International Information Technologies University

The paper proposes an adaptive architecture of an electromechanical drive with internally controlled energy exchange, implemented through the integration of an inertial flywheel and a controlled clutch into the structure of a planetary transmission. A multi-mass dynamic and energy model of the system is developed, and the power balance is verified. Based on the energy formulation, adaptive energy and predictive energy control strategies are implemented. The results of numerical simulation confirm that the use of the internal energy exchange loop increases system stability, reduces peak motor torque by 30–40%, decreases maximum output speed deviations by 35–45% under step load conditions, and reduces the root-mean-square tracking error by 20–30% compared with reactive energy-based control, demonstrating improved tracking performance and reduced actuator load compared to the classical drive architecture.

Machine translation for low-resource languages such as Kazakh remains a complex task due to the scarcity of training data, intricate morphological structures, and culturally specific linguistic characteristics. This study presents the first extensive exploration of fine-tuning large language models for automated post-editing of Kazakh translations. We introduce KazPE, a carefully curated and annotated dataset that includes 10,008 training sentences and 311 test sentences spanning six domains: the medical, scientific, journalistic, oral, fiction, and legal. The dataset features detailed error classifications across 9 linguistic categories. Our method fine-tunes GPT-4.1 mini using supervised learning to enhance translation quality by systematically correcting targeted errors. According to human evaluations, conducted on a continuous 0–1 scale, the fine-tuned model achieves an average quality score of 0.84, surpassing the baseline score of 0.80, corresponding to a 5% relative improvement. The greatest improvements are observed in handling morphological and lexical errors, as well as in domain-specific texts—particularly in legal (+17%) and medical (+22%) domains. In addition, the translations were evaluated using the automatic metrics: BLEU, TER and METEOR. The fine-tuned model shows improvements across all automatic metrics (BLEU, TER, METEOR), which confirms better n-gram overlap with reference texts, fewer edits needed, and enhanced lexical and semantic alignment with the reference texts. Comprehensive error analysis shows that the fine-tuning process effectively mitigates challenges related to Kazakh’s agglutinative morphology and specialized terminology, while preserving accuracy on already correct sentences. This research establishes the first structured evaluation framework for Kazakh translation post-editing and offers valuable guidance for enhancing machine translation in morphologically rich, low-resource languages. To facilitate further progress in Turkic language processing, we publicly release the KazPE dataset, trained models, and evaluation framework.

The object of the study was a modular large-scale urban VR system designed for real-time standalone execution. The study addressed the absence of an engineering asset based on modular design and rule-based geometric reduction for developing large-scale urban virtual reality (VR) environments that preserve navigation-relevant realism under standalone real-time performance constraints. Existing study primarily focused on small-scale scenes or treated realism as a global aesthetic property without systematic resource allocation strategies for city-scale environments. To solve this problem, a rule-based geometric reduction asset grounded in selective realism was developed and experimentally validated. Architectural objects were classified by spatial and functional significance (Classes A-C), and interior accessibility levels were introduced to regulate geometric complexity. A modular urban prototype comprising more than thirty architectural assets was implemented on a unified metric grid. Geometric reduction decreased vertex count from 49,114 to 4,033 and polygon count from 89,840 to 5,615, representing more than a fifteenfold complexity reduction while preserving object hierarchy. Experimental validation (n = 20) demonstrated high perceived spatial clarity (5.9/7), low navigation error rate (1.3 errors), mean completion time of 4.8 ± 1.2 minutes, and 18% average route deviation. The framework proved applicable to standalone VR education scenarios under strict rendering constraints

This study focuses on the experimental selection of the optimal Global Navigation Satellite System (GNSS) antenna for integration into a domestically developed navigation seal designed for transport logistics and digital customs control tasks in the Republic of Kazakhstan. The relevance of this study stems from the need to enhance technological independence and improve cargo monitoring efficiency to support domestic manufacturing and technological self-reliance. The scientific novelty lies in a comprehensive comparative assessment of commercial antennas not only in laboratory settings but also under field conditions, including installation on metallic surfaces, which reflects real operational scenarios. The objective of the research is to select an antenna that ensures the best balance between cost and performance for integration into the locally developed navigation seal. To achieve this, laboratory measurements and field experiments were conducted, analyzing parameters such as cold start time, carrier-to-noise density ratio (C/N₀, dB-Hz) for each satellite, the number of visible and used satellites, accuracy indicators (Position Dilution of Precision (PDOP), Horizontal Dilution of Precision (HDOP), Vertical Dilution of Precision (VDOP)), and tracking stability under internal and external shielding effects. As a result, the 1575R-A antenna was identified as having the most favorable characteristics among the tested samples and can be recommended for integration into the developed navigation seal. The practical significance of the study lies in providing recommendations for local manufacturers, whereas its theoretical contribution is the expansion of knowledge on the impact of structural and operational factors on the efficiency of GNSS antennas in compact devices.

This paper presents the design, modeling and control of a flywheel actuator for mobile platform stabilization. A Lagrangian-based model couples platform mechanics with DC-motor electromechanics. Analytical calculations estimate natural frequencies, damping and actuator limits. Numerical simulations in Python 3.12 evaluate cascade and state-feedback controllers for suppressing free oscillations and rejecting external disturbances. Additional studies examine filtering to improve measurement quality and unloading strategies to avoid actuator saturation. The results validate the proposed control architecture and demonstrate its applicability to robotic and energy systems operating under dynamic loads.

