A hydrogen energy three-state-thermal energy gradient collaborative electric hydrogen heat multi-time scale optimization method

Abstract

This invention provides a multi-timescale optimization method for hydrogen energy in three states and thermal energy in a cascaded manner, relating to the field of multi-energy system technology. Photovoltaic and wind power grid-connected generation devices utilize renewable energy to generate electricity, which is then uniformly connected to the energy management platform. The platform coordinates and distributes the generated electricity to the electric energy storage unit, hydrogen energy storage unit, and thermal energy storage unit, and performs multi-scale scheduling based on the stratified energy storage combined with the synergistic utilization of hydrogen states and the cascaded utilization of thermal energy. The hydrogen state conversion and transportation module enables bidirectional controllable conversion of hydrogen between solid, liquid, and gaseous states, facilitating cross-unit and cross-timescale hydrogen energy transportation and energy quality matching. The conjugate thermal coupling heat exchange module establishes heat input and output channels with the hydrogen energy storage unit and the thermal energy storage unit, respectively. The energy management platform, based on a multi-timescale coordination optimization framework, performs unified and coordinated control of the electric energy storage unit, hydrogen energy storage unit, thermal energy storage unit, and conjugate thermal coupling heat exchange module.

Description

Technical Field

[0001] This invention relates to the field of multi-energy system technology, and in particular to a multi-timescale optimization method for hydrogen energy tri-state-thermal energy cascade synergy. Background Technology

[0002] In the context of the new power system, the proportion of renewable energy sources such as wind power and photovoltaics is rapidly increasing. However, their output exhibits significant volatility and randomness, easily leading to frequent peak shaving and energy curtailment issues in the power system. To enhance the absorption capacity of new energy sources and maintain system stability, various energy storage forms, such as electrochemical energy storage, hydrogen energy storage, and thermal energy storage, are widely introduced into integrated energy systems to achieve the spatiotemporal migration and multi-form coupling of energy. Among them, hydrogen energy storage, due to its high energy density, long storage period, and flexible cross-domain application, has become a crucial hub for the coordinated regulation of integrated electricity-hydrogen-thermal energy.

[0003] However, existing multi-energy systems still have several shortcomings in terms of energy storage levels and optimized scheduling. First, hydrogen energy systems mostly adopt a single gaseous hydrogen storage method, failing to fully utilize the high safety and high density characteristics of solid-state hydrogen storage, and also failing to combine the energy advantages of liquid hydrogen storage in long-distance transportation. This results in insufficient matching of hydrogen storage forms across different time scales, making it difficult to achieve optimal energy efficiency throughout the entire storage-transportation-use process. Second, the grade differences of thermal energy in integrated energy systems are not fully utilized. High-grade heat sources such as electrolysis by-product heat, liquefaction heat release, and fuel cell waste heat cannot be effectively utilized through staged recovery and conjugate heat exchange, resulting in low thermal energy utilization efficiency and limiting the overall energy quality utilization level of the system.

[0004] Furthermore, current time-scale scheduling mechanisms for multi-energy storage systems are relatively fragmented, primarily focusing on single-layer or single-time-domain optimization. They lack a multi-time-scale collaborative optimization framework that spans from annual planning to real-time control, making it difficult to achieve cross-level energy coordination and dynamic feedback. Simultaneously, under multiple uncertainties such as wind and solar power output, load demand, and hydrogen form conversion energy consumption, traditional deterministic scheduling methods lack robustness, making the system prone to energy deviations and performance degradation in complex operating scenarios.

[0005] Therefore, it is urgent to construct an electric-hydrogen-thermal multi-energy storage system that integrates the three-state synergistic mechanism of hydrogen energy and the cascade utilization structure of thermal energy. A multi-timescale coordinated optimization model from season to minute should be established, and the electric-hydrogen-thermal layered energy storage and dual-domain conjugate regulation of form-thermal energy should be realized in the energy management platform to improve the system's energy quality utilization rate, achieve cross-seasonal energy balance and robust optimization control under uncertain environments, and provide support for the efficient operation of new integrated energy systems. Summary of the Invention

[0006] To address the problems of single hydrogen storage form, extensive thermal energy utilization, fragmented time scales across multiple energy storage systems, and insufficient operational robustness in existing integrated electricity-hydrogen-thermal energy systems, this invention proposes a multi-timescale optimization method for electricity, hydrogen, and thermal energy storage that integrates three states of hydrogen energy and a cascaded thermal energy system. Using photovoltaic and wind power as the main energy inputs, the system achieves coordinated operation and multi-scale optimization of the three types of energy storage units (electricity, hydrogen, and thermal) through an energy management platform (EMS). The system introduces a three-state hydrogen energy synergy mechanism and a cascaded thermal energy conjugate control mechanism to achieve dynamic matching and energy quality coordination of energy storage in both the form and thermal energy domains.

[0007] The technical solution of this invention is as follows:

[0008] On the one hand, a multi-timescale optimization method for hydrogen energy tri-state-thermal energy cascade synergy is realized through a multi-timescale optimization system for hydrogen energy tri-state-thermal energy cascade synergy, specifically including: photovoltaic and wind power grid-connected power generation device, energy management platform EMS, electric energy storage unit, hydrogen energy storage unit, thermal energy storage unit, hydrogen form conversion and transportation module, and conjugate thermal coupling heat exchange module CHT;

[0009] The photovoltaic and wind power grid-connected power generation device uses renewable energy to generate electricity, and the generated electricity is uniformly connected to the energy management platform EMS. The energy management platform EMS coordinates and distributes the electricity to the electric energy storage unit, hydrogen energy storage unit, and thermal energy storage unit, and performs multi-scale scheduling based on the tiered energy storage combined with hydrogen form coordination and thermal energy cascade utilization.

[0010] The electrical energy storage unit uses lithium batteries, and the hydrogen energy storage unit includes an electrolytic hydrogen production device, a reversible solid oxide battery (RSOC), and a solid-liquid-gas multi-form hydrogen storage device. Electricity is converted into hydrogen through the electrolytic hydrogen production device, and the hydrogen storage form is adaptively selected according to the operating cycle and energy demand. Specifically, during the storage stage, a solid, liquid, or gaseous hydrogen storage device is selected according to the operating conditions. During the transportation stage, the hydrogen is transported in a normal temperature chemical state or a low-pressure liquid state through a liquefaction device, an amino carrier, or a LOHC system. When the output is insufficient or the demand is high, energy feedback is achieved through gasification heating, dehydrogenation reaction, or through fuel cells or reversible solid oxide batteries (RSOC), and a cascade utilization with the thermal energy ladder utilization submodule of the thermal energy storage unit is formed to match the thermal grade. The heat generated during the power generation process through gasification heating, dehydrogenation reaction, or through fuel cells or reversible solid oxide batteries (RSOC) is recovered in layers according to the temperature grade.

[0011] The thermal energy storage unit absorbs byproduct heat from electrolysis and liquefaction through a heat tank, heat pump, and waste heat recovery device, recovering waste heat from the fuel cell. Internally, it constructs three temperature zones: a high-temperature zone, a medium-temperature zone, and a low-temperature zone. The thermal energy ladder utilization submodule, based on these three temperature zones, forms a multi-stage energy quality cascading channel from high temperature to medium temperature and from medium temperature to low temperature through a staged heat exchanger and valve-controlled loop. Specifically, the high-temperature zone is used for liquefaction heat recovery and high-grade heating; the medium-temperature zone is used for vaporization heat absorption and district heating; and the low-temperature zone is used for waste heat recovery and recirculation preheating. Each temperature zone achieves orderly heat transfer between different grades through heat exchange paths.

[0012] The hydrogen form conversion and transportation module enables bidirectional controllable conversion of hydrogen between solid, liquid, and gaseous states, and completes hydrogen energy transportation and energy quality matching across units and time scales. Specifically, based on the system operating cycle, load demand, and thermal energy grade constraints, it automatically selects appropriate form switching paths, including hydrogen absorption or dehydrogenation reactions of solid hydrogen storage materials, vaporization and heating processes of liquid hydrogen, and compression and storage processes of gaseous hydrogen. By adjusting the form conversion method and hydrogen pressure and temperature, it achieves orderly transportation of hydrogen between electrolytic hydrogen production devices, reversible solid oxide batteries (RSOCs), fuel cells, transportation pipelines, and solid-liquid-gas multi-form hydrogen storage devices. In addition, the hydrogen form conversion and transportation module works in conjunction with the thermal energy cascade utilization submodule of the thermal energy storage unit to recover high, medium, and low grade heat generated during the form conversion process according to temperature stratification and return it to the corresponding thermal energy storage unit.

[0013] The conjugate thermally coupled heat exchange module CHT establishes heat input and output channels with the hydrogen energy storage unit and the thermal energy storage unit, respectively. Specifically, it consists of a bidirectional heat exchange loop composed of a liquefier, a vaporizer, a heat pump, and a thermal storage tank. The high-temperature heat released during the liquefaction stage is input to the upper layer of the thermal storage tank or the regional heating network through the high-temperature heat exchange channel. The medium-temperature heat is used for vaporization or heating, while the low-temperature heat is recovered through the heat pump or preheating loop. The heat required during the vaporization or hydrogen absorption reaction stage is supplied in layers by the thermal energy storage unit or the heat pump. Through the multi-layer rolling optimization of the energy management platform EMS, the coordinated control of medium- and long-term, short-term, and real-time is achieved: the minute level corresponds to the rapid adjustment of electric energy storage, the hour level corresponds to the dynamic adjustment of hydrogen energy form and the coordination of medium-temperature heat load, the weekly level corresponds to the cross-temperature zone energy migration and buffering of thermal energy storage, and the seasonal level corresponds to the cross-form hydrogen storage and thermal energy step balance correction.

[0014] The Energy Management System (EMS) is based on a multi-timescale coordination and optimization framework, used for unified coordination and control of electric energy storage units, hydrogen energy storage units, thermal energy storage units, and conjugate thermal coupling heat exchange modules across levels and cycles. The multi-timescale coordination and optimization framework specifically includes three levels: an annual level (i.e., a medium- and long-term planning level), a short-term level, and a real-time level. Among them, the medium- and long-term level constructs an annual optimization model with constraints on hydrogen form conversion and thermal energy grade stratification, determines the ratio of solid, liquid, and gaseous hydrogen storage capacity and the energy quality boundaries of each temperature zone of thermal storage, and achieves cross-seasonal energy balance and adaptive quota correction. The short-term layer uses a CNN-LSTM and TimesNet fusion prediction model to predict renewable energy output and load changes over the next 24 hours. A rolling time-domain optimization strategy generates an hourly power baseline and introduces a tiered energy storage synergy of electricity, hydrogen, and heat and a thermal energy cascade utilization mechanism to complete intraday power allocation and energy quality degradation regulation. The real-time layer uses the Model Predictive Control (MPC) algorithm as the core scheduling strategy, combined with a Warm-start fast calculation strategy, to perform 15-minute rolling optimization, achieving rapid power correction, morphological response, and adaptive regulation of thermal quality to maintain the dynamic steady-state operation of the system.

[0015] On the other hand, a multi-timescale optimization method for hydrogen energy tri-state-thermal energy gradient coordination includes the following steps:

[0016] Step 1: First, determine the annual scale input parameters, establish hourly decision variables and discretize them over time, construct an annual optimization model with hydrogen form conversion constraints and thermal energy grade stratification constraints, introduce cross-seasonal inventory smoothing and form switching penalty terms as well as conjugate heat balance and thermal energy grade stratification constraints, and form a dual-domain annual optimization framework for hydrogen form synergy and thermal energy tiered utilization.

[0017] Specifically, the system adopts a “form-stage-grade coordination strategy” in the annual layer, that is, it conducts joint planning among the three hydrogen storage forms of solid, liquid and gaseous, seasonal time stages and heat grades of high temperature zone, medium temperature zone and low temperature zone; specifically, the following variables are solved through the annual optimization model: (1) the capacity ratio of solid, liquid and gaseous hydrogen storage; (2) the form switching threshold of quarter or bimonthly; (3) the heat storage capacity boundary of high temperature zone, medium temperature zone and low temperature zone; (4) the heat energy quality transfer efficiency of high temperature zone → medium temperature zone and medium temperature zone → low temperature zone;

[0018] The above four annual optimization models are jointly determined under the following constraints: cross-seasonal inventory smoothing term, liquefaction, gasification, hydrogen absorption and desorption energy consumption and lifetime cost, thermal energy grade decline conservation constraint, and conjugate thermal energy exchange relationship; through solving, the optimal hydrogen storage form capacity configuration is obtained; at the same time, the upper and lower bounds of capacity in high temperature zone, medium temperature zone, and low temperature zone, as well as the unidirectional energy and mass transfer ratio from high temperature zone to medium temperature zone to low temperature zone are determined. The form capacity ratio, form switching threshold, energy and mass boundary of the three temperature zones, and cross-temperature zone transfer efficiency given by the annual layer will be used as constraints and sent to the short-term layer and real-time layer; the system establishes a quarterly or bi-monthly rolling correction mechanism, dynamically updating the capacity ratio, switching threshold, and energy and mass boundary based on the form response deviation, thermal energy balance deviation, and inventory deviation returned from the lower layer, thereby forming a cross-seasonal adaptive closed-loop optimization, so that the multi-form hydrogen storage and thermal energy cascade structure maintain the optimal balance between energy security and economy;

[0019] Step 1.1: Collect wind power data Photovoltaic power generation Electricity load demand Heat load demand In addition to hydrogen demand, regional electricity purchase and sale prices, regional heat recovery prices, and meteorological environmental data, combined with equipment rated parameters and operating efficiency, an annual input set and representative time series are established.

[0020] Then, the data compression stage is performed: K-means clustering algorithm is used to extract typical days and compress the hourly data throughout the year. Within the annual input set, the energy consumption and heat transfer characteristics of liquefaction and gasification processes are considered, and conjugate heat transfer efficiency parameters are defined, including thermal storage and charging efficiency. With heat release efficiency This describes the heat exchange efficiency between the exothermic liquefaction, endothermic vaporization, and hydrogen absorption / desorption reaction heat and the thermal energy storage system.

[0021] Within the annual input set, three heat grade ranges—high temperature, medium temperature, and low temperature—are defined, corresponding to the liquefaction heat release, high-grade heating, and medium-to-low temperature waste heat recovery stages, respectively; and a cross-grade energy and mass transfer efficiency parameter is introduced. , This describes the decreasing utilization relationship of thermal energy from high-grade to low-grade; among which... This refers to the energy and mass transfer efficiency from the high-temperature region to the medium-temperature region. The energy and mass transfer efficiency from the medium-temperature region to the low-temperature region;

[0022] Step 1.2: Establish The set of time-series decision variables at each moment;

[0023] Specifically, this includes PEM electrolyzers. Power consumed at all times Solid oxide electrolyzer (SOEC) in Constantly consuming electrical power The electrical power output during fuel cell power generation Power consumption of reversible solid oxide batteries (RSOCs) in electrolysis mode Power generation of reversible solid oxide batteries (RSOCs) in fuel cell mode Gaseous hydrogen storage unit in Hydrogen charging power at any time Gaseous hydrogen storage unit in Hydrogen release power at any time Liquid hydrogen storage unit in Hydrogen charging power at any time Liquid hydrogen storage unit in Hydrogen release power at any time Solid-state hydrogen storage units in Hydrogen absorption power at any time Solid-state hydrogen storage units in Dehydrogenation power at time The exothermic power of the hydrogen liquefaction process The heat absorption power during hydrogen vaporization Chemical endothermic power of solid hydrogen storage formation process Waste heat power of solid-state hydrogen storage desorption reaction ;

[0024] After the seasonal data is compressed to represent the time series, a typical quarterly set is formed, and annual layer optimization is performed with quarters as the rolling cycle.

