Category: Journal Automatica and Informatics

Key Words: Large language models; management information systems; artificial intelligence; machine learning; local fine-tuning.

Abstract. The study presents a platform designed for local training and inference of popular large language models (LLMs), addressing the growing need for solutions that preserve the privacy of training data. The platform enables researchers and developers to further fine-tune base models using locally stored and processed data through a modular software architecture. It supports a wide range of models and tasks, offering seamless integration with existing machine learning workflows and tools within enterprise management information systems. Two alternative approaches to contextualization of LLMs’ knowledge bases are implemented within the platform i.e., indirectly through model tuning (parameter-efficient fine-tuning, reinforcement learning with human feedback and tutorial fine-tuning) and directly through extending the original knowledge base via retrieval-augmented generation. The proposed hybrid solution for knowledge contextualization addresses existing challenges within the generative artificial intelligence field, namely accuracy of generated information (hallucinations), as well as transparency and traceability of generated information to source data. Benchmark results from the platform utilization demonstrate competitive performance in terms of latency, context accuracy, and resource utilization compared to cloud-based alternatives. By decentralizing model development and inference, the platform empowers institutions to maintain data sovereignty, reduce operational costs, and promote sustainable practices in the field of artificial intelligence.

Read More

Key Words: Differentil topology; smooth manifolds; smooth maps; singularities; matrix computations; regularization.

Abstract. This survey briefly reviews some fundamental concepts of differential topology and discusses their application to the solution of problems arising in engineering disciplines. The use of manifolds to represent families of mathematical models makes it possible to describe parameter-dependent problems and to study their global properties. The study of mappings between manifolds includes the investigation of singular properties of these mappings at certain points of the manifolds. Although such points are not typical, they can have a significant influence on the behavior of the corresponding mapping. The application of manifolds and the associated bifurcation diagrams in parameter space to the analysis of parameter-dependent problems is presented in brief. Examples are provided to illustrate the introduced concepts.

Read More

Key Words: Logs; Table Curve 2D; frozen bound water; frozen free water; defrosting; energy consumption.

Abstract. A mathematical description of the thermal energy Qw-defr required for the complete defrosting of logs, i.e. for the complete melting of the ice in them, formed in the wood in winter by the natural freezing of both free and bound water in it, using the software package Table Curve 2D has been presented. This package allows for the selection of equations, which provide the best similarity between the calculated with them values of Qw-defr during thermal treatment of logs containing ice and the respective data obtained experimentally or using an adequate temperature-energy model of Qw-defr. An approach for computing the energy required for heating to defrost logs subjected to thermal treatment before their subsecuent mechanical processing was applied. The approach is based on the use of two personal models. With the help of the non-stationary model, the defrosting times of beech logs with a diameter of 0.4 m, initial temperatures of –1 °C, –10 ˚C, –20 °C, –30 °C, –40 °C, and moisture content u of 0.4 kg∙kg-1, 0.6 kg∙kg-1, and 0.8 kg∙kg-1 were determined at operating temperatures in the boiling or steaming equipment equal to 80 °C. For different combination of the specified three values of the moisture content and five values of the initial log temperature there are significant differences between the amount of both free and bound frozen water in the logs, which ensures sufficiently large differences between the calculated values of the energy required for defrosting of the logs.

Read More

Key Words: Energy Management Systems; Reference Architectures; Reference Frameworks; Conceptual Models; Semantic Interoperability; Unified Information Models; Industry 4.0; ISO 50001; SGAM; IIRA; OPC UA; IEC 62264.

Abstract. Achieving the objectives of the Industry 4.0 initiative, particularly in relation to the implementation of Energy Management Systems (EnMS) in industry, depends on the structured adoption of reference frameworks, architectures, and models capable of ensuring interoperability, scalability, and continuous enhancement of energy efficiency. In recent years, concepts such as reference frameworks and architectures, conceptual models, ontologies, meta-models, and unified information models have become increasingly prominent in the development and deployment of Energy Management Systems (EnMS). Nevertheless, their simultaneous application frequently results in conceptual uncertainty and overlapping interpretations. The growing use of these concepts requires a clear understanding of their meaning, scope, and interrelationships. This paper presents an overview of intelligent reference frameworks, architectures, and models applied in the development and implementation of industrial energy management systems.
The relationships between models, reference frameworks, and architectures are analyzed, with a particular emphasis on clarifying the conceptual distinctions between conceptual models, reference models, meta-models, and meta-meta-models. Building on this analysis, the paper reviews several widely recognized standards and architectural frameworks, including ISO 50001, the Smart Grid Architecture Model (SGAM), the Industrial Internet Reference Architecture (IIRA), OPC Unified Architecture (OPC UA), and the IEC 62264 standard for enterprise control system integration.

