Oleksandr Veligorskyi

The article addresses the critical challenge of ensuring the reliability of power electronics, specifically buck converters, by implementing a diagnostic methodology based on physically-informed digital twin technology. Unexpected failures in DC-DC converters lead to massive economic losses due to equipment downtime. Traditional diagnostic methods face significant limitations: signal-based techniques are highly sensitive to noise, model-oriented approaches struggle with complex system modeling, and machine learning methods often lack the interpretability required for critical systems. The proposed information technology bridges these gaps by combining mathematical modeling with real-time data analysis. Using PySpice and Ngspice, the authors developed a high-fidelity model that incorporates parasitic parameters, including inductor resistance and capacitor ESR. This model serves as the digital twin's core, enabling the transition from reactive to proactive maintenance. A key contribution of the study is the systematic construction of a correspondence table. By simulating six degrees of relative parameter changes (from "---" to "+++") for nine input variables, the authors established how deviations in internal components affect four output indicators: mean and pulse values of output voltage and inductor current. The correspondence table building algorithm utilizes a threshold value θ = 0.06. The methodology was validated using 6,705 experimental measurements obtained through the remote laboratory system based on the Lab Discovery Engine technology. Results demonstrated 100% efficiency in fault detection across all test scenarios, including deviations in PWM duty cycle, input voltage, and load resistance. The accuracy of identifying specific fault causes ranged between 85.1% and 100%, with minor inaccuracies attributed to hardware measurement noise rather than methodological flaws. This implementation provides a robust, interpretable, and computationally efficient framework for the real-time health monitoring of power electronic systems.

This paper investigates the application of genetic algorithms for optimizing parasitic parameters of mathematical models of DC-DC buck and boost converters to create high-precision digital twins for hardware diagnostics. The research addresses a critical gap in power electronics where parasitic parameters (winding resistance, equivalent series resistance of capacitors, parasitic inductance, transistor switching delays, and diode characteristics) introduce discrepancies between nominal component values and actual circuit behavior, making accurate fault detection and component degradation monitoring challenging. An approach for identifying seven parasitic parameters was proposed based on 6705 experimental measurements collected over four hours from a physical buck converter operating under constant conditions (12V input, 50% PWM duty cycle, 44 μF capacitance, 3.8Ω load resistance). The implementation utilizes PySpice for circuit simulation and employs a genetic algorithm with 100 generations and population size of 50 to optimize parameters including gate rise/fall times (0.1-100 ns), transistor on-resistance (1 pΩ-1 mΩ), diode parameters, capacitor ESR (1 mΩ-0.5Ω), and parasitic inductance (1 pH-100 nH). A midpoint-based procedure was implemented to identify target signals for optimization, ensuring the model produces outputs equidistant from maximum and minimum measured values. Moving average filtering (17-sample window) reduced current ripple variance by factor of 2.83. The multi-objective fitness function incorporated weighted coefficients (0.35 for average voltage, 0.3 for voltage ripple, 0.2 for average current, 0.15 for current ripple) to prioritize critical parameters. Logarithmic scaling was applied to six parameters spanning multiple orders of magnitude, ensuring uniform exploration of the solution space. Optimization results demonstrated symmetric deviations of ±41.69% for voltage ripples and ±42.25% for current ripples, enabling establishment of unified threshold values for diagnostic methodology. Unlike previous research predominantly using Particle Swarm Optimization (PSO), this study provides the first comprehensive application of genetic algorithms specifically for parasitic parameter identification in buck/boost converter digital twins based on experimental measurements. The proposed approach enables non-invasive creation of accurate digital twins without requiring specialized measurement equipment or taking systems offline, making it suitable for predictive maintenance applications. The method can be extended to boost topology and integrated into real-time monitoring systems for early detection of component degradation in critical power electronics applications.

