Dynamic balance regulation and control method and system for hydrogen energy preparation-storage-utilization system

Abstract

The invention discloses a hydrogen energy'preparation-storage-use 'system dynamic balance regulation and control method and system, and relates to the field of dynamic regulation and control, and the method comprises the steps: collecting full-chain data in real time through building a'preparation-storage-use' cooperative monitoring network, and constructing a historical-real-time-prediction dynamic database; predicting future productivity, load and external energy fluctuation based on the improved LSTM model; a supply-demand difference value is calculated and a dynamic balance target is set, so that a model prediction control algorithm is adopted to generate a multi-main-body cooperation strategy, hydrogen production energy input is optimized, the hydrogen storage charging and discharging rate is adjusted, and hydrogen load peak shifting is implemented; and after simulation verification and constraint verification, a disassembling strategy is issued to each unit as an execution instruction, and dynamic iterative optimization is performed through real-time feedback data comparison, so that closed-loop regulation and control of the system are realized. The system has the advantages that a full-chain cooperative monitoring network is established, the supply and demand of hydrogen energy and the output of energy are accurately predicted through the improved LSTM model, and the dynamic balance of hydrogen energy preparation-storage-use and the stable and efficient operation of the system are realized.

Description

TECHNICAL FIELD

[0001] The present application relates to the field of dynamic regulation, in particular to a hydrogen energy "production-storage-use" system dynamic balance regulation method and system. BACKGROUND

[0002] With the acceleration of global energy structure transformation to clean and low carbon, hydrogen energy has become a key strategic energy carrier due to its zero emission, high heat value and wide application scenarios. The core link of hydrogen energy industry chain - "hydrogen production" (green hydrogen technology such as water electrolysis using renewable energy), "hydrogen storage and transportation" (high-pressure gaseous, liquid, solid or pipeline transportation), "hydrogen use" (transportation, industry, power generation and other fields) - constitutes a complex and mutually coupled system.

[0003] The current hydrogen energy "production-storage-use" balance regulation method on the market is mostly limited to a single unit rather than the whole chain in terms of data coverage, and the database update cycle is long, making it difficult to synchronize with real-time system fluctuations. Moreover, there is a lack of structured integration of "historical-real-time-predictive" data, and the data support is weak. The prediction link mostly uses ordinary LSTM or random optimization models without combining attention mechanisms, which cannot capture the coupling rules of new energy output and energy load, and the prediction confidence is often below 90%, making it difficult to cope with high-frequency fluctuations of wind and light energy. The regulation strategy is mostly single subject optimization, or only focuses on hydrogen production load adjustment, without realizing the multi-subject collaboration of "energy input optimization-hydrogen storage rate regulation-hydrogen use priority scheduling", and lacking balance consideration of clean energy proportion and cost. In addition, there is a general lack of full-system simulation verification link, which is prone to equipment risks due to unanticipated constraint conflicts, and the feedback iteration mechanism lags behind, with a deviation threshold of more than 5%, and the regulation accuracy and stability are far from the dynamic iterative optimization mode. SUMMARY

[0004] In order to improve the existing method and system, a hydrogen energy "production-storage-use" system dynamic balance regulation method and system is provided. The method builds a whole-chain collaborative monitoring network and a dynamic database to accurately predict hydrogen energy supply and demand and energy output with an improved LSTM model, and then realizes dynamic balance of hydrogen energy "production-storage-use" and stable and efficient operation of the system through multi-subject collaborative regulation, simulation verification and deviation iteration.

[0005] To achieve the above purposes, the technical solution adopted by the present application is as follows:

[0006] The hydrogen energy "production-storage-use" system dynamic balance regulation method comprises:

[0007] Building a hydrogen energy "production-storage-use" collaborative monitoring network, real-time acquisition of real-time production data of hydrogen production unit, state data of hydrogen storage unit, load data of hydrogen use unit, acquisition of output prediction data and real-time supply data of external energy;

[0008] The collected data is preprocessed, and the preprocessed data is classified and stored based on time sequence association rules to construct a dynamic database containing "historical data-real-time data-predicted data";

[0009] Based on the constructed dynamic database, an improved long short-term memory network is used to construct a supply and demand prediction model, historical data is used as a training sample to iteratively train the model, and the fluctuation curve of the hydrogen production unit, the load demand curve of the hydrogen consumption unit, and the output fluctuation interval of the external energy in the future time period are predicted respectively;

[0010] Based on the prediction results, the hydrogen energy supply and demand difference in the current time and each future time slice is calculated, and a dynamic balance target is set based on the hydrogen energy supply and demand difference;

[0011] Based on the supply and demand difference and the balance target, a multi-agent collaborative control strategy is generated, the energy input power is adjusted to optimize the hydrogen production unit using model predictive control algorithm, the hydrogen charging and discharging rate is dynamically adjusted to optimize the hydrogen storage unit according to the supply and demand difference, and for the hydrogen consumption unit, the load peak shaving or peak shifting control is implemented to adjust the operation time sequence of the low-priority equipment;

[0012] The generated control strategy is input into the preset hydrogen energy system simulation model to simulate the system operation state after the strategy is executed, and to verify whether the control strategy meets the constraint conditions;

[0013] The verified control strategy is decomposed into specific execution instructions for each unit, which is sent to the unit controller, and the instruction execution feedback data of each unit is compared with the preset control target, the response deviation is calculated, and the dynamic iterative optimization is performed based on the deviation.