Read More

Key Words: Adaptation; integration; Machine Learning; modeling; optimization; Reinforcement Learning; tuning.

Abstract. The article examines the main directions for integrating the Reinforcement Machine Learning (RL) as an extension of the already existing advanced industrial automation systems based on the classical PID and MPC regulators. The appropriate areas of application, functionality, advantages and disadvantages of the application of RL in direct process control, hybrid regulation, as well as in local coordination of complex multidimensional, nonlinear, with parametric variability and large number of constraints objects with frequent changes of the control objectives are analyzed. The goal of integrating RL is to achieve optimality, adaptability and security in all possible operational situations from technological and business aspects that are unattainable for traditional systems with PID and MPC regulators. An analysis was made and the prospects for the gradual transformation of RL into the third main pillar of industrial automation along with PID and MPC were assessed.

Read More

Key Words: Neural PID control; adaptive control; electro-hydraulic system; nonlinear systems; MATLAB/Simulink; robustness.

Abstract. This article presents an adaptive Neural PID Optimizer for control of a nonlinear electro-hydraulic servo system. Conventional PID controllers with fixed parameters ensure satisfactory performance only in a limited operating range and exhibit degraded behavior at low rotational speeds due to dead-zone effects and at high speeds due to actuator saturation. To address these limitations, a neural network is employed to adapt the PID gains Kp, Ki, and Kd in real time based on the system state. Simulation studies conducted in MATLAB/Simulink for different operating regimes and load variations demonstrate that the proposed adaptive controller improves dynamic performance, reduces settling time, eliminates steady-state error, and enhances robustness to disturbances. The time evolution of the controller parameters confirms the effective operation of the neural adaptation mechanism. The results validate the suitability of the proposed approach for control of nonlinear electrohydraulic systems with variable operating conditions.

Read More

Key Words: Signal sources; PLC; Arduino; Raspberry Pi; Data Base; data server; programming software.

Abstract. In this article are considered several variants of systems for capturing data from measuring devices, their processing and their transmission for further storage in a database. Databases are considered, also the formation of records in them. In the modern digital world, automation acquires new dimensions and new qualities of receiving, processing and storing results. This article considers: measuring devices – sources of digital signals; devices/controllers for signal processing; methods and means for storing measured information; software tools for processing and storing data. As an example, elements of a program for receiving, processing and storing data received via the Modbus protocol are shown.

Read More

Key Words: Bit manipulations; hardware optimizations; Edge devices; lookup tables (LUT); CORDIC; fast inverse square root; embedded machine learning.

Abstract. Fast and energy-efficient evaluation of transcendental functions (for example, logarithm, exponential, square root, and normalization) is critical for real-time signal processing and on-device AI inference in Edge and IoT systems. This paper surveys and experimentally validates a set of lightweight techniques – primarily IEEE-754 based bit-level manipulations (bit tricks) – alongside code- and hardware-level optimizations (loop unrolling, DMA usage, and other low-level optimizations) that dramatically accelerate these functions on resource-constrained microcontrollers. We describe simple but effective bit hacks for fast approximations of log₂(x), 2ˣ, √x and 1/√x (including the well-known “Quake” inverse square root), explain their theoretical basis in floating-point representation, and show how modest corrections to the mantissa (linear and low-order polynomial) can substantially improve accuracy while retaining most performance gains. Benchmarking across representative platforms (AVR, Cortex-M variants with/without FPU, ESP32, and RISC-V boards) demonstrates up to 10× speedups relative to standard library implementations, with typical accuracy in the range of 5-6 bits for raw bit-trick estimates and much improved error after correction terms. We discuss practical integration patterns, trade-offs between latency, memory footprint, and numerical fidelity, and recommend when to combine bit tricks with LUTs or CORDIC for broader function coverage. Finally, we outline use cases (embedded ML feature scaling, entropy computation, normalization in DSP and robotics) and caveats (unsuitability for high-precision scientific/financial calculations and for periodic functions like sin/cos), providing guidelines for safe deployment in production Edge/IoT projects.