[0025] Step 1.3: Construct an annual optimization model with constraints on hydrogen-containing gas form conversion and thermal energy grade stratification, where the annual optimization objective function is expressed as:

[0026] ;

[0027] in, For equipment operating costs, For equipment capacity investment cost, As a penalty for smoothing out cross-seasonal inventory, ; This represents the total amount of hydrogen energy within the system. This represents the average inventory level for the entire year. The inventory smoothing weighting coefficient is used. This is a penalty for form switching. , The energy consumption weighting coefficient for form switching. This refers to the energy required for hydrogen to switch between its gaseous and liquid states. The energy required for hydrogen to switch between liquid and solid states. The energy required for hydrogen to switch between gaseous and solid states; For conjugate heat energy terms, , For time step, The weighting coefficient for the benefits of liquefaction waste heat recovery. This represents the weighting coefficient for the system operating cost of the vaporization endothermic process. The weighting factor for the CHT heat charge power term. This is the weighting factor for the heat release power term in CHT. For stratification of thermal energy grade:

[0028] ;

[0029] in, As a penalty weight for thermal energy grade deviation, It serves as a metric for thermal energy grade in mixed-layer structures.

[0030] Step 1.4: Establish physical and operational constraints, as follows:

[0031] (1) Hydrogen energy conservation equation:

[0032] ;

[0033] in, for The total amount of hydrogen energy at any given moment. The efficiency of PEM electrolyzer in converting electrical energy into hydrogen energy. The efficiency of converting electrical energy into hydrogen energy in a solid oxide electrolyzer (SOEC). To optimize the time interval length, The efficiency of converting hydrogen energy into electrical energy in fuel cells. The efficiency of a reversible solid oxide battery (RSOC) in converting electrical energy into hydrogen energy in electrolysis mode. The efficiency of reversible solid oxide batteries (RSOCs) in converting hydrogen energy into electrical energy in fuel cell mode. For hydrogen in The inevitable loss of time;

[0034] (2) Shape conservation constraints: ;

[0035] in, This refers to the inventory of gaseous hydrogen. This refers to the liquid hydrogen inventory. This refers to the inventory of solid hydrogen.

[0036] (3) Conjugate thermal energy conservation constraint:

[0037] ;

[0038] in for The total amount of thermal energy in the thermal energy storage system at any given time. for The total amount of thermal energy in the thermal energy storage system at any given time. for The heat loss power of the thermal energy storage system at all times;

[0039] (4) Constraints on the conservation of thermal energy quality:

[0040] , ;

[0041] in, This refers to the heat power that can be transferred from the high-temperature region to the medium-temperature region. This represents the actual heat power received in the mid-temperature region. This refers to the heat power that can be transferred from the medium temperature region to the low temperature region. This represents the actual heat power received in the low-temperature region.

[0042] (5) Quarterly heat balance correction factor:

[0043] , ;

[0044] in, This is the heat balance correction factor for the qth quarter. This represents the net equivalent heat generated through the CHT module in the qth quarter. This is the reference heat value for the qth quarter, i.e., the baseline seasonal heat value;

[0045] Step 1.5: Establish a rolling correction mechanism based on quarterly or bi-monthly cycles; specifically: the system dynamically updates the annual energy quota, form capacity ratio and thermal energy grade boundary parameters based on the operating indicators returned by the short-term layer and the real-time layer, including hydrogen form response deviation and thermal energy balance deviation.

[0046] When the actual operating data is found to deviate from the annual plan by more than a set threshold, the lightweight re-optimization process is triggered.

[0047] The lightweight re-optimization process is specifically as follows:

[0048] For hydrogen energy storage units, the ratio of solid, liquid, and gaseous hydrogen storage capacity is adjusted and the form switching threshold is updated;

[0049] For thermal energy storage units, adjust the high, medium, and low temperature thermal energy capacity boundaries and conjugate heat transfer weights;

[0050] Update the smoothing penalty coefficient and related parameters of the energy-mass conservation constraint, specifically including: inventory smoothing weight coefficient. Heat quality deviation penalty weight Weighting coefficients for the benefits of liquefaction waste heat recovery Weighting coefficients of the vaporization endothermic process on the system operating cost Weighting coefficient of CHT heat charging power term Weighting factor for the CHT heat exothermic power term Energy and mass transfer efficiency from high-temperature region to medium-temperature region Energy and mass transfer efficiency from the intermediate temperature region to the low temperature region ;

[0051] The optimization results are distributed to the short-term layer for execution after the end of each quarter or bi-monthly period, and the annual layer operating baseline is updated to achieve cross-seasonal energy balance and multi-domain adaptive scheduling.

[0052] Step 2: Under the constraints of annual quotas and operational boundaries issued by the medium- and long-term layers, the short-term layer undertakes rolling optimization scheduling tasks with a 24-hour prediction window and a 7-day feedback cycle. Here, 24 hours represents the prediction window length for short-term rolling optimization, and 7 days represents the weekly feedback cycle of operational indicators to the medium- and long-term layers. Together, they constitute the dual-time-scale scheduling mechanism of the short-term layer. The inputs to the short-term layer are functionally divided into two categories: one is time-series data used for the prediction model, i.e., historical operational data. This type of input is fed into the short-term composite time-series prediction model, i.e., a fusion model of CNN-LSTM and TimesNet, to generate power and load prediction sequences for the next 24 hours. The other category is constraints and state inputs used for optimization solutions, including the form capacity boundary, heat grade range, form switching threshold, and penalty coefficient issued by the medium- and long-term layers, as well as operational deviations and instantaneous state quantities of each energy storage unit from the real-time layer. This type of data, along with the prediction results... The results are input into a 24-hour rolling optimization model to construct and solve short-term scheduling problems. Joint prediction is performed based on a short-term composite time-series prediction model, and a rolling time-domain optimization strategy is used to achieve hourly dynamic solutions, executing only the first hour's scheduling to form a continuous power baseline. During operation, an electric-hydrogen-thermal stratified energy storage synergy mechanism and a thermal energy tiered utilization mechanism are introduced. Specifically, the electric energy storage system achieves minute-level rapid response, thermal energy storage performs hourly balance adjustment based on the grade requirements of high, medium, and low temperature zones, and hydrogen energy storage undertakes cross-day energy migration and form compensation. Through a hydrogen solid-liquid-gas three-state coupling mechanism, a dynamic conversion relationship is established between solid-state stable storage, liquid-state buffering, and gaseous rapid adjustment. Furthermore, a conjugate thermal coupling and thermal grade stratified synergy mechanism are integrated to achieve multi-level recovery of latent heat and tiered reuse of thermal energy during liquefaction, gasification, and hydrogen absorption / desorption reactions. When the system experiences power or inventory deviations, phase switching and thermal energy grade redistribution are triggered. Finally, the 7-day cycle operation indicators are fed back to the medium- and long-term layers.

[0053] Step 2.1: Establish a short-term composite time series forecasting model. Specifically, a composite time series forecasting structure is adopted, which consists of a fusion forecasting framework composed of a CNN-LSTM model and a TimesNet model. The short-term composite time series forecasting model uses historical operating data to predict the wind power, photovoltaic power output and load demand for the next 24 hours and generates confidence intervals. The forecast results are used as inputs to the subsequent 24-hour rolling optimization model to form a rolling scheduling problem.

[0054] Step 2.2: Based on the forecast results and medium-to-long-term boundaries, establish a 24-hour rolling optimization model with the goal of minimizing system operating costs. Decision variables include grid power purchase and sale, power output of each dispatchable unit; charging power, discharging power, and state change of energy storage; hydrogen electrolysis power and fuel cell power generation; bidirectional state switching between solid hydrogen, liquid hydrogen, and gaseous hydrogen; heat power distribution in high-temperature, medium-temperature, and low-temperature zones; conjugate heat energy flow rates of liquefaction heat release, gasification heat absorption, and hydrogen absorption / desorption reaction heat; heat pump input power and output heat power; and power smoothing, inventory deviation compensation, and electricity-hydrogen-heat coupling adjustment variables. By updating the forecast and constraints every hour and solving the above decision variables, only the results of the first hour are executed to form a continuous short-term dispatch baseline.

[0055] The specific optimization objectives are as follows:

[0056] ;

[0057] in, To optimize the objective function, representing the overall system operating cost over the entire rolling time domain, To predict and optimize time sets, Weighting for inventory deviation penalties. for The target or benchmark inventory at any given time is issued by the mid-to-long-term layer. This is the hydrogen three-state switching vector. Indicates time The equivalent energy change between different phases.

[0058] The constraints include:

[0059] (1) Conservation of hydrogen energy and phase transformation:

[0060] ;

[0061] ;

[0062] , ;

[0063] (2) Electrical power and thermal balance:

[0064] ;

[0065] ;

[0066] in: For electrical energy storage discharge power, Power for charging electric energy storage, For thermal energy storage heat release power, Heating capacity from external renewable energy sources, Power for charging thermal energy storage;

[0067] (3) Power and inventory boundary constraints:

[0068] ;

[0069] ; ;

[0070] ; ; ;

[0071] ; ;

[0072] ; ;

[0073] ; ;

[0074] ; ;

[0075] ; ;

[0076] in: The minimum permissible inventory for hydrogen energy storage, The maximum permissible inventory for hydrogen energy storage, This represents the maximum permissible hydrogen charging power for a gaseous hydrogen storage unit. This represents the maximum permissible hydrogen release power for a gaseous hydrogen storage unit. This represents the maximum permissible hydrogen charging power for a liquid hydrogen storage unit. This represents the maximum permissible hydrogen release power for a liquid hydrogen storage unit. This represents the maximum allowable hydrogen absorption power for a solid-state hydrogen storage unit. This represents the maximum permissible dehydrogenation power for a solid-state hydrogen storage unit. This represents the maximum exothermic power during the hydrogen liquefaction process. This represents the maximum endothermic power during the hydrogen vaporization process. The maximum chemical endothermic power for the solid-state hydrogen storage formation process is... Maximum waste heat power of solid hydrogen storage desorption reaction.

[0077] Step 2.3: Adopt a rolling time-domain optimization strategy, update the prediction every hour and resolve the scheduling for the next 24 hours, and only execute the solution for the first hour to form a continuous power baseline;

[0078] The power baseline includes: electrical power, thermal power, hydrogen power, and constraint baseline;

[0079] When prediction errors or external disturbances cause baseline deviations, the system automatically triggers gas-liquid-solid state switching and heat flow redistribution in high-temperature, medium-temperature, and low-temperature zones.

[0080] Step 2.4: Construct a synergistic mechanism for stratified energy storage of electricity, hydrogen, and heat;

[0081] In short-term rolling scheduling, electrical energy storage units, thermal energy storage units, and hydrogen energy storage units are prioritized from top to bottom according to their response speed, energy storage capacity level, and duration scale, specifically "electrical energy storage → thermal energy storage → hydrogen energy storage". Based on the operable range and real-time inventory status issued by the medium- and long-term layers, when the state quantity of a certain energy storage unit reaches a preset threshold, the EMS triggers the hierarchical compensation logic, causing energy to be transferred to the next layer of energy storage units in a fixed order:

[0082] When the distance to the boundary of the electrical energy storage is less than a set threshold, the thermal energy storage is activated to participate in the balance.

[0083] When the thermal energy storage boundary distance is less than a set threshold, hydrogen energy storage is activated to participate in regulation.

[0084] The energy changes of each energy storage unit are uniformly represented as follows:

[0085] ;

[0086] in: This is an index for energy storage unit types, used to distinguish different types of energy storage subsystems. Indicates an energy storage unit. Indicates thermal energy storage unit, Indicates a hydrogen energy storage unit; For energy storage units exist Energy reserves at any given moment For energy storage units exist Energy reserves at any given moment For energy storage units The charging efficiency, For energy storage units exist The charging power at any given moment For energy storage units Energy release efficiency, For energy storage units exist The power output at any given moment;

[0087] The decision variables in the 24-hour rolling optimization model satisfy the upper and lower bound constraints of SOC for electric energy storage, the upper limit constraint of charging and discharging power for electric energy storage, the SOC change rate constraint for electric energy storage, the upper and lower bound constraints of energy inventory in the three temperature zones for thermal energy storage, the upper limit constraint of thermal power in each temperature zone, the thermal power ramp-up rate constraint in the three temperature zones, the upper and lower bound constraints of total inventory and gaseous, liquid, and solid partial inventory for hydrogen energy storage, the upper limit constraint of hydrogen form switching power, the form switching rate constraint, and the operating constraints of the power upper limit and power change rate of each hydrogen-related device.

[0088] Step 2.5: The EMS sets explicit energy migration trigger thresholds in the optimization solution: When any of the following conditions are met, the system immediately starts the electrolytic hydrogen production or hydrogen form conversion operation:

[0089] , ;

[0090] in: for The state-of-charge ratio of the energy storage unit at any given time. The upper limit buffer band parameter for triggering electric energy storage, This is the upper limit of the maximum permissible state of charge for the energy storage unit. For time t, the three-temperature zone thermal energy storage unit is in the temperature zone The available thermal energy below Temperature zone The upper limit of the maximum permissible thermal energy capacity It serves as an upper limit buffer zone for three-temperature zone thermal energy storage;

[0091] When the load rise rate exceeds the set threshold or hydrogen energy is insufficient, hydrogen energy cross-form recirculation is triggered: liquid hydrogen is vaporized, solid hydrogen is desorbed if necessary, and gaseous hydrogen enters the RSOC or fuel cell to feed back electricity.

[0092] , ;

[0093] in for Real-time renewable energy output The load gap threshold used to trigger hydrogen energy cross-mode recirculation. The parameters for triggering the lower limit buffer band of hydrogen energy storage;

[0094] Step 2.6: The short-term layer calls the conjugate thermally coupled heat exchange module CHT set in the system architecture; in the short-term rolling scheduling, the heat exchange power of CHT is used as... The optimized variables are used in the 24-hour rolling optimization model to dynamically manage the heat flow during latent heat recovery and morphological transformation.

[0095] The thermal power constraint of the conjugate thermally coupled heat exchange module CHT is:

[0096] ;

[0097] ;

[0098] ;

[0099] in for The overall heat exchange power of the CHT conjugate thermal coupling module at all times. and These are the minimum and maximum heat transfer capacities of CHT, respectively. The temperature at the heat exchange interface. To ensure safety, set the temperature. The allowable temperature deviation range;

[0100] Step 2.7: To achieve flexible coordination and dynamic coupling among multiple energy storage systems (electric, thermal, and hydrogen), a hierarchical priority soft penalty term is introduced into the objective function of the 24-hour rolling optimization model for the short-term layer:

[0101] ;

[0102] ;

[0103] ;

[0104] in: Indicates that the system is in Net power exchange on the electrical side at any given time. This refers to the conjugate heat power exchange between the thermal energy storage unit and the hydrogen energy storage unit. , , These are the priority penalty coefficients for electrical energy storage, thermal energy storage, and hydrogen energy storage, respectively.

[0105] Step 2.8: The short-term layer outputs hourly power allocation curves and constraint baselines, which are then sent to the real-time layer for execution;

[0106] The system summarizes operational indicators every 7 days, including energy storage utilization rate, mode switching frequency, thermal energy cascade efficiency, conjugate thermal energy utilization rate, power tracking error index, hydrogen energy cross-period balance performance, thermal energy storage cross-interval synergy performance, and energy storage equipment operating pressure index, and feeds them back to the medium and long term layer for quarterly or bi-monthly boundary correction and parameter re-optimization.

