Scientific Papers of Donetsk National Technical University. Series: “Computer Engineering and Automation"

USING AGENTSCRIPT TO PRODUCE MULTI-LEVEL AGENT-BASED MODELLING MODELS

2024-12-28T14:27:17+02:00

The relevance of the study is driven by the need to find affordable and effective tools for modelling complex systems that would be suitable for both research and educational purposes. Existing platforms, although providing significant opportunities, are often difficult to learn, which creates barriers to their widespread use. The purpose of the article is to study the capabilities of the AgentScript platform for creating agent-based models in the web environment, as well as to analyse its advantages and limitations in comparison with traditional tools. The practical significance of the work lies in highlighting the ways in which AgentScript can be used for rapid prototyping, development of models of medium complexity and interactive display. The tool allows to implement modelling without the need to install specialised software, which contributes to its popularisation among students and researchers. The scientific significance of the work lies in determining the prospects of using AgentScript for modelling multilevel systems. Particular attention is paid to the analysis of the platform architecture based on the Model-View-Controller template, which ensures efficient modelling. The article provides an overview of AgentScript functionality, describes the process of creating agents, environments and behavioural rules, demonstrates examples of implementing models for studying the spread of diseases, simulating forest fires and interactions in ecosystems, and compares the platform with other tools, which emphasises its advantages, such as accessibility and ease of integration with modern web technologies. The results show that AgentScript is an effective tool for research and education in the field of agent-based modelling, which is able to meet the needs of users with different levels of training.

EFFICIENT NEURAL NETWORK ALGORITHMS FOR MICROCONTROLLERS IN AUDIO DETECTION SYSTEMS

2024-12-23T09:33:31+02:00

This study focuses on the development and deployment of optimized neural network algorithms for real-time audio-based drone detection on resource-constrained microcontroller systems. Addressing the growing need for efficient drone detection, the research aims to design lightweight models that balance accuracy, low latency, and energy efficiency for edge devices such as STM32 and ESP32 microcontrollers.

The study evaluates neural network architectures, including Convolutional Neural Networks (CNNs), Recurrent Neural Networks (RNNs), and lightweight models like MobileNetV2 and TinyCNN. GRU-based RNNs achieved the highest accuracy of 98%, while MobileNetV2 offered a balance between performance and efficiency. Optimization techniques such as quantization and pruning were applied, enabling quantized MobileNetV2 to achieve inference speeds of 45 FPS and energy consumption of just 3 mW per inference. These results underscore the practicality of deploying such models in real-world scenarios.

Integration of these models onto microcontrollers was facilitated by frameworks like TensorFlow Lite and STM32Cube.AI. Field tests demonstrated the system’s robustness in diverse environments, including noisy urban areas. Detection accuracy exceeded 90% within a 100-meter range, even under adverse conditions. The results highlight the system’s potential for low-power, autonomous surveillance applications.

Key contributions include demonstrating the effectiveness of neural network optimization for edge systems, creating a scalable framework for audio-based detection, and advancing lightweight models for energy-efficient tasks. Future research will focus on advanced optimizations, expanding datasets, and integrating multi-modal detection.

This study lays a foundation for practical, efficient, and scalable drone detection technologies, addressing key challenges in energy use, accuracy, and real-world deployment.

CREATION AND APPLICATION OF A SENSOR NETWORK IN THE INFORMATION SYSTEM OF ENERGY ACCOUNTING OF A RESIDENTIAL COMPLEX

2024-12-15T15:13:57+02:00

The article is devoted to a comprehensive study of the possibilities of using wireless sensor networks for energy consumption metering in residential complexes. The existing solutions are analyzed, appropriate mathematical models are developed, and a simulation experiment is conducted to evaluate the effectiveness of the proposed routing and network self-organization algorithms. The dependence of the duration of operation, stability, and throughput of a wireless sensor network on the ratio of coverage and communication radii is analyzed. A way to improve the accuracy of monitoring the energy consumption system of a neighborhood by researching and developing an information system for accounting for electricity consumption is found.

The technical requirements of the network organization to the system and components of the neighborhood energy management system are considered, recommendations for the installation of the main components and their configuration are given, and installation diagrams are presented. The clustering algorithm for wireless sensor networks is improved, which is characterized by the integrated use of the previously known combined forecasting criterion and the value of the suitability of the sensor node to act as the master, which, due to the integrated use of the above values, provides a longer life cycle and an increase in the duration of the stability period compared to the known algorithms.

The study of the BSM of a 5-storey building of the central gateways of the building with the selection of the main gateway (node) and the main gateways of the houses of the neighborhood was carried out. To obtain a positive result, the network clustering method was applied using the Kohanen layer and neural networks. This work was carried out to increase the life cycle of the WSN, which was proved in theoretical studies.