[0014] Preferably, the hydrogen energy "production-storage-consumption" collaborative monitoring network is built, real-time production data of the hydrogen production unit, state data of the hydrogen storage unit, load data of the hydrogen consumption unit, and output prediction data and real-time supply data of the external energy are collected, which specifically includes:

[0015] The monitoring network covers the hydrogen production unit, the hydrogen storage unit, the hydrogen consumption unit and the external energy supply;

[0016] The hydrogen production unit collects real-time production-related information, including electrolytic cell current density, raw water purity, and energy input power;

[0017] The hydrogen storage unit determines and collects state-related information, including hydrogen storage tank pressure, temperature, liquid level, hydrogen purity, and leakage rate;

[0018] The hydrogen consumption unit determines and collects load-related information, including real-time power of each hydrogen-consuming device, hydrogen demand fluctuation period, and priority level;

[0019] The external energy supply determines the collection of output prediction data and real-time supply data, including wind power, photovoltaic, power grid, and biomass energy.

[0020] The various types of data collected by each unit are summarized to the edge computing node to complete preliminary caching.

[0021] Preferably, the collected data is preprocessed, and the preprocessed data is classified and stored based on time sequence association rules to build a dynamic database containing "historical data-real-time data-prediction data", which specifically includes:

[0022] Abnormal fluctuation data is removed by Kalman filtering algorithm, and the data after removing the abnormality is normalized to eliminate the dimension difference of different dimensions.

[0023] The preprocessed data is divided into historical data, real-time data, and prediction data based on time sequence association rules, and is stored in the corresponding data module to build a dynamic database.

[0024] The database update period is not more than 10 seconds, and the new preprocessed data is received in real time and the storage content is updated, so that the database data is synchronized with the actual running state of the system.

[0025] Preferably, based on the constructed dynamic database, an improved long short-term memory network is used to build a supply and demand prediction model, the historical data is used as a training sample, the model is iteratively trained, and the production capacity fluctuation curve of the hydrogen production unit, the load demand curve of the hydrogen consumption unit, and the output fluctuation interval of the external energy in the future time period are predicted respectively. Specifically includes:

[0026] Through the improved long short-term memory network, a supply and demand prediction model is built by combining the attention mechanism, historical supply and demand data, external energy output data, and meteorological data are extracted from the dynamic database to form a model training sample set;

[0027] Based on the sample set, the model is iteratively trained, and the model time step and the number of neurons are adjusted to optimize the model's ability to capture the supply and demand change pattern;

[0028] Set 90% as the prediction result confidence threshold, test the trained model with the verification data set, if the confidence is lower than the threshold, supplement the sample and adjust the model parameters until the requirements are met;

[0029] The real-time collected system running data is input into the qualified supply and demand prediction model to predict the production capacity fluctuation curve of the hydrogen production unit, the load demand curve of the hydrogen consumption unit, and the output fluctuation interval of the external energy in the future time period.

[0030] By comparing the prediction results with the actual running data, if the deviation is large, supplement the near 12 hours real-time data to retrain the model, and dynamically optimize the prediction ability.

[0031] Preferably, the calculation of the hydrogen energy supply and demand difference in the current time and future time slices based on the prediction results comprises:

[0032] Based on the supply and demand prediction results, the predicted production capacity of the hydrogen production unit, the predicted supply amount of external energy, and the predicted load of the hydrogen consumption unit in the future time period are obtained, and the system conventional transportation loss and hydrogen storage loss data are also counted.

[0033] Based on the actual fluctuation degree of the system, an adaptive time slice interval is set to divide the future time period into multiple continuous time slices. For each time slice, the hydrogen energy supply and demand difference in each time period is calculated through the logic of hydrogen production prediction capacity + external energy supply amount - hydrogen consumption prediction load - system loss amount.

[0034] The balance indicators of the hydrogen storage unit, the hydrogen consumption unit, and the hydrogen production unit are set, the limit condition that the hydrogen production capacity fluctuation amplitude in a single control does not exceed 20% of the rated production capacity is set, and it is checked whether the balance indicators and the limit condition conflict.

[0035] Preferably, the multi-agent collaborative control strategy is generated based on the supply and demand difference and the balance target, the model predictive control algorithm is used to adjust the energy input power for optimizing the hydrogen production unit, the hydrogen charging and discharging rate is dynamically adjusted according to the supply and demand difference for optimizing the hydrogen storage unit, and for the hydrogen consumption unit, the load peak shaving or peak shifting control is implemented for low-priority equipment to adjust its operation timing, specifically comprising:

[0036] For the hydrogen production unit, the model predictive control algorithm is used to adjust the energy input power. When the supply and demand difference is positive, the proportion of high-cost energy input is reduced, the proportion of clean energy input is increased, and the electrolyzer operation load is reduced. When the supply and demand difference is negative, standby energy is called to supplement hydrogen production or the electrolyzer rated load is temporarily increased;

[0037] For the hydrogen storage unit, the hydrogen charging and discharging rate is dynamically adjusted according to the supply and demand difference. When the supply is greater than the demand, the hydrogen charging rate is increased. When the supply is less than the demand, the hydrogen is released according to the priority level of the hydrogen consumption equipment, and the hydrogen consumption of high-priority equipment is preferentially guaranteed.

[0038] For the hydrogen consumption unit, the load peak shaving or peak shifting control is implemented for low-priority equipment to adjust its operation timing.

[0039] Preferably, the generated control strategy is input into a preset hydrogen energy system simulation model to simulate the system operation state after the strategy is executed, and to verify whether the control strategy meets the constraint conditions, specifically comprising:

[0040] The generated hydrogen production, hydrogen storage, and hydrogen consumption unit control strategy is imported into a hydrogen energy system simulation model containing dynamic characteristic sub-models of electrolyzer, hydrogen storage tank, fuel cell, and hydrogen refueling machine core equipment.

