New energy consumption measuring and calculating method considering hydrogen production process

Abstract

The invention relates to the technical field of new energy consumption, and discloses a new energy consumption calculation method considering a hydrogen production process, and the method comprises the steps: building a hydrogen production coupling model which comprises a new energy power generation power prediction sub-model and a hydrogen production efficiency dynamic sub-model, and obtaining a new energy power generation prediction value and a hydrogen production real-time efficiency coefficient; designing a multi-energy collaborative optimization framework, and generating a scheduling scheme based on a dynamic game theory algorithm; executing a dynamic correction strategy, and adjusting the scheme according to real-time conditions; and establishing a data feedback mechanism, and collecting data to update model parameters. According to the method, the accuracy of new energy consumption measurement and calculation is improved, multi-energy collaborative scheduling is optimized, the dynamic adaptability of the system is enhanced, model self-optimization is realized, the cost can be effectively reduced, wind curtailment and light curtailment are reduced, and remarkable economic benefits and environmental benefits are achieved.

Description

Technical Field

[0001] The present invention relates to the technical field of new energy consumption, and specifically to a method for calculating new energy consumption considering the hydrogen production process. Background Art

[0002] With the increasing global demand for clean energy, the proportion of new energy in power supply is continuously rising. New energy sources such as solar energy and wind energy have the characteristics of strong intermittency and volatility, which pose great challenges to the stable operation of the power system and the efficient consumption of new energy.

[0003] In the traditional power system, when the new energy generation power exceeds the grid load demand, the excess electric energy cannot be effectively stored and utilized, resulting in a large amount of wind and light abandonment. This not only causes energy waste, but also restricts the sustainable development of the new energy industry, resulting in huge economic losses. Moreover, the instability of new energy generation will also affect the frequency and voltage stability of the grid, threatening the safe and reliable operation of the grid. For example, in areas with concentrated wind power generation, sudden changes in wind speed will cause power fluctuations in the grid, increasing the difficulty of grid dispatching.

[0004] Existing new energy consumption methods have many limitations. On the one hand, relying solely on the grid's own regulation ability is difficult to cope with the large-scale access of new energy. The grid's peak shaving and frequency modulation means are limited, and the traditional thermal power peak shaving has a slow response speed and cannot quickly track the power changes of new energy. On the other hand, although energy storage technology can alleviate the volatility problem of new energy to a certain extent, the current energy storage cost is relatively high, and large-scale application is restricted. The high cost makes it difficult for many new energy power generation plants to bear.

[0005] In terms of new energy consumption considering the hydrogen production process, relevant research and practice are still in the exploratory stage. Hydrogen production, as a potential large-scale energy storage and new energy consumption approach, has broad application prospects. However, there is currently a lack of a systematic and accurate method for calculating new energy consumption considering the hydrogen production process. Existing calculation methods often ignore the influence of various factors in the hydrogen production process on new energy consumption, such as the dynamic influence of electrolyzer temperature, electrolysis pressure, and catalyst activity on hydrogen production efficiency, resulting in inaccurate calculation results and unable to provide a reliable basis for new energy consumption decisions.

[0006] In addition, the insufficient accuracy of new energy power generation prediction is also a key issue. Existing prediction methods are difficult to fully consider the complexity of meteorological data and the uncertainty of historical output data, resulting in large prediction errors. Inaccurate power generation prediction will make the operation arrangements of grid dispatching and hydrogen production equipment unreasonable, further reducing the new energy consumption efficiency. To sum up, developing a method that can comprehensively consider various factors and accurately calculate new energy consumption is of great practical significance for improving new energy utilization efficiency and ensuring the stable operation of the power system. Summary of the Invention

[0007] The purpose of the present invention is to provide a new energy consumption measurement method considering the hydrogen production process to solve the problems raised in the above-mentioned background technology.

[0008] To achieve the above purpose, the present invention provides the following technical solutions: A new energy consumption measurement method considering the hydrogen production process, the method includes: Construct a hydrogen production coupling model, including establishing a new energy power generation prediction sub-model and a hydrogen production efficiency dynamic sub-model; the new energy power generation prediction sub-model outputs a predicted value of new energy power generation based on meteorological data and historical output data; the hydrogen production efficiency dynamic sub-model integrates electrolyzer temperature, electrolysis pressure and catalyst activity parameters and outputs the real-time efficiency coefficient of the hydrogen production process; Design a multi-energy collaborative optimization framework, including inputting the predicted value of new energy output, grid load demand and hydrogen production efficiency coefficient into the collaborative optimization framework, and generating a multi-energy collaborative scheduling plan through a dynamic game theory algorithm; the dynamic game theory algorithm includes defining the grid operator, new energy power station and hydrogen production equipment as game participants and solving the optimal consumption strategy based on Nash equilibrium; Execute a dynamic correction strategy, including adjusting the collaborative scheduling plan through rolling horizon optimization according to the real-time new energy output fluctuation and the operating state of the hydrogen production equipment, and outputting a corrected energy distribution instruction; Establish a data feedback mechanism, including collecting the actual new energy consumption rate and the operating data of the hydrogen production equipment, and updating the parameters of the new energy power generation prediction sub-model and the hydrogen production efficiency dynamic sub-model.

