Hydrogen production system optimization control method and device considering purification constraints

Abstract

The application provides a hydrogen production system optimization control method and device considering purification constraints. The application designs a hydrogen production system optimization control method considering purification constraints. The renewable energy output is accurately predicted through a prediction model. A mixed integer programming model considering purification cycle constraints is established according to the renewable energy output prediction value and the hydrogen demand. In the model, the cycle of the purification system is introduced as a hard constraint through binary state variables and time sequence constraint conditions, so that the electrolytic cell can be stably operated in the heating cycle, the hydrogen production efficiency is improved, and the negative influence caused by power interruption is avoided.

Description

Technical Field

[0001] This invention relates to the field of hydrogen production technology, and more specifically to a method and apparatus for optimizing the control of a hydrogen production system that takes into account purification constraints. Background Technology

[0002] As the global energy structure accelerates its transition to a low-carbon model, hydrogen energy, as a key clean energy carrier for achieving carbon neutrality, is increasingly demonstrating its strategic importance. Hydrogen production technology based on water electrolysis from renewable energy sources such as wind and solar power, due to its zero-carbon emission characteristics, has become the core pathway for large-scale green hydrogen production. Achieving efficient hydrogen production from renewable energy sources has become a research hotspot and challenge in the energy field. However, the inherent intermittency and volatility of renewable energy sources lead to frequent fluctuations in the input power of hydrogen production systems, affecting not only electrolysis efficiency but also posing a severe challenge to the system's operational stability and economic viability. In the hydrogen production process chain, hydrogen purification is an indispensable key step, requiring the removal of impurity gases through a purification system. However, the operation of the purification system is subject to strict periodic constraints, further increasing the complexity of system control.

[0003] In existing technologies, most studies simplify the purification system as an independent processing unit, lacking systematic modeling of its periodic operating characteristics and failing to incorporate its operational constraints into a global optimization framework. This results in a significant disconnect between the control scheme and actual process requirements. This fragmented control approach may cause two problems: First, when renewable energy output is insufficient, forcing the electrolyzer to shut down during the purification heating phase not only violates process specifications and accelerates equipment aging but also significantly increases maintenance costs. Second, due to the mismatch between the purification sequence and hydrogen production fluctuations, the purity of the product hydrogen may not meet standards, seriously affecting the quality of hydrogen energy.

[0004] In existing hydrogen production system control technologies, the critical process constraint of the periodic operation characteristics of the purification system is generally overlooked during system modeling. This modeling deficiency leads to serious operational problems: when renewable energy generation is insufficient, the system may force the electrolyzer to shut down during the heating phase of the purification system. This not only violates the basic operating specifications of the purification process but also triggers multiple negative effects: 1) Unplanned shutdowns of the purification system increase system maintenance costs; 2) Frequent start-up and shutdown operations of the electrolyzer significantly accelerate the irreversible aging of electrode materials; 3) Process interruptions may lead to excessive levels of impurities such as oxygen in the hydrogen. Furthermore, existing methods do not adequately consider the dynamic coupling characteristics of electrolyzer operating constraints, particularly the interaction between ramp rate limits and power balance constraints, which may make scheduling schemes difficult to implement in actual operation. Summary of the Invention

[0005] The present invention aims to at least partially solve one of the technical problems in the related art.

[0006] Therefore, the first objective of this invention is to propose an optimized control method for a hydrogen production system that takes into account purification constraints.

[0007] The second objective of this invention is to provide an optimized control device for a hydrogen production system that takes into account purification constraints.

[0008] To achieve the above objectives, a first aspect of the present invention proposes an optimized control method for a hydrogen production system considering purification constraints, comprising:

[0009] S1, based on the historical output data of renewable energy generator sets, uses a long short-term memory neural network to build a prediction model to predict the output power of renewable energy in future dispatch cycles;

[0010] S2, based on real-time monitoring data from the hydrogen storage tank, calculates the hydrogen demand during the future scheduling cycle and converts it into hydrogen production demand under standard conditions;

[0011] S3. Construct a mixed integer linear programming model with the goal of minimizing the total cost of hydrogen production. Embed the periodic operation characteristics of the purification system as a hard constraint in the model. The purification system cycle includes a heating sub-cycle and a cooling sub-cycle. During the heating sub-cycle, the electrolyzer must run continuously, while during the cooling sub-cycle, the electrolyzer can be shut down.