[0041] Simulate the execution process of the regulation strategy under actual working conditions based on the hydrogen energy system simulation model, and restore the running state of each unit after adjusting according to the strategy;

[0042] Check whether the simulation data meets the preset constraint requirements. If there is a constraint that is not met, record the substandard data and the corresponding regulation link. By adjusting the regulation parameters, reimport the model simulation verification until all constraint conditions are met.

[0043] Preferably, the regulation strategy that passes the verification is disassembled into specific execution instructions for each unit, which are issued to each unit controller. Real-time collection of each unit's instruction execution feedback data is compared with the preset regulation target, the response deviation is calculated, and dynamic iterative optimization is performed based on the deviation, which specifically includes:

[0044] The regulation strategy that passes the simulation verification is refined into parameters that need to be adjusted for hydrogen production units, hydrogen storage units, and hydrogen use units. The corresponding instructions are transmitted to the unit controllers according to unit categories;

[0045] The edge computing node continuously collects actual running data after each unit executes the instructions. The collected feedback data is compared with the preset regulation target, and the response deviation of the feedback data and the regulation target is calculated;

[0046] If the deviation is within 5%, the current regulation state is maintained. If the deviation exceeds 5%, a secondary regulation mechanism is triggered immediately to re-optimize the instructions and issue them until the execution result meets the regulation target.

[0047] Further, a hydrogen energy "production-storage-use" system dynamic balance regulation system is proposed, which includes:

[0048] Collaborative monitoring network module: Real-time collection of hydrogen production, hydrogen storage, hydrogen use unit and external energy running data, realizing full system data coverage and edge caching;

[0049] Dynamic database module: Based on time sequence association rules, historical, real-time and predicted data are classified and stored, supporting 10-second-level updates to synchronize system status;

[0050] Supply and demand prediction model module: Improved LSTM network is used to predict energy production, load and energy output curve, and dynamically optimize model confidence to more than 90%;

[0051] Supply and demand difference calculation module: Adaptive time slicing is divided, and the hydrogen energy supply and demand difference in each period is calculated and associated with system loss data;

[0052] Multi-agent regulation strategy module: Based on model predictive control algorithm, hydrogen production energy input, hydrogen storage charging and discharging rate, and hydrogen use equipment priority scheduling are optimized;

[0053] Simulation verification module: verify the feasibility of the hydrogen energy system simulation model, adjust the parameters until all the constraints are met;

[0054] Processor: the processor is used for processing the calculation process of each formula and the construction calculation process of each model.

[0055] Compared with the prior art, the advantages of the present application are:

[0056] The whole chain dynamic coordination and precise controllability are realized, and the multi-variable and strong fluctuation characteristics of the hydrogen energy system are effectively adapted. By building a collaborative monitoring network covering hydrogen production, hydrogen storage, hydrogen use and external energy, combined with a 10-second-level dynamic database, data preprocessing is used to ensure data quality and provide real-time and comprehensive data support for regulation and control; an improved LSTM network combined with an attention mechanism is used to construct a supply and demand prediction model, which is dynamically iteratively optimized with a 90% confidence threshold, accurately predicting energy production, load and energy output fluctuations, and providing a scientific basis for balancing regulation. Through multi-agent collaborative strategy, the hydrogen production link optimizes energy input with model predictive control, the hydrogen storage link dynamically adjusts the charging and discharging rate, and the hydrogen link cuts peaks and valleys according to priority, achieving precise matching of supply and demand; at the same time, simulation verification is introduced to avoid risks in advance, combined with a 5% deviation threshold feedback iteration mechanism to ensure that the regulation and control instructions are reliable, ensuring system stability with hydrogen production fluctuation not exceeding the rated value by 20%, adapting to wind power, photovoltaic and other new energy output fluctuations, ultimately improving the overall balancing ability and operation efficiency of the system and reducing the overall cost. BRIEF DESCRIPTION OF DRAWINGS

[0057] Figure 1 The method proposed by the present application is shown in the schematic diagram;

[0058] Figure 2 The data collection schematic diagram proposed by the present application is shown in the schematic diagram;

[0059] Figure 3 The data preprocessing and database construction schematic diagram proposed by the present application is shown in the schematic diagram;

[0060] Figure 4 The supply and demand prediction model construction schematic diagram proposed by the present application is shown in the schematic diagram;

[0061] Figure 5 The dynamic balancing target setting schematic diagram proposed by the present application is shown in the schematic diagram;

[0062] Figure 6 The multi-agent collaborative regulation strategy generation schematic diagram proposed by the present application is shown in the schematic diagram;

[0063] Figure 7 The hydrogen energy system simulation model schematic diagram proposed by the present application is shown in the schematic diagram;

[0064] Figure 8 The dynamic iterative optimization schematic diagram proposed by the present application is shown in the schematic diagram. Detailed Implementation

[0065] The following description is intended to disclose the invention and enable those skilled in the art to implement it. The preferred embodiments described below are merely examples, and other obvious variations will occur to those skilled in the art.

[0066] The dynamic balance control system for hydrogen energy production-storage-use includes:

[0067] Collaborative monitoring network module: Real-time collection of operational data from hydrogen production, storage, and utilization units, as well as external energy sources, to achieve full system data coverage and edge caching;

[0068] Dynamic database module: Based on time-series association rules, it classifies and stores historical, real-time, and predictive data, and supports 10-second-level updates to synchronize system status;

[0069] Supply and demand forecasting model module: An improved LSTM network is used to predict capacity, load and energy output curves, and the model confidence level is dynamically optimized to over 90%.

