches capable of ensuring robust operation under highly variable conditions remains an important challenge in modern automation and mechatronic systems. In this work, an advanced adaptive control framework based on an Adaptive Neuro-Fuzzy Inference System (ANFIS) is proposed for a cabinet-type solar drying system. The controller is formulated using a first-order Sugeno fuzzy inference structure with Gaussian membership functions and trained through a hybrid learning algorithm that combines least-squares estimation with gradient-based optimization. To further enhance adaptability and robustness, the ANFIS parameters are optimized using a particle swarm optimization (PSO) algorithm, while physical constraints derived from heat and mass transfer principles are explicitly incorporated into the control design. The complete control architecture is implemented in a MATLAB/Simulink environment, enabling digital-twin-based simulation and comprehensive performance evaluation. The simulation results demonstrate that the proposed optimization-enhanced adaptive ANFIS controller significantly improves temperature and humidity regulation compared to conventional control approaches. Faster setpoint tracking, reduced overshoot, and improved disturbance rejection are achieved under variable solar irradiance and ambient conditions. Quantitative analysis indicates a reduction in settling time of approximately 30% and a decrease in energy consumption by about 12-18% while maintaining stable and efficient drying regimes. The obtained results confirm that integrating adaptive neuro-fuzzy control, metaheuristic optimization, and physics-informed constraints provides an effective and scalable solution for complex nonlinear solar drying processes. The proposed framework can be extended to other renewable energy-driven thermal and mechatronic systems.

Introduction: Cascaded PID control remains popular in quadcopter platforms because it is simple to implement and certify; however, tuning the coupled attitude–altitude loops is often time‑consuming and sensitive to disturbances, actuator limits, and model mismatch. This work targets the inner-loop controller of an existing MATLAB/Simulink quadcopter model, where four PIDs regulate roll (φ), pitch (θ), yaw (ψ), and altitude (z) under an outer-loop command generator.
Methods: Three tuning strategies are compared under an equal simulation budget: (i) a classical baseline using Simulink PID Tuner followed by manual refinement, (ii) particle swarm optimization (PSO) directly optimizing the 12 PID gains, and (iii) reinforcement-learning‑guided PSO (RL‑PSO), where PSO searches the gain vector while an RL agent adapts PSO hyperparameters (inertia weight and acceleration coefficients) online based on swarm progress and diversity features. The objective function combines integrated time‑weighted absolute error (ITAE) tracking terms for φ, θ, ψ, and z with penalties on overshoot, control effort, actuator saturation, and unstable responses.
Results: A robustness benchmark is defined using two disturbance-focused scenarios: (1) feedback-path perturbations and (2) plant-side disturbances. Performance will be reported using Monte‑Carlo statistics of RMS error, overshoot, settling time, control effort, and constraint violations, together with convergence curves (best cost versus iteration and total model evaluations). The study is designed to test the hypothesis that RL‑PSO improves worst‑case disturbance rejection and reduces constraint violations relative to standard PSO and classical tuning.
Conclusions: RL‑PSO provides a practical, simulation-based route to robust multi-loop PID tuning for quadcopter attitude–altitude control without altering the overall cascaded control structure.

The effective control of industrial wastewater treatment processes based on ion-exchange resins is challenging due to their nonlinear dynamics, time-varying operating conditions, and uncertainty in water quality parameters. In particular, variations in water hardness and total dissolved solids (TDS) strongly influence purification efficiency, necessitating adaptive and intelligent control strategies beyond conventional approaches. In this study, an adaptive neuro-fuzzy inference system (ANFIS)-based intelligent control model is developed to regulate the opening degree of a control valve governing the wastewater flow rate in an ion-exchange treatment process. A laboratory-scale experimental setup was designed and implemented, and a dataset of 300 experimental samples was collected under diverse operating conditions. Water hardness and TDS were selected as input variables, while the valve opening degree was defined as the output control variable. A clustering-based rule extraction method was employed to construct the ANFIS structure, and model parameters were optimized using the Adam optimizer and Particle Swarm Optimization (PSO). The performance of the ANFIS–Adam and ANFIS–PSO models was evaluated using regression and control performance metrics, including RMSE, MAE, R², settling time, and integral absolute error (IAE). The results indicate that both optimization algorithms significantly enhance ANFIS performance while exhibiting complementary strengths. The ANFIS–Adam model achieves faster convergence and improved dynamic response, reducing settling time by approximately 15–20%, making it suitable for real-time control applications. In contrast, the ANFIS–PSO model demonstrates superior robustness and global search capability, achieving up to 10–15% lower RMSE and improved steady-state accuracy. Both models attain high prediction accuracy (R² > 0.96), suggesting that the choice of optimizer should be guided by specific control objectives, such as real-time responsiveness or robust offline tuning.

The increasing demand for energy-efficient and high-quality tire manufacturing requires advanced control strategies capable of handling the nonlinear and time-varying characteristics of the vulcanization process. Conventional control methods often struggle to maintain optimal operating conditions and minimize energy consumption under varying technological and material parameters, which has led to growing interest in intelligent data-driven control approaches. In this study, an adaptive neuro-fuzzy inference system (ANFIS)–based model is developed for intelligent modeling and optimization of the tire vulcanization process in a steam-heated vulcanization press. Four key process variables, vulcanization temperature, steam pressure, tire mass, and process time, are selected as input parameters, while the gas consumption rate is considered as the output control variable. These variables are chosen based on the physical characteristics of the process and practical operating conditions in light-vehicle tire production. A dataset consisting of 250 experimentally consistent process samples is used to train and validate the ANFIS model using the Python programming environment. The model is optimized using Particle Swarm Optimization (PSO), Genetic Algorithm (GA), and Gradient Boosting-based tuning strategies. Model performance is evaluated using root mean square error (RMSE), mean absolute error (MAE), and coefficient of determination (R²). The results demonstrate that the optimized ANFIS model effectively captures the nonlinear relationship between the process parameters and gas consumption. The best-performing configuration achieves a high predictive accuracy with R² ≈ 0.945, while maintaining low prediction errors (RMSE = 0.018–0.025 m³/min, MAE = 0.012–0.017 m³/min). These findings confirm the effectiveness of the proposed approach and its potential for intelligent and energy-efficient control of tire vulcanization processes.