[0012] S4, based on a mixed-integer linear programming model, introduces the electrolyzer operating power, grid power purchase power, and purification system state variables as decision variables, and adds power balance constraints, electrolyzer ramp rate constraints, and hydrogen production constraints to perform optimization and solve the problem, generating a control scheme for the electrolyzer operating power and purification system state.

[0013] In one embodiment of the present invention, S1 includes:

[0014] S11, normalize the historical power output data, and use the Min-Max standardization method to linearly transform the original power data to the [0,1] interval;

[0015] S12, construct a time series prediction network containing two LSTM layers, use a linear activation function for the output layer, and use a 5-fold cross-validation strategy for model training. Use the Adam optimizer with mean squared error (MSE) as the loss function to optimize the parameters until the MSE on the test set drops below 0.02.

[0016] In one embodiment of the present invention, S2 includes:

[0017] S21, reads temperature and pressure sensor data from the hydrogen storage tank in real time via the Modbus protocol, with sensor accuracy not less than 0.5%;

[0018] S22. Based on the ideal gas law, the real-time monitoring data is converted to standard conditions. Using 0°C and 1 atmosphere as standard conditions, the current hydrogen storage ratio is calculated, and the hydrogen production demand in the future scheduling cycle is determined in combination with the designed hydrogen storage capacity of the hydrogen storage tank.

[0019] In one embodiment of the present invention, S3 includes:

[0020] S31, The specific cycle constraints of the purification system include: the purification system has a large cycle of 2N hours, and each large cycle is divided into a heating sub-cycle and a cooling sub-cycle, each of N hours;

[0021] S32, the electrolytic cell must run continuously during the heating sub-cycle and must not be stopped. During the cooling sub-cycle, the electrolytic cell can be stopped, and the cycle interval Gk is the interval between the (k-1)th cycle and the kth cycle.

[0022] In one embodiment of the present invention, it further includes:

[0023] S5. Based on grid electricity price data, when there is a demand for hydrogen production, the output power of renewable energy is prioritized during the cooling sub-cycle. When renewable energy cannot meet the demand for hydrogen production, the power purchased from the grid is considered. This is to reduce the proportion of high-priced grid electricity used in the heating sub-cycle while meeting the demand for hydrogen production, thereby minimizing the total cost of hydrogen production.

[0024] To achieve the above objectives, a second aspect of the present invention provides an optimized control device for a hydrogen production system considering purification constraints, comprising:

[0025] The renewable energy output prediction module uses a long short-term memory neural network to build a prediction model based on historical output data of renewable energy generators to predict the output power of renewable energy in future dispatch cycles.

[0026] The hydrogen demand calculation module calculates the hydrogen demand in the future scheduling cycle based on real-time monitoring data of the hydrogen storage tank, and converts it into hydrogen production demand under standard conditions.

[0027] The model building module is optimized to construct a mixed integer linear programming model with the goal of minimizing the total cost of hydrogen production. The periodic operation characteristics of the purification system are embedded into the model as hard constraints. The purification system cycle includes a heating sub-cycle and a cooling sub-cycle. During the heating sub-cycle, the electrolyzer must run continuously, while during the cooling sub-cycle, the electrolyzer can be shut down.

[0028] The control scheme generation module, based on a mixed-integer linear programming model, introduces the electrolyzer operating power, grid power purchase, and purification system state variables as decision variables, and adds power balance constraints, electrolyzer ramp rate constraints, and hydrogen production constraints to perform optimization and solve the problem, generating a control scheme for the electrolyzer operating power and purification system state.

[0029] The beneficial effects of the technical solution of this invention are as follows:

[0030] 1) This invention integrates renewable energy output forecasting, hydrogen storage demand analysis, and periodic operation constraints of the purification system into an optimization model to achieve collaborative optimization and intelligent hydrogen production system control strategy.

[0031] 2) This invention takes the minimum total cost of hydrogen production within the control cycle as the objective function, and optimizes the start-up and shutdown strategy of the electrolyzer by combining the purification cycle constraint, so as to avoid using high-priced grid electricity during the heating cycle and reduce operating costs.