Read More

Key Words: Concrete pits; veneer logs; steaming; thermal balance; energy efficiency.

Abstract. A methodology for computing the thermal balance and energy efficiency of concrete pits when steaming veneer logs in them has been presented. The methodology is based on the use of two personal mathematical models: 1D non-linear model of the unsteady temperature distribution along the radius in the central cross section of non-frozen logs at conductive boundary conditions, and a stationary model of the thermal balance of concrete pits when steaming wood materials in them. For numerical solving of the models and practical application of the suggested methodology, a software packages were prepared in the calculation environment of Visual FORTRAN Professional and in Excel respectively. With the help of the first model, the steaming times of beech logs with a diameter of 0.4 m, initial temperature of 0, 10, and 20 °C and moisture content of 0.4, 0.6, and 0.8 kg∙kg-1 were determined at operating temperatures in the pit equal to 70, 80, and 90 °C. Using the determined logs’ steaming durations, with the help of the second model the change in the energy required for the entire steaming process and that for each of the components of the pit’s thermal balance was calculated. The thermal efficiency of a concrete pit was also determined under different combinations of factors affecting the steaming of the logs. Computer simulations were performed for a well-insulated pit with working volume of 20 m3 and degree of filling it with logs 45%, 60%, and 75%. It has been established that the energy consumption of the pit decreases from 122.8 to 106.3 kWh·m-3 when the initial temperature of logs with a moisture content of 0.6 kg∙kg-1 is increased from 0 °C to 20 °C, and their logs heating is realized at a temperature of the steaming medium of 80 °C and a maximum possible in practice degree of filling the pit with logs of 75%. The increase in the temperature in the pit from 70 to 90 °C at its 75% filling with logs with an initial temperature of 10 °C and a moisture content of 0.6 kg∙kg-1 causes an increase in its energy consumption from 103.4 to 127.0 kWh·m-3. In these two cases, the thermal efficiency of the pit decreases – in the first case from 41.8% to 35.7% and in the second case – from 39.1% to 38.9%. The increase in the moisture content of the logs from 0.4 to 0.8 kg·kg-1 causes an increase in the energy consumption of the pit from 104.3 to 125.2 kWh∙m-3 at its 75% filling and steaming at 80 °C of logs with an initial temperature of 10 °C. In this case, the thermal efficiency of the pit increases from 34.3% to 42.9%. The presented methodology can be used to calculate the thermal balance and efficiency of concrete pits with different geometric and thermophysical parameters when steaming logs of different wood species to any desired degree of heating and plasticization in the production of veneer

Read More

Key Words: EBSI; EUROPEUM-EDIC; EBSI-NE; VeloxWallet; EBSI OnePass.

Abstract. The aim of the article is to present the challenges, experience and solutions obtained by the teams in which the authors work for the architecture of EBSI nodes, processes and roles in verification and financing of start-ups and SMEs in the European projects EBSI-NE and EBSI OnePass. The European Blockchain Infrastructure for Services (EBSI) is presented as a common cross-border framework in Europe – need and opportunities, technologies and a single legal framework to support growth and business. The architectures and basic functionalities of the EBSI-NE projects, which develop the EBSI network by adding 18 new validator nodes to the production network and providing comprehensive support services to all stakeholders, and the EBSI OnePass project, which develops decentralized services for the processes of innovative financing of SMEs, are described. The developed digital wallet VeloxWallet, which offers various functionalities, such as KYC (Know Your Customer) and KYB (Know Your Business) schemes for verifiable certificates and the verification process of the participants in the OnePass project, are examined. Empirical results from the work on the two projects are also discussed: a developed environment for implementation in the EBSI pre-production stage, ecommendations for testing the OnePass project, and future directions for the application and development of the OnePass platform.

Read More

Key Words: Artificial intelligence; expert systems; industrial automation; operation control; predictive maintenance.