[0070] Supply and demand difference calculation module: Divide into adaptive time slices, calculate the hydrogen energy supply and demand difference for each time period and correlate with system loss data;

[0071] Multi-entity regulation strategy module: Optimizes hydrogen production energy input, hydrogen storage charging and discharging rate, and priority scheduling of hydrogen consumption equipment based on model predictive control algorithm;

[0072] Simulation verification module: Verifies the feasibility of the strategy through a hydrogen energy system simulation model, and adjusts parameters until all constraints are met;

[0073] Processor: The processor is used to handle the calculation process of each formula and the construction calculation process of each model.

[0074] See Figure 1 As shown, the dynamic balance control method for hydrogen energy production-storage-use systems includes:

[0075] Step 1: Establish a collaborative monitoring network for hydrogen energy production, storage, and utilization, and collect real-time production capacity data of hydrogen production units, status data of hydrogen storage units, and load data of hydrogen utilization units, as well as obtain output forecast data and real-time supply data of external energy sources.

[0076] Step 2: Preprocess the collected data, classify and store the preprocessed data based on time-series association rules, and build a dynamic database containing "historical data - real-time data - predicted data";

[0077] Step three: based on the constructed dynamic database, a supply and demand prediction model is constructed using an improved long short-term memory network. The model is iteratively trained using historical data as training samples to predict the capacity fluctuation curve of the hydrogen production unit, the load demand curve of the hydrogen consumption unit, and the output fluctuation interval of external energy in the future time period;

[0078] Step four: based on the prediction results, the hydrogen energy supply and demand difference in the current time and future time slices is calculated, and a dynamic balance target is set based on the hydrogen energy supply and demand difference;

[0079] Step five: based on the supply and demand difference and the balance target, a multi-agent collaborative control strategy is generated. The model predictive control algorithm is used to adjust the energy input power to optimize the hydrogen production unit, the charging and discharging rate is dynamically adjusted to optimize the hydrogen storage unit according to the supply and demand difference, and for the hydrogen consumption unit, the load peak shaving or peak shifting control is implemented to adjust the operation timing of low-priority equipment;

[0080] Step six: input the generated control strategy into the preset hydrogen energy system simulation model to simulate the system operation state after the strategy is executed, and verify whether the control strategy meets the constraint conditions;

[0081] Step seven: the verified control strategy is decomposed into specific execution instructions for each unit, which are sent to the unit controllers. The instruction execution feedback data of each unit is collected in real time and compared with the preset control target to calculate the response deviation, and the dynamic iterative optimization is performed based on the deviation.

[0082] Referring to Figure 2 The hydrogen energy "production-storage-consumption" collaborative monitoring network is built to collect real-time capacity data of the hydrogen production unit, state data of the hydrogen storage unit, load data of the hydrogen consumption unit, and obtain output prediction data and real-time supply data of external energy, which specifically includes:

[0083] The monitoring network covers the hydrogen production unit, the hydrogen storage unit, the hydrogen consumption unit, and external energy supply;

[0084] The hydrogen production unit collects real-time capacity-related information, including electrolyzer current density, raw water purity, and energy input power;

[0085] The hydrogen storage unit determines and collects state-related information, including hydrogen storage tank pressure, temperature, liquid level, hydrogen purity, and leakage rate;

[0086] The hydrogen consumption unit determines and collects load-related information, including real-time power of each hydrogen-consuming device, hydrogen demand fluctuation period, and priority level;

[0087] The external energy supply determines and collects output prediction data and real-time supply data, including wind power, photovoltaic power, grid power, and biomass energy;

[0088] The various types of data collected by each unit are summarized to the edge computing node to complete preliminary caching.

[0089] Referring to Figure 3 The collected data is preprocessed, and the preprocessed data is classified and stored based on a time sequence correlation rule to construct a dynamic database containing "historical data-real-time data-predicted data", which specifically includes:

[0090] Abnormal fluctuation data is removed by Kalman filtering algorithm, and the data after removing the abnormality is normalized to eliminate the dimension difference of different dimension data.

[0091] The preprocessed data is divided into three categories of historical data, real-time data and predicted data based on a time sequence correlation rule, and is stored in the corresponding data module to construct a dynamic database.

[0092] The database update period is not more than 10 seconds, and new preprocessed data is received in real time to update the storage content, so that the database data is synchronized with the actual running state of the system.

[0093] Specifically, for the preliminarily classified data, abnormal data is identified in combination with the operation law of the hydrogen energy system: first, pay attention to data mutation, such as jump of hydrogen production capacity due to instantaneous voltage impact, and value exceeding the normal operation range of hydrogen storage pressure due to sensor failure; second, investigate data stagnation, such as abnormal state of no power fluctuation of hydrogen using equipment within 5 minutes and constant hydrogen storage liquid level data; by comparing the normal fluctuation range of the data within 10 minutes before and after, and combining the historical operation data trend of similar equipment at the same period, the abnormal data is judged and removed to avoid interference of abnormal data on subsequent analysis.

[0094] Referring to Figure 4 Based on the constructed dynamic database, an improved long short-term memory network is used to construct a supply and demand prediction model, historical data is used as a training sample to iteratively train the model, and the production capacity fluctuation curve of the hydrogen production unit, the load demand curve of the hydrogen using unit, and the output fluctuation interval of the external energy in the future time period are predicted, which specifically includes:

[0095] A supply and demand prediction model is built by an improved long short-term memory network combined with an attention mechanism, historical supply and demand data, external energy output data and meteorological data are extracted from the dynamic database to form a model training sample set;

[0096] The model is iteratively trained based on the sample set, and the model's ability to capture the supply and demand change pattern is optimized by adjusting the model time step and the number of neurons.

[0097] Set 90% as the prediction result confidence threshold, test the trained model with the verification data set, if the confidence is lower than the threshold, supplement the sample and adjust the model parameters until the requirements are met.