[0032] 3) This invention optimizes the model to coordinate the purification system cycle with the fluctuation of renewable energy, concentrates the start-up and shutdown of the electrolyzer in the cooling cycle, and reduces the mechanical stress and electrochemical loss of the equipment; at the same time, the ramp constraint limits the rate of change of the electrolyzer power, avoiding damage to the electrode material caused by sudden rises and falls in current.

[0033] Additional aspects and advantages of the invention will be set forth in part in the description which follows, and in part will be obvious from the description, or may be learned by practice of the invention. Attached Figure Description

[0034] The above and / or additional aspects and advantages of the present invention will become apparent and readily understood from the following description of the embodiments taken in conjunction with the accompanying drawings, wherein:

[0035] Figure 1 This is a flowchart of an optimized control method for a hydrogen production system considering purification constraints according to an embodiment of the present invention.

[0036] Figure 2 This is an architecture diagram of an optimized control method for a hydrogen production system considering purification constraints according to an embodiment of the present invention;

[0037] Figure 3 This is a framework diagram of an optimized control system for a hydrogen production system considering purification constraints according to an embodiment of the present invention.

[0038] Figure 4 This is a rendering of an optimized control method for a hydrogen production system considering purification constraints according to an embodiment of the present invention.

[0039] Figure 5 This is a structural diagram of an optimized control device for a hydrogen production system considering purification constraints according to an embodiment of the present invention. Detailed Implementation

[0040] It should be noted that, unless otherwise specified, the embodiments and features described in the present invention can be combined with each other. The present invention will now be described in detail with reference to the accompanying drawings and embodiments.

[0041] To enable those skilled in the art to better understand the present invention, the technical solutions of the present invention will be clearly and completely described below with reference to the accompanying drawings of the embodiments of the present invention. Obviously, the described embodiments are only some embodiments of the present invention, and not all embodiments. Based on the embodiments of the present invention, all other embodiments obtained by those skilled in the art without creative effort should fall within the scope of protection of the present invention.

[0042] The following description, with reference to the accompanying drawings, describes an optimized control method and apparatus for a hydrogen production system considering purification constraints, according to an embodiment of the present invention.

[0043] Figure 1 This is a flowchart of an optimized control method for a hydrogen production system considering purification constraints according to an embodiment of the present invention, as shown below. Figure 1 As shown, it includes:

[0044] S1, based on the historical output data of renewable energy generator sets, uses a long short-term memory neural network to build a prediction model to predict the output power of renewable energy in future dispatch cycles;

[0045] S2, based on real-time monitoring data from the hydrogen storage tank, calculates the hydrogen demand during the future scheduling cycle and converts it into hydrogen production demand under standard conditions;

[0046] S3. Construct a mixed integer linear programming model with the goal of minimizing the total cost of hydrogen production. Embed the periodic operation characteristics of the purification system as a hard constraint in the model. The purification system cycle includes a heating sub-cycle and a cooling sub-cycle. During the heating sub-cycle, the electrolyzer must run continuously, while during the cooling sub-cycle, the electrolyzer can be shut down.

[0047] S4, based on a mixed-integer linear programming model, introduces the electrolyzer operating power, grid power purchase power, and purification system state variables as decision variables, and adds power balance constraints, electrolyzer ramp rate constraints, and hydrogen production constraints to perform optimization and solve the problem, generating a control scheme for the electrolyzer operating power and purification system state.

[0048] This invention discloses an optimized control method for a hydrogen production system that considers purification constraints. This method first accurately predicts the output of renewable energy through a prediction model, and then establishes a mixed integer programming model that considers purification cycle constraints based on the predicted output of renewable energy and the hydrogen demand. In the model, the cycle of the purification system is introduced as a hard constraint through binary state variables and time-series constraints, thereby ensuring that the electrolyzer can operate stably during the heating cycle, improving hydrogen production efficiency, and avoiding negative impacts caused by power interruption.

[0049] The following describes in detail, with reference to the accompanying drawings, an embodiment of the present invention for optimizing the control of a hydrogen production system considering purification constraints.

