DIGITAL ALGORITHMIC AND SOFTWARE SUPPORT OF DESIGN OF PORTSIDE RAILWAY TRANSPORT TECHNOLOGIC SYSTEMS

Danish Scientific Journal No102, 2025 107 TECHNICAL SCIENCES DIGITAL ALGORITHMIC AND SOFTWARE SUPPORT OF DESIGN OF PORTSIDE RAILWAY TRANSPORT TECHNOLOGIC SYSTEMS Luganchenko N. Postgraduate student of Chair «Stations and cargo work», Rostov State Transport University (RSTU), Rostov-on-Don Chislov O. Doctor of technical sciences, professor, head of Chair «Stations and cargo work», Rostov State Transport University (RSTU), Rostov-on-Don https://doi.org/10.5281/zenodo.17741009 Abstract The paper addresses the problem of designing portside railway transport systems in the digital economy era. The research focuses on developing AI-based algorithmic and software support for complex transport systems. The authors propose an integrated approach combining methods from system theory, operations research, artificial intelligence, mathematical modeling, and economic geography. The study presents modern digital methods for designing portside railway systems, including economic-geographical algorithms for car flow distribution, logistics centers network formation, technozenosis modeling, digital axiomatic modeling, and genetic layout algorithms. The research demonstrates practical applicability through registered software solutions implementation. The key findings include developing a unified framework for integrating various design methods into a single digital platform. This approach enables creating digital twins of transport nodes and conducting in-depth optimization at all project lifecycle stages. Keywords: port-side railway systems, digital design, digital algorithms, transport modeling, technozenosis, logistics optimization, axiomatic modeling, genetic algorithms, transport infrastructure, railway transport technology. Uninterrupted functioning of portside railway hubs, which play a critically important role in global logistics chains, is a fundamental condition for their efficiency. These facilities represent complex technological systems that integrate multiple elements of infrastructure, management, and logistics. In essence, such a hub is a point of convergence for several modes of transport, where any delays or inefficiencies are multiplied, creating a "bottleneck" effect and negatively impacting the competitiveness of the regional economy. The design of such systems is associated with numerous challenges stemming from their organizational structure: it is necessary to account for the pronounced stochastic nature of freight flows, dependent on global trends and seasonality; rigid spatial constraints of logistics flows; the need to minimize wagon turnover time while simultaneously maximizing the hub's capacity. Traditional design methods, based on regulatory heuristics, are often unable to find an optimal benchmark solution given such multi-criteria nature, high dimensionality, and large scale of the transport problem. Such methods often lead to suboptimal, redundant, or inflexible solutions that are not resilient to changing operational conditions. This objectively necessitates the urgent implementation of advanced AI-algorithmic and software suites, which enable multi-variant modeling, in-depth optimization, and "what-if" analysis at all stages of the project lifecycle—from strategic planning to detailed design and operational management. Modern approaches to designing technological systems for portside railway transport are characterized by the integration of methods from various fields of knowledge: system theory and operations research, artificial intelligence, mathematical modeling, and economic geography. A new paradigm, based on data and algorithmic optimization, is emerging. Among