[0009] Preferably, the construction of the new energy power generation prediction sub-model includes: using a time series network based on the attention mechanism to process the wind speed, light intensity and temperature sequences in the meteorological data; when inputting historical output data, filling in the missing values through spatio-temporal interpolation and extracting the frequency domain characteristics of the output curve using wavelet transform.

[0010] Preferably, the calculation formula of the hydrogen production efficiency dynamic sub-model is:

[0011] Where is the hydrogen production efficiency coefficient at time , is the real-time temperature of the electrolyzer, is the optimal temperature threshold, is the electrolysis pressure, is the reference pressure, is the starting time of catalyst activity decay, , , , is the weight coefficient.

[0012] Preferably, the dynamic game theory algorithm includes defining the utility function of each game participant: the grid operator aims to minimize the absorption cost, the new energy station aims to minimize the wind and solar power abandonment rate, and the hydrogen production equipment aims to maximize efficiency; and the Pareto optimal solution of the three-party game is solved by mixed integer programming.

[0013] Preferably, the execution process of the rolling time domain optimization includes: taking 15 minutes as a time window, predicting the new energy output deviation of the next window based on the Kalman filter, and adjusting the power allocation priority of the hydrogen production equipment using fuzzy logic rules.

[0014] Preferably, in the data feedback mechanism, the calculation formula for the actual new energy consumption rate is:

[0015] in, For time The amount of electricity connected to the grid by renewable energy sources, The amount of electricity consumed by the hydrogen production equipment, Preferably, the spatiotemporal interpolation filling adopts Kriging interpolation method, combining the historical output data of neighboring stations with the meteorological spatial correlation, to generate the estimated output value of the new energy in the missing period.

[0016] Preferably, the multi-energy coordinated scheduling scheme includes defining a flexible operating range of the hydrogen production equipment: when the output of new energy is in excess, the high-efficiency mode of the hydrogen production equipment is started first; when the load demand of the power grid increases, the hydrogen production equipment is switched to a low-power standby mode.

[0017] Preferably, the input variables of the fuzzy logic rule include the new energy output deviation rate, the hydrogen production efficiency coefficient and the peak load demand intensity of the power grid, and the output variable is the membership function of the power adjustment range of the hydrogen production equipment.

[0018] Preferably, the parameter updating process adopts an online reinforcement learning algorithm, takes the deviation between the actual absorption rate and the predicted value as a reward signal, and dynamically optimizes the network weight of the new energy power generation prediction submodel and the weight coefficient of the hydrogen production efficiency dynamic submodel.

[0019] Compared with the prior art, the present invention has the following beneficial effects: The present invention constructs a new energy power generation prediction sub-model and a hydrogen production efficiency dynamic sub-model, comprehensively considering meteorological data, historical output data, and key parameters in the hydrogen production process, such as electrolyzer temperature, electrolysis pressure, and catalyst activity. In terms of new energy power generation prediction, a time series network based on the attention mechanism is used to process meteorological data, and at the same time, spatio-temporal interpolation filling and wavelet transform are performed on historical output data to extract features, significantly improving the prediction accuracy. This enables a more accurate estimation of new energy power generation, providing a reliable data basis for subsequent consumption measurement. In the hydrogen production efficiency dynamic sub-model, an accurate calculation formula can reflect the efficiency changes in the hydrogen production process in real time, thereby more accurately calculating the consumption of new energy by hydrogen production equipment and overall improving the accuracy of new energy consumption measurement.

[0020] The designed multi-energy collaborative optimization framework uses the dynamic game theory algorithm to define grid operators, new energy power stations, and hydrogen production equipment as game participants. Each participant, based on its own objective function, such as the grid operator pursuing the minimization of consumption cost, the new energy power station pursuing the lowest curtailment rate of wind and light, and the hydrogen production equipment pursuing the maximization of efficiency, solves the Pareto optimal solution through mixed integer programming to generate a multi-energy collaborative scheduling plan. This approach fully considers the interests of all parties and realizes the optimal allocation of energy resources. For example, when the new energy output is excessive, the high-efficiency mode of the hydrogen production equipment is preferentially started to convert the excess electric energy into hydrogen energy for storage; when the grid load demand increases, the hydrogen production equipment is switched to the low-power standby mode to ensure the power supply of the grid, effectively improving the energy utilization efficiency.