[0098] The real-time collected system operation data is input into a validated supply and demand forecasting model to predict the hydrogen production unit capacity fluctuation curve, the hydrogen consumption unit load demand curve, and the external energy output fluctuation range in the future time period.

[0099] By comparing the predicted results with the actual operating data, if the deviation is large, the model is retrained by supplementing it with real-time data from the past 12 hours, and the prediction capability is dynamically optimized.

[0100] Specifically, based on the basic LSTM network framework, a feature attention module is added to the input layer to assign higher attention weights to key feature data that affect supply and demand forecasting. For example, wind power output data is given 1.2-1.5 times the weight of ordinary data, and data from the peak hydrogen load period is marked as key feature samples. The weights of secondary features are appropriately reduced. Two to three neuron nodes are set in the LSTM hidden layer, and the initial number of neurons is determined according to the dimension of the input features. The output layer is divided into three prediction branches, corresponding to the prediction results of hydrogen production capacity, hydrogen load, and external energy output, respectively. At the same time, a reliability calculation module is added to output the reliability index of each prediction result.

[0101] The first round of model training begins by inputting training set data into the model sequentially, aiming to minimize the deviation between predicted and actual values. Model parameters are then gradually adjusted: if a large deviation in hydrogen production capacity prediction is found, the proportion of hydrogen production data samples under different energy input combinations is increased, while the LSTM time step is fine-tuned; if significant deviations occur in hydrogen load predictions during holidays, hydrogen load-specific data from the last three holidays are added for retraining; after each round of training, validation set data is used to test model performance. If the validation set prediction deviation decreases by less than 2% for three consecutive rounds, the model learning rate is reduced to avoid parameter oscillations; if the validation set deviation increases for five consecutive rounds, an early stopping mechanism is triggered, the current optimal model parameters are saved, and training is stopped.

[0102] See Figure 5 As shown, the hydrogen supply and demand difference is calculated based on the prediction results for the current time and each future time slice, and a dynamic balance target is set based on the hydrogen supply and demand difference. Specifically, this includes:

[0103] Based on the supply and demand forecast results, the predicted capacity of hydrogen production units, the predicted amount of external energy replenishment, and the predicted load of hydrogen consumption units are obtained for the future time period. At the same time, the system's conventional transportation loss and hydrogen storage loss data are statistically analyzed.

[0104] Based on the actual fluctuation of the system, an adaptive time slice interval is set, and the future period is divided into multiple continuous time segments. For each time slice, the hydrogen supply and demand difference of each period is calculated through the logic of hydrogen production prediction capacity + external energy supply - hydrogen usage prediction load - system loss.

[0105] Set balance indicators for hydrogen storage units, hydrogen consumption units, and hydrogen production units. Set a limit that the fluctuation range of hydrogen production capacity in a single adjustment shall not exceed 20% of the rated capacity, and check whether there is any conflict between the balance indicators and the limit.

[0106] Specifically, the formula for calculating the hydrogen supply-demand difference within a time slice is as follows:

[0107]

[0108] in, Let be the difference between hydrogen supply and demand in the t-th time slice. Let be the predicted capacity of the hydrogen production unit within the t-th time slice. Let be the predicted external energy supply within the t-th time slice. The predicted load of the hydrogen unit within the t-th time slice, Let t be the total system loss within the t-th time slice, where t is the time slice number.

[0109] Combining hydrogen storage safety and emergency needs, the target liquid level is set at 30%-80% of the total capacity. The lower limit of 30% serves as an emergency reserve line to ensure that at least 30% of the hydrogen storage capacity can be urgently called upon in the event of a sudden increase in hydrogen load or failure of hydrogen production equipment. The upper limit of 80% serves as a safety capacity line to prevent the pressure in the hydrogen storage tank from exceeding the design threshold due to excessive liquid level.

[0110] The hydrogen supply guarantee rate target is defined by a two-dimensional approach of priority and time: the hydrogen supply guarantee rate is ≥95%; the guarantee rate of high-priority equipment must reach 100%, that is, the load demand must be met at any time and no supply interruption is allowed; the guarantee rate of medium-priority equipment is ≥98%, and a load gap of ≤2% is only allowed in the event of extreme system anomalies; the guarantee rate of low-priority equipment is ≥90%, and some load fluctuations can be handled through peak-shifting regulation.

[0111] The energy utilization efficiency target calibration of the hydrogen production unit combines equipment characteristics and energy structure, setting the energy utilization efficiency of the hydrogen production unit to ≥85%. The efficiency calculation is based on the theoretical energy consumption corresponding to the actual hydrogen production, while also linking it to the clean energy ratio optimization target. The efficiency loss caused by insufficient clean energy is compensated by optimizing the way the power grid is used, so as to ensure the stability of the overall energy utilization level.

[0112] See Figure 6 As shown, a multi-entity coordinated control strategy is generated based on the supply-demand difference and the balance target. Model predictive control algorithms are used to adjust the energy input power to optimize the hydrogen production unit. The hydrogen storage unit is optimized by dynamically adjusting the hydrogen charging and discharging rates according to the supply-demand difference. For the hydrogen consumption unit, load shaving or staggered peak control is implemented for low-priority equipment to adjust its operating sequence. Specifically, this includes:

[0113] For the hydrogen production unit, a model predictive control algorithm is used to adjust the energy input power. When the supply-demand difference is positive, the proportion of high-cost energy input is reduced, the proportion of clean energy input is increased, and the operating load of the electrolyzer is reduced. When the supply-demand difference is negative, the backup energy is prioritized to supplement hydrogen production or the rated load of the electrolyzer is temporarily increased.

[0114] For hydrogen storage units, the hydrogen charging and discharging rates are dynamically adjusted according to the supply-demand difference. When the supply exceeds the demand, the hydrogen charging rate is increased. When the supply is less than the demand, hydrogen is released in stages according to the priority of the hydrogen-using equipment, and priority is given to ensuring the hydrogen supply of high-priority equipment.