[0050] This invention provides an optimized control method for a hydrogen production system considering purification constraints, such as... Figure 2 and Figure 3 As shown, this method achieves efficient and stable operation of the hydrogen production system through collaborative optimization. The specific scheme is as follows:

[0051] S10: Renewable Energy Output Forecast.

[0052] In one embodiment of the present invention, the historical power output data is normalized by using the Min-Max normalization method to linearly transform the original power data to the [0,1] interval; a time series prediction network containing two LSTM layers is constructed, the output layer adopts a linear activation function, and a 5-fold cross-validation strategy is used for model training. The Adam optimizer is used to optimize the parameters with the mean squared error (MSE) as the loss function until the MSE of the test set drops below 0.02.

[0053] Specifically, historical output data of renewable energy generator sets are collected. Since the data is collected at intervals, it conforms to the characteristics of time series, and a long short-term memory neural network (LSTM) is used for prediction.

[0054] First, the collected historical data is normalized to eliminate the influence of units. Then, a prediction model is constructed using a deep learning framework, building an LSTM time series prediction network with a linear activation function in the output layer. The dataset is divided into training and testing sets, and the model is trained using cross-validation. The parameters are optimized using the Adam optimizer with mean squared error (MSE) as the loss function. The weights are iteratively updated using the backpropagation algorithm until the model reaches the expected accuracy. This model is the final model. The historical data is then input into the final model to obtain the renewable energy output prediction results.

[0055] S20: Hydrogen production demand calculation.

[0056] In one embodiment of the present invention, temperature and pressure sensor data of the hydrogen storage tank are read in real time via the Modbus protocol, with the sensor accuracy level not less than 0.5%; the real-time monitoring data is converted to standard conditions based on the ideal gas law, and the current hydrogen storage ratio is calculated using 0°C and 1 atmosphere as standard conditions, and the hydrogen production demand in the future scheduling cycle is determined in combination with the designed hydrogen storage capacity of the hydrogen storage tank.

[0057] Specifically, real-time monitoring data of hydrogen storage tanks is obtained to calculate hydrogen production demand.

[0058]

[0059] Where: h need This represents the hydrogen demand during the future scheduling cycle; h max The hydrogen storage capacity is designed for the hydrogen storage tank; h now This represents the current hydrogen storage ratio (0-1 per unit). It is important to note that all parameters must be converted to values ​​under standard conditions (0°C, 1 atmosphere) before processing.

[0060] S30: Optimize model construction.

[0061] The purification system cycle constraints specifically include: the purification system has a large cycle of 2N hours, and each large cycle is divided into a heating sub-cycle and a cooling sub-cycle, each lasting N hours; the electrolytic cell must run continuously during the heating sub-cycle and must not be shut down, while the electrolytic cell can be shut down during the cooling sub-cycle, and the cycle interval Gk is the interval between the (k-1)th cycle and the kth cycle.

[0062] Specifically, a mixed-integer linear programming model is constructed, which includes two parts: an objective function and constraints.

[0063] With the goal of minimizing hydrogen production costs, the objective function is:

[0064]

[0065] In the formula: f represents the total cost; C w C represents the total cost of wind power. pv C represents the total cost of photovoltaic power. net This represents the total cost of grid electricity.

[0066]

[0067] In the formula: i represents the time period; T is the total number of time periods; P w (i), P pv (i), P net (i) represent the power output of wind power, photovoltaic power, and grid power respectively during time period i; ρ w ρ pv ρ netThese are the unit power costs for wind power, solar power, and grid power, respectively.

[0068] The constraints of the model include:

[0069] 1) Renewable energy output constraints:

[0070]

[0071] In the formula: P w,f (i), P pv,f (i) represents the predicted power output of wind power and photovoltaic power during time period i.

[0072] 2) Electrolytic cell operating power constraints:

[0073]

[0074] In the formula: Onoff j (i) is a binary state variable, representing the state of the purification system in the j-th hydrogen production system during time period i, with heating state being 1 and cooling state being 0; P ele,jmin P ele,jmax P represents the lower and upper limits of the operating power of the electrolyzer in the j-th hydrogen production system, respectively; ele,j (i) represents the operating power of the electrolyzer in the j-th hydrogen production system during time period i.