[0021] Execute the dynamic correction strategy. According to the real-time new energy output fluctuation and the operating state of the hydrogen production equipment, the collaborative scheduling plan is adjusted through rolling horizon optimization. Taking 15 minutes as the time window, the deviation of the new energy output in the next window is predicted based on the Kalman filter, and the fuzzy logic rule is used to adjust the power allocation priority of the hydrogen production equipment. This dynamic adjustment mechanism can quickly respond to the uncertainty of new energy power generation and the real-time working conditions of the hydrogen production equipment, ensuring that the system is always in the optimal operating state. For example, when the new energy output suddenly increases, the fuzzy logic rule can timely judge and increase the power allocation priority of the hydrogen production equipment to quickly consume the excess electric energy and avoid the occurrence of wind and light curtailment phenomena, enhancing the adaptability of the system to new energy fluctuations.

[0022] Establish a data feedback mechanism to collect the actual new energy consumption rate and the operating data of the hydrogen production equipment, and update the parameters of the new energy power generation prediction sub-model and the hydrogen production efficiency dynamic sub-model. An online reinforcement learning algorithm is used, taking the deviation between the actual consumption rate and the predicted value as the reward signal to dynamically optimize the network weights and weight coefficients of the model. As time goes by and data accumulates, the model can continuously learn and adapt to the changes in the actual situation, improving its own prediction and measurement capabilities, and making the new energy consumption measurement method more accurate and reliable.

[0023] By optimizing the consumption of new energy and the coordinated dispatching of multiple energy sources, the phenomena of wind and light curtailment are reduced, the utilization efficiency of new energy is improved, and the economic losses of new energy power generation plants are decreased. At the same time, the use of traditional fossil energy is reduced, carbon emissions are lowered, which is of positive significance to environmental protection. Brief Description of the Drawings

[0024] Figure 1 It is the working principle diagram of the new energy consumption measurement method described in the present invention; Figure 2 It is the schematic diagram of the construction process of the new energy power generation power prediction sub-model; Figure 3 It is the schematic diagram of the process of solving the optimal consumption strategy by the dynamic game theory algorithm. Detailed Embodiments

[0025] Next, the technical solutions in the embodiments of the present invention will be clearly and completely described in conjunction with the drawings in the embodiments of the present invention. Obviously, the described embodiments are only a part of the embodiments of the present invention, rather than all the embodiments. Based on the embodiments of the present invention, all other embodiments obtained by those of ordinary skill in the art without creative efforts shall fall within the protection scope of the present invention.

[0026] Please refer to Figures 1-3 , the present invention provides a new energy consumption measurement method considering the hydrogen production process, and its overall implementation scheme is as follows: Construct a hydrogen production coupling model: respectively establish a new energy power generation power prediction sub-model and a dynamic hydrogen production efficiency sub-model. Using meteorological data and historical output data, a new energy power generation power prediction sub-model is constructed through a specific algorithm to output the predicted value of new energy power generation. At the same time, the electrolyzer temperature, electrolysis pressure, catalyst activity parameters, etc. are integrated into the dynamic hydrogen production efficiency sub-model to output the real-time efficiency coefficient of the hydrogen production process.

[0027] Design a multi-energy coordinated optimization framework: input the predicted value of new energy output, grid load demand, and hydrogen production efficiency coefficient into the coordinated optimization framework. In this framework, using the dynamic game theory algorithm, the grid operator, new energy power station, and hydrogen production equipment are defined as game participants, and the optimal consumption strategy is solved based on the Nash equilibrium, and then a multi-energy coordinated dispatching scheme is generated.

[0028] Execute the dynamic correction strategy: According to the real-time new energy output fluctuation and the operation state of the hydrogen production equipment, the coordinated dispatching scheme is adjusted by means of rolling horizon optimization, and finally the corrected energy distribution instruction is output.

[0029] Establish a data feedback mechanism: Collect the actual new energy consumption rate and the operation data of hydrogen production equipment, and use this data to update the parameters of the new energy power generation prediction sub-model and the hydrogen production efficiency dynamic sub-model, so as to continuously optimize the calculation method.

[0030] The following further describes the implementation of the present invention in conjunction with Embodiments 1 to 5.