[0107] Step 3: The real-time layer uses the real-time optimization control module of Model Predictive Control (MPC) as its core, employing a Warm-start fast calculation strategy for 15-minute rolling optimization. Under the power baseline and boundary constraints issued by the short-term layer, the system coordinates and schedules energy based on the response speed and energy quality hierarchy of the electric-thermal-hydrogen energy storage: the electric energy storage unit undertakes rapid balancing at the millisecond to minute level, the thermal energy storage unit performs hourly heat load regulation and maintains thermal energy quality stratification, and the hydrogen energy storage unit achieves cross-day scale energy migration and emergency compensation through a three-phase dynamic switching mechanism. When the state quantity of electric or thermal energy storage is less than the set threshold, the system automatically triggers and recovers the liquefaction heat release through the conjugate heat exchange module CHT and injects it into the thermal energy storage system or regional heat network. When the load increases or when the system's hydrogen energy inventory falls below the minimum safe inventory threshold, the real-time layer triggers fuel cell power generation or solid-state hydrogen release, with CHT providing gasification heat absorption. The liquefaction, gasification, and hydrogen absorption / desorption processes are all bidirectionally controlled by CHT. Heat exchange channel regulation; within each rolling cycle, the real-time layer ensures the conservation of electrical, thermal, and hydrogen power and form energy, and satisfies the following constraints: upper and lower bounds of SOC, maximum charge / discharge power, and ramp rate constraints for the electrical energy storage unit; upper and lower bounds of energy in the three temperature zones, thermal power limits, temperature zone switching rates, and thermal energy grade constraints for the thermal energy storage unit; solid-liquid-gas three-form storage boundaries, form switching rates, liquefaction / gasification power limits, and hydrogen consumption constraints for the hydrogen energy storage unit; power balance constraints for the electrical-thermal-hydrogen coupling; power boundaries of the CHT conjugate heat exchange and latent heat recovery efficiency constraints; and operating boundaries issued by the short-term layer. The system combines prediction and feedback to achieve power correction, phase response, and thermal grade self-adaptation. When disturbances or deviations occur, dual-domain self-adjustment is automatically triggered: power domain, i.e., electrical domain self-adjustment: maintaining electrical power balance through fast charging and discharging of electrical energy storage and power correction of fuel cells or electrolyzers; thermal domain self-adjustment: maintaining thermal field stability through bidirectional heat exchange in the CHT, redistribution of thermal power in the three temperature zones, and step adjustment of thermal grade.

[0108] The real-time layer outputs 15-minute level correction values, state switching commands, and thermal power adjustment signals, which are then sent to the execution control unit. The execution control unit includes an electric energy storage power conversion controller, a battery management system (BMS), and power control modules for the electrolyzer and fuel cell; a three-temperature zone distribution valve regulator, a circulating pump speed controller, and a heat pump or heat exchanger power adjustment module for the thermal energy storage unit; and a liquefaction controller, a vaporization and pressurization flow control valve, a solid hydrogen absorption and desorption regulator, and a three-phase state switching actuator for the hydrogen energy storage unit. Simultaneously, operating indicators are periodically fed back to the short-term layer.

[0109] Step 3.1: The real-time layer uses the power baseline issued by the short-term layer. Baseline heat load Using the shape boundary as input and 15-minute adjustment step size, a short-term composite time series prediction model is used to generate power output and load estimates in the future time domain, providing input for real-time rolling optimization.

[0110] The optimization objective is to minimize operating costs and baseline deviation.

[0111] ;

[0112] in: For the system in Actual power output at any given time Power baseline issued from the short-term layer; This is the power deviation penalty coefficient;

[0113] Step 3.2: Hierarchical fast response and power correction mechanism;

[0114] EMS establishes a multi-time-constant fast response mechanism based on the dynamic characteristics of energy storage units:

[0115] The energy storage unit is responsible for correcting fluctuations at the second to minute level;

[0116] The thermal energy storage unit performs hourly-level thermal buffering and waste heat absorption, and achieves a tiered thermal energy utilization mechanism through temperature-zoned heat exchange: the system divides the interior of the thermal energy storage unit into three temperature zones: a high-temperature zone, a medium-temperature zone, and a low-temperature zone. Each temperature zone corresponds to a temperature setpoint. The thermal energy ladder utilizes the temperature zone, temperature set point, and inventory level of the sub-module to dynamically adjust according to the high → medium → low ladder rule; when the heat release in the high temperature zone approaches the lower limit threshold, the medium temperature zone automatically replenishes heat; when load fluctuations cause heat loss in the low temperature zone, the medium temperature zone prioritizes energy supply, realizing the orderly migration and dynamic balance of heat between different temperature layers.

[0117] The hydrogen energy storage unit introduces a rapid solid-liquid-gas phase switching mechanism to complete trans-diurnal energy transfer and flexible compensation; specifically, when the electrical energy storage unit or the thermal energy storage unit meets the trigger threshold condition... At this time, the energy management system (EMS) automatically triggers electrolysis to produce hydrogen or liquefy hydrogen for storage, realizing the energy transfer from electricity to hydrogen; the heat released during liquefaction is injected into the medium-temperature or low-temperature zone via the conjugate heat exchange module (CHT); when the load increases or the hydrogen storage is sufficient... This triggers fuel cell power generation or solid hydrogen release, i.e., hydrogen energy to electricity return, in which the heat absorption of liquid hydrogen vaporization or solid hydrogen desorption is provided by the high temperature zone or heat pump;

[0118] The conjugate heat power exchange between the thermal energy storage unit and the hydrogen energy storage unit is defined as follows:

[0119] ;

[0120] Step 3.3: Energy Conservation and Safety Constraints;

[0121] The real-time layer satisfies the conservation relationships of electrical, thermal, and hydrogen energy flows in each rolling cycle:

[0122] ;

[0123] ;

[0124] Meanwhile, the thermal energy reduction constraint is met in each temperature zone of the thermal storage:

[0125] , , ;

[0126] in, , These represent the heat transfer power from high temperature to medium temperature and from medium temperature to low temperature, respectively.

[0127] The phase switching equation and the energy conservation equation together constitute the state evolution constraint of the hydrogen storage unit, which is used to limit the phase switching frequency and morphological energy balance in rolling optimization.

[0128] The regulation process of the energy storage unit is considered as a single-step approximation of the MPC target, and the dynamic equation is:

[0129] ;

[0130] in: For electrical energy storage discharge power, Power for charging electric energy storage, , These represent the charging and discharging efficiencies of the energy storage unit, respectively. for State of electrical energy storage at any given time. for State of electrical energy storage at any given time. The charging efficiency of the energy storage unit The discharge efficiency of the energy storage unit;

[0131] Step 3.4: Dual-domain self-adjustment mechanism and dynamic steady-state maintenance;

[0132] During the rolling solution process, the system combines short-term forecast data with local state feedback to achieve rapid correction of power deviations, sensitive response to phase switching, and adaptive allocation of thermal energy grade. When external disturbances or forecast errors cause deviations in the operating state, the real-time layer automatically triggers a dual-domain self-adjustment mechanism of phase and grade: the electric-hydrogen channel balances the energy gap through phase switching, and the thermal channel maintains the stability of the temperature gradient through valve control, thus achieving dual protection of electric power balance and thermal field stability.

[0133] Step 3.5: Execution and Cross-Layer Feedback;

[0134] The real-time layer outputs 15-minute level power correction values, form switching commands, and thermal energy adjustment signals, and sends the CHT valve opening control to the execution unit in conjunction with these signals. The EMS continuously monitors the thermal potential difference ΔT and form inventory in each temperature zone, updating self-adjustment parameters and thresholds in a closed loop. The system summarizes operational indicators every 7 days, including power deviation residuals, CHT energy reuse rate, form switching frequency, and thermal energy quality utilization efficiency, and feeds them back to the short-term and medium-to-long-term layers for quarterly boundary correction and retraining of the composite time-series forecasting model, forming a quarterly boundary correction mechanism.

[0135] Step 4: After completing the medium- and long-term and short-term layer optimizations and summarizing the real-time layer operation indicators, the system enters the sensitivity and robustness analysis phase. Multi-scenario perturbation samples are generated using Latin hypercube sampling (LHS) combined with block bootstrap to maintain the temporal correlation between wind and solar power output, load, and hydrogen storage and conjugate heat exchange behavior. A representative set of operating scenarios is formed through clustering and scenario reduction. A sub-Bruker optimization model based on Wasserstein distance is established to uniformly incorporate uncertainties into the evaluation, including wind and solar fluctuations, load deviations, hydrogen phase switching energy consumption, efficiency fluctuations due to heat energy grade decline in the three temperature zones, CHT heat exchange efficiency perturbations, and heat pump coefficient of performance (COP) changes. The sub-Bruker optimization model, through synergistic constraints and sensitivity trade-offs in the morphology-thermal energy dual domains, outputs the robust operating domain and the corrected hydrogen morphology ratio and thermal storage capacity quota, providing adaptive boundaries and parameter calibration for the next round of annual optimization.

[0136] Step 4.1: Based on the quarterly boundary correction mechanism of the annual collaborative optimization model, the system enters the sensitivity assessment and robustness optimization stage. First, a multi-scenario disturbance sample set is generated using Latin hypercube sampling (LHS) combined with the block bootstrap method to maintain the temporal correlation between wind power, photovoltaic output, load, hydrogen phase switching, and thermal energy grade decline behavior. Four types of disturbance variables are introduced into the disturbance sample set: renewable power output disturbance. Load forecasting error Uncertain deviation between energy storage phase switching rate and energy consumption Conjugate heat exchange power fluctuation term ;

[0137] Step 4.2: Cluster and reduce the scenes from the perturbation samples to obtain a representative set of scenes. By maintaining the correlation between form energy and thermal energy grade, a representative expression of the dynamic synergy of multi-energy systems can be achieved.

[0138] Step 4.3: Construct a DRO model based on Wasserstein distance and the split-bar optimization method.

[0139] ;

[0140] in Represented by empirical distribution Centered on, the distance from Wasserstein does not exceed A set of uncertain distributions; This is a vector of decision variables, containing all controllable quantities of the system within the scheduling period; In distribution Expected operating cost under the perturbation scenario; For the perturbation scenario, For the system in disturbed scenarios The multi-energy cooperative operation cost function includes:

[0141] ;

[0142] in: This is the power deviation penalty coefficient. This is the comprehensive penalty coefficient for thermal power fluctuations in the CHT module. The power baseline is from the short-term layer. The weighting coefficient for the penalty of mixed-layer thermal grade;

[0143] Step 4.4: Solve the robust optimization model by reducing the weight of the scenario, and then further correct the hydrogen three-state ratio and thermal energy level boundary constraints by combining the results of sensitivity analysis.

[0144] The system outputs a robust operating domain across layers. And the revised morphology and thermal storage synergy quota: , Together, they serve as the initial boundary and parameter input for the next round of annual layer optimization, enabling closed-loop updates and full-cycle adaptive scheduling in both the morphology and thermal energy domains.

[0145] The beneficial effects of adopting the above technical solution are as follows:

[0146] This invention provides a multi-timescale optimization method for hydrogen energy tri-state and thermal energy cascade synergy. By constructing a hydrogen energy tri-state synergy mechanism, a thermal energy cascade conjugate control mechanism, and a multi-timescale hierarchical optimization system, this invention achieves energy coupling and coordinated scheduling of multiple energy storage units (electric, hydrogen, and thermal), enabling dynamic energy matching and multi-domain synergy across different timescales. This method significantly improves renewable energy utilization and system energy quality efficiency, enhances the robustness and adaptability of integrated energy systems under uncertain environments, and has high engineering promotion value and application prospects. By embedding the electric-hydrogen-thermal hierarchical energy storage synergy mechanism and the tri-state-thermal dual-domain conjugate control mechanism into the energy management platform (EMS), it achieves efficient renewable energy absorption, energy matching of energy storage units, and robust cross-scale operation of the system, offering the following outstanding advantages:

[0147] (1) Constructing a three-state hydrogen storage and thermal energy cascade collaborative structure to improve the overall energy efficiency of the system. This invention is the first to collaboratively model and dynamically switch between solid, liquid, and gaseous hydrogen storage methods. In the storage stage, solid or liquid form is used to improve safety and energy density. In the transportation stage, liquefaction and gasification are adaptively switched according to distance and load requirements. In the utilization stage, electro-thermal energy feedback is achieved through fuel cells or exothermic reactions. A conjugate thermal coupling heat exchange module (CHT) is set between the hydrogen storage unit and the thermal storage unit to realize bidirectional recovery of latent heat in the liquefaction exothermic process and gasification endothermic process. It forms an energy quality cascade transfer path with the high-medium-low temperature multi-grade structure of the thermal storage, which significantly improves the waste heat utilization rate and overall energy quality efficiency of the system.

[0148] (2) Establish a multi-timescale coordinated optimization system from season to minute to achieve cross-layer energy synergy and adaptive feedback. This invention proposes a multi-level coordinated optimization framework of season-day-hour-minute: the medium-to-long-term layer realizes the planning of annual energy quota and form capacity boundary; the short-term layer uses a composite model of CNN-LSTM and TimesNet to perform intraday rolling prediction and power allocation; and the real-time layer uses model predictive control (MPC) combined with a warm-start fast calculation strategy to perform 15-minute-level rapid regulation, forming a "plan-execution-feedback" closed-loop control structure. The system can automatically allocate regulation priority according to the response speed of the energy storage unit, realize minute-level rapid adjustment of electric storage, hour-level thermal storage balance and cross-day hydrogen storage energy migration, and ensure the dynamic balance and stable operation of energy flow and power flow.

[0149] (3) A conjugate heat-grade dual-domain regulation mechanism is introduced to achieve synergistic optimization of energy quality among electricity, heat, and hydrogen. By establishing a three-layer heat quality range of high, medium, and low temperature for thermal storage, combined with bidirectional channels for liquefaction heat release and gasification heat absorption, the heat energy can be cascaded and reused at different temperature levels. By-product heat generated during electrolysis, liquefaction, and fuel cell operation can be recovered and injected into the corresponding temperature zone, improving the closed-loop energy utilization rate and preventing heat backflow and energy quality mixing. This structure not only improves the heat recovery efficiency but also enhances the synergistic response capability of the hydrogen storage unit under different operating conditions.

[0150] (4) A sensitivity assessment mechanism for multi-energy systems based on sub-Brühl bar optimization is proposed to enhance system robustness and reliability. In the later stages of system operation, this invention introduces Latin hypercube sampling (LHS) and the Wasserstein distance sub-Brühl bar optimization model to uniformly model and assess uncertainties such as wind and solar power output, load prediction errors, hydrogen phase switching energy consumption, and thermal efficiency degradation. Through scene reduction and robust domain correction, adaptive adjustment of the operating boundary is achieved, effectively suppressing performance degradation caused by prediction errors and external disturbances, and improving the system's stability and safety margin in variable environments. Attached Figure Description

[0151] Figure 1 This is a schematic diagram of the multi-energy storage and multi-timescale coordinated optimization system of the electric-hydrogen-thermal according to an embodiment of the present invention;

[0152] Figure 2 This is a flowchart of a medium-to-long-term optimization method for the annual layer of this invention, which includes morphological transformation constraints and thermal energy grade stratification.

[0153] Figure 3 This is a flowchart of the short-term scheduling layer of the present invention based on the CNN-LSTM and TimesNet fusion prediction model for rolling optimization of multiple energy storage systems (electricity-hydrogen-thermal).