[0075] 3) Electrolytic cell ramp-up speed constraint:

[0076]

[0077] In the formula: r up,j r down,j Δt represents the maximum uphill and downhill ramp power of the electrolyzer in the j-th hydrogen production system; Δt is the time interval between the two sampling points.

[0078] 4) Power balance constraints:

[0079]

[0080] In the formula: M represents the number of hydrogen production systems.

[0081] 5) Hydrogen production constraints:

[0082]

[0083] In the formula: v ele,j Let be the hydrogen production rate of the j-th hydrogen production system.

[0084] 6) Purification system cycle constraints:

[0085] The purification system operates on a 2N-hour cycle, with periods between cycles not necessarily discontinuous. Each major cycle is further divided into a heating sub-cycle and a cooling sub-cycle, each lasting N hours. The electrolytic cell must not be shut down during the heating cycle, but it can be shut down during the cooling cycle.

[0086]

[0087] In the formula: T start,j,k T end,j,k Let G be the start and end times of the k-th major cycle of the j-th purification system, respectively; when k≥2, G k The interval between the (k-1)th period and the kth period is given.

[0088] S40: Solve the model.

[0089] The optimization model is solved using Python's PuLP library. The specific process is as follows:

[0090] 1) Input the initial parameter dataset, including the predicted output power of renewable energy, grid electricity price, hydrogen storage demand, etc.; create decision variables, including the actual output power of renewable energy, the operating power of the electrolyzer, and the binary state variables of the purification system, etc.

[0091] 2) Taking the minimization of the total cost of hydrogen production throughout the entire cycle as the optimization objective, an objective function expression is constructed, and six types of constraints are added in sequence: renewable energy output constraint, electrolyzer operating power constraint, electrolyzer ramp-up speed constraint, power balance constraint, hydrogen production constraint, and purification system cycle constraint;

[0092] 3) Call the CBC solver, set the solution parameters, and perform mixed-integer linear programming solution.

[0093] S50: Output control scheme.

[0094] In one embodiment of the present invention, based on grid electricity price data, when there is a demand for hydrogen production, the output power of renewable energy is prioritized during the cooling sub-cycle. When renewable energy cannot meet the demand for hydrogen production, the power purchased from the grid is considered, so as to reduce the proportion of high-priced grid electricity used in the heating sub-cycle while meeting the demand for hydrogen production, thereby minimizing the total cost of hydrogen production.

[0095] Specifically, after the solver completes the calculation, the system automatically processes and outputs the required results, such as instructions on the purification system operating status, electrolytic cell power, and grid interaction power, forming a complete control scheme.

[0096] The core of this invention is to provide an optimized control method for a hydrogen production system that takes into account purification constraints. The invention will be further illustrated below through embodiments, but is not limited to these embodiments.

[0097] S100: Taking a photovoltaic power station as the research object, its historical power output data for the past year was collected (sampling interval is 15 minutes). First, the data was preprocessed, and the original power data was linearly transformed to the [0,1] interval using the Min-Max normalization method. Then, a prediction model was constructed using the Keras deep learning framework, building a time-series prediction network containing two LSTM layers (64 neurons per layer), with a linear activation function in the output layer. The model was trained using a 5-fold cross-validation strategy, and the parameters were optimized using the Adam optimizer (initial learning rate 0.001), with mean squared error (MSE) as the loss function. The weights were iteratively updated using the backpropagation algorithm until the MSE on the test set dropped below 0.02. After the model reached the expected accuracy requirement, the complete dataset was input into the trained model, and the photovoltaic power prediction results for the next 24 hours were output. The final results are as follows: Figure 4 As shown by the green curve.

[0098] S200: Reads data from the temperature and pressure sensors of the hydrogen storage tank in real time via the Modbus protocol (accuracy level not less than 0.5%), performs standard condition conversion based on the ideal gas law, and calculates the hydrogen production demand for the next 24 hours based on the converted standard condition hydrogen storage ratio and the design capacity of the hydrogen storage tank.