В статье рассмотрены механизмы государственной поддержки инновационного предпринимательства в контексте перехода к инновационной модели развития экономики, а также выявлена роль модернизации как одного из определяющих факторов устойчивого экономического роста. Определены основные направления государственной отраслевой поддержки инновационной деятельности и представлено их комплексное обоснование. В работе проанализированы меры государственной поддержки инновационного предпринимательства с учетом теоретических положений и практических инструментов их реализации. Рассмотрен полный спектр инструментов стимулирования индустриально-инновационного развития Республики Казахстан, включая финансовые механизмы, инфраструктурную поддержку и налоговые преференции. Проанализированы ключевые приоритеты Государственной программы «Дорожная карта бизнеса – 2025», определена ее структура, охватывающая поддержку новых предпринимательских инициатив, развитие приоритетных отраслей и снижение валютных рисков. Показаны основные механизмы реализации программы, включая субсидирование процентных ставок, предоставление гарантий и грантовое финансирование, направленные на стимулирование предпринимательской активности. Отмечена ориентация программы на поддержку субъектов бизнеса в моногородах, сельских населенных пунктах и в сфере индустриально-инновационной деятельности. Приняты во внимание ключевые инструменты государственной поддержки, такие как гранты на технологическую модернизацию, льготное кредитование и субсидирование внедрения энергоэффективных технологий. Показано, что государственная поддержка способствует повышению качества продукции, расширению ассортимента и росту конкурентоспособности предприятий пищевой промышленности. Выявлены основные барьеры развития инноваций и сформулированы рекомендации по повышению эффективности мер государственной поддержки инновационного предпринимательства в сфере производства хлебобулочных изделий. This article examines mechanisms of state support for innovative entrepreneurship and the transition to innovative economic development in the current context, highlighting the importance of modernization as a key factor in economic growth. The paper presents a detailed analysis of state sectoral support measures for innovative entrepreneurship. Two main aspects are examined: theoretical foundations and practical tools. The full range of measures aimed at stimulating industrial and innovative development in Kazakhstan is analyzed, including financing, infrastructure support, and tax incentives. This summary analyzes the key priorities of the State Program "Business Roadmap 2025." It examines the program's structure, including support for new initiatives, industry development, and currency risk mitigation. It also highlights mechanisms for subsidies, guarantees, and grant financing for entrepreneurs aimed at stimulating growth. The program is focused on supporting entrepreneurs in single-industry towns and rural communities, as well as providing industry-specific support for industrial and innovative entities. Key tools taken into account:: grants for technological modernization, preferential lending, and subsidies for the implementation of energy-efficient technologies. It is shown that government support contributes to improved product quality, expanded product range, and increased competitiveness of food industry enterprises. Barriers are identified and recommendations are proposed for increasing the effectiveness of government support for innovation in bread production.

Simultaneous Localization and Mapping (SLAM) enable autonomous mobile robots to build a map of an unknown environment while estimating their own position within it. Unstructured terrain, dynamic environmental conditions, and sensor limitations, often encountered in off-road areas can limit the effectiveness of SLAM. This review analyzes SLAM methodologies applicable to off-road mobile robots, categorizing them into LiDAR-based, visual, multi-sensor fusion, and learning-based or semantic approaches. Each category is examined in terms of algorithmic principles, performance characteristics, and suitability for varying terrain and environmental conditions. Furthermore, to assess the performance of these categories evaluation metrics are utilized, including accuracy, drift rate, robustness, computational efficiency, and benchmarking datasets. The comparative analysis highlights trade-offs between geometric precision, adaptability, and computational demands, with multi-sensor fusion and semantic integration. Real-time operation under limited onboard computation, scalability to large unstructured terrains, resilience in GPS-denied and feature-scarce environments, and integration with autonomous navigation systems are some of the identified research gaps. The findings emphasize the need for hybrid, computation-aware SLAM frameworks and standardized off-road benchmarks to accelerate the deployment of reliable autonomous systems in challenging outdoor environments.

Timely identification and accurate delineation of ultra-early ischemic stroke lesions in non-contrast computed tomography (CT) scans of the human brain are of paramount importance for prompt medical intervention and improved patient outcomes. In this study, we propose a deep learning-driven methodology specifically designed for segmenting ultra-early ischemic regions, with a particular emphasis on both the ischemic core and the surrounding penumbra during the initial stages of stroke progression. We introduce a lightweight preprocessing model based on convolutional filtering techniques, which enhances image clarity while preserving the structural integrity of medical scans, a critical factor when detecting subtle signs of ultra-early ischemic strokes. Unlike conventional preprocessing methods that directly modify the image and may introduce artifacts or distortions, our approach ensures the absence of neural network-induced artifacts, which is especially crucial for accurate diagnosis and segmentation of ultra-early ischemic lesions. The model employs predefined differentiable filters with trainable parameters, allowing for artifact-free and precision-enhanced image refinement tailored to the challenges of ultra-early stroke detection. In addition, we incorporated into the combined preprocessing pipeline a newly proposed trainable linear combination of pretrained image filters, a concept first introduced in this study. For model training and evaluation, we utilize a publicly available dataset of acute ischemic stroke cases, focusing on the subset relevant to ultra-early stroke manifestations, which contains annotated non-contrast CT brain scans from 112 patients. The proposed model demonstrates high segmentation accuracy for ultra-early ischemic regions, surpassing existing methodologies across key performance metrics. The results have been rigorously validated on test subsets from the dataset, confirming the effectiveness of our approach in supporting the early-stage diagnosis and treatment planning for ultra-early ischemic strokes.