This paper presents a novel methodology for diagnosing hardware systems based on the use of the digital twin by using a vector-based representation of both input and output values. The method involves comparing the output vector of a real electronic device with a corresponding reference vector derived from a mathematical model (digital twin). Deviations between these vectors are analyzed using a specially designed table of discrete qualitative correspondences, which enables identification of the likely causes of incorrect behavior in specific components without requiring intrusive measurements or additional sensors.The approach models the behavior of an electronic system as a transformation from an input vector to an output vector, where each deviation in input causes a deterministic change in output. This vector-based abstraction allows for the construction of decision rules and classification zones using relative relationships between components rather than absolute values, making the system robust to noise, parasitic effects, and measurement uncertainties. The proposed method is computationally efficient, allowing deployment on low-power embedded platforms, and does not rely on deep physical modeling or extensive machine learning training datasets. The methodology introduces the concept of a “gray zone” in the multidimensional output space, where small deviations are treated as acceptable tolerances, and only significant directional changes trigger diagnostic logic. Additionally, a normal-ization step and the introduction of threshold vectors allow for scalable precision and adaptation to non-linear dependencies in hardware behavior.Advantages of this approach include simplicity of implementation, human readability of diagnostic outputs (e.g., “+”, “++”, “–”, “--”), low computational requirements, and high resilience in noisy or constrained environments. Moreover, the system enables real-time monitoring and provides a foundation for future extensions such as remaining useful life (RUL) pre-diction and adaptive table generation through machine learning.The methodology is particularly suited for power electronics and embedded systems, but can be generalized to any ap-plication where the device can be abstracted as a system with measurable input/output behavior. This work provides a flexible, scalable, and interpretable diagnostic framework that relies on vector mappings within a model-based digital twin approach, enabling robust and low-cost fault detection without the need for high-precision measurements, extensive sensor networks, or machine learning.

Electric energy consumption is increasing much faster than the predicted growth in energy generation. Although the installed capacity of renewable energy sources is also expanding, grid congestion remains unavoidable without adopting smart energy management systems (EMS) and flexible power electronics structures. Given the significant installed capacity of photovoltaic (PV) systems in the residential sector, moving towards zero-emission buildings (ZEBs) through the use of storage systems and smart power electronics is essential. This article provides a detailed review of power electronics solutions for ZEBs and offers strategies to address related challenges. By exploring the promising future of the low-voltage dc (LVDC) industry in ZEBs, it presents and compares grid connection scenarios and evaluates their overall efficiencies across hybrid, dc, and ac technologies. Furthermore, it addresses the integration of dc and ac systems in energy resources (ER), proposing solutions for challenges related to protection, grounding, and leakage currents. Finally, it examines the latest EMS solutions, emphasizing the shift to full digitalization through a combination of cloud-based and edge-computing platforms.

The rapid development of information technologies and global challenges of recent years have significantly altered approaches to organizing educational processes, especially in technical sciences. The need for remote hardware laboratories has become particularly acute in Ukraine due to the ongoing military conflict, where many students and teachers are forced to work and study remotely. This research addresses the critical need for effective remote laboratory systems that can provide quality technical education regardless of physical location. The study focuses on analyzing and comparing modern remote hardware laboratory systems, proposing a methodology for their evaluation, and determining the most optimal solution for implementation. The problem lies in the traditional methods of hardware study requiring physical access to equipment, which becomes impossible in remote learning conditions. This creates a significant gap between the theoretical knowledge and practical skills of students. The research proposed a new methodology that involves a systematic analysis of existing remote laboratory systems based on six key criteria: technology level, updates and support, Learning Management System (LMS) integration capabilities, simplicity, real-time performance, and ease of configuration. The study examines several major systems, including WebLabDeusto, LabDiscoveryEngine, VISIR, iLab, and FREE, analyzing their architectures, advantages, and limitations. Through comparative analysis, the research revealed that LabDiscoveryEngine demonstrates the highest overall effectiveness among the examined systems, scoring 29 out of 30 possible points across all evaluation criteria. The study identified key architectural approaches used in existing systems and analyzed their respective strengths and weaknesses. Special attention was paid to aspects such as the user interface, managing access to limited hardware resources, and security mechanisms. The results of the study allow the use of the new proposed methodology for comparing remote laboratory systems and improve the overall understanding of the development and implementation of remote laboratory systems, providing valuable information for universities seeking to implement or improve their remote laboratory capabilities. The article emphasizes the critical importance of remote laboratory systems for ensuring quality technical education in modern conditions and outlines future research directions, including the investigation of long-term effects, economic efficiency, and standardization of remote laboratory systems. The research results can be practically applied in selecting key technologies for creating new remote laboratories and improving existing ones.

Injection plastic moulding is a leading technology of plastic parts on a mass production scale. Intelligent features, implemented on edge or cloud level, can significantly extend the borders of quality and reproducibility of such technology. This paper demonstrates an approach for the design of an Industrial loT-based monitoring system for mould controllers for injection plastic moulding technology. Relying on Raspberry Pi as an edge computing device, the hardware of the monitoring system was proposed. Based on the parameters of monitoring data, a two-layer structure with SQL-based for operation mode and time-series databases for measurement data was developed. Finally, possible solutions for remote data access and visualization were identified, and web-application based on the Giraffe visualization library was introduced.