Improving the Operating Modes of the Electrotechnical Complex of Hydrogen Supply Using Intelligent Control in the Electric Transport System
Вантажиться...
Дата
2026
Автори
Науковий керівник
Назва журналу
Номер ISSN
Назва тому
Видавець
Igor Sikorsky Kyiv Polytechnic Institute
Анотація
Chen Linfei. Improving the Operating Modes of the Electrotechnical Complex of Hydrogen Supply Using Intelligent Control in the Electric Transport System. Dissertation for the degree of Doctor of Philosophy (PhD) in Specialty 141 – Electrical Power Engineering, Electrical Engineering, and Electromechanics (Field of Knowledge 14 – Electrical Engineering). – National Technical University of Ukraine «Igor Sikorsky Kyiv Polytechnic Institute», Kyiv, 2026.
The dissertation «Improvement of Operating Modes of the Electrotechnical Hydrogen Supply Complex Using Intelligent Control in the Electric Transport System» is dedicated to the urgent scientific task of enhancing the energy efficiency, operational stability, and economic viability of «green» hydrogen infrastructure. Given the rapid electrification of the global transport sector, the local Hydrogen Refueling Station (HRS) has emerged as a critical electrotechnical complex bridging renewable energy source (RES) grids with Fuel Cell Electric Vehicles (FCEVs). However, its efficient operation is severely hindered by a complex «double-stochastic» environment, where the significant intermittency of renewable generation constantly conflicts with the high volatility of refueling demand. Purpose. To enhance the energy efficiency and operational reliability of the electrotechnical hydrogen supply complex for electric transport by developing an intelligent automated control system for electrolyzer operating modes and energy conversion systems based on dynamic spatio-temporal «Source-Grid-Load-Storage» coordination. Relevance. This research aims to eliminate the structural mismatch between intermittent renewable energy and the rigid processes of hydrogen production. By addressing the issues of significant thermal fatigue and degradation of the Membrane Electrode Assembly (MEA) caused by frequent start-stop cycles, this study extends the operational lifespan of megawatt-class electrolyzers and ensures the economic security of large-scale green transport networks. Research Hypothesis. The application of a continuous electrothermal control topology in «hot standby» mode, based on a specialized type of Recurrent Neural Network (RNN) with deep long-term memory (LSTM – Long Short-Term Memory), will minimize the degradation of electrolysis cells and reduce the duration of transient processes during load changes. LSTM was specifically designed to overcome the «gradient vanishing» problem, allowing the system to effectively store and process long data sequences without losing the context of previous steps. This ensures the stability of the hydrogen supply system under the dynamic demand conditions of electric transport. Object of Research – conversion and consumption of electrical energy in hydrogen production and supply systems operating in stochastic conditions for the needs of electric transport. Subject of Research – the mode parameters of intelligent control of electrical equipment of hydrogen filling stations. Research Tasks: 1. To comprehensively describe the electrotechnical system (from the power grid to the electrolyzer) and systematically analyze the dynamic characteristics, degradation mechanisms, and operational constraints of megawatt-class PEM electrolyzers under intermittent power supply conditions, accounting for the non-linearity of characteristics during transient regimes. 2. To perform a critical analysis with comparative examples and characteristics of strengths and weaknesses based on a SWOT analysis. 3. To study the stochastic behavior patterns of FCEV fleet electrical networks and develop a multi-level load forecasting structure. 4. To formulate a macroscopic M/M/S/N queueing model integrated with microscopic vehicle thermodynamics for the quantitative estimation of instantaneous hydrogen mass flow. 5. To design an Automated Electrolyzer Control System (AECS) with a hierarchical architecture for real-time dynamic state management. 6. To justify the selection and adapt Long Short-Term Memory (LSTM) neural network algorithms for high-precision short-term (next-hour) demand forecasting. 7. To create a multiphysics (electrothermal) dynamic model of a multi-PEM electrolyzer cluster using equivalent circuit topologies. 8. To develop a Demand Response strategy allowing the complex to adapt to RES generation schedules (solar/wind) and electricity market prices. 9. To calculate the Levelized Cost of Hydrogen (LCOH) and perform a technoeconomic comparative analysis of the AECS and traditional Rule-Based Control. Chapter 1 analyzes the current state of hydrogen infrastructure development and the critical limitations of existing control methods. It is determined that modern HRS operate in an extremely complex «double-stochastic» environment. A comprehensive literature review indicates that traditional hysteresis (On/Off) control strategies suffer from «information blindness» – they react only to static pressure thresholds in storage tanks. This rigid approach causes significant thermal fatigue, mechanical stress in the MEA, and massive curtailment of renewable energy, failing to resolve the fundamental structural conflict between system efficiency and equipment lifespan. Chapter 2 characterizes a pioneering multi-level load forecasting structure. Departing from the oversimplified static assumptions of previous studies, this section synthesizes macroscopic M/M/S/N queueing theory with microscopic non-linear vehicle thermodynamics (utilizing real gas equations of state) to generate highprecision dynamic load profiles. Mathematical modeling and statistical analysis revealed an acute «supply-demand deficit» driven by traffic surges, quantifying an extreme peak-to-trough ratio of 15:1 during morning and evening rush hours. The results mathematically prove that smoothing these surges solely through passive buffer storage is economically unfeasible due to the exponential increase in CAPEX, confirming the rigorous necessity for active, demand-oriented management. Chapter 3 proposes the core architecture of the