[0154] Figure 4 This is a flowchart of the 15-minute rolling scheduling method for real-time layer operation based on Model Predictive Control (MPC) and Warm-start fast calculation strategy of the present invention;

[0155] Figure 5 This is a flowchart of the sensitivity analysis based on multi-scenario perturbation and sub-Bruker optimization of the present invention. Detailed Implementation

[0156] The specific implementation methods of this application will be further described in detail below with reference to the accompanying drawings and embodiments.

[0157] Example 1: This invention proposes a coordinated optimization system for multiple energy storage and time scales (electricity-hydrogen-thermal) considering the synergistic effects of hydrogen energy in its three states and thermal energy levels, such as... Figure 1 As shown.

[0158] On the one hand, a multi-timescale optimization method for hydrogen energy tri-state-thermal energy cascade synergy is realized through a multi-timescale optimization system for hydrogen energy tri-state-thermal energy cascade synergy, specifically including: photovoltaic and wind power grid-connected power generation device, energy management platform EMS, electric energy storage unit, hydrogen energy storage unit, thermal energy storage unit, hydrogen form conversion and transportation module, and conjugate thermal coupling heat exchange module CHT;

[0159] The photovoltaic and wind power grid-connected power generation device uses renewable energy to generate electricity, and the generated electricity is uniformly connected to the energy management platform EMS. The energy management platform EMS coordinates and distributes the electricity to the electric energy storage unit, hydrogen energy storage unit, and thermal energy storage unit, and performs multi-scale scheduling based on the tiered energy storage combined with hydrogen form coordination and thermal energy cascade utilization.

[0160] The energy storage unit uses lithium batteries to perform rapid regulation and peak shaving and valley filling from minutes to days, to achieve short-term balance between grid load and renewable energy output, to alleviate power fluctuations, reduce peak loads, and fill in low-valley periods, in order to maintain stable grid operation and achieve rapid regulation of power fluctuations at the minute level.

[0161] The hydrogen energy storage unit includes an electrolytic hydrogen production device and a reversible solid oxide cell (RSOC). RSOC (Resistant Solid Oxide Battery) and solid-liquid-gas multi-form hydrogen storage devices, electricity is converted into hydrogen by electrolysis hydrogen production devices (PEM or SOEC), and the hydrogen storage form is adaptively selected according to the operating cycle and energy demand; specifically: during the storage stage, solid, liquid or gaseous hydrogen storage devices are selected according to operating conditions; during the transportation stage, hydrogen is transported in a normal temperature chemical state or low pressure liquid state through liquefaction devices, amino carriers or LOHC systems; when the output is insufficient or the demand is high, energy feedback is achieved through gasification heating, dehydrogenation reaction, or through fuel cells or reversible solid oxide batteries RSOC, and cascade utilization with the thermal energy ladder utilization sub-module of the thermal energy storage unit is formed to match the thermal grade. The heat generated in the power generation process of gasification heating, dehydrogenation reaction, or through fuel cells or reversible solid oxide batteries RSOC is recovered in layers according to temperature grade: high temperature waste heat is preferentially sent to high-grade thermal storage tank to drive solid hydrogen storage material dehydrogenation or provide high temperature industrial heat; medium temperature waste heat enters the medium temperature phase change thermal storage unit after heat exchange; low temperature waste heat is heated by heat pump and supplied to low temperature thermal storage or district heating network. Through the aforementioned stepwise flow from high to medium to low, matching and coupling between different heat grades are achieved, constituting graded recovery and cascaded utilization of thermal energy, improving the overall energy efficiency of the system, realizing the switching between liquefaction and gasification, and achieving electro-thermal conversion through fuel cells or exothermic reactions during the utilization stage; in this embodiment, the dynamic optimal matching of the entire process of hydrogen storage, hydrogen transportation, and hydrogen release is achieved through a liquefaction device, a gasification heating device, a solid hydrogen storage reactor, and a form switching control valve.

[0162] The thermal energy storage unit absorbs byproduct heat from electrolysis and liquefaction through a heat tank, heat pump, and waste heat recovery device, recovering waste heat from the fuel cell. Internally, it constructs three temperature zones: a high-temperature zone, a medium-temperature zone, and a low-temperature zone. Based on these temperature zones, the thermal energy ladder utilization submodule uses a staged heat exchanger and valve-controlled loop to form a multi-stage energy quality descent channel from high temperature to medium temperature and from medium temperature to low temperature, achieving coupled reuse of liquefaction heat release and vaporization heat absorption. Specifically, the high-temperature zone is used for liquefaction heat release recovery and high-grade heating, the medium-temperature zone for vaporization heat absorption and district heating, and the low-temperature zone for waste heat recovery and recirculation preheating. Each temperature zone achieves orderly heat transfer between different grades through heat exchange paths.

[0163] The hydrogen form conversion and transport module enables bidirectional controllable conversion of hydrogen between solid, liquid, and gaseous states, and completes hydrogen energy transport and energy quality matching across units and time scales. Specifically, based on the system operating cycle, load demand, and thermal energy grade constraints, it automatically selects appropriate form switching paths, including hydrogen absorption or dehydrogenation reactions of solid hydrogen storage materials, vaporization and heating processes of liquid hydrogen, and compression and storage processes of gaseous hydrogen. By adjusting the form conversion method and hydrogen pressure and temperature, it achieves orderly transport of hydrogen between electrolytic hydrogen production devices, reversible solid oxide batteries (RSOCs), fuel cells, transport pipelines, and solid-liquid-gas multi-form hydrogen storage devices. In addition, the hydrogen form conversion and transport module works in conjunction with the thermal energy cascade utilization submodule of the thermal energy storage unit to recover high, medium, and low grade heat generated during the form conversion process according to temperature stratification and return it to the corresponding thermal energy storage unit, thereby ensuring the grade matching of hydrogen energy flow and thermal energy flow, improving overall energy efficiency and system regulation capability.

[0164] The conjugate thermally coupled heat exchange module CHT establishes heat input and output channels with the hydrogen energy storage unit and the thermal energy storage unit, respectively. Specifically, it consists of a bidirectional heat exchange loop composed of a liquefier, a vaporizer, a heat pump, and a thermal storage tank. The high-temperature heat released during the liquefaction stage is input to the upper layer of the thermal storage tank or the regional heating network through the high-temperature heat exchange channel. The medium-temperature heat is used for vaporization or heating, while the low-temperature heat is recovered through the heat pump or preheating loop. The heat required during the vaporization or hydrogen absorption reaction stage is supplied in layers by the thermal energy storage unit or the heat pump. Through the multi-layer rolling optimization of the energy management platform EMS, the coordinated control of medium- and long-term, short-term, and real-time is achieved: the minute level corresponds to the rapid adjustment of electric energy storage, the hour level corresponds to the dynamic adjustment of hydrogen energy form and the coordination of medium-temperature heat load, the weekly level corresponds to the cross-temperature zone energy migration and buffering of thermal energy storage, and the seasonal level corresponds to the cross-form hydrogen storage and thermal energy step balance correction.

[0165] The Energy Management System (EMS) platform is implemented based on a multi-timescale coordinated optimization framework. It is used for unified, cross-level, and cross-cycle coordinated control of electrical energy storage units, hydrogen energy storage units, thermal energy storage units, and conjugate thermally coupled heat exchange modules. This multi-timescale coordinated optimization framework specifically includes three levels: an annual level (i.e., a medium-to-long-term planning level), a short-term level, and a real-time level. The medium-to-long-term level constructs an annual optimization model with constraints on hydrogen form conversion and thermal energy grade stratification, determining the ratio of solid, liquid, and gaseous hydrogen storage capacity and the energy-quality boundaries of each temperature zone in the thermal storage system, achieving cross-seasonal energy balance and adaptive quota correction. The short-term level is based on CNN-... The LSTM and TimesNet fusion prediction model predicts renewable energy output and load changes for the next 24 hours. A rolling time-domain (Receding-Horizon) optimization strategy generates an hourly power baseline and introduces a tiered energy storage mechanism for electricity, hydrogen, and heat, along with a cascaded utilization mechanism for thermal energy, to complete intraday power allocation and energy quality degradation regulation. The real-time layer uses the Model Predictive Control (MPC) algorithm as the core scheduling strategy, combined with a Warm-start fast calculation strategy, to perform 15-minute rolling optimization, achieving rapid power correction, morphological response, and adaptive regulation of thermal quality, thus maintaining the dynamic steady-state operation of the system.

[0166] In this embodiment, the system includes a photovoltaic and wind power grid-connected generation device, an energy management platform (EMS), an electrical energy storage unit, a hydrogen energy storage unit, a thermal energy storage unit, a hydrogen form conversion and transport module, and a conjugate thermal coupling heat exchange module. The electrical energy storage unit uses lithium-ion batteries and is responsible for rapid adjustment of power fluctuations from minutes to days. The hydrogen energy storage unit consists of a PEM / SOEC electrolyzer, an RSOC fuel cell, and a solid-liquid-gas multi-form hydrogen storage device, achieving dynamic optimal matching of the entire hydrogen storage, transport, and release process through a liquefier, a vaporization heating device, and a form switching valve. The thermal energy storage unit consists of a heat tank, a heat pump, and a waste heat recovery device, internally constructing a three-level heat quality structure of high temperature, medium temperature, and low temperature. A conjugate thermal coupling heat exchange module (CHT) is installed between the hydrogen storage unit and the thermal storage unit, recovering released heat during the liquefaction stage and providing absorbed heat during the vaporization stage, realizing bidirectional flow of latent heat and multi-level energy and mass transfer, thereby achieving dual-domain coupling and dynamic balance of the electrical-hydrogen-thermal energy chain.

[0167] Example 2: A multi-timescale optimization method for hydrogen energy tri-state-thermal energy cascade synergy, comprising the following steps:

[0168] Step 1: First, determine the annual scale input parameters, establish hourly decision variables and discretize them over time, construct an annual optimization model with hydrogen form conversion constraints and thermal energy grade stratification constraints, introduce cross-seasonal inventory smoothing and form switching penalty terms as well as conjugate heat balance and thermal energy grade stratification constraints, and form a dual-domain annual optimization framework for hydrogen form synergy and thermal energy tiered utilization.

[0169] Specifically, the system adopts a “form-stage-grade coordination strategy” in the annual layer, that is, it conducts joint planning among the three hydrogen storage forms of solid, liquid and gaseous, seasonal time stages and heat grades of high temperature zone, medium temperature zone and low temperature zone; specifically, the following variables are solved through the annual optimization model: (1) the capacity ratio of solid, liquid and gaseous hydrogen storage; (2) the form switching threshold of quarter or bimonthly; (3) the heat storage capacity boundary of high temperature zone, medium temperature zone and low temperature zone; (4) the heat energy quality transfer efficiency of high temperature zone → medium temperature zone and medium temperature zone → low temperature zone;

[0170] The above four annual optimization models are jointly determined under the following constraints: cross-seasonal inventory smoothing term, liquefaction, gasification, hydrogen absorption and desorption energy consumption and lifetime cost, thermal energy grade decline conservation constraint, and conjugate heat energy exchange relationship; through solving, the optimal hydrogen storage form capacity configuration is obtained; at the same time, the upper and lower bounds of capacity in high temperature zone, medium temperature zone, and low temperature zone and the unidirectional energy and mass transfer ratio of high temperature zone → medium temperature zone → low temperature zone are determined to ensure the reasonable distribution of liquefaction heat release, gasification heat absorption and heat pump recovery heat in different temperature zones, forming a stable energy and mass ladder chain. Since the operational boundaries of electric energy storage and thermal energy storage are jointly determined by the hydrogen form capacity and thermal grade structure, the form capacity ratio, form switching threshold, energy quality boundary of the three temperature zones, and cross-temperature zone transfer efficiency given by the annual layer will be issued as constraints to the short-term layer and the real-time layer. The system establishes a quarterly or bi-monthly rolling correction mechanism, which dynamically updates the capacity ratio, switching threshold and energy quality boundary based on the form response deviation, thermal energy balance deviation and inventory deviation returned by the lower layer, thereby forming a cross-seasonal adaptive closed-loop optimization, so that the multi-form hydrogen storage and thermal energy cascade structure maintains the optimal balance between energy security and economy.

[0171] Step 1.1: Collect wind power data Photovoltaic power generation Electricity load demand Heat load demand In addition to hydrogen demand, regional electricity purchase and sale prices, regional heat recovery prices, and meteorological environmental data, combined with equipment rated parameters and operating efficiencies (electrolyzers, fuel cells, reversible solid oxide cells (RSOCs), lithium batteries, heat pumps, three-temperature zone heat storage tanks, conjugate heat exchangers, liquefiers, vaporizers, compressors, pipeline transportation efficiency, etc.), an annual input set and representative time series are established.

[0172] The data compression phase then proceeds: K-means clustering is used to extract typical daily data and compress the time series data for each hour throughout the year, ensuring consistency in seasonal distribution characteristics and fluctuations, thus providing a representative data foundation for annual optimization. In the annual input set, the energy consumption and heat transfer characteristics of liquefaction and gasification processes are considered, and conjugate heat transfer efficiency parameters are defined, including thermal storage charging efficiency. With heat release efficiency This describes the heat exchange efficiency between the exothermic liquefaction, endothermic vaporization, and hydrogen absorption / desorption reaction heat and the thermal energy storage system.

[0173] Furthermore, to reflect the cascaded utilization characteristics of thermal energy, three thermal grade ranges—high temperature, medium temperature, and low temperature—are defined within the annual input set, corresponding to the liquefaction heat release, high-grade heating, and medium- and low-temperature waste heat recovery stages, respectively; and a cross-grade energy and mass transfer efficiency parameter is introduced. , This describes the decreasing utilization relationship of thermal energy from high-grade to low-grade; among which... This refers to the energy and mass transfer efficiency from the high-temperature region to the medium-temperature region. The energy and mass transfer efficiency from the medium-temperature region to the low-temperature region;

[0174] Step 1.2: Establish The set of hourly decision variables for each moment; after the seasonal data is compressed to a representative time series, a typical quarterly set is formed, which is the representative quarterly data formed by the annual input set plus the compression of typical days. The annual layer optimization is carried out with the quarter as the rolling cycle to maintain the cross-seasonal correlation between hydrogen energy and thermal energy and the continuity of thermal quality.

[0175] Step 1.3: Construct an annual optimization model with constraints on hydrogen form conversion and thermal energy grade stratification;

[0176] Step 1.4: Establish physical and operational constraints;

[0177] Step 1.5: To maintain consistency between the annual planning and the operation of the lower layer, a rolling correction mechanism based on quarterly or bi-monthly cycles is established; specifically, the system dynamically updates the energy quota, form capacity ratio and thermal energy grade boundary parameters of the annual layer based on the operating indicators returned by the short-term layer and the real-time layer, including hydrogen form response deviation and thermal energy balance deviation.

[0178] When the actual operating data is found to deviate from the annual plan by more than a set threshold, the lightweight re-optimization process is triggered.

[0179] The lightweight re-optimization process is specifically as follows:

[0180] For hydrogen energy storage units, the ratio of solid, liquid, and gaseous hydrogen storage capacity is adjusted and the form switching threshold is updated;

[0181] For thermal energy storage units, adjust the high, medium, and low temperature thermal energy capacity boundaries and conjugate heat transfer weights;

[0182] Update the smoothing penalty coefficient and related parameters of the energy-mass conservation constraint, specifically including: inventory smoothing weight coefficient. Heat quality deviation penalty weight Weighting coefficients for the benefits of liquefaction waste heat recovery Weighting coefficients of the vaporization endothermic process on the system operating cost Weighting coefficient of CHT heat charging power term Weighting factor for the CHT heat release power term Energy and mass transfer efficiency from high-temperature region to medium-temperature region Energy and mass transfer efficiency from the intermediate temperature region to the low temperature region ;

[0183] The optimization results are distributed to the short-term layer for execution after the end of each quarter or bi-monthly period, and the annual layer operating baseline is updated to achieve cross-seasonal energy balance and multi-domain adaptive scheduling.