[0115] For hydrogen-using units, load shaving or staggered control is implemented for low-priority equipment by issuing control commands to adjust their operating sequence.

[0116] Specifically, the energy input optimization strategy for hydrogen production units prioritizes adjusting the energy structure based on photovoltaic, wind power, biomass energy, off-peak grid electricity, and grid-normal electricity. When supply exceeds demand, the proportion of grid electricity input is reduced, and the proportion of clean energy is increased to over 80%. When supply is less than demand, if the output of clean energy is insufficient, biomass energy is used first to supplement the energy supply. Only if there is still a gap will grid electricity be used, and the proportion of grid electricity shall not exceed 40%.

[0117] When the supply of the electrolytic cell load exceeds the demand, the load of the electrolytic cell shall be reduced by 5% per cycle, with a maximum single reduction of 20%, and the load shall not be lower than 60% of the rated value; when the supply is less than the demand, the load shall be increased briefly, with a duration not exceeding 1 hour, and the equipment shall be kept in stable operation for more than 30 minutes after each increase to prevent overheating.

[0118] Hydrogen storage unit strategy for hydrogen charging and discharging rate control: When supply exceeds demand, the charging rate is set to 60%-80% of the maximum charging rate of the hydrogen storage tank, while monitoring pressure changes. If the pressure approaches 80% of the design pressure, the charging rate is reduced to below 50%. When supply is less than demand, hydrogen is released in stages according to hydrogen usage priority. High-priority equipment needs are met 100%, medium-priority needs are met above 95%, and low-priority needs are met 80%-90%, and the hydrogen storage level is not lower than 30% after hydrogen release.

[0119] Hydrogen storage tank switching management: If the hydrogen level in a single tank is close to the upper or lower limit, switch to the standby hydrogen storage tank to ensure continuous hydrogen charging and discharging. When the supply exceeds the demand, prioritize using the tank with the lower level to charge hydrogen, and when the supply is less than the demand, prioritize using the tank with the higher level to release hydrogen, so as to balance the usage frequency of each tank.

[0120] Load shaving and staggering using hydrogen unit strategies: When supply is less than demand, implement peak shaving for low-priority equipment, such as reducing the load of non-essential heating equipment by 20%-40% or staggering it to nighttime operation; when supply exceeds demand, encourage low-priority equipment to operate at full load during the day to improve hydrogen energy utilization.

[0121] Equipment start-up and shutdown coordination: Communicate with hydrogen users in advance to adjust the planned shutdown time of low-priority equipment to the period when supply is less than demand, and adjust the planned start-up time to the period when supply is greater than demand, so as to reduce supply and demand contradictions.

[0122] See Figure 7 As shown, the generated control strategy is input into a preset hydrogen energy system simulation model to simulate the system's operating state after the strategy is executed, and to verify whether the control strategy meets the constraints. Specifically, this includes:

[0123] The generated hydrogen production, storage, and utilization unit control strategies are imported into a hydrogen energy system simulation model that includes dynamic characteristic sub-models of core equipment such as electrolyzers, hydrogen storage tanks, fuel cells, and hydrogen refueling machines.

[0124] The simulation model of the hydrogen energy system simulates the execution process of the control strategy under actual working conditions, and restores the operating state of each unit after adjustment according to the strategy.

[0125] The simulation data is checked to see if it meets the preset constraints. If any constraints are not met, the non-compliant data and corresponding control steps are recorded. The control parameters are then adjusted, and the model is re-imported for simulation verification until all constraints are met.

[0126] Specifically, verify the simulation results against the preset constraints one by one: Hydrogen storage tank pressure: Check the simulated pressure values ​​for all time periods. It must be ≤ 90% of the design pressure. If the pressure reaches 18.5MPa in a certain period, it is determined that the constraint is not met; Electrolyzer temperature: Ensure that the simulated temperature is ≤ 95% of the rated operating temperature. If a simulated value of 77℃ appears, mark it as a temperature exceeding the limit; Hydrogen-using equipment voltage: Check the simulated voltage fluctuation of all hydrogen-using equipment. It must be ≤ ±5%. If it exceeds this, it is determined as a voltage anomaly; External energy supply: Verify that the simulated energy supply amount for each time period is ≤ the maximum external energy supply capacity to avoid over-consumption.

[0127] See Figure 8 As shown, the verified control strategy is broken down into specific execution instructions for each unit, which are then sent to the controller of each unit. Real-time feedback data from the execution of these instructions is collected and compared with the preset control target to calculate the response deviation. Dynamic iterative optimization is then performed based on this deviation. Specifically, this includes:

[0128] The control strategy verified by simulation is refined into parameters that need to be adjusted for hydrogen production unit, hydrogen storage unit, and hydrogen consumption unit, and the corresponding instructions are transmitted to the unit controller according to the unit category.

[0129] The edge computing nodes continuously collect the actual operating data after each unit executes the instructions, compare the collected feedback data with the preset control target, and calculate the response deviation between the feedback data and the control target.

[0130] If the deviation is within 5%, the current control status is maintained. If the deviation exceeds 5%, the secondary control mechanism is immediately triggered, the instructions are re-optimized and issued, until the execution result meets the control target.