the most promising and practically effective methods are: economic-geographical analysis, which sets the macroeconomic and territorial framework; the techno-zenosis method, ensuring systemic stability and balance of elements; the axiomatic approach, formalizing the requirement decomposition process; genetic algorithms for solving high-level layout tasks; and neural network methods for forecasting, classification, and adaptive control under uncertainty. The synergistic application of these methods, organized within a unified software suite, opens up new possibilities for creating intelligent, efficient, and robust transport and logistics hubs. Picture 1 presents a generalized structural and logical diagram of the application of these methods in the design process. 108 Danish Scientific Journal No102, 2025 Pic. 1 Structural and logical diagram of the application of digital methods in the design of port railway systems The method of economic-geographical algorithmization for wagon flow distribution is based on integrating economic-geographical factors into the process of managing wagon flows in portside transport and technological systems (TTS). It allows for optimizing the distribution of empty wagons between unloading stations and accumulation stations or junction points, taking into account the territorial location of freight flows, section capacity, and the competitive environment [1]. Key stages of the method are: - Assessing the transport efficiency of polygon sections using the Coefficient of Transport Efficiency (CTE); - Constructing a Geometric Euclidean Model (GEM) of the railway polygon; - Forming a Geometric Routing Model (GRM) to delineate stations' "areas of influence"; - Solving a modified transport problem considering empty runs and cost parameters. The method helps reduce empty runs, decrease wagon turnover time, and increase wagon fleet productivity. The presented method can be integrated into AIalgorithmic systems for designing technological systems of portside railway transport. Its strength lies in combining economic, geographical, and technological factors, enabling not only the optimization of runs but also accounting for the competitive environment. However, to enhance the model's accuracy and flexibility, it is recommended to supplement it with machine learning methods and real data monitoring. The method for forming a network of logistics freight distribution centers, based on modified economic-geographical principles, represents a development of classical approaches to locating logistics facilities by incorporating multi-agent considerations (the presence of several competing or interacting terminals) and economic-geographical factors. It allows for optimizing the service areas of LGRCs in transport hubs, considering the actual configuration of communication routes, the client base, transportation costs, and environmental constraints [10]. The main idea involves building a geometric Euclidean model of the transport hub, which considers not only straight-line distances but also coefficients of nonrectilinearity reflecting the actual route lengths. The method also includes criteria of economic, temporal, and environmental efficiency, enabling a comprehensive assessment of terminal placement options. The method of techno-zenosis of a self-organizing ecosystem of transport processes is understood as a stable, self-organizing system of technological, production, and economic entities operating within a transport polygon. In the context of the Southern region's transport system, the techno-zenosis unites railway stations, ports, logistics centers, rolling stock, and other elements into a unified ecosystem capable of adapting and self-regulating under changing loads and