[0184] Step 2: Under the constraints of annual quotas and operational boundaries issued by the medium- and long-term layers, the short-term layer undertakes rolling optimization scheduling tasks with a 24-hour prediction window and a 7-day feedback cycle. Here, 24 hours represents the prediction window length for short-term rolling optimization, and 7 days represents the weekly feedback cycle of operational indicators to the medium- and long-term layers. Together, they constitute the dual-time-scale scheduling mechanism of the short-term layer. The inputs to the short-term layer are divided into two categories according to function: one category is time-series data used for the prediction model, including wind power output, photovoltaic power output, electrical load, heat load, hydrogen demand, and temperature zone thermal energy changes, etc. Historical operating data is used as input to a short-term composite time-series prediction model, which is a fusion model of CNN-LSTM and TimesNet, to generate power and load prediction sequences for the next 24 hours. Another type of input is the constraints and state inputs used for optimization, including the form capacity boundary, heat grade range, form switching threshold and penalty coefficient issued by the medium and long term layer, as well as the operating deviation and instantaneous state of each energy storage unit from the real-time layer. This type of data and prediction results are input into the 24-hour rolling optimization model to construct and solve the short-term scheduling problem. Joint forecasting is performed based on a short-term composite time-series forecasting model, and a rolling time-domain (Receding-Horizon) optimization strategy is adopted to achieve hourly dynamic solutions, executing only the first hour's scheduling to form a continuous power baseline. During operation, a tiered energy storage mechanism of electricity, hydrogen, and heat and a tiered utilization mechanism of thermal energy are introduced. Specifically, the electricity storage system achieves minute-level rapid response, the thermal energy storage performs hourly balance adjustment according to the grade requirements of high, medium, and low temperature zones, and the hydrogen storage undertakes cross-day energy migration and form compensation. Through the hydrogen solid-liquid-gas three-state coupling mechanism, a dynamic conversion relationship between solid-state stable storage, liquid-state buffering, and gaseous rapid adjustment is established. Furthermore, the conjugate thermal coupling and thermal grade tiered synergistic mechanism are integrated to achieve multi-level recovery of latent heat and tiered reuse of thermal energy during liquefaction, gasification, and hydrogen absorption and desorption reactions. When the system experiences power or inventory deviations, phase switching and thermal grade redistribution are triggered. Finally, the 7-day cycle operation indicators are fed back to the medium- and long-term layers.

[0185] Step 2.1: Establish a short-term composite time-series forecasting model. Specifically, a composite time-series forecasting structure is adopted, consisting of a fusion forecasting framework composed of a CNN-LSTM model and a TimesNet model. The short-term composite time-series forecasting model uses historical operating data to predict wind power, photovoltaic power output, and load demand for the next 24 hours and generates confidence intervals to provide a reference for model optimization. The forecast results are used as input to the subsequent 24-hour rolling optimization model to form a rolling scheduling problem.

[0186] Step 2.2: Based on the forecast results and medium-to-long-term boundaries, establish a 24-hour rolling optimization model with the goal of minimizing system operating costs. Decision variables include grid power purchase and sale, power output of each dispatchable unit; charging power, discharging power, and state change of energy storage; hydrogen electrolysis power and fuel cell power generation; bidirectional state switching between solid hydrogen, liquid hydrogen, and gaseous hydrogen; heat power distribution in high-temperature, medium-temperature, and low-temperature zones; conjugate heat energy flow rates of liquefaction heat release, gasification heat absorption, and hydrogen absorption / desorption reaction heat; heat pump input power and output heat power; and power smoothing, inventory deviation compensation, and electricity-hydrogen-heat coupling adjustment variables. By updating the forecast and constraints every hour and solving the above decision variables, only the results of the first hour are executed to form a continuous short-term dispatch baseline.

[0187] Step 2.3: Adopt a receding-horizon optimization strategy, update the forecast every hour and resolve the scheduling for the next 24 hours, and only execute the solution for the first hour of the current time to form a continuous power baseline;

[0188] The power baseline includes: electrical power, thermal power, hydrogen power, and constraint baseline;

[0189] When prediction errors or external disturbances cause baseline deviations, the system automatically triggers gas-liquid-solid state switching and heat flow redistribution in high-temperature, medium-temperature, and low-temperature zones.

[0190] Step 2.4: Construct a synergistic mechanism for stratified energy storage of electricity, hydrogen, and heat;

[0191] In short-term rolling scheduling, electrical energy storage units, thermal energy storage units, and hydrogen energy storage units are prioritized from top to bottom according to their response speed, energy storage capacity level, and duration scale, specifically "electrical energy storage → thermal energy storage → hydrogen energy storage". Based on the operable range and real-time inventory status issued by the medium- and long-term layers, when the state quantity of a certain energy storage unit reaches a preset threshold, the EMS triggers the hierarchical compensation logic, causing energy to be transferred to the next layer of energy storage units in a fixed order:

[0192] When the distance to the boundary of the electrical energy storage is less than a set threshold, the thermal energy storage is activated to participate in the balance.

[0193] When the thermal energy storage boundary distance is less than a set threshold, hydrogen energy storage is activated to participate in regulation.

[0194] The physical response constants of the electrical energy storage unit, thermal energy storage unit, and hydrogen energy storage unit correspond to different time scales, specifically:

[0195] Electrical energy storage: used for power smoothing and peak reduction at the second to minute level;

[0196] Thermal energy storage: used for energy balance on an hourly-daily scale;

[0197] Hydrogen energy storage: used for energy migration and cross-seasonal compensation on a diurnal-seasonal scale;

[0198] The thermal energy storage system is further divided into high-temperature, medium-temperature, and low-temperature zones according to their energy grade, with the following functions:

[0199] High-temperature zone: Responsible for rapid power buffering and instantaneous compensation of peak heat load;

[0200] Medium-temperature zone: responsible for intraday thermal balance and medium-cycle heat regulation;

[0201] Low-temperature zone: used for trans-day heat recovery and energy degradation storage, providing a bottom heat source or heat sink for processes such as hydrogen liquefaction / vaporization;

[0202] The decision variables in the 24-hour rolling optimization model satisfy the upper and lower bound constraints of SOC for electric energy storage, the upper limit constraint of charging and discharging power for electric energy storage, the SOC change rate constraint for electric energy storage, the upper and lower bound constraints of energy inventory in the three temperature zones for thermal energy storage, the upper limit constraint of thermal power in each temperature zone, the thermal power ramp-up rate constraint in the three temperature zones, the upper and lower bound constraints of total inventory and gaseous, liquid, and solid partial inventory for hydrogen energy storage, the upper limit constraint of hydrogen form switching power, the form switching rate constraint, and the operating constraints of the power upper limit and power change rate of each hydrogen-related device.

[0203] Step 2.5: To enhance the system's adaptability, EMS sets explicit energy migration trigger thresholds in the optimization solution: When any of the following conditions are met, the system immediately initiates electrolytic hydrogen production or hydrogen form conversion operations:

[0204] , ;

[0205] in: for The state-of-charge ratio of the energy storage unit at any given time. The upper limit buffer band parameter for triggering electric energy storage, This is the upper limit of the maximum permissible state of charge for the energy storage unit. For time t, the three-temperature zone thermal energy storage unit is in the temperature zone The available thermal energy below Temperature zone The upper limit of the maximum permissible thermal energy capacity It serves as an upper limit buffer zone for three-temperature zone thermal energy storage;

[0206] When the load rise rate exceeds the set threshold or hydrogen energy is insufficient, hydrogen energy cross-form recirculation is triggered: liquid hydrogen is vaporized, solid hydrogen is desorbed if necessary, and gaseous hydrogen enters the RSOC or fuel cell to feed back electricity.

[0207] , ;

[0208] in for Real-time renewable energy output The load gap threshold used to trigger hydrogen energy cross-mode recirculation. The lower limit buffer band parameters for hydrogen energy storage are set; two different buffer bands are used. and This creates a hysteresis zone, preventing frequent start-ups and shutdowns of the hydrogen production or reflux process.

[0209] Step 2.6: To achieve energy-mass complementarity between hydrogen energy and thermal energy, the conjugate thermally coupled heat exchange module CHT set in the short-term layer call system structure is used; in short-term rolling scheduling, the heat exchange power of CHT is used as... The optimized variables are used in the 24-hour rolling optimization model to dynamically manage the heat flow during latent heat recovery and morphological transformation.

[0210] Step 2.7: To achieve flexible coordination and dynamic coupling among multiple energy storage systems (electric, thermal, and hydrogen), a hierarchical priority soft penalty term is introduced into the objective function of the 24-hour rolling optimization model for the short-term layer: This mechanism embodies a flexible and collaborative logic of "prioritizing electricity storage, followed by thermal storage, and with hydrogen storage as a backup," ensuring that multiple energy storage units maintain dynamic balance and stable energy flow on an hourly-daily scale.

[0211] Step 2.8: The short-term layer outputs hourly power allocation curves and constraint baselines, which are then sent to the real-time layer for execution. The system summarizes operational indicators every 7 days, including energy storage utilization rate, mode switching frequency, thermal energy cascade efficiency, conjugate thermal energy utilization rate, power tracking error indicators, hydrogen energy cross-period balance performance, thermal energy storage cross-interval synergy performance, and energy storage equipment operating pressure indicators, and feeds them back to the medium- and long-term layers for quarterly or bi-monthly boundary correction and parameter re-optimization.

[0212] Step 3: The real-time layer uses the Model Predictive Control (MPC) real-time optimization control module as its core, employing a warm-start fast calculation strategy for 15-minute rolling optimization. Under the power baseline and boundary constraints issued by the short-term layer, the system coordinates and schedules energy based on the response speed and energy quality hierarchy of the electric-thermal-hydrogen energy storage: the electric energy storage unit undertakes rapid balancing at the millisecond to minute level, the thermal energy storage unit performs hourly fine-tuning of heat load and maintains thermal energy quality stratification, and the hydrogen energy storage unit achieves cross-day scale energy migration and emergency compensation through a three-phase dynamic switching mechanism; when the state quantity of electric or thermal energy storage is less than the set threshold, the system automatically triggers and recovers the liquefaction heat release through the conjugate heat exchange module CHT and injects it into the thermal energy storage system or regional heat network. When the load increases or when the system's hydrogen energy inventory falls below the minimum safe inventory threshold, the real-time layer triggers fuel cell power generation or solid-state hydrogen release, with CHT providing gasification heat absorption. The liquefaction, gasification, and hydrogen absorption / desorption processes are all regulated by the CHT bidirectional heat exchange channel; within each rolling cycle, the real-time layer... The time layer ensures the conservation of electrical, thermal, and hydrogen power and energy in all states, while satisfying the following constraints: upper and lower bounds of the State of Charge (SOC) for the electrical energy storage unit, as well as constraints on the maximum charge / discharge power and ramp rate; upper and lower bounds of the three temperature zones for the thermal energy storage unit, limits on thermal power, temperature zone switching rates, and thermal energy grade gradient constraints; storage boundaries for the solid, liquid, and gaseous states of the hydrogen energy storage unit, limits on the state switching rate, limits on liquefaction / vaporization power, and hydrogen consumption constraints; power balance constraints for the electro-thermal-hydrogen coupling; and power boundaries for the CHT conjugate heat exchange and latent heat recovery efficiency. The system maintains consistency constraints with the operating boundaries (power baseline, phase threshold, and heat grade range) issued by the short-term layer. By combining prediction and feedback, the system achieves power correction, phase response, and heat grade self-adaptation. When disturbances or deviations occur, it automatically triggers dual-domain self-adjustment: power domain, i.e., electric domain self-adjustment: maintaining electric power balance through fast charging and discharging of electric energy storage, power correction of fuel cells or electrolyzers; thermal domain self-adjustment: maintaining thermal field stability through bidirectional heat exchange in the CHT, redistribution of heat power in three temperature zones, and step adjustment of heat grade.

[0213] The real-time layer outputs 15-minute level correction values, state switching commands, and thermal power adjustment signals, which are then sent to the execution control unit. The execution control unit includes an electric energy storage power conversion controller, a battery management system (BMS), and power control modules for the electrolyzer and fuel cell; a three-temperature zone distribution valve regulator, a circulating pump speed controller, and a heat pump or heat exchanger power adjustment module for the thermal energy storage unit; and a liquefaction controller, a vaporization and pressurization flow control valve, a solid hydrogen absorption and desorption regulator, and a three-phase state switching actuator for the hydrogen energy storage unit. Simultaneously, operating indicators are periodically fed back to the short-term layer.

[0214] Step 3.1: The real-time layer uses the power baseline issued by the short-term layer. Baseline heat load And the shape boundary is used as input, with a 15-minute adjustment step size. Based on a short-term composite time-series forecasting model, it generates output and load estimates for the future time domain, providing input for real-time rolling optimization; by using the Warm-start fast calculation strategy as the initial value of the optimal solution of the previous time period, it ensures that MPC is solved within a few seconds, realizing rolling fast calculation;

[0215] Step 3.2: Hierarchical fast response and power correction mechanism;

[0216] EMS establishes a multi-time-constant fast response mechanism based on the dynamic characteristics of energy storage units:

[0217] The energy storage unit is responsible for correcting fluctuations at the second to minute level;

[0218] The thermal energy storage unit performs hourly-level thermal buffering and waste heat absorption, and achieves a tiered thermal energy utilization mechanism through temperature-zoned heat exchange: the system divides the interior of the thermal energy storage unit into three temperature zones: a high-temperature zone, a medium-temperature zone, and a low-temperature zone. Each temperature zone corresponds to a temperature setpoint. The thermal energy ladder utilizes the temperature zone, temperature set point, and inventory level of the sub-module to dynamically adjust according to the high → medium → low ladder rule; when the heat release in the high temperature zone approaches the lower limit threshold, the medium temperature zone automatically replenishes heat; when load fluctuations cause heat loss in the low temperature zone, the medium temperature zone prioritizes energy supply, realizing the orderly migration and dynamic balance of heat between different temperature layers.

[0219] The hydrogen energy storage unit introduces a rapid solid-liquid-gas phase switching mechanism to complete trans-diurnal energy transfer and flexible compensation; specifically, when the electrical energy storage unit or the thermal energy storage unit meets the trigger threshold condition... At this time, the energy management system (EMS) automatically triggers electrolysis to produce hydrogen or liquefy hydrogen for storage, realizing the energy transfer from electricity to hydrogen; the heat released during liquefaction is injected into the medium-temperature or low-temperature zone via the conjugate heat exchange module (CHT); when the load increases or the hydrogen storage is sufficient... This triggers fuel cell power generation or solid hydrogen release, i.e., hydrogen energy to electricity return, in which the heat absorption of liquid hydrogen vaporization or solid hydrogen desorption is provided by the high temperature zone or heat pump;

[0220] Step 3.3: Energy Conservation and Safety Constraints;

[0221] The real-time layer satisfies the conservation relationship of electrical, thermal, and hydrogen energy flows in each rolling cycle. At the same time, each temperature zone of the thermal storage satisfies the thermal energy reduction constraint. The phase switching equation and the energy storage conservation equation together constitute the state evolution constraint of the hydrogen storage unit, which is used to limit the phase switching frequency and morphological energy balance in the rolling optimization.