[0131] Specifically, the core execution requirements of each unit are extracted from the verified control strategies and broken down into concrete instructions: for the hydrogen production unit, these are broken down into instructions for adjusting the energy input ratio and adjusting the electrolyzer load; for the hydrogen storage unit, these are broken down into instructions for hydrogen charging and discharging rates and instructions for switching storage tanks; for the hydrogen consumption unit, these are broken down into instructions for adjusting the load and instructions for staggered operation. At the same time, the instruction format is adapted according to the execution characteristics of each unit to ensure that the instructions for the hydrogen production unit match its energy scheduling logic, the instructions for the hydrogen storage unit meet the pressure-level coordination requirements, and the instructions for the hydrogen consumption unit are connected to the equipment operation sequence, so as to avoid conflicts between instructions and unit operation logic.

[0132] It should be noted that the order of the above embodiments of the present invention is merely for descriptive purposes and does not represent the superiority or inferiority of the embodiments. Furthermore, the above description focuses on specific embodiments of this specification. Additionally, the processes depicted in the accompanying drawings do not necessarily require a specific or sequential order to achieve the desired results. In some embodiments, multitasking and parallel processing are possible or may be advantageous.

[0133] The various embodiments in this specification 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.

[0134] The above description is only a preferred embodiment of the present invention and is not intended to limit the present invention. Any modifications, equivalent substitutions, improvements, etc., made within the principles of the present invention should be included within the protection scope of the present invention.

Claims

1. A dynamic balance control method for a hydrogen energy production-storage-use system, characterized in that, include: Establish a collaborative monitoring network for hydrogen energy production, storage, and utilization to collect real-time production capacity data of hydrogen production units, status data of hydrogen storage units, and load data of hydrogen utilization units, and obtain output forecast data and real-time supply data of external energy sources. The collected data is preprocessed, and the preprocessed data is classified and stored based on time-series association rules to build a dynamic database containing "historical data, real-time data and predicted data". Based on the constructed dynamic database, an improved long short-term memory network is used to build a supply and demand forecasting model. Historical data is used as training samples to iteratively train the model and predict the capacity fluctuation curve of hydrogen production units, the load demand curve of hydrogen consumption units, and the output fluctuation range of external energy in the future time period. The hydrogen supply and demand difference is calculated based on the prediction results for the current time and each future time slice, and a dynamic balance target is set based on the hydrogen supply and demand difference. Based on the supply-demand difference and balance target, a multi-entity collaborative regulation strategy is generated. Model predictive control algorithm is used to adjust the energy input power to optimize the hydrogen production unit. The hydrogen storage unit is optimized by dynamically adjusting the hydrogen charging and discharging rate according to the supply-demand difference. For the hydrogen consumption unit, load peak shaving or staggered peak regulation is implemented to adjust the operation sequence of low-priority equipment. The generated control strategy is input into the preset hydrogen energy system simulation model to simulate the system operation state after the strategy is executed, and to verify whether the control strategy meets the constraints. The verified control strategy is broken down into specific execution instructions for each unit and sent to each unit controller. The instruction execution feedback data of each unit is collected in real time and compared with the preset control target to calculate the response deviation. Based on the deviation, dynamic iterative optimization is performed.

2. The dynamic balance control method for a hydrogen energy "production-storage-use" system according to claim 1, characterized in that, The aforementioned establishment of a hydrogen energy "production-storage-use" collaborative monitoring network, which collects real-time production capacity data of hydrogen production units, status data of hydrogen storage units, and load data of hydrogen use units, and obtains external energy output forecast data and real-time supply data, specifically includes: The monitoring network covers hydrogen production units, hydrogen storage units, hydrogen consumption units, and external energy supply. The hydrogen production unit collects real-time production capacity-related information, including electrolyzer current density, feed water purity, and energy input power. The hydrogen storage unit determines and collects status-related information, including hydrogen storage tank pressure, temperature, liquid level, hydrogen purity, and leakage rate. The hydrogen-using unit determines and collects load-related information, including the real-time power of each hydrogen-using device, the hydrogen demand fluctuation cycle, and the priority level. The external energy supply is determined by collecting power output prediction data and real-time supply data, including wind power, photovoltaic power, power grid, and biomass energy; The various types of data collected by each unit are aggregated to the edge computing node to complete the initial caching.

3. The dynamic balance control method for a hydrogen energy "production-storage-use" system according to claim 1, characterized in that, The process of preprocessing the collected data, classifying and storing the preprocessed data based on time-series association rules, and constructing a dynamic database containing "historical data - real-time data - predicted data" specifically includes: The Kalman filter algorithm is used to remove abnormal fluctuation data, and the data after removing the abnormalities is normalized to eliminate the difference in the dimensions of data in different dimensions. Based on time-series association rules, the preprocessed data is divided into three categories: historical data, real-time data, and predicted data, and stored in the corresponding data modules to build a dynamic database. The database update cycle does not exceed 10 seconds, and it receives new preprocessed data in real time and updates the stored content to synchronize the database data with the actual operating status of the system.

4. The dynamic balance control method for a hydrogen energy "production-storage-use" system according to claim 1, characterized in that, The aforementioned dynamic database, based on an improved long short-term memory network, constructs a supply and demand forecasting model. Using historical data as training samples, the model is iteratively trained to predict the capacity fluctuation curve of hydrogen production units, the load demand curve of hydrogen consumption units, and the output fluctuation range of external energy sources within future time periods. Specifically, this includes: By using an improved long short-term memory network and combining it with an attention mechanism, a supply and demand forecasting model is built. Historical supply and demand data, external energy output data, and meteorological data are extracted from a dynamic database to form a model training sample set. The model is iteratively trained based on a sample set. By adjusting the model's time step and the number of neurons, the model's ability to capture the changing patterns of supply and demand is optimized. Set 90% as the confidence threshold for the prediction results, test the trained model with a validation dataset, and if the confidence is lower than the threshold, add samples and adjust the model parameters until the requirements are met; The real-time collected system operation data is input into a validated supply and demand forecasting model to predict the hydrogen production unit capacity fluctuation curve, the hydrogen consumption unit load demand curve, and the external energy output fluctuation range in the future time period. By comparing the predicted results with the actual operating data, if the deviation is large, the model is retrained by supplementing it with real-time data from the past 12 hours, and the prediction capability is dynamically optimized.