external factors [3]. This approach allows to: - Assess the maturity level of the transport system; - Identify the stability of its operation; - Determine the optimal parameters for the interaction of elements; - Minimize external control through self-organization. The method is based on the concept of a transport polygon as a cluster of regional development and a techno-zenosis, where each element occupies its own Danish Scientific Journal No102, 2025 109 ecological niche and interacts with others according to laws similar to biological systems. Digital Axiomatic Modeling of a transport infrastructure facility is a formalized method for describing and optimizing technological processes based on a system of axioms. In this context, an "axiom" is understood as a minimal, logically complete unit of a technological process (e.g., "train reception," "movement of wagons to the cargo front"), represented as a formal rule or data structure. The totality of interconnected axioms describing the functioning of the entire facility (e.g., a port station) forms an Axiomatic Model of Transport and Technological Processes (AMTTP) [5]. The methodology is implemented in two key stages. In the first stage, a Basic Axiomatic Model (AMTTP-B) is formed, which describes in detail the infrastructure objects (tracks, yards, cargo fronts) and the technological operations between them in their normative or current state. This model serves as a digital twin of the system, allowing for structural analysis and identification of bottlenecks. In the second stage, a transition is made to a Modernized Model (AMTTP-M), designed for intelligent control and forecasting. This transition is achieved by processing the AMTTP-B using Recurrent Neural Networks (RNN), Fuzzy Set Theory (FST), and Active Systems Theory (AST). RNNs, in particular, allow for accounting temporal dependencies between successive states of the axioms, which is critically important for forecasting the time parameters of station operation under various management scenarios, resolving transport conflicts, and selecting rational options for servicing cargo fronts. The method of the Genetic Layout Algorithm for placing objects in a transport hub is a modification of the classical genetic algorithm, adapted for solving infrastructure object placement problems under conditions of fuzzy parameterization and multi-criteria. The algorithm operates with concepts such as "population" (a set of hub zones), "chromosome" (a list of zone coordinates or parameters), "crossover," and "mutation," which are used to generate new layout variants [8]. The main goal of the genetic layout algorithm for transport hubs is to minimize transport costs and capital expenditures while ensuring rational zoning of the hub's territory. For this purpose, a system of Linear Diophantine Equations (LDE) is used, allowing for the assessment of the "survivability" of descendant zones based on their parameters. The LDE solutions are filtered using the Pareto rule, the center of gravity method, and the minimization of zero solutions. The algorithm supports hub development strategies: merging, transferring, modifying, or liquidating zones. For example, "parent" zones can be merged into an "offspring" zone with parameter inheritance via crossover schemes. The method has been successfully tested on 17 transport hubs in the south of Russia, showing improved score ratings and reduced transport costs. It can be adapted for the design of port railway systems, taking into account their specificities – the presence of sea terminals, logistics zones, and intensive cargo flows. A more detailed description of the methods is presented in Table 1. Table 1 – Characteristics of digital methods in the design of port railway systems Method Economic-mathematical model Limits 1 2 3 Economic-Geographical Algorithms for Car Flow Distribution [1] 1. Transport Efficiency Coefficient of a section (TEC): 𝐾эф =∑𝑘𝑟𝑖𝑗⋅𝛽𝑟𝑖𝑗 𝑛 𝑟=1 Where: 𝑘𝑟𝑖𝑗=𝑝𝑟𝑖𝑗 𝑝max – normalized indicator of section performance; 𝛽𝑟𝑖𝑗=𝜎𝑟𝑖𝑗 𝑅𝑖𝑗 – indicator weight; 𝜎𝑟𝑖𝑗= ∑ 𝑥𝑠,𝑟 𝑖𝑗 𝑛 𝑠=1 – sum of indicator ranks; 𝑅𝑖𝑗=∑ 𝜎𝑟𝑖𝑗 𝑚 𝑟=1 – sum of all indicator ranks. 