[0099] S300: Construct a mixed-integer linear programming model with the objective of minimizing hydrogen production cost. The objective function is to minimize cost, and includes six types of constraints: 1) Photovoltaic output does not exceed the predicted value; 2) The electrolyzer is forced to operate during the heating cycle, and the power range is limited; 3) The ramp-up and ramp-down rates of the electrolyzer power are limited; 4) The photovoltaic power generation, grid power purchase, and electrolyzer power are balanced in each time period; 5) The total hydrogen production is not less than the hydrogen demand; 6) Cyclic constraints on the purification system, where the heating cycle must be continuous.

[0100] S400: The PuLP library is used to call the CBC solver to solve the model, and the calculation time is less than 10 seconds.

[0101] S500: Outputs the electrolytic cell operating power for the next 24 hours, such as Figure 4 As shown by the blue curve in the figure, the electrolyzer power exhibits the following characteristics: 1) During peak photovoltaic output periods, the electrolyzer power changes synchronously with the photovoltaic output; 2) During periods of insufficient photovoltaic output, the minimum operating power of the electrolyzer is maintained by purchasing electricity from the grid to ensure continuous operation of the purification heating cycle; 3) Power changes are strictly controlled within the ramp-up limits.

[0102] This invention discloses an optimized control method for a hydrogen production system that considers purification constraints. This method first accurately predicts the output of renewable energy through a prediction model, and then establishes a mixed integer programming model that considers purification cycle constraints based on the predicted output of renewable energy and the hydrogen demand. In the model, the cycle of the purification system is introduced as a hard constraint through binary state variables and time-series constraints, thereby ensuring that the electrolyzer can operate stably during the heating cycle, improving hydrogen production efficiency, and avoiding negative impacts caused by power interruption.

[0103] To achieve the above embodiments, such as Figure 5 As shown, this embodiment also provides a hydrogen production system optimization control device 10 that considers purification constraints. The device 10 includes a renewable energy output prediction module 100, a hydrogen demand calculation module 200, an optimization model construction module 300, and a control scheme generation module 400.

[0104] The renewable energy output prediction module 100 uses a long short-term memory neural network to construct a prediction model based on the historical output data of renewable energy generator sets, and predicts the output power of renewable energy in future dispatch cycles.

[0105] The hydrogen demand calculation module 200 calculates the hydrogen demand in the future scheduling cycle based on the real-time monitoring data of the hydrogen storage tank, and converts it into the hydrogen production demand under standard conditions.

[0106] The optimization model building module 300 constructs a mixed integer linear programming model with the goal of minimizing the total cost of hydrogen production. The periodic operation characteristics of the purification system are embedded in the model as hard constraints. The purification system cycle includes a heating sub-cycle and a cooling sub-cycle. During the heating sub-cycle, the electrolyzer must run continuously, while during the cooling sub-cycle, the electrolyzer can be shut down.

[0107] The control scheme generation module 400, based on a mixed integer linear programming model, introduces the electrolyzer operating power, grid power purchase power, and purification system state variables as decision variables, and adds power balance constraints, electrolyzer ramp rate constraints, and hydrogen production constraints to perform optimization and solve the problem, generating a control scheme for the electrolyzer operating power and purification system state.

[0108] Furthermore, the aforementioned renewable energy output prediction module 100 is also used for:

[0109] The historical power output data is normalized, and the original power data is linearly transformed to the [0,1] interval using the Min-Max normalization method;

[0110] A time series prediction network consisting of two LSTM layers was constructed. The output layer adopted a linear activation function, and a 5-fold cross-validation strategy was used for model training. The Adam optimizer was used to optimize the parameters with mean squared error (MSE) as the loss function until the MSE on the test set dropped below 0.02.

[0111] Furthermore, the hydrogen demand calculation module 200 is also used for:

[0112] The temperature and pressure sensor data of the hydrogen storage tank are read in real time using the Modbus protocol, with a sensor accuracy level of no less than 0.5%.

[0113] The real-time monitoring data is converted to standard conditions based on the ideal gas law, with 0°C and 1 atmosphere as standard conditions. The current hydrogen storage ratio is calculated, and the hydrogen production demand in the future scheduling cycle is determined in combination with the designed hydrogen storage capacity of the hydrogen storage tank.