Abstract. Expert systems were the first field at the dawn of artificial intelligence, in the mid-20th century. They also represent the first real practical applications in areas such as the synthesis of chemical compounds, healthcare, and computer system design, thanks to their remarkable ability to mimic human abilities for logical inference in the presence of in-depth expert knowledge. Industrial automation, in turn, reached a high level of development in the 1980s, but technological innovations posed a number of new problems, the solution of which by applying expert systems proved to be a suitable alternative in combination with the significant achievements of computing and control theory. This synergy of rapidly developing fields is particularly productive currently in the era of exponentially advancing artificial intelligence technologies, part of which are expert systems. The paper examines the features and capabilities of a large number of modern industrial automation systems with the integration of expert systems and artificial intelligence modules. Particular emphasis is placed on expanding the scope of industrial automation with the inclusion of predictive maintenance of the technical condition of technological objects as an important part of the operational management of objects from “end to end”. The main challenges and unresolved problems of the application of expert systems in industrial automation are examined with confidence in the possibilities for realizing their potential.

Read More

Key Words: Artificial Intelligence; business analytics; Machine Learning; optimization; Decision Making.

Abstract. The paper discusses the development and implementation of artificial intelligence algorithms for business analytics. The study focuses on the potential of AI in predictive analytics, decision making, and optimization of business processes. Key technologies such as machine learning, natural language processing, and data mining are analyzed. Practical applications in sectors like finance, marketing, and supply chain management are explored. The impact of AI-driven solutions on improving business efficiency and decision accuracy is highlighted. The results from real-world case studies are presented, confirming the effectiveness of AI in complex business environments.

Read More

Key Words: Generalized predictive control; disturbance compensation; feedforward control; time-delay; nitrogen oxides (NOx).

Abstract. This paper describes a NOx control system in a combined cycle power plant. The aim is to increase the control performance of a NOx decomposition process by improving measurable disturbance compensation. Extended Generalized Predictive Control algorithm is used. The typical predictive model in GPC algorithms (ARIMAX) is extended by a measurable disturbance vector.

Read More

Key Words: Waste valorisation; bioenergy production; containerised biogas system; climate change mitigation.

Abstract. The EU is fostering new set of policies in all sectors, so as to achieve climate neutrality in 2050. The generation of new practices for improved exploitation of available waste streams in the agricultural sector holds significant potential for minimization of the generated greenhouse gases. This requires innovations towards more efficient waste valorisation systems for the production of bioenergy. To tackle the issue, a containerised biogas system is proposed to prevent the free release of methane and other GHG, while producing renewable energy. The experience from the development of containerised biogas system in Dimitrievo, Bulgaria is described.

Read More

Key Words: Applied Artificial Intelligence; Generative Artificial Intelligence (GenAI); Large Language Models; Digital transformation.

Abstract. The paper gives a condensed overview of the industry’s current state-of-the-art artificial intelligence (AI) applications. From a business perspective, the critical issue of understanding the differences between basic AI and the new popular Generative AI (GenAI) is emphasized. Special attention is given to current and future technology trends, such as Causal AI, automated reasoning with knowledge graphs, Large Language Model (LLM)-based agents, Agentic AI, and their value-creation potential. The application landscape, defined mainly by the impressive record of basic AI after 2010 and the potential for mass-scale use of GenAI, is discussed. The paper also suggests a roadmap for big and small businesses to introduce an appropriate form of basic AI or GenAI.

Read More

Key Words: Case-Base Reasoning; Risk Management; Fault Tree; System Control; Power Plant.

Abstract. The management of Energy Power Plants (PP) is a multifaceted technological process influenced by various factors, including environmental changes, the state of system control elements, and potential faults in sensors and other critical components. During the control of Power Plants, some failures in system devices (such as sensors, valves, and other essential components) can occur. These failures can lead to damage and malfunctions within the system control loops, ultimately causing the system to work unstably and rise risk situations. Early diagnosis and prediction of these failures are crucial for maintaining the safe and efficient operation of Energy Power Plants. The research introduces a new approach for diagnosing the state of the system and presents a mechanism for predicting some failures of control loop elements at an early stage. The proposed approach combines Fault Tree (FT) analysis with Case-Based Reasoning (CBR) in a way to provide a comprehensive system for prediction of fault devices detection. Using the previous knowledge for the technological characteristics of control devices elements – such as material wear, working time under high load, and contamination of sensors and photocells is used for forming of knowledge base-oriented system (CBR).

Read More

Key Words: Steaming autoclaves; thermal balance; wooden prisms; energy consumption; energy efficiency.