2. Integral function for assessing the operator company's efficiency: 𝑊КО−𝑖=𝛼1⋅𝑅1𝑖 ∗+𝛼2⋅𝑅2𝑖 ∗+⋯+𝛼6⋅𝑅6𝑖 ∗ Where: 𝑅𝑖∗=𝑅𝑖𝑝𝑟 𝑅𝑖𝑠𝑣 – ratio of project and existing parameter values; 𝛼𝑖 – weight coefficients, ∑𝛼𝑖=1. 3. bjective function of the transport problem considering empty runs: 𝑧=𝑎⋅∑∑𝜃𝑒𝑚𝑝𝑡 𝑖𝑗 ⋅𝑥𝑖𝑗→min 𝑛 𝑗=1 𝑚 𝑖=1 1. Dependence on expert assessments – the CTE calculation is based on the Delphi method, which introduces subjectivity; 2. Model staticity – parameters 𝑎𝑖,𝑏𝑗,𝑥𝑖𝑗 can change over time, but the model assumes they are fixed at the time of calculation; 3. Simplification of the geometric model – the GEM and GRM do not always accurately reflect the actual track configuration and logistic routes; 4. Data intensity – a large volume of statistical and infrastructural data is required to build the models; 5. Limited adaptability – the model poorly accounts for dynamic market changes and external factors (e.g., seasonality); 110 Danish Scientific Journal No102, 2025 Where: 𝜃empt 𝑖𝑗 =𝜆𝑖𝑗⋅𝜃empt – coefficient of empty turnover; 𝜆𝑖𝑗=𝛾 𝑘𝑖𝑗 – reduced coefficient of empty run; 𝛾 – harmonic mean 𝑘𝑖𝑗. 4. Economic-geographical distribution mode: 𝐼(𝐿)=∑(𝑝𝑘fr +𝑞𝑘fr ⋅𝐿GEM ∗) 𝑛 𝑘=1 +∑(𝑞𝑗𝑒𝑚𝑝𝑡⋅𝐿FMMmin ∗) 𝑚 𝑗=1 →min где: • 𝑝𝑘fr,𝑞𝑘fr – costs for initial-terminal and movement operations for a loaded run, • 𝑞𝑗𝑒𝑚𝑝𝑡 – costs for movement operations for an empty run, • 𝐿GEM ∗,𝐿GMMmin ∗ – section lengths in geometric models. 6. Computational complexity – requires the use of specialized software (e.g., based on Java + JavaFX). Formation of a Network of Logistics Freight Distribution Centers Based on Modified EconomicGeographical Principles [10] 1. Criterion for minimizing transport costs Π=arg min 𝑗∑(𝑝𝑘𝑟+𝐾нп 𝑗⋅𝑞𝑘𝑟⋅𝑙𝑗𝑘) 𝑛 𝑘=1 Where: 𝑝𝑘𝑟 – costs for initial-terminal operations for the 𝑘-th consumer; 𝑞𝑘𝑟 – cost of movement operations per 1 km; 𝐾np 𝑗 – non-rectilinearity coefficient for the 𝑗 -th LFDC; 𝑙𝑗𝑘 – distance from the 𝑗th LFDC to the 𝑘-th consumer. 2. Service area boundaries Equation for determining boundaries between the zones of influence of two LFDCs: 𝑝𝑖+𝑞𝑖⋅√(𝑥−𝑥𝑖)2+(𝑦−𝑦𝑖)2 =𝑝𝑗+𝑞𝑗⋅√(𝑥−𝑥𝑗)2+(𝑦−𝑦𝑗)2 3. Non-rectilinearity coefficient 𝑘np =𝐿𝑚 𝐿𝑒 Где: 𝐿𝑚 – actual (tariff) distance; 𝐿𝑒 – Euclidean distance. Average coefficient for a group of consumers: 𝐾avg.nl=1𝑛∑𝑘𝑛𝑙.𝑖 𝑛 𝑖=1 4. Placement efficiency criteria • Economic: ∑∑𝑐𝑖𝑗⋅𝑥𝑖𝑗→min • Temporal: 𝑇= ∑𝐿 𝑣𝑚𝑎𝑟+𝑛⋅𝑡𝑝𝑒+𝑛⋅𝑡𝑎𝑢𝑥 • cological: Environmental damage from emissions (e.g., for road transport): Уа=3,88⋅𝑙⋅𝛽⋅𝑔𝛾 • System organization level: 𝑅=1− 𝐻 𝐻𝑚𝑎𝑥 Where 𝐻 – system entropy. 1. Dependence on the quality of initial data – the accuracy of calculations heavily depends on the reliability of distance matrices, cost data, and coordinates; 2. Simplification of the transport network – despite accounting for the non-rectilinearity coefficient, the model remains Euclidean and may not capture all the specifics of the terrain and road network; 3. Model staticity – the method does not always account for dynamic changes in cargo flows and tariffs in real-time; 4. Sensitivity to the number of agents – computational complexity increases sharply with a large number of LFRCs and consumers; 5. Limited consideration of environmental factors – environmental criteria are represented in a simplified manner and may require enhancement; 6. Does not account for all aspects of competition – the