[0114] Furthermore, the aforementioned optimization model construction module 300 is also used for:

[0115] The purification system cycle constraints specifically include: the purification system has a large cycle of 2N hours, and each large cycle is divided into a heating sub-cycle and a cooling sub-cycle, each of N hours;

[0116] The electrolytic cell must run continuously during the heating sub-cycle and must not be stopped. During the cooling sub-cycle, the electrolytic cell can be stopped, and the cycle interval Gk is the interval between the (k-1)th cycle and the kth cycle.

[0117] Furthermore, the aforementioned device 10 also includes:

[0118] The scheduling optimization module is used to prioritize the scheduling of renewable energy output power during the cooling sub-cycle when there is a demand for hydrogen production, based on grid electricity price data. When renewable energy cannot meet the demand for hydrogen production, the power purchased from the grid is considered, so as to reduce the proportion of high-priced grid electricity used in the heating sub-cycle while meeting the demand for hydrogen production, thereby minimizing the total cost of hydrogen production.

[0119] This invention discloses an optimized control device for a hydrogen production system that considers purification constraints. This device first accurately predicts the output of renewable energy through a prediction model, and then establishes a mixed integer programming model that considers purification cycle constraints based on the predicted output of renewable energy and the hydrogen demand. In the model, the cycle of the purification system is introduced as a hard constraint through binary state variables and time-series constraints, thereby ensuring that the electrolyzer can operate stably during the heating cycle, improving hydrogen production efficiency, and avoiding negative impacts caused by power interruption.

[0120] In the description of this specification, the references to terms such as "one embodiment," "some embodiments," "example," "specific example," or "some examples," etc., refer to specific features, structures, materials, or characteristics described in connection with that embodiment or example, which are included in at least one embodiment or example of the present invention. In this specification, the illustrative expressions of the above terms do not necessarily refer to the same embodiment or example. Furthermore, the specific features, structures, materials, or characteristics described may be combined in any suitable manner in one or more embodiments or examples. Moreover, without contradiction, those skilled in the art can combine and integrate the different embodiments or examples described in this specification, as well as the features of different embodiments or examples.

[0121] Furthermore, the terms "first" and "second" are used for descriptive purposes only and should not be construed as indicating or implying relative importance or implicitly specifying the number of technical features indicated. Thus, a feature defined as "first" or "second" may explicitly or implicitly include at least one of that feature. In the description of this invention, "a plurality of" means at least two, such as two, three, etc., unless otherwise explicitly specified.

Claims

1. A method for optimizing the control of a hydrogen production system considering purification constraints, characterized in that, include: S1, based on the historical output data of renewable energy generator sets, uses a long short-term memory neural network to build a prediction model to predict the output power of renewable energy in future dispatch cycles; S2, based on real-time monitoring data from the hydrogen storage tank, calculates the hydrogen demand during the future scheduling cycle and converts it into hydrogen production demand under standard conditions; S3. Construct a mixed integer linear programming model with the goal of minimizing the total cost of hydrogen production. Embed the periodic operation characteristics of the purification system as a hard constraint in the model. The purification system cycle includes a heating sub-cycle and a cooling sub-cycle. During the heating sub-cycle, the electrolyzer must run continuously, while during the cooling sub-cycle, the electrolyzer can be shut down. S4, based on a mixed integer linear programming model, introduces the electrolyzer operating power, grid power purchase power and purification system state variables as decision variables, and adds power balance constraints, electrolyzer ramp rate constraints and hydrogen production constraints to perform optimization and solve the problem, generating a control scheme for the electrolyzer operating power and purification system state; S5. Based on grid electricity price data, when there is a demand for hydrogen production, the output power of renewable energy is prioritized during the cooling sub-cycle. When renewable energy cannot meet the demand for hydrogen production, the power purchased from the grid is considered. This is to reduce the proportion of high-priced grid electricity used in the heating sub-cycle while meeting the demand for hydrogen production, thereby minimizing the total cost of hydrogen production.

2. The method as described in claim 1, characterized in that, S1 includes: S11, normalize the historical power output data, and use the Min-Max standardization method to linearly transform the original power data to the [0,1] interval; S12, construct a time series prediction network containing two LSTM layers, use a linear activation function for the output layer, and use a 5-fold cross-validation strategy for model training. Use the Adam optimizer with mean squared error (MSE) as the loss function to optimize the parameters until the MSE on the test set drops below 0.