Abstract. An approach for computing the energy consumption and energy efficiency of autoclaves during steaming of non-frozen wooden prisms intended for veneer production has been presented. The approach is based on the use of two personal mathematical models: 2D non-linear model of the unsteady distribution of the temperature in the central cross section of non-frozen prisms subjected to steaming at conductive boundary conditions, and stationary model of the thermal balance of autoclaves during steaming of wood materials in them. For numerical solving of the models and practical application of the suggested approach, a software packages were prepared in the calculation environment of Visual FORTRAN Professional and in Excel respectively. With the help of the first model, the durations of all of the 5 stages of the regimes for steaming beech prisms with cross section dimensions of 0.3 × 0.3 m, 0.4 × 0.4 m, 0.5 × 0.5 m, initial temperature of 0 °C, 10 °C, 20 °C and moisture content of 0.6 kg∙kg-1 were determined at maximum temperatures of the steaming medium equal to 130 °C.

Read More

Key Words: Generalized predictive controller; multi-objective optimization; constant step scanning; tuning parameters.

Abstract. A multi-objective and multi-parametric optimization problem with interval constraints, linear inequality constraints and constraints on the values of the objective functions is formulated to determination of the tuning parameters in generalized predictive controller design. Quality indicators of the desired control system are optimization objectives. The problem is solved by modified methods of constant step scanning. The tunable parameters optimization problem is transformed into a single-objective constraint problem by methods of the vector criterion scalarization.

Read More

Key Words: Artificial intelligence; control theory; data driven; decision making; industry; industrial automation; integration; learning; modeling; operation control; process control; security; system theory.

Abstract. Industrial automation has entered a new stage in its long historical development by integrating the remarkable achievements of AI, accelerating exponentially in the last decade. The present work aims to summarize the prospects and expected main directions of the evolution of industrial automation in the era of artificial intelligence. It is based on the realistic assumption that innovations in industrial automation will be implemented primarily by upgrading existing and proven industrial facility management systems. One of the main directions is the multifaceted expansion of modern capabilities of industrial automation through the adaptation and further development of leading artificial intelligence technologies. The functional scope of industrial automation systems is expanding through “end-to-end automation” of industrial systems with a focus on operational management of the final production stages; the increasing volume of automation of particularly important procedures such as decision-making, dynamic optimization, coordination in unforeseen and extreme conditions (damages, accidents, disasters); predictive maintenance of the technical condition of facilities. Modern industrial automation, integrating elements of artificial intelligence, will implement a new approach in the strategy of industrial management through the transformation of approaches to achieving high efficiency, production quality, energy efficiency, environmental friendliness, and cybersecurity. Artificial intelligence methods will expand the capabilities of modern management of complex objects with large dimensions, nonlinearity, uncertainty, non-stationary variability of structures, models, parameters, and disturbances. The learning methods and generative capabilities of artificial intelligence systems will allow the implementation of complex hierarchical, networked and multi-connected structures with autonomous agents, necessary for a number of critical industrial infrastructure systems. Integrating elements of artificial intelligence into industrial automation systems still poses a major challenge for researchers, designers, operators and managers. Therefore, any implementation of artificial intelligence technologies so far has a unique character, represents an innovative solution and requires significant efforts, a holistic approach and conviction. The undoubted rapid development of artificial intelligence in the coming years and the increased needs of industrial systems will simulate rapid progress in industrial automation systems.

Read More

Key Words: Anaerobic co-digestion; complex mixtures; two-stage anaerobic co-digestion; Extremum Seeking Control; combination of AD with photovoltaic energy.

Abstract. Anaerobic digestion is a biotechnological method by which various organic wastes can be used as feedstock for biogas production. Anaerobic co-digestion can be considered as the simultaneous anaerobic digestion of a mixture of two or more waste types as substrate and co-substrate. Over the past two decades, our multidisciplinary team has performed many experiments on anaerobic co-digestion of various organic wastes. Different ratios of organic waste mixtures, in binary and ternary mixtures, have been investigated in order to maximize the resulting methane. Experimental studies confirm that biogas yields and process stability can be significantly increased when using two or three-component mixtures depending on the feedstock types and the ratio between them. Another option for increasing energy yields is the two-stage process, in which hydrogen and methane are produced simultaneously. A very good opportunity in this direction theoretically provides also the so-called extremum seeking control, on which our team has achieved significant results. Our research group is currently working on a project exploring the possibilities of combining solar energy with photovoltaic energy for improvement of energy balance of model facilities representing livestock farms.

Read More