model is focused on cost indicators but may not reflect all the market and strategic aspects of multi-agent interactions. Techno-zenosis of a Self-Organizing Ecosystem of Transport Processes [3] 1. Approximation of the distribution of techno-zenosis elements A power function is used to describe the distribution of the number of stations of different types in the polygon: 𝑁(𝑟)=𝐴⋅𝑟−𝛽 1. Dependence on data quality – the method requires reliable statistics on the elements of the techno-zenosis; Danish Scientific Journal No102, 2025 111 Where 𝑁(𝑟) – number of stations of rank 𝑟; 𝐴 – normalization constant; 𝛽 – parameter characterizing the degree of system balance. 2. Assessment of techno-zenosis efficiency Assessment is done by comparing empirical data with the theoretical model. The smaller the deviation, the higher the system's balance. 3. Enterprise profitability model in the techno-zenosis A parabolic dependence of revenue on transshipment volume is used: 𝑦=−𝑎𝑥2+𝑏𝑥−𝑐 Where: 𝑦 – revenue; 𝑥 – freight turnover; 𝑎,𝑏,𝑐 – parameters identified from statistical data. Optimal freight turnover is found via the derivative: 𝑑𝑦 𝑑𝑥=−2𝑎𝑥+𝑏=0⇒𝑥opt =𝑏 2𝑎 2. Stationarity of the environment – under conditions of abrupt changes (crises, shifts in logistics routes), the model may produce inaccurate forecasts; 3. Complexity of parameter identification – requires expert assessment and calibration for a specific polygon; 4. Limited applicability for small systems – the method is most effective for large transport polygons with a significant number of elements; 5. Neglect of non-economic factors – political, environmental, and social aspects may not be accounted for. Digital Axiomatic Modeling of a Transport Infrastructure Facility [5] 1. Basic Axiomatic Model (AMTTP-B) Formalizes the technological process as a set of interconnected axioms. For example, for station processes: AMTTP: { GL1 ↔ PG1 ↔ θ(SU) ↔ PS ↔ GFP1 } where each element (GL1, PG1, θ(SU), etc.) represents an axiom -- a formal description of a technological block or operation. 2. Recurrent transformation for forecasting (AMTTPM) For transitioning from the basic to the modernized model, a Recurrent Neural Network (RNN) architecture is used. Calculation of hidden layer values at the previous step: H(AMTTP){t−1} = F { W(ASTTP){XH} · X(AMTTP-B){t−1} + b(ASTTP){H} } where: X(AMTTP-B){t−1} -- input vector of basic model parameters at step t-1. W(ASTTP){XH} -- weight matrix between input and hidden layers. b(ASTTP){H} -- hidden layer bias vector. F - activation function. Predicted value at the current step: Y(AMTTP){t} = W(ASTTP){HY} · H(AMTTP){t−1} + b(ASTTP){Y} Iterative calculation of the hidden layer for forecasting the future step (t+1): H(AMTTP){t} = F{ W(ASTTP){XH} · X(AMTTPB){t} + W(ASTTP){HH} · H(AMTTP){t−1} + b(ASTTP){H} } Final forecast of the modernized model: Y(AMTTP-M){t+1} = W(ASTTP){HY} · H(AMTTP){t} + b(ASTTP)_{Y} 1. Static network parameters: The weight coefficients (W_{XH}, W_{HH}, W_{HY}) and biases (b_H, b_Y) are constant for all calculation steps within a single model, which can reduce adaptability to rapidly changing conditions. 2. Dependence on axiom length: The number of calculation cycles (iterations) of the neural network directly depends on the number of technological blocks in the axiom (process "length"), which affects computational complexity. 3. Discrete activation of axioms: Each RNN cell corresponding to an axiom can either output a value Y(AMTTP-M)_{t+1} or remain unused (value ∅), depending on whether that specific technological block is involved in a particular control scenario. This requires complex data flow control logic. 