[0222] Step 3.4: Dual-domain self-adjustment mechanism and dynamic steady-state maintenance;

[0223] During the rolling solution process, the system combines short-term forecast data with local state feedback to achieve rapid power deviation correction, sensitive phase switching response, and adaptive allocation of thermal energy grade. When external disturbances or prediction errors cause deviations in the operating state, the real-time layer automatically triggers a dual-domain self-adjustment mechanism of phase and grade: the electric-hydrogen channel balances the energy gap through phase switching, and the thermal channel maintains temperature gradient stability through valve control, achieving dual protection of electric power balance and thermal field stability. This mechanism is integrated into the objective function as a feedback correction term within the MPC optimization framework, jointly correcting power deviation, phase energy difference, and temperature thermal deviation.

[0224] Step 3.5: Execution and Cross-Layer Feedback;

[0225] The real-time layer outputs 15-minute level power correction values, form switching commands, and thermal energy adjustment signals, and sends the CHT valve opening control to the execution unit in conjunction with these signals. The EMS continuously monitors the thermal potential difference ΔT and form inventory in each temperature zone, updating self-adjustment parameters and thresholds in a closed loop. The system summarizes operational indicators every 7 days, including power deviation residuals, CHT energy reuse rate, form switching frequency, and thermal energy quality utilization efficiency, and feeds them back to the short-term and medium-to-long-term layers for quarterly boundary correction and retraining of the composite time-series forecasting model, forming a quarterly boundary correction mechanism.

[0226] Step 4: After completing the medium-to-long-term and short-term layer optimizations and summarizing real-time layer operating indicators, the system enters the sensitivity and robustness analysis phase. Multi-scenario perturbation samples are generated using Latin hypercube sampling (LHS) combined with block bootstrap to maintain the temporal correlation between wind and solar power output, load, and hydrogen storage and conjugate heat exchange behavior. A representative set of operating scenarios is formed through clustering and scenario reduction. A sub-Bruker optimization model based on Wasserstein distance is established to uniformly incorporate uncertainties into the evaluation, including wind and solar fluctuations, load deviations, hydrogen phase switching energy consumption, efficiency fluctuations due to heat energy grade decline in the three temperature zones, CHT heat exchange efficiency perturbations, and heat pump coefficient of performance (COP) changes. The sub-Bruker optimization model, through synergistic constraints and sensitivity trade-offs in the morphology-thermal energy dual domains, outputs the robust operating domain and the corrected hydrogen morphology ratio and thermal storage capacity quota, providing adaptive boundaries and parameter calibration for the next round of annual optimization.

[0227] Step 4.1: Based on the quarterly boundary correction mechanism of the annual collaborative optimization model, the system enters the sensitivity assessment and robustness optimization stage. First, a multi-scenario disturbance sample set is generated using Latin hypercube sampling (LHS) combined with the block bootstrap method to maintain the temporal correlation between wind power, photovoltaic output, load, hydrogen phase switching, and thermal energy grade decline behavior. Four types of disturbance variables are introduced into the disturbance sample set: renewable power output disturbance. Load forecasting error Uncertain deviation between energy storage phase switching rate and energy consumption Conjugate heat exchange power fluctuation term This reflects the uncertainty of latent heat and the deviation of thermal storage response during the hydrogen liquefaction or gasification process.

[0228] Step 4.2: Cluster and reduce the scenes from the perturbation samples to obtain a representative set of scenes. By maintaining the correlation between form energy and thermal energy grade, a representative expression of the dynamic synergy of multi-energy systems can be achieved.

[0229] Step 4.3: Construct a DRO model based on Wasserstein distance and the split-bar optimization method.

[0230] ;

[0231] in Represented by empirical distribution Centered on, the distance from Wasserstein does not exceed A set of uncertain distributions; This is a vector of decision variables, containing all controllable quantities of the system within the scheduling period; In distribution Expected operating cost under the perturbation scenario; For the perturbation scenario, For the system in disturbed scenarios The multi-energy cooperative operation cost function includes:

[0232] ;

[0233] in: This is the power deviation penalty coefficient. It is a comprehensive penalty coefficient for thermal power fluctuations in the CHT module, used to suppress latent heat fluctuations and the effects of thermal grade mixing between the thermal storage and hydrogen storage. The power baseline is from the short-term layer. The weighting coefficient for the penalty of mixed-layer thermal grade;

[0234] Step 4.4: Solve the robust optimization model by reducing the weight of the scenario, and then further correct the hydrogen three-state ratio and thermal energy level boundary constraints by combining the results of sensitivity analysis.

[0235] The system outputs a robust operating domain across layers. And the revised morphology and thermal storage synergy quota: , Together, they serve as the initial boundary and parameter input for the next round of annual layer optimization, enabling closed-loop updates and full-cycle adaptive scheduling in both the morphology and thermal energy domains.

[0236] Example 2: This invention proposes a medium- to long-term optimization method for annual energy quota and morphological boundary optimization, incorporating morphological transformation constraints and thermal energy grade stratification, such as... Figure 2 As shown.

[0237] First, determine the annual input parameters, including the output of photovoltaic and wind power, load demand, and hydrogen energy demand sequences; then use K-means clustering to extract typical days to form a representative time series set for the year.

[0238] Based on this, an annual optimization model is constructed, which includes cross-seasonal inventory smoothing constraints, form switching penalty terms, and conjugate heat balance equations, to comprehensively determine the capacity ratio of solid, liquid, and gaseous hydrogen storage and the high-medium-low temperature thermal energy storage mass boundary.

[0239] Through a quarterly rolling correction mechanism, based on operational indicators fed back from short-term and real-time layers, energy quotas and boundary parameters are adaptively updated to achieve cross-seasonal energy balance and form-thermal energy synergistic optimization.

[0240] Example 3: This invention proposes a rolling optimization method for multiple energy storage systems (electricity-hydrogen-thermal) based on a fusion prediction model of CNN-LSTM and TimesNet for the short-term scheduling layer, such as... Figure 3 As shown.

[0241] (1) Short-term prediction: By using the fusion of CNN-LSTM and TimesNet structure, feature extraction and prediction of the next 24 hours are performed on historical wind and solar power output and load data, and prediction curves and confidence intervals are generated.

[0242] (2) Rolling optimization modeling: With the goal of minimizing the system operating cost, an optimization model is established that includes constraints on electrical power, thermal power, morphological switching amount and thermal grade. The rolling time domain strategy is used to solve the problem hour by hour, and only the scheduling results of the first hour are executed to form a power baseline.

[0243] (3) Layered energy storage coordination: EMS sets priorities "electric storage → thermal storage → hydrogen storage" based on response speed. Electric storage achieves minute-level adjustment, thermal storage achieves hour-level thermal balance, and hydrogen storage achieves cross-day migration.

[0244] (4) Synergistic optimization of conjugate heat and heat grade: In the processes of liquefaction, gasification and hydrogen absorption and desorption, latent heat is recovered in multiple stages and thermal energy is reused in stages through the CHT module to achieve synergistic optimization of electricity, hydrogen and heat energy. The short-term layer summarizes the operating indicators every 7 days and feeds them back to the medium and long-term layer for quarterly boundary correction and parameter re-optimization.

[0245] Example 4: This invention proposes a 15-minute rolling scheduling method based on Model Predictive Control (MPC) and a Warm-start fast computation strategy for real-time operation, such as... Figure 4 As shown.

[0246] In this embodiment, the real-time layer uses the power baseline and shape boundary issued by the short-term layer as input, and combines a lightweight prediction model to achieve power and load prediction in the future time domain. The EMS establishes a tiered rapid response mechanism based on the time constant of the energy storage unit: electrical energy storage handles second- to minute-level fluctuation correction; thermal energy storage performs hourly-level heat load regulation and maintains high-, medium-, and low-temperature grade stratification; hydrogen energy storage completes inter-day energy migration and emergency compensation through three-phase state switching. When electrical or thermal storage approaches its upper limit, electrolytic hydrogen production or liquefaction hydrogen storage is automatically triggered, with the heat released during liquefaction injected into the thermal storage system via the CHT module; when the load increases or the hydrogen storage is low, the fuel cell or solid-state hydrogen storage releases energy, and the heat absorbed during vaporization is supplied by the thermal storage, achieving a closed-loop flow of electricity, hydrogen, and heat. The system maintains power balance and thermal field stability through a dual-domain self-adjustment mechanism, outputting real-time power correction amounts, shape switching commands, and thermal power adjustment signals to the execution unit, achieving dynamic steady-state control from milliseconds to hours.

[0247] Example 5: After completing multi-timescale optimization, this invention proposes a method for biblical robustness optimization and sensitivity analysis based on Wasserstein distance, used for system robustness assessment, such as... Figure 5 As shown.

[0248] 1) Latin hypercube sampling (LHS) and block bootstrap are used to generate multi-scenario perturbation samples to maintain the temporal correlation between wind and solar power output, load, morphological switching and thermal grade characteristics.

[0249] 2) A representative perturbation set is formed through clustering and scene reduction, and a sub-Bruker optimization model is established;

[0250] The model takes into account the combined effects of power deviation, energy consumption during mode switching, and thermal efficiency degradation, and solves the robust operating domain.

[0251] Based on the sensitivity analysis results, the ratio of hydrogen's three states and the boundary of thermal storage capacity are corrected, resulting in a new robust operating domain across layers. This provides parameter calibration and boundary feedback for optimization in the following year. This method effectively enhances the system's resilience to fluctuations in renewable power output and load uncertainties, enabling adaptive operation and robust optimization of the multi-energy system in complex environments.

[0252] The various embodiments in this application are described in a progressive manner. The same or similar parts between the various embodiments can be referred to each other. Each embodiment focuses on describing the differences from other embodiments.

[0253] The scope of protection of this application is not limited to the embodiments described above. Obviously, those skilled in the art can make various modifications and variations to this disclosure without departing from the scope and spirit of this disclosure. If such modifications and variations fall within the scope of the disclosed solution and its equivalents, then the intent of this disclosure also includes these modifications and variations.

Claims

1. A multi-timescale optimization method for hydrogen energy tri-state-thermal energy gradient coordination, implemented through a multi-timescale optimization system for hydrogen energy tri-state-thermal energy gradient coordination, characterized in that... Specifically, it includes photovoltaic and wind power grid-connected power generation devices, energy management platform (EMS), electric energy storage unit, hydrogen energy storage unit, thermal energy storage unit, hydrogen form conversion and transportation module, and conjugate thermal coupling heat exchange module (CHT). The photovoltaic and wind power grid-connected power generation device uses renewable energy to generate electricity, and the generated electricity is uniformly connected to the energy management platform EMS. The energy management platform EMS coordinates and distributes the electricity to the electric energy storage unit, hydrogen energy storage unit, and thermal energy storage unit, and performs multi-scale scheduling based on the tiered energy storage combined with hydrogen form synergy and thermal energy cascade utilization. The electrical energy storage unit uses lithium batteries, and the hydrogen energy storage unit includes an electrolytic hydrogen production device, a reversible solid oxide battery (RSOC), and a solid-liquid-gas multi-form hydrogen storage device. Electricity is converted into hydrogen by the electrolytic hydrogen production device, and the hydrogen storage form is adaptively selected according to the operating cycle and energy demand. Specifically, during the storage stage, a solid, liquid, or gaseous hydrogen storage device is selected according to the operating conditions. During the transportation stage, low-pressure liquid or room-temperature chemical state transportation is carried out through a liquefaction device, an amino carrier, or a LOHC system. When the output is insufficient or the demand is high, energy feedback is achieved through gasification heating, dehydrogenation reaction, or through fuel cells or reversible solid oxide batteries (RSOC), and a cascade utilization with the thermal energy cascade utilization submodule of the thermal energy storage unit is formed with matching thermal grade. The heat generated during the power generation process through gasification heating, dehydrogenation reaction, or through fuel cells or reversible solid oxide batteries (RSOC) is recovered in layers according to temperature grade. The thermal energy storage unit absorbs by-product heat from electrolysis and liquefaction through a thermal storage tank, heat pump, and waste heat recovery device, recovers waste heat from the fuel cell, and internally constructs three temperature zones: a high-temperature zone, a medium-temperature zone, and a low-temperature zone. The thermal energy cascade utilization submodule is based on these three temperature zones, forming a multi-stage energy quality descending channel from high temperature to medium temperature and from medium temperature to low temperature through a staged heat exchanger and valve control loop. Specifically, the high-temperature zone is used for liquefaction heat recovery and high-grade heating, the medium-temperature zone is used for vaporization heat absorption and district heating, and the low-temperature zone is used for waste heat recovery and recirculation preheating. Each temperature zone achieves orderly heat transfer between different grades through heat exchange paths. The hydrogen form conversion and transportation module is used to realize the bidirectional controllable conversion of hydrogen between solid, liquid and gaseous states, and to complete hydrogen energy transportation and energy quality matching across units and time scales. Specifically, based on the system operating cycle, load demand and thermal energy grade constraints, it automatically selects the appropriate form switching path, including the hydrogen absorption or dehydrogenation reaction of solid hydrogen storage materials, the vaporization heating process of liquid hydrogen, and the compression storage and transportation process of gaseous hydrogen. By adjusting the form conversion method and hydrogen pressure and temperature, it realizes the orderly transportation of hydrogen between electrolytic hydrogen production devices, reversible solid oxide batteries (RSOCs), fuel cells, transportation pipelines, and solid-liquid-gas multi-form hydrogen storage devices. In addition, the hydrogen form conversion and transportation module works in conjunction with the thermal energy cascade utilization submodule of the thermal energy storage unit to recover the high, medium and low grade heat generated during the form conversion process to the corresponding thermal energy storage unit according to temperature stratification. The conjugate thermally coupled heat exchange module CHT establishes heat input and output channels with both the hydrogen energy storage unit and the thermal energy storage unit. Specifically, it consists of a bidirectional heat exchange loop formed by a liquefier, a vaporizer, a heat pump, and a thermal storage tank. The high-temperature heat released during the liquefaction stage is input to the upper layer of the thermal storage tank or the regional heating network through the high-temperature heat exchange channel. The medium-temperature heat is used for vaporization or heating, while the low-temperature heat is recovered through the heat pump or preheating loop. The heat required during the vaporization or hydrogen absorption reaction stage is supplied in layers by the thermal energy storage unit or the heat pump. Through the multi-layer rolling optimization of the energy management platform EMS, the coordinated control of medium- and long-term, short-term, and real-time is achieved: the minute level corresponds to the rapid adjustment of electric energy storage, the hour level corresponds to the dynamic adjustment of hydrogen energy form and the coordination of medium-temperature heat load, the weekly level corresponds to the cross-temperature zone energy migration and buffering of thermal energy storage, and the seasonal level corresponds to the cross-form hydrogen storage and thermal energy step balance correction. The Energy Management System (EMS) is based on a multi-timescale coordination and optimization framework, and is used for unified coordination and control of electric energy storage units, hydrogen energy storage units, thermal energy storage units, and conjugate thermal coupling heat exchange modules across levels and cycles.