Automated Electrolyzer Control System (AECS). The system features a hierarchical structure that organically combines an AI-based load forecasting module (LSTM) with a Deterministic Finite Automata (DFA/FSM). By analyzing historical operational matrices, the LSTM network accurately predicts the aggregate demand for the subsequent hour. As an innovation, a «hot standby» operating mode is added to the FSM logic to dynamically decouple the thermal state of the electrolyzer stack from its electrical load. This enables proactive state transitions based on real-time predictive demand data and solar energy availability, transforming the electrolyzer from a passive load into a highly flexible energy router. Chapter 4 describes the physical validation of the proposed AECS through rigorous multiphysics dynamic simulation in the MATLAB/Simulink environment. Using a 2RC equivalent circuit model combined with a feedback thermal integrator, the electrochemical transient processes of a megawatt-class PEM cluster were simulated. Experimental results demonstrated that providing a minimum holding current density (yielding a production rate of 3,0 kg/h, or approximately 3% of nominal load) balances internal heat generation with environmental dissipation. This precisely maintains the stack temperature above the critical 60°C threshold during idle periods. Consequently, this mechanism completely eliminates the thermal stress associated with cold starts and endows the PEM electrolyzer with the capability to ramp up power in milliseconds to meet sudden load surges. Chapter 5 is dedicated to a comprehensive techno-economic evaluation of the commercial feasibility of the AECS. By comparing the intelligent AECS with standard Rule-Based Control (RBC) under a hybrid Time-of-Use (TOU) tariff, simulations showed exceptionally favorable economic indicators. The AECS successfully synchronized flexible hydrogen production with intermittent solar availability, increasing solar energy utilization from an initial 0% to 38,5%. Through intelligent tariff arbitrage and maximum asset protection, the system achieved a specific production cost to 2,92$/kg (approximately 19,96 CNY/kg based on current exchange rates), representing a significant reduction of 13,20%. Crucially, the implementation of the «hot standby» mode reduced harmful start-stop cycles by 98,3%, substantially decreasing associated depreciation costs. Furthermore, the validated fast transient characteristics confirm the system’s ability to participate in the electricity market as a Virtual Power Plant (VPP), providing critical active power support to the grid. Scientific Novelty: 1. Further developed the methodology of intelligent control for electrolyzer operating modes. The use of an adaptive algorithm based on the LSTM neural network architecture is proposed, allowing for the forecasting of hydrogen demand from electric transport and real-time optimization of the fuel production schedule. 2. Developed a comprehensive mathematical model of the electrotechnical hydrogen supply complex which, unlike existing models, accounts for the interconnection between power converter dynamic characteristics and non-linear electrochemical processes in the electrolyzer during operation from unstable renewable energy sources (RES). 3. Proposed a dual-mode intelligent optimization strategy integrating the Arithmetic Optimization Algorithm (AOA) and the African Vulture Optimization Algorithm (AVOA), and constructed an AO–AVOA–BP neural network model for the high-precision state-of-health (SOH) forecasting of lithium-ion batteries. Practical Significance: 1. The developed AECS architecture, multi-level load forecasting algorithms, and electrothermal models provide a ready-to-use control paradigm for megawatt-scale commercial green hydrogen projects. 2. The proposed «hot standby» strategy reduces start-stop depreciation costs by over 98% and lowers the Levelized Cost of Hydrogen to 2,92$/kg (19,96 CNY/kg). Furthermore, the established dynamic thermal regulation principles will directly facilitate the transformation of hydrogen refueling stations from rigid industrial loads into flexible grid assets. A new approach to integrating a hydrogen refueling complex into the Smart Grid system provides the possibility of the complex's participation in regulating the frequency and power of the power system, which increases the overall stability of the electric vehicle power supply network. 3. The proposed integration of AO and AVOA optimization mechanisms significantly improves global search capabilities and local convergence performance of the BP neural network. A dynamic coordination strategy that solves the «impossible trinity» of hydrogen infrastructure (simultaneously achieving low cost, high RES penetration, and extended service life) has been mathematically interpreted and verified against traditional baseline models. On NASA and CALCE datasets, the AO–AVOA–BP model achieves reductions in MAE, RMSE, and MAPE of 72,9–85,7%, 69,6–85,2%, and 72,6–85,8%, respectively, compared to the baseline BP model. 4. A matching method for hybrid vehicle power plants is proposed, combining orthogonal tests with Cruise software, supplemented by the development of control strategies for critical vehicle components on the MATLAB/Simulink platform. This can significantly reduce the number of experiments and the cost of developing electric vehicle powertrain prototypes. The dissertation consists of 190 pages. The main body contains 24 figures and 20 tables, 51 formulas.
Опис
Ключові слова
green hydrogen, renewable energy sources, energy efficiency, electrotechnical hydrogen supply complex, intelligent control, Automated Electrolyzer Control System (AECS), hot standby, multi-level stochastic modeling, зелений водень, відновлювані джерела енергії, енергоефективність, електротехнічний комплекс забезпечення воднем, інтелектуальне керування, автоматизована система керування електролізером, гарячий резерв, багаторівневе стохастичне моделювання
Бібліографічний опис
Chen Linfei. Improving the Operating Modes of the Electrotechnical Complex of Hydrogen Supply Using Intelligent Control in the Electric Transport System : dissertation submitted for the Doctor of Philosophy degree : 141 – Electric power engineering / Chen Linfei. – Kyiv, 2026. – 190 p.