4. Architecture type: Although various RNN architectures are possible ("many-toone", "one-to-many", "manyto-many"), the "many-tomany" architecture is predominantly used for AMTTP modeling, which is the most complex to configure and train. 112 Danish Scientific Journal No102, 2025 Genetic Layout Algorithm for Placing Objects in a Transport Hub [8] 1. Objective function -- minimization of total transport costs: 𝑃=arg min 𝑇∑∑∑(𝐶𝑖𝑗𝑘 kd ⋅𝐴𝑇{𝑢(𝐿(𝐶𝐷) ∗)}𝑖𝑗𝑘 kd ⋅𝑄𝑖𝑗𝑘 𝑘𝑑 𝑝 𝑘=1 𝑚 𝑗=1 𝑛 𝑖=1+𝐶𝑖𝑗𝑘 𝑎𝑢𝑡𝑜⋅𝐴𝑇{𝑢(𝐿(𝐶𝐷) ∗)}𝑖𝑗𝑘 𝑎𝑢𝑡𝑜⋅𝑄𝑖𝑗𝑘 𝑎𝑢𝑡𝑜 +𝐶𝑖𝑗𝑘 𝑤𝑎𝑡⋅𝐴𝑇{𝑢(𝐿(𝐶𝐷) ∗)}𝑖𝑗𝑘 𝑤𝑎𝑡⋅𝑄𝑖𝑗𝑘 wat +𝐶𝑖𝑗𝑘 𝐺𝐺⋅𝐴𝑇{𝑢(𝐿(𝐶𝐷) ∗)}𝑖𝑗𝑘 ГГ ⋅𝑎𝑖𝑗𝑘 𝑇+⋯ +𝐵𝑖𝑗𝑘⋅𝐸𝑛) Where: 𝐶 – cost of transportation; 𝑄 – volume of traffic; 𝑎 – number of passengers; 𝐴𝑇{𝑢(𝐿(𝐶𝐷) ∗)} – matrix of modified distances considering the divergence coefficient 𝐶𝐷; 𝐵 – capital expenditures; 𝐸𝑛 – normative payback ratio. 2. Divergence coefficient (system homogeneity): 𝐶𝐷=1𝑝{∑(𝑥𝑘𝑖−𝑥𝑘𝑗 𝑥𝑘𝑖+𝑥𝑘𝑗)2} 𝑝 𝑘=1 1/2 3. Degree of infrastructure developmen: 𝜓= 𝑅 (𝐶𝐷)2 where 𝑅 – hub's score rating. 4. Linear Diophantine Equation (LDE) for assessing zone parameters: 5. 𝑇⋅𝑎𝑖+𝐿⋅𝑏𝑖+𝐵⋅𝑐𝑖+𝑆⋅𝑑𝑖+𝐴⋅𝑒𝑖+𝑃⋅𝑓𝑖= 𝑅ТY𝑖 Where 𝑇,𝐿,𝐵,𝑆,𝐴,𝑃 – zone parameters (transport, length, width, area, population, urban planning indicator). 1. Computational complexity with a large number of zones and parameters; 2. Dependence on the quality of initial data (statistics, standards, expert assessments); 3. Sensitivity to the choice of development strategies (merging, transferring, liquidating zones); 4. Need for manual adjustment at the stage of interpreting LDE solutions; 5. Orientation towards railway-urban hubs, which may require adaptation for other types of transport systems (e.g., portside); 6. Conditional nature of score ratings, which depend on the weighting coefficients of the criteria. The presented methods, despite differences in their mathematical apparatus and immediate areas of application, are not mutually exclusive. On the contrary, they form a complementary toolkit whose synergistic application allows for the comprehensive solution of multifaceted design tasks for portside railway systems. A comparative analysis of their key characteristics, presented in Table 1, shows that the limitations of one method can often be compensated for by the advantages of another. Important evidence of the practical value and technical readiness of the considered algorithmic solutions is their official registration as computer programs. Specifically, the software implementations of the economic-geographical algorithm for wagon flow distribution (GEM v.1), the digital axiomatic modeling method (Axiomatic v.1), the genetic layout algorithm (OP GKU 1.0), and the complex for designing a logistics centers network (LGRC v.1) have undergone state registration [2, 6, 9, 11]. This not only confirms their novelty and uniqueness but also demonstrates the possibility of their direct implementation into design and management practice. Thus, the economic-geographical algorithms and the method for forming the LGRC network set the macroeconomic and spatial framework of the future system, defining optimal zones of influence, routes, and placement points for key objects. However, their static nature and dependence on expert assessments can be mitigated through integration with dynamic and adaptive models. For example, digital axiomatic modeling, based on RNNs, allows for imbuing this framework with "intelligence" – the ability to forecast the operational situation, adapt to changes in cargo flows in realtime, and resolve technological conflicts. In turn, the techno-zenosis method provides a systemic view of the stability and balance of the entire transport hub ecosystem, assessing its maturity and capacity for self-organization. This approach helps verify whether local optimizations achieved by other methods undermine the overall system