5. The dynamic balance control method for a hydrogen energy "production-storage-use" system according to claim 1, characterized in that, The calculation of the hydrogen supply-demand difference based on the prediction results for the current moment and each future time slice, and the setting of a dynamic balance target based on the hydrogen supply-demand difference, specifically includes: Based on the supply and demand forecast results, the predicted capacity of hydrogen production units, the predicted amount of external energy replenishment, and the predicted load of hydrogen consumption units are obtained for the future time period. At the same time, the system's conventional transportation loss and hydrogen storage loss data are statistically analyzed. Based on the actual fluctuation of the system, an adaptive time slice interval is set, and the future period is divided into multiple continuous time segments. For each time slice, the hydrogen supply and demand difference of each period is calculated through the logic of hydrogen production prediction capacity + external energy supply - hydrogen usage prediction load - system loss. Set balance indicators for hydrogen storage units, hydrogen consumption units, and hydrogen production units. Set a limit that the fluctuation range of hydrogen production capacity in a single adjustment shall not exceed 20% of the rated capacity, and check whether there is any conflict between the balance indicators and the limit.

6. The dynamic balance control method for a hydrogen energy "production-storage-use" system according to claim 1, characterized in that, The multi-entity coordinated control strategy, based on the supply-demand difference and balance target, employs model predictive control algorithms to adjust energy input power for hydrogen production units, dynamically adjusts hydrogen charging and discharging rates to optimize hydrogen storage units based on the supply-demand difference, and for hydrogen consumption units, implements load shaving or staggered control to adjust the operating sequence of low-priority equipment. Specifically, this includes: For the hydrogen production unit, a model predictive control algorithm is used to adjust the energy input power. When the supply-demand difference is positive, the proportion of high-cost energy input is reduced, the proportion of clean energy input is increased, and the operating load of the electrolyzer is reduced. When the supply-demand difference is negative, the backup energy is prioritized to supplement hydrogen production or the rated load of the electrolyzer is temporarily increased. For hydrogen storage units, the hydrogen charging and discharging rates are dynamically adjusted according to the supply-demand difference. When the supply exceeds the demand, the hydrogen charging rate is increased. When the supply is less than the demand, hydrogen is released in stages according to the priority of the hydrogen-using equipment, and priority is given to ensuring the hydrogen supply of high-priority equipment. For hydrogen-using units, load shaving or staggered control is implemented for low-priority equipment by issuing control commands to adjust their operating sequence.

7. The dynamic balance control method for a hydrogen energy "production-storage-use" system according to claim 1, characterized in that, The step of inputting the generated control strategy into a preset hydrogen energy system simulation model, simulating the system operating state after the strategy is executed, and verifying whether the control strategy meets the constraints specifically includes: The generated hydrogen production, storage, and utilization unit control strategies are imported into a hydrogen energy system simulation model that includes dynamic characteristic sub-models of core equipment such as electrolyzers, hydrogen storage tanks, fuel cells, and hydrogen refueling machines. The simulation model of the hydrogen energy system simulates the execution process of the control strategy under actual working conditions, and restores the operating state of each unit after adjustment according to the strategy. The simulation data is checked to see if it meets the preset constraints. If any constraints are not met, the non-compliant data and corresponding control steps are recorded. The control parameters are then adjusted, and the model is re-imported for simulation verification until all constraints are met.

8. The dynamic balance control method for a hydrogen energy "production-storage-use" system according to claim 1, characterized in that, The process of breaking down the verified control strategy into specific execution instructions for each unit, sending them to the controller of each unit, collecting the instruction execution feedback data of each unit in real time, comparing it with the preset control target, calculating the response deviation, and performing dynamic iterative optimization based on the deviation specifically includes: The control strategy verified by simulation is refined into parameters that need to be adjusted for hydrogen production unit, hydrogen storage unit, and hydrogen consumption unit, and the corresponding instructions are transmitted to the unit controller according to the unit category. The edge computing nodes continuously collect the actual operating data after each unit executes the instructions, compare the collected feedback data with the preset control target, and calculate the response deviation between the feedback data and the control target. If the deviation is within 5%, the current control status is maintained. If the deviation exceeds 5%, the secondary control mechanism is immediately triggered, the instructions are re-optimized and issued, until the execution result meets the control target.

9. A dynamic balance control system for a hydrogen energy production-storage-use system, used to implement the dynamic balance control method for a hydrogen energy production-storage-use system as described in any one of claims 1-8, characterized in that, include: Collaborative monitoring network module: Real-time collection of operational data from hydrogen production, storage, and utilization units, as well as external energy sources, to achieve full system data coverage and edge caching; Dynamic database module: Based on time-series association rules, it classifies and stores historical, real-time, and predictive data, and supports 10-second-level updates to synchronize system status; Supply and demand forecasting model module: An improved LSTM network is used to predict capacity, load and energy output curves, and the model confidence level is dynamically optimized to over 90%. Supply and demand difference calculation module: Divide into adaptive time slices, calculate the hydrogen energy supply and demand difference for each time period and correlate with system loss data; Multi-entity regulation strategy module: Optimizes hydrogen production energy input, hydrogen storage charging and discharging rate, and priority scheduling of hydrogen consumption equipment based on model predictive control algorithm; Simulation verification module: Verifies the feasibility of the strategy through a hydrogen energy system simulation model, and adjusts parameters until all constraints are met; Processor: The processor is used to handle the calculation process of each formula and the construction calculation process of each model.

Images