APPLIED MATHEMATICS
AND CONTROL SCIENCES

The work is devoted to mathematical modeling of the dynamics of solid or liquid dispersed inclusions suspended in gas – gas suspensions. The study numerically simulated the dynamics of a gas suspension in a channel with and without taking into account the effect of coagulation of dispersed inclusions. It was assumed that a dusty medium moves in the channel, through the side surface of the channel there is an injection of droplet fractions coagulating with dispersed inclusions of the dusty medium. The work presents a mathematical model that implements a continuum technique for modeling the dynamics of multiphase media, which involves solving a complete system of dynamic equations for each of the phases of the mixture. The system of equations for the dynamics of gas, as well as dispersed phase fractions, included the density conservation equation, the equations for the conservation of spatial components of momentum, and the energy conservation equation. For the dispersed phase fractions, the "average density" function was introduced, which is the product of the fraction volume content, which is a function of spatial and temporal variables, and its physical density, which is a constant value.The carrier medium was described as a viscous compressible and heat-conducting gas. The interphase momentum exchange and interphase heat exchange were also taken into account. The interphase momentum exchange included the dynamic Archimedes force, the force of added masses, and the force of aerodynamic drag of particles. Homogeneous Dirichlet boundary conditions were specified for the velocity components of the carrier medium and the dispersed phase at the boundaries of the computational domain, modeled as solid surfaces. The dispersed phase of the gas suspension was described as multifractional, the fractions of which differ in the size of dispersed inclusions and the density of the particle material.The system of equations of the mathematical model was integrated by the finite-difference method of the second order of accuracy. To suppress numerical oscillations, a nonlinear correction scheme for the grid function was used. The mathematical model assumed that the interaction between particles was taken into account, through the absorption of smaller particles by larger particles due to collisional coagulation. Comparison of the results with and without taking into account the coagulation effect of the droplet and dust fractions of the gas suspension demonstrates that the absence of consideration of the coagulation effect has a significant effect both on the distribution of concentrations of the gas suspension fractions and on the physical fields of the fractions and the carrier medium.

Применение алгоритмов искусственного интеллекта в рутинных задачах, где имеются большие данные, играет ключевую роль в развитии любой отрасли, в том числе и в строительной. Алгоритмы машинного обучения широко применяются в зарубежных странах, а также набирают популярность научные исследования в данном направлении. Несмотря на интерес к применению искусственного интеллекта, к сожалению, как в российской практике, так и в научных исследованиях в строительной области его применение практически не затрагивается. Цель данного исследования состоит в рассмотрении существующих российских и зарубежных практик применения алгоритмов искусственного интеллекта в инженерных расчетах и наметка архитектуры будущей автоматизированной системы перевода ведомостей объемов работ и дефектных ведомостей в сметную документацию. Эта работа является одним из первых исследований в области применения алгоритмов искусственного интеллекта в дефектно-сметной документации. В будущем это поможет существенно сократить издержки компаниям строительной отрасли и внесет вклад как в практическую, так и в научно-исследовательскую деятельность.

The paper presents the development of an automated information retrieval system of media content on user's request in natural language. It describes the current problems of search in the context of rapid growth of media content volume and diversity, and also reveals the limitations of traditional search methods based on the exact match of keywords and metadata. Special attention is paid to the analysis of natural language processing (NLP) methods, such as tokenization, lemmatization, vectorization and semantic proximity computation, as well as the use of neural network models based on the “transformer” architecture. A comparative analysis of modern open language models, including Qwen3, Vikhr, Saiga and YandexGPTLite 5, in terms of their applicability for multilingual and multimodal text retrieval and generation tasks is carried out. The paper proposes solutions for integrating modern NLP-methods and neural network algorithms into the server side of the search system, which allows to improve the relevance, accuracy and convenience of media content search for unstructured and imprecise user queries. The presented approaches provide a balance between search quality, performance and versatility of the system, and open up prospects for further development of intelligent search services.

A hybrid approach combining machine learning and expert assessment is proposed for mass assessment of the market value of objects presented on the secondary market. The approach was tested using the example of assessment of used passenger vehicles. The developed prototype of an intelligent system is described, including a module for collecting and processing market data, the CatBoost model, and a mechanism for "secondary labeling" of data with the participation of professional experts. The novelty lies in the organizational model of interaction with experts - professional market participants and dynamic refinement of the model due to the assessment of readiness for bargaining. As a result of testing the prototype, an accuracy of MAPE ≈ 9.8% was achieved, which allows using the system in banking, insurance and appraisal activities. Directions for further development are discussed, including NLP and CV analysis of ads.