stability. It answers the question "how harmonious is the system as a whole?", while other methods focus on "how to optimally perform a specific operation?". The genetic layout algorithm serves as a powerful tool for solving high-level design tasks – the optimal placement of functional zones and objects under multiple conflicting criteria. Its ability to generate and select Pareto-optimal layout variants perfectly complements the economic-geographical calculations, providing the designer with a set of rational scenarios for subsequent detailed analysis. Danish Scientific Journal No102, 2025 113 The synergy effect manifests when these methods are organized within a unified software suite implementing the principle of end-to-end digital design. The sequence can be as follows: 1. The techno-zenosis method and economic-geographical analysis define the strategic framework and constraints; 2. The genetic algorithm generates hub layout variants; 3. For each layout variant, a digital twin is created using the axiomatic model; 4. Multi-variant experiments are conducted on the digital twin using forecasting methods (RNNs in the axiomatic model) and optimization methods (adapted economic-geographical algorithms) to assess operational efficiency under various scenarios; 5. The obtained results are evaluated from the perspective of techno-zenosis stability. Such an iterative process allows for a transition from seeking suboptimal solutions for isolated tasks to designing intelligent, robust, and economically efficient transport-technological systems capable of adapting to the challenges of the digital economy. Conclusion Thus, the modern challenges of designing technological systems for portside railway transport cannot be effectively solved by traditional normative-heuristic methods. The high dimensionality, multi-criteria nature, and inherent stochasticity of these systems necessitate a transition to a new paradigm based on digital algorithmic and software support. A comprehensive review and comparative analysis of advanced methods—economic-geographical distribution algorithms, modified principles for forming logistics center networks, techno-zenosis of self-organizing ecosystems, digital axiomatic modeling, and genetic layout algorithms—confirm their significant potential. Each method possesses unique advantages for solving specific classes of problems, from strategic spatial planning to operational forecasting and adaptive management. The key conclusion of this work is that the greatest synergistic effect is achieved not by the isolated application of these methods, but by their integration within a unified digital design platform. Such integration enables the creation of a "digital twin" of a transport hub, allows for multi-variant modeling, "what-if" analysis, and in-depth optimization at all stages of the project lifecycle. This comprehensive approach compensates for the limitations of individual methods and ensures that local optimizations do not undermine the overall stability, balance, and economic efficiency of the system. The practical implementation of the proposed AIalgorithmic suite paves the way for creating intelligent, data-driven, and robust portside railway hubs. Such next-generation systems will be characterized by reduced wagon turnover time, minimized empty runs, maximized throughput capacity, and increased adaptability to fluctuations in market conditions and cargo flows. Ultimately, this will significantly enhance the competitiveness of regional economies within global logistics chains. Acknowledgements The research was supported by the Russian Science Foundation grant No. 24-29-00869, https://rscf.ru/project/24-29-00869/ References: 1. Zadorozhny, V. M., Bakalov, M. V., & Shlyapnikov, S. V. (2022). Economic-Geographical Schemes and Algorithms for Wagon Flow Distribution in the "Station-Port" System. *Transport: Nauka, Obrazovanie, Proizvodstvo (Transport: Science, Education, Production): Proceedings of the International Scientific-Practical Conference, Rostov-on-Don, April 2527, 2022* (pp. 194-197). Rostov-on-Don: Rostov State Transport University. EDN ALGYBC. 