2. The method for multi-timescale optimization of hydrogen energy across three states and thermal energy gradients according to claim 1, characterized in that, The multi-timescale coordinated optimization framework comprises three levels: an annual level (i.e., a medium-to-long-term level), a short-term level, and a real-time level. The medium-to-long-term level constructs an annual optimization model with constraints on hydrogen form conversion and thermal energy grade stratification, determining the ratio of solid, liquid, and gaseous hydrogen storage capacity and the energy-quality boundaries of each temperature zone in the thermal storage system, achieving cross-seasonal energy balance and adaptive quota correction. The short-term level, based on a CNN-LSTM and TimesNet fusion prediction model, predicts renewable energy output and load changes over the next 24 hours, generates hourly power baselines using a rolling time-domain optimization strategy, and introduces a synergistic energy storage mechanism for electricity-hydrogen-thermal stratification and a thermal energy cascade utilization mechanism to complete intraday power allocation and energy quality degradation adjustment. The real-time level uses Model Predictive Control (MPC) as the core scheduling strategy, combined with a Warm-start fast calculation strategy, to perform 15-minute rolling optimization, achieving rapid power correction, form response, and adaptive thermal grade control, maintaining the system's dynamic steady-state operation.

3. The method for multi-timescale optimization of hydrogen energy across three states and thermal energy gradients according to claim 2, characterized in that, The method specifically includes the following steps: Step 1: First, determine the annual scale input parameters, establish hourly decision variables and discretize them over time, construct an annual optimization model with hydrogen form conversion constraints and thermal energy grade stratification constraints, introduce cross-seasonal inventory smoothing and form switching penalty terms as well as conjugate heat balance and thermal energy grade stratification constraints, and form a dual-domain annual optimization framework for hydrogen form synergy and thermal energy tiered utilization. Step 2: Under the constraints of annual quotas and operational boundaries issued by the medium- and long-term layers, the short-term layer undertakes rolling optimization scheduling tasks with a 24-hour prediction window and a 7-day feedback cycle. Here, 24 hours represents the prediction window length for short-term rolling optimization, and 7 days represents the weekly feedback cycle of operational indicators to the medium- and long-term layers. Together, they constitute the dual-time-scale scheduling mechanism of the short-term layer. The inputs to the short-term layer are functionally divided into two categories: one is time-series data used for the prediction model, i.e., historical operational data. This type of input is fed into the short-term composite time-series prediction model, i.e., a fusion model of CNN-LSTM and TimesNet, to generate power and load prediction sequences for the next 24 hours. The other category is constraints and state inputs used for optimization solutions, including the form capacity boundary, heat grade range, form switching threshold, and penalty coefficient issued by the medium- and long-term layers, as well as operational deviations and instantaneous state quantities of each energy storage unit from the real-time layer. This type of data, along with the prediction results, is input into the 24-hour layer. A rolling optimization model is used to construct and solve short-term scheduling problems. Joint prediction is performed based on a short-term composite time-series prediction model, and a rolling time-domain optimization strategy is adopted to achieve hourly dynamic solutions, executing only the first hour's scheduling to form a continuous power baseline. During operation, an electric-hydrogen-thermal stratified energy storage synergy mechanism and a thermal energy cascade utilization mechanism are introduced. Specifically, electric energy storage units achieve minute-level rapid response, thermal energy storage performs hourly balance adjustment based on the grade requirements of high, medium, and low temperature zones, and hydrogen energy storage undertakes cross-day energy migration and form compensation. Through a hydrogen solid-liquid-gas three-state coupling mechanism, a dynamic conversion relationship between solid-state stable storage, liquid-state buffering, and gaseous rapid adjustment is established. Furthermore, a conjugate thermal coupling and thermal grade stratified synergy mechanism are integrated to achieve multi-level recovery of latent heat and cascade reuse of thermal energy during liquefaction, gasification, and hydrogen absorption / desorption reactions. When the system experiences power or inventory deviations, phase switching and thermal energy grade redistribution are triggered. Finally, the 7-day cycle operation indicators are fed back to the medium- and long-term layers. Step 3: The real-time layer uses the real-time optimization control module of Model Predictive Control (MPC) as its core, employing a Warm-start fast calculation strategy for 15-minute rolling optimization. Under the power baseline and boundary constraints issued by the short-term layer, the system coordinates and schedules energy based on the response speed and energy quality hierarchy of the electric-thermal-hydrogen energy storage: the electric energy storage unit undertakes rapid balancing at the millisecond to minute level, the thermal energy storage unit performs hourly heat load regulation and maintains thermal energy quality stratification, and the hydrogen energy storage unit achieves cross-day scale energy migration and emergency compensation through a three-phase dynamic switching mechanism. When the state quantity of electric or thermal energy storage is less than the set threshold, the system automatically triggers and recovers the liquefaction heat release through the conjugate thermal coupling heat exchange module CHT and injects it into the thermal energy storage unit or regional heat network. When the load increases or when the system's hydrogen energy inventory falls below the minimum safe inventory threshold, the real-time layer triggers fuel cell power generation or solid-state hydrogen release, with CHT providing gasification heat absorption. The liquefaction, gasification, and hydrogen absorption / desorption processes are all controlled by the CHT dual-phase heat exchange module. The system regulates the heat exchange channel. Within each rolling cycle, the real-time layer ensures the conservation of electrical, thermal, and hydrogen power and energy in different states, while satisfying the following constraints: upper and lower bounds of SOC, maximum charge / discharge power, and ramp rate constraints for the electrical energy storage unit; upper and lower bounds of energy in the three temperature zones, thermal power limits, temperature zone switching rates, and thermal energy grade constraints for the thermal energy storage unit; solid-liquid-gas three-state storage boundaries, state switching rates, liquefaction / gasification power limits, and hydrogen consumption constraints for the hydrogen energy storage unit; power balance constraints for the electrical-thermal-hydrogen coupling; power boundaries of the CHT conjugate heat exchange and latent heat recovery efficiency constraints; and operating boundaries issued by the short-term layer. The system combines prediction and feedback to achieve power correction, phase response, and thermal grade self-adaptation. When disturbances or deviations occur, dual-domain self-adjustment is automatically triggered: power domain, i.e., electrical domain self-adjustment: maintaining electrical power balance through fast charging and discharging of electrical energy storage and power correction of fuel cells or electrolyzers; thermal domain self-adjustment: maintaining thermal field stability through bidirectional heat exchange in the CHT, redistribution of thermal power in the three temperature zones, and step adjustment of thermal grade. The real-time layer outputs 15-minute level correction values, state switching commands, and thermal power adjustment signals, which are then sent to the execution control unit. The execution control unit includes an electric energy storage power conversion controller, a battery management system (BMS), and power control modules for the electrolyzer and fuel cell; a three-temperature zone distribution valve regulator, a circulating pump speed controller, and a heat pump or heat exchanger power adjustment module for the thermal energy storage unit; and a liquefaction controller, a vaporization and pressurization flow control valve, a solid hydrogen absorption and desorption regulator, and a three-phase state switching actuator for the hydrogen energy storage unit. Simultaneously, operating indicators are periodically fed back to the short-term layer. Step 4: After completing the mid-to-long-term and short-term layer optimizations and summarizing the real-time layer operation indicators, the system enters the sensitivity and robustness analysis stage. Multi-scenario perturbation samples are generated using Latin hypercube sampling (LHS) combined with block bootstrap to maintain the temporal correlation between wind and solar power output, load, and hydrogen storage and conjugate heat exchange behavior. A representative set of operating scenarios is formed through clustering and scenario reduction. A sub-Bruker optimization model based on Wasserstein distance is established to uniformly incorporate uncertainties into the evaluation, including wind and solar fluctuations, load deviations, hydrogen phase switching energy consumption, efficiency fluctuations in heat energy grade reduction across the three temperature zones, CHT heat exchange efficiency perturbations, and heat pump coefficient of performance (COP) changes. The sub-Bruker optimization model, through synergistic constraints and sensitivity trade-offs between the morphology and thermal energy domains, outputs the robust operating domain and the corrected hydrogen morphology ratio and thermal storage capacity quota, providing adaptive boundaries and parameter calibration for the next round of annual optimization.

4. The method for multi-timescale optimization of hydrogen energy across three states and thermal energy gradients according to claim 3, characterized in that, Step 1 is specifically as follows: The system adopts a "form-stage-grade coordination strategy" in the annual layer, that is, it conducts joint planning among the three hydrogen storage forms of solid, liquid and gaseous, seasonal time stages and heat grades of high temperature zone, medium temperature zone and low temperature zone; Specifically, the following variables are solved through the annual optimization model: (1) the capacity ratio of solid, liquid and gaseous hydrogen storage; (2) the form switching threshold of quarter or bimonthly; (3) the heat storage capacity boundary of high temperature zone, medium temperature zone and low temperature zone; (4) the heat energy quality transfer efficiency of high temperature zone → medium temperature zone and medium temperature zone → low temperature zone; The above four variables are jointly determined under the following constraints: cross-seasonal inventory smoothing term, liquefaction, gasification, hydrogen absorption and desorption energy consumption and lifetime cost, thermal energy grade decline conservation constraint, and conjugate thermal energy exchange relationship; through solving, the optimal hydrogen storage form capacity configuration is obtained; at the same time, the upper and lower bounds of capacity in high temperature zone, medium temperature zone, and low temperature zone, as well as the unidirectional energy and mass transfer ratio from high temperature zone to medium temperature zone to low temperature zone are determined. The form capacity ratio, form switching threshold, energy and mass boundary of the three temperature zones, and cross-temperature zone transfer efficiency given by the annual layer will be used as constraints and sent to the short-term layer and real-time layer; the system establishes a quarterly or bi-monthly rolling correction mechanism, dynamically updating the capacity ratio, switching threshold, and energy and mass boundary based on the form response deviation, thermal energy balance deviation, and inventory deviation returned from the lower layer, thereby forming a cross-seasonal adaptive closed-loop optimization, so that the multi-form hydrogen storage and thermal energy cascade structure maintains the optimal balance between energy security and economy.

5. The method for multi-timescale optimization of hydrogen energy across three states and thermal energy gradients according to claim 3, characterized in that, Step 1 specifically includes the following steps: Step 1.1: Collect wind power data Photovoltaic power generation Electricity load demand Heat load demand In addition to hydrogen demand, regional electricity purchase and sale prices, regional heat recovery prices, and meteorological environmental data, combined with equipment rated parameters and operating efficiency, an annual input set and representative time series are established. Then, the data compression stage is performed: K-means clustering algorithm is used to extract typical days and compress the hourly data throughout the year. Within the annual input set, the energy consumption and heat transfer characteristics of liquefaction and gasification processes are considered, and conjugate heat transfer efficiency parameters are defined, including thermal storage and charging efficiency. With heat release efficiency This describes the heat exchange efficiency between the heat of liquefaction (exothermic), the heat of vaporization (endothermic), the heat of hydrogen absorption and desorption, and the thermal energy storage unit. Within the annual input set, three heat grade ranges—high temperature, medium temperature, and low temperature—are defined, corresponding to the liquefaction heat release, high-grade heating, and medium-to-low temperature waste heat recovery stages, respectively; and a cross-grade energy and mass transfer efficiency parameter is introduced. , This describes the decreasing utilization relationship of thermal energy from high-grade to low-grade; among which... This refers to the energy and mass transfer efficiency from the high-temperature region to the medium-temperature region. The energy and mass transfer efficiency from the medium-temperature region to the low-temperature region; Step 1.2: Establish The time-series decision variable set at each moment; Specifically, this includes PEM electrolyzers. Power consumed at all times Solid oxide electrolyzer (SOEC) in Constantly consuming electrical power The electrical power output during fuel cell power generation Power consumption of reversible solid oxide batteries (RSOCs) in electrolysis mode Power generation of reversible solid oxide batteries (RSOCs) in fuel cell mode Gaseous hydrogen storage devices in Hydrogen charging power at any time Gaseous hydrogen storage devices in Hydrogen release power at any time Liquid hydrogen storage devices in Hydrogen charging power at any time Liquid hydrogen storage devices in Hydrogen release power at any time Solid-state hydrogen storage devices in Hydrogen absorption power at any time Solid-state hydrogen storage devices in Dehydrogenation power at time The exothermic power of the hydrogen liquefaction process The heat endothermic power of hydrogen gasification process Chemical endothermic power of solid hydrogen storage formation process Waste heat power of solid hydrogen storage desorption reaction ; After the seasonal data is compressed to represent the time series, a typical quarterly set is formed, and annual layer optimization is performed with quarters as the rolling cycle. Step 1.3: Construct an annual optimization model with constraints on hydrogen-containing gas form conversion and thermal energy grade stratification, where the annual optimization objective function is expressed as: ； in, For equipment operating costs, For equipment capacity investment costs, As a penalty for smoothing out cross-seasonal inventory, ; This represents the total amount of hydrogen energy within the system. This represents the average inventory level for the entire year. The inventory smoothing weighting coefficient is used. This is a penalty for form switching. , The energy consumption weighting coefficient for form switching. This refers to the energy required for hydrogen to switch between its gaseous and liquid states. The energy required for hydrogen to switch between liquid and solid states. This refers to the energy required for hydrogen to switch between its gaseous and solid states. For conjugate heat energy terms, , For time step, The weighting coefficient for the benefits of liquefaction waste heat recovery. This represents the weighting coefficient of the gasification endothermic process on the system operating cost. The weighting factor for the CHT heat charge power term. This is the weighting factor for the heat release power term in CHT. For stratification of thermal energy grade: ； in, As a penalty weight for thermal energy grade deviation, It serves as a metric for thermal energy grade in mixed-layer structures. Step 1.4: Establish physical and operational constraints, as follows: (1) Hydrogen energy conservation equation: ； in, for The total amount of hydrogen energy at any given moment. The efficiency of PEM electrolyzer in converting electrical energy into hydrogen energy. The efficiency of converting electrical energy into hydrogen energy in a solid oxide electrolyzer (SOEC). To optimize the time interval length, The efficiency of converting hydrogen energy into electrical energy in fuel cells. The efficiency of a reversible solid oxide battery (RSOC) in converting electrical energy into hydrogen energy in electrolysis mode. The efficiency of reversible solid oxide batteries (RSOCs) in converting hydrogen energy into electrical energy in fuel cell mode. For hydrogen in The inevitable loss of time; (2) Shape conservation constraints: ; in, This refers to the inventory of gaseous hydrogen. This refers to the liquid hydrogen inventory. This refers to the inventory of solid hydrogen. (3) Conjugate thermal energy conservation constraint: ； in for The total amount of thermal energy within the thermal energy storage unit at any given time. for The total amount of thermal energy within the thermal energy storage unit at any given time. for The heat loss power of the thermal energy storage unit at all times; (4) Constraints on the conservation of thermal energy quality: ， ； in, This refers to the heat power that can be transferred from the high-temperature region to the medium-temperature region. This represents the actual heat power received in the mid-temperature region. This refers to the heat power that can be transferred from the medium temperature region to the low temperature region. This represents the actual heat power received in the low-temperature region. (5) Quarterly heat balance correction factor: ， ； in, This is the heat balance correction factor for the qth quarter. This represents the net equivalent heat generated through the CHT module in the qth quarter. This is the reference heat value for the qth quarter, i.e., the baseline seasonal heat value; Step 1.5: Establish a rolling correction mechanism based on quarterly or bi-monthly cycles; specifically: the system dynamically updates the annual energy quota, form capacity ratio and thermal energy grade boundary parameters based on the operating indicators returned by the short-term layer and the real-time layer, including hydrogen form response deviation and thermal energy balance deviation. When the actual operating data is found to deviate from the annual plan by more than a set threshold, the lightweight re-optimization process is triggered. The lightweight re-optimization process is specifically as follows: For hydrogen energy storage units, the ratio of solid, liquid, and gaseous hydrogen storage capacity is adjusted and the form switching threshold is updated; For thermal energy storage units, adjust the high, medium, and low temperature thermal energy capacity boundaries and conjugate heat transfer weights; Update the smoothing penalty coefficient and related parameters of the energy-mass conservation constraint, specifically including: inventory smoothing weight coefficient. Heat quality deviation penalty weight Weighting coefficients for the benefits of liquefaction waste heat recovery Weighting coefficients of the gasification endothermic process on the system operating cost Weighting coefficient of CHT heat charging power term Weighting factor for the CHT heat exothermic power term Energy and mass transfer efficiency from high-temperature region to medium-temperature region Energy and mass transfer efficiency from the intermediate temperature region to the low temperature region ; The optimization results are distributed to the short-term layer for execution after the end of each quarter or bi-monthly period, and the annual layer operating baseline is updated to achieve cross-seasonal energy balance and multi-domain adaptive scheduling.