02.

3. The method as described in claim 1, characterized in that, The S2 includes: S21, reads temperature and pressure sensor data from the hydrogen storage tank in real time via the Modbus protocol, with sensor accuracy not less than 0.5%; S22. Based on the ideal gas law, the real-time monitoring data is converted to standard conditions. Using 0°C and 1 atmosphere as standard conditions, the current hydrogen storage ratio is calculated, and the hydrogen production demand in the future scheduling cycle is determined in combination with the designed hydrogen storage capacity of the hydrogen storage tank.

4. The method as described in claim 1, characterized in that, The S3 includes: S31, The specific cycle constraints of the purification system include: the purification system has a large cycle of 2N hours, and each large cycle is divided into a heating sub-cycle and a cooling sub-cycle, each of N hours; S32, the electrolytic cell must run continuously during the heating sub-cycle and must not be stopped. During the cooling sub-cycle, the electrolytic cell can be stopped, and the cycle interval Gk is the interval between the (k-1)th cycle and the kth cycle.

5. An optimized control device for a hydrogen production system considering purification constraints, characterized in that, include: The renewable energy output prediction module uses a long short-term memory neural network to build a prediction model based on historical output data of renewable energy generators to predict the output power of renewable energy in future dispatch cycles. The hydrogen demand calculation module calculates the hydrogen demand in the future scheduling cycle based on real-time monitoring data of the hydrogen storage tank, and converts it into hydrogen production demand under standard conditions. The model building module is optimized to construct a mixed integer linear programming model with the goal of minimizing the total cost of hydrogen production. The periodic operation characteristics of the purification system are embedded into the model as hard constraints. The purification system cycle includes a heating sub-cycle and a cooling sub-cycle. During the heating sub-cycle, the electrolyzer must run continuously, while during the cooling sub-cycle, the electrolyzer can be shut down. The control scheme generation module, based on a mixed integer linear programming model, introduces the electrolyzer operating power, grid power purchase power, and purification system state variables as decision variables, and adds power balance constraints, electrolyzer ramp rate constraints, and hydrogen production constraints to perform optimization and solve the problem, generating a control scheme for the electrolyzer operating power and purification system state. The scheduling optimization module is used to prioritize the scheduling of renewable energy output power during the cooling sub-cycle when there is a demand for hydrogen production, based on grid electricity price data. When renewable energy cannot meet the demand for hydrogen production, the power purchased from the grid is considered, so as to reduce the proportion of high-priced grid electricity used in the heating sub-cycle while meeting the demand for hydrogen production, thereby minimizing the total cost of hydrogen production.

6. The apparatus as claimed in claim 5, characterized in that, The renewable energy output prediction module is also used for: The historical power output data is normalized, and the original power data is linearly transformed to the [0,1] interval using the Min-Max normalization method; A time series prediction network consisting of two LSTM layers was constructed. The output layer adopted a linear activation function, and a 5-fold cross-validation strategy was used for model training. The Adam optimizer was used to optimize the parameters with mean squared error (MSE) as the loss function until the MSE on the test set dropped below 0.

02.

7. The apparatus as claimed in claim 5, characterized in that, The hydrogen demand calculation module is also used for: The temperature and pressure sensor data of the hydrogen storage tank are read in real time using the Modbus protocol, with a sensor accuracy level of no less than 0.5%. The real-time monitoring data is converted to standard conditions based on the ideal gas law, with 0°C and 1 atmosphere as standard conditions. The current hydrogen storage ratio is calculated, and the hydrogen production demand in the future scheduling cycle is determined in combination with the designed hydrogen storage capacity of the hydrogen storage tank.

8. The apparatus as claimed in claim 5, characterized in that, The optimization model construction module is also used for: The purification system cycle constraints specifically include: the purification system has a large cycle of 2N hours, and each large cycle is divided into a heating sub-cycle and a cooling sub-cycle, each of N hours; The electrolytic cell must run continuously during the heating sub-cycle and must not be stopped. During the cooling sub-cycle, the electrolytic cell can be stopped, and the cycle interval Gk is the interval between the (k-1)th cycle and the kth cycle.

Images