2. Certificate of State Registration of Computer Program No. 2022681809, Russian Federation. GEM v.1 - a program for automated optimization of wagon flow distribution on a portside transport polygon: No. 2022681060: filed 07.11.2022: published 16.11.2022 / O. N. Chislov, V. M. Zadorozhny, V. A. Bogachev [et al.]. EDN PIOAWD. 3. Bakalov, M. V., & Zadorozhny, V. M. (2021). Development of Mathematical Methods for Intellectualizing Logistic Interaction on the Railway Polygon. In A. F. Borodin (Ed.), R. A. Efimov (Comp.), *Fyodor Petrovich Kochnev - an Outstanding Organizer of Transport Education and Science in Russia: Proceedings of the International Scientific-Practical Conference, Moscow, April 22-23, 2021* (pp. 48-52). Moscow: Russian University of Transport. EDN IZSMUY. 4. Chislov, O. N., Bezusov, D. S., & Luganchenko, N. M. (2023). On the Issues of Intellectualization in the Mathematical Modeling of Railway Transport and Technological Systems and Processes. *Digital Transformation of Transport: Problems and Prospects: Materials of the III International Scientific-Practical Conference, Moscow, September 27, 2023* (pp. 248-255). Moscow: Russian University of Transport. EDN ICRDGR. 5. Chislov, O. N., Luganchenko, N. M., Khan, V. V., & Magomedova, N. M. (2024). Development of a Methodology for Digital Axiomatic Modeling of a Transport Infrastructure Facility. Vestnik of the Siberian State University of Railway Engineering, 1(68), 22-32. https://doi.org/10.52170/18159265_2024_68_22. EDN VKTSSE. 6. Certificate of State Registration of Computer Program No. 2022681955, Russian Federation. Axiomatic v. 1 - a program for calculating transport process parameters: No. 2022681124: filed 07.11.2022: published 17.11.2022 / O. N. Chislov, N. M. Luganchenko, D. S. Bezusov [et al.]. EDN PBZIMB. 7. Chislov, O. N., Luganchenko, N. M., & Bezusov, D. S. (2025). Current Directions for Developing the Principles of Managing Transport Operations at Portside Freight Stations. Izvestia of the Petersburg University of Railway Transport, 22(3), 685697. https://doi.org/10.20295/1815-588X-2025-3-685697. EDN KXSPBW. 8. Chislov, O. N., Mizgireva, E. E., & Luganchenko, N. M. (2025). Genetic Layout Algorithm for 114 Danish Scientific Journal No102, 2025 Placing Objects in a Transport Hub: Problem Statement, Data Array Formation and Use, Solution Methodology. Izvestia of the Petersburg University of Railway Transport, 22(1), 3246. https://doi.org/10.20295/1815-588X-2025-1-3246. EDN RYTEYQ. 9. Certificate of State Registration of Computer Program No. 2025663676, Russian Federation. OP GKU 1.0 - Optimization of Genetic Layout Algorithm Parameters: filed 12.05.2025: published 29.05.2025 / O. N. Chislov, E. E. Mizgireva, N. M. Luganchenko. EDN BNMSEM. 10. Trapenov, V. V. (2024). A Complex for Designing a Network of Nodal Logistics Freight Distribution Centers under Conditions of Transport Service Multi-Agentness. Vestnik of the Rostov State University of Railway Transport, 3(95), 90102. https://doi.org/10.46973/0201-727X2024390. EDN VWJVES. 11. Certificate of State Registration of Computer Program No. 2024690984, Russian Federation. LGRC v. 1 - a program for automated optimization of cargo flow distribution in a transport hub: No. 2024683276: filed 07.10.2024: published 18.12.2024 / V. V. Trapenov, O. N. Chislov, S. V. Polyakov [et al.]. EDN UGBTFP.