6. The method for multi-timescale optimization of hydrogen energy across three states and thermal energy gradients according to claim 5, characterized in that, Step 2 specifically includes the following steps: Step 2.1: Establish a short-term composite time series forecasting model. Specifically, a composite time series forecasting structure is adopted, which is a fusion forecasting framework composed of CNN-LSTM model and TimesNet model. The short-term composite time series forecasting model uses historical operating data to predict the wind power, photovoltaic power output and load demand in the next 24 hours and generates confidence intervals. The forecast results are used as input for the subsequent 24-hour rolling optimization model to form a rolling scheduling problem. Step 2.2: Based on the prediction results and medium-to-long-term boundaries, establish a 24-hour rolling optimization model with the goal of minimizing system operating costs. Decision variables include grid power purchase and sale, power output of each dispatchable unit; charging power, discharging power, and state change of energy storage; hydrogen electrolysis power and fuel cell power generation; bidirectional state switching between solid hydrogen, liquid hydrogen, and gaseous hydrogen; heat power distribution in high-temperature, medium-temperature, and low-temperature zones; conjugate heat energy flow rates of liquefaction heat release, gasification heat absorption, and hydrogen absorption / desorption reaction heat; heat pump input power and output heat power; and power smoothing, inventory deviation compensation, and electricity-hydrogen-heat coupling adjustment variables. By updating the prediction and constraints every hour and solving the above decision variables, only the results of the first hour are executed to form a continuous short-term dispatch baseline. The specific optimization objectives are as follows: ； in, To optimize the objective function, representing the overall system operating cost over the entire rolling time domain, To predict and optimize time sets, Weighting for inventory deviation penalties. for The target or benchmark inventory at any given time is issued by the mid- to long-term management level. This is the hydrogen three-state switching vector. Indicates time The equivalent energy change between each phase; The constraints include: (1) Conservation of hydrogen energy and phase transformation: ； ； ， ； (2) Electrical power and thermal balance: ； ； in: For electrical energy storage discharge power, Power for charging electric energy storage, For thermal energy storage heat release power, Heating capacity from external renewable energy sources, Power for charging thermal energy storage; (3) Power and inventory boundary constraints: ； ； ； ; ; ； ； ； ； ； ； ； ； ； ； ； in: The minimum permissible inventory for hydrogen energy storage, The maximum permissible inventory for hydrogen energy storage, This represents the maximum permissible hydrogen charging power for a gaseous hydrogen storage device. This represents the maximum permissible hydrogen release power for a gaseous hydrogen storage device. This represents the maximum permissible hydrogen charging power for a liquid hydrogen storage device. This represents the maximum permissible hydrogen release power for a liquid hydrogen storage device. This represents the maximum permissible hydrogen absorption power for a solid-state hydrogen storage device. This represents the maximum permissible dehydrogenation power for a solid-state hydrogen storage device. This represents the maximum exothermic power during the hydrogen liquefaction process. This represents the maximum endothermic power during the hydrogen vaporization process. The maximum chemical endothermic power for the solid-state hydrogen storage formation process is... Maximum waste heat power of solid-state hydrogen storage desorption reaction; Step 2.3: Adopt a rolling time-domain optimization strategy, update the prediction every hour and resolve the scheduling for the next 24 hours, and only execute the solution for the first hour to form a continuous power baseline; The power baseline includes: electrical power, thermal power, hydrogen power, and constraint baseline; When prediction errors or external disturbances cause baseline deviation, the system automatically triggers gas-liquid-solid state switching and heat flow redistribution in high-temperature, medium-temperature, and low-temperature zones. Step 2.4: Construct a synergistic mechanism for stratified energy storage of electricity, hydrogen, and heat; In short-term rolling scheduling, electrical energy storage units, thermal energy storage units, and hydrogen energy storage units are prioritized from top to bottom according to their response speed, energy storage capacity level, and duration scale, specifically "electrical energy storage → thermal energy storage → hydrogen energy storage". Based on the operable range and real-time inventory status issued by the medium- and long-term layers, when the state quantity of a certain energy storage unit reaches a preset threshold, the EMS triggers the hierarchical compensation logic, causing energy to be transferred to the next layer of energy storage units in a fixed order: When the distance to the boundary of the electrical energy storage is less than a set threshold, the thermal energy storage is activated to participate in the balance. When the thermal energy storage boundary distance is less than a set threshold, hydrogen energy storage is activated to participate in regulation. The energy changes of each energy storage unit are uniformly represented as follows: ； in: This is an index for energy storage unit types, used to distinguish different types of energy storage subsystems. Indicates an energy storage unit. Indicates thermal energy storage unit, Indicates a hydrogen energy storage unit; For energy storage units exist Energy reserves at any given moment For energy storage units exist Energy reserves at any given moment For energy storage units The charging efficiency, For energy storage units exist The charging power at any given moment For energy storage units Energy release efficiency, For energy storage units exist The power output at any given moment; The decision variables in the 24-hour rolling optimization model satisfy the upper and lower bound constraints of SOC for electric energy storage, the upper limit constraint of charging and discharging power for electric energy storage, the SOC change rate constraint for electric energy storage, the upper and lower bound constraints of energy inventory in the three temperature zones for thermal energy storage, the upper limit constraint of thermal power in each temperature zone, the thermal power ramp-up rate constraint in the three temperature zones, the upper and lower bound constraints of total inventory and gaseous, liquid, and solid partial inventory for hydrogen energy storage, the upper limit constraint of hydrogen form switching power, the form switching rate constraint, and the operating constraints of the power upper limit and power change rate of each hydrogen-related device. Step 2.5: The EMS sets explicit energy migration trigger thresholds in the optimization solution: When any of the following conditions are met, the system immediately starts the electrolytic hydrogen production or hydrogen form conversion operation: ， ； in: for The state-of-charge ratio of the energy storage unit at any given time. The upper limit buffer band parameter for triggering electrical energy storage, This is the upper limit of the maximum permissible state of charge for the energy storage unit. For time t, the three-temperature zone thermal energy storage unit is in the temperature zone The available thermal energy below Temperature zone The upper limit of the maximum permissible thermal energy capacity, It serves as an upper limit buffer zone for three-temperature zone thermal energy storage; When the load rise rate exceeds the set threshold or hydrogen energy is insufficient, hydrogen energy cross-form recirculation is triggered: liquid hydrogen is vaporized, solid hydrogen is desorbed, and gaseous hydrogen enters the RSOC or fuel cell to feed back electricity. ， ； in for Real-time renewable energy output The load gap threshold used to trigger hydrogen energy cross-mode recirculation. The parameters for triggering the lower limit buffer band of hydrogen energy storage; Step 2.6: The short-term layer calls the conjugate thermally coupled heat exchange module CHT set in the system architecture; in the short-term rolling scheduling, the heat exchange power of CHT is used as... The optimized variables are used in the 24-hour rolling optimization model to dynamically manage the heat flow during latent heat recovery and morphological transformation. The thermal power constraint of the conjugate thermally coupled heat exchange module CHT is: ； ； ； in for The overall heat transfer power of the CHT conjugate thermal coupling heat exchange module at all times. and These are the minimum and maximum heat transfer capacities of CHT, respectively. The temperature of the heat exchange interface. To ensure safety, set the temperature. The allowable temperature deviation range; Step 2.7: To achieve flexible coordination and dynamic coupling among multiple energy storage systems (electric, thermal, and hydrogen), a hierarchical priority soft penalty term is introduced into the objective function of the 24-hour rolling optimization model for the short-term layer: ； ； ； in: Indicates that the system is in Net power exchange on the electrical side at any given time. This refers to the conjugate heat power exchange between the thermal energy storage unit and the hydrogen energy storage unit. , , These are the priority penalty coefficients for electrical energy storage, thermal energy storage, and hydrogen energy storage, respectively. Step 2.8: The short-term layer outputs hourly power allocation curves and constraint baselines, which are then sent to the real-time layer for execution; The system summarizes operational indicators every 7 days, including energy storage utilization rate, mode switching frequency, thermal energy cascade efficiency, conjugate thermal energy utilization rate, power tracking error index, hydrogen energy cross-period balance performance, thermal energy storage cross-interval synergy performance, and energy storage equipment operating pressure index, and feeds them back to the medium and long term layer for quarterly or bi-monthly boundary correction and parameter re-optimization.

7. The method for multi-timescale optimization of hydrogen energy across three states and thermal energy gradients according to claim 6, characterized in that, Step 3 specifically includes the following steps: Step 3.1: The real-time layer uses the power baseline issued by the short-term layer. Baseline heat load Using the shape boundary as input and 15-minute adjustment step size, a short-term composite time series prediction model is used to generate power output and load estimates in the future time domain, providing input for real-time rolling optimization. The optimization objective is to minimize operating costs and baseline deviation. ； in: For the system in Actual power output at any given time Power baseline issued from the short-term layer; Power deviation penalty coefficient Step 3.2: Hierarchical fast response and power correction mechanism; EMS establishes a multi-time-constant fast response mechanism based on the dynamic characteristics of energy storage units: The energy storage unit is responsible for correcting fluctuations at the second to minute level; The thermal energy storage unit performs hourly-level thermal buffering and waste heat absorption, and achieves a cascaded utilization mechanism of thermal energy through temperature-zoned tiered heat exchange: the system divides the interior of the thermal energy storage unit into three temperature zones: a high-temperature zone, a medium-temperature zone, and a low-temperature zone. Each temperature zone corresponds to a temperature setpoint. The thermal energy cascade utilization submodule is dynamically adjusted according to the temperature zone, temperature set point and inventory level of the submodule, according to the high → medium → low cascade rule; when the heat release in the high temperature zone approaches the lower limit threshold, the medium temperature zone automatically replenishes heat; when load fluctuations cause heat loss in the low temperature zone, the medium temperature zone gives priority to energy supply, realizing the orderly migration and dynamic balance of heat between different temperature layers. The hydrogen energy storage unit introduces a rapid solid-liquid-gas phase switching mechanism to complete trans-diurnal energy transfer and flexible compensation; specifically, when the electrical energy storage unit or the thermal energy storage unit meets the trigger threshold condition... At this time, the Energy Management System (EMS) automatically triggers electrolysis to produce hydrogen or liquefy hydrogen for storage, realizing the energy transfer from electricity to hydrogen. The heat released during liquefaction is injected into the medium-temperature or low-temperature zone via the conjugate thermal coupling heat exchange module (CHT). When the load increases or the hydrogen storage capacity is sufficient... This triggers fuel cell power generation or solid hydrogen release, i.e., hydrogen energy to electricity return, in which the heat absorption of liquid hydrogen vaporization or solid hydrogen desorption is provided by the high temperature zone or heat pump; The conjugate heat power exchange between the thermal energy storage unit and the hydrogen energy storage unit is defined as follows: ； Step 3.3: Energy Conservation and Safety Constraints; The real-time layer satisfies the conservation relationships of electrical, thermal, and hydrogen energy flows in each rolling cycle: ； ； Meanwhile, the thermal energy reduction constraint is met in each temperature zone of the thermal storage: ， ， ； in, , These represent the heat transfer power from high temperature to medium temperature and from medium temperature to low temperature, respectively. The phase switching equation and the energy conservation equation together constitute the state evolution constraint of the hydrogen storage unit, which is used to limit the phase switching frequency and morphological energy balance in rolling optimization. The regulation process of the energy storage unit is considered as a single-step approximation of the MPC target, and the dynamic equation is: ； in: For electrical energy storage discharge power, Power for charging electric energy storage, , These represent the charging and discharging efficiencies of the energy storage unit, respectively. for State of electrical energy storage at any given time. for The state of electrical energy storage at any given moment; Step 3.4: Dual-domain self-adjustment mechanism and dynamic steady-state maintenance; During the rolling solution process, the system combines short-term prediction data with local state feedback to achieve rapid correction of power deviation, sensitive response to phase switching, and adaptive allocation of thermal energy grade. When external disturbances or prediction errors cause deviations in the operating state, the real-time layer automatically triggers a dual-domain self-adjustment mechanism of phase and grade: the electric-hydrogen channel balances the energy gap through phase switching, and the thermal channel maintains the stability of the temperature gradient through valve control, thus achieving dual protection of electric power balance and thermal field stability. Step 3.5: Execution and Cross-Layer Feedback; The real-time layer outputs 15-minute level power correction, form switching commands, and thermal energy adjustment signals, and sends the CHT valve opening control to the execution unit in conjunction with these signals. The EMS continuously monitors the thermal potential difference ΔT and form inventory in each temperature zone, and updates the self-adjustment parameters and thresholds in a closed loop. The system summarizes the operating indicators every 7 days, including power deviation residual, CHT energy reuse rate, form switching frequency, and thermal energy quality utilization efficiency, and feeds them back to the short-term and medium-to-long-term layers for quarterly boundary correction and retraining of the composite time series prediction model, thus forming a quarterly boundary correction mechanism.

8. The method for multi-timescale optimization of hydrogen energy across three states and thermal energy gradients according to claim 3, characterized in that, Step 4 specifically includes the following steps: Step 4.1: Based on the quarterly boundary correction mechanism of the annual collaborative optimization model, the system enters the sensitivity assessment and robustness optimization stage. First, a multi-scenario disturbance sample set is generated using Latin hypercube sampling (LHS) combined with the block bootstrap method to maintain the temporal correlation between wind power, photovoltaic output, load, hydrogen phase switching, and thermal energy grade decline behavior. Four types of disturbance variables are introduced into the disturbance sample set: renewable power output disturbance. Load forecasting error Uncertain deviation between energy storage phase switching rate and energy consumption Conjugate heat exchange power fluctuation term ; Step 4.2: Cluster and reduce the scenes from the perturbation samples to obtain a representative set of scenes. By maintaining the correlation between form energy and thermal energy grade, a representative expression of the dynamic synergy of multi-energy systems can be achieved; Step 4.3: Construct a DRO model based on Wasserstein distance and the split-bar optimization method. ； in Represented by empirical distribution Centered on, the distance from Wasserstein does not exceed A set of uncertain distributions; This is a vector of decision variables, containing all controllable quantities of the system within the scheduling period; In distribution Expected operating cost under the perturbation scenario; For the perturbation scenario, For the system in disturbed scenarios The multi-energy cooperative operation cost function includes: ； in: This is the power deviation penalty coefficient. This is the comprehensive penalty coefficient for thermal power fluctuations in the CHT module. The power baseline is from the short-term layer. The weighting coefficient for the penalty of mixed-layer thermal grade; Step 4.4: Solve the robust optimization model by reducing the weight of the scenario, and then further correct the hydrogen three-state ratio and thermal energy level boundary constraints by combining the results of sensitivity analysis. The system outputs a robust operating domain across layers. And the revised morphology and thermal storage synergy quota: , Together, they serve as the initial boundary and parameter input for the next round of annual layer optimization, enabling closed-loop updates and full-cycle adaptive scheduling in both the morphology and thermal energy domains.

Images