INTEGRATION OF NODE-RED AND OPENAI API FOR INTELLIGENT ANALYTICS

2024-12-24T14:21:32+02:00

The purpose of this work is determined by the need to create an affordable and functional intelligent analytical system for automating the processes of data collection, processing and analysis. The proposed system is based on the integration of the Node-RED platform, the OpenAI API, and the Raspberry Pi minicomputer. This approach allows to effectively solve the tasks of monitoring, forecasting and optimisation in various industries, providing flexibility and affordability.

The paper analyses the advantages and limitations of using low-code platforms to create data processing flows, which greatly simplifies the integration of automation systems. Node-RED provides the ability to quickly build workflows, while the OpenAI API adds tools for deep data analysis using artificial intelligence models. The Raspberry Pi is a cost-effective hardware base suitable for small and medium-sized businesses.

The experiments allowed us to identify the key factors of system performance and stability. The test results showed that the system is capable of processing large amounts of data in real time, detecting anomalies and providing operators with prompt information about possible deviations. Particular attention was paid to the cost-effectiveness analysis, which confirmed a significant advantage over traditional solutions that require high implementation costs.

Taking into account the possibility of widespread use of the developed system, the paper also offers recommendations for its scaling, integration with cloud services and application to solve specific problems in various industries. This approach provides enterprises with the ability to quickly adapt to changes in technological conditions, reducing dependence on expensive and complex solutions.

The development of this system is an important step towards making modern technologies accessible to enterprises of all sizes. The proposed solution is aimed at improving the efficiency of production processes and creating cost-effective automation tools that meet modern market requirements.

INTELLIGENT SYSTEM FOR ASSISTING PEOPLE WITH VISUAL IMPAIRMENTS

2024-11-29T02:40:50+02:00

The article is dedicated to the analysis of modern technologies used to create intelligent systems for supporting people with visual impairments. The relevance of this topic is determined by the need to develop effective and accessible solutions that facilitate the daily lives of people with disabilities, particularly those with vision impairments. The aim of the article is to provide a comprehensive analysis of existing technologies, examine the advantages and disadvantages of available solutions, and identify potential ways to improve such systems. The article discusses leading solutions in this field, such as Aira, Be My Eyes, BlindSquare, and Seeing AI, which enable navigation and support for people with visual impairments through mobile devices, cameras, and other sensory systems.

The authors also explore the needs of the users of these systems, identifying the main challenges faced by people with limited vision, such as obstacle detection, providing timely information about the surrounding environment, and ensuring ease of use. Based on this analysis, the requirements for future support systems have been formulated, and the architecture of the intelligent system, which includes components for object recognition, obstacle detection, navigation, and voice assistance, has been developed.

Particular attention is given to the integration of object and obstacle recognition components, which are implemented using computer vision technologies such as OpenCV and neural networks. Additionally, the possibility of integrating voice assistants to provide convenient and accessible feedback is considered. The article also outlines the methods used for developing and testing such systems, including structural and comparative analysis of existing solutions, as well as experimental methods to assess the effectiveness of the software.

The research and development of intelligent systems for supporting people with visual impairments is an important step towards creating an accessible and barrier-free environment for people with disabilities, opening up new opportunities to improve their quality of life.

SYSTEM FOR MONITORING AND ANALYSIS OF ELECTRIC ENERGY CONSUMPTION IN HOUSEHOLDS BASED ON IOT TECHNOLOGIES

2024-11-29T17:11:26+02:00

This paper discusses the development of an energy consumption monitoring and analysis system for households based on IoT technologies. With the increasing demand for energy and the need for its rational use, automating the monitoring process is crucial, allowing users to efficiently manage energy resources. The proposed system integrates modern hardware and software tools for collecting, processing, storing, and visualizing real-time energy consumption data. It consists of a data collection module based on the ESP32-CAM board, which captures images of electricity meters, and software for processing these images using the OpenCV library. For data storage, the reliable PostgreSQL database is chosen, and for data visualization, Power BI and Looker platforms are used to create interfaces for convenient energy consumption analysis.

Functional and non-functional system requirements have been investigated, including real-time monitoring, data reliability and security, and user-friendliness. Existing technologies and solutions for energy consumption monitoring, such as EcoBee Smart Thermostat and EnergyHub, were also reviewed, allowing for the identification of the main advantages and disadvantages of such systems. To verify the effectiveness of the proposed system, its components were tested under real-life conditions, which helped assess the accuracy of data collection and processing.

The developed system has high potential for implementation in various households, helping reduce energy consumption costs, improve control over electricity use, and ensure its rational utilization. The practical significance of the work lies in creating an accessible and efficient tool for energy consumption monitoring, adaptable to the needs of different users and infrastructures. Further research will focus on expanding the system's functionality and integrating it with other devices to enhance its capabilities.