[0031] Embodiment 1: When constructing the new energy power generation prediction sub-model, the wind speed, light intensity and temperature sequences in meteorological data have a significant impact on new energy power generation. A time series network based on the attention mechanism is used to process these meteorological data sequences. This network can automatically learn the importance of meteorological data at different times and highlight the data features that have a greater impact on power generation. For example, in the wind power generation scenario, the change trend of wind speed is crucial for power generation, and the attention mechanism can strengthen the capture of key information in the wind speed sequence.

[0032] For historical output data, there will inevitably be missing value situations. At this time, the Kriging interpolation method is used for spatio-temporal interpolation filling. The Kriging interpolation method combines the historical output data of adjacent stations and the meteorological spatial correlation to generate the estimated value of new energy output in the missing period. Suppose there is a new energy station A, and the output data of a certain period is missing. By collecting the historical output data of adjacent stations B, C, etc., and considering the meteorological spatial correlation between these stations and station A, such as the similarity degree of wind speed and light intensity in space, a mathematical model is constructed using the Kriging interpolation method. Let the output data of adjacent stations be , are the spatial positions of each station. By calculating the weights of different positions to the missing value position , then the estimated value of the missing value . After the filling is completed, wavelet transform is used to extract the frequency domain features of the output curve. Wavelet transform can decompose the output curve into different frequency components, extract the low-frequency trend information and high-frequency fluctuation information, and provide richer data features for subsequent power generation prediction.

[0033] Embodiment 2: The purpose of this embodiment is to clarify the specific calculation formula of the hydrogen production efficiency dynamic sub-model and the meaning of each parameter, and accurately calculate the real-time efficiency coefficient of the hydrogen production process. The calculation formula of the hydrogen production efficiency dynamic sub-model is:

[0034] Among them, is the hydrogen production efficiency coefficient at time , which reflects the efficiency of the hydrogen production process at time . is the real-time temperature of the electrolyzer, which changes during the hydrogen production process. Real-time monitoring of its temperature value is crucial for accurately calculating the hydrogen production efficiency coefficient. is the optimal temperature threshold, which is the temperature reference value when the hydrogen production efficiency reaches the best state. is the electrolysis pressure, and the change in pressure will also affect the hydrogen production efficiency. is the reference pressure, which serves as the reference value for pressure comparison. is the starting time of catalyst activity decay. As time goes by, the catalyst activity will gradually decay, affecting the hydrogen production efficiency. This parameter is used to measure this change. , , , are the weight coefficients, which are determined through a large amount of experimental data and optimization algorithms and are used to adjust the relative importance of each influencing factor in the calculation of the hydrogen production efficiency coefficient. For example, after multiple experimental verifications, under specific hydrogen production equipment and working conditions, it is determined that , , , , to ensure the accuracy of the calculation of the hydrogen production efficiency coefficient.

[0035] Example 3: In the dynamic game theory algorithm, clarify the utility functions of each game participant. The grid operator aims to minimize the consumption cost, which includes but is not limited to the transmission loss cost of the grid, the peak shaving cost for balancing the fluctuations of new energy output, etc. Assume that the consumption cost function of the grid operator is , where is the power of the grid to consume new energy at time , , are the cost coefficients related to time . The new energy power station aims to minimize the curtailment rate of wind and light. The curtailment rate of wind and light , where is the power generation of the new energy power station at time , is the power of new energy consumed at time . The hydrogen production equipment aims to maximize the efficiency, that is, to maximize the efficiency coefficient calculated by the dynamic sub-model of hydrogen production efficiency.

[0036] Solve the Pareto optimal solution of the three-party game through mixed integer programming. Mixed integer programming is a mathematical method for solving the optimal value of the objective function under a series of constraints, where part of the decision variables are integers. In this scenario, the constraints include the power transmission limit of the grid, the power generation capacity limit of the new energy power station, the power regulation range limit of the hydrogen production equipment, etc. For example, the upper limit of the power transmission of a certain transmission line in the grid is , it is necessary to meet , , which is the number of new energy power generation points accessing the line. By solving this mixed-integer programming problem, the optimal strategies of the power grid operator, new energy power stations and hydrogen production equipment in different situations can be obtained, realizing the collaborative optimization scheduling of multiple energies.

[0037] Example 4: This example mainly describes the specific execution process of rolling horizon optimization and the application of fuzzy logic rules to realize the dynamic adjustment of the collaborative scheduling scheme.

[0038] The rolling horizon optimization uses a 15-minute time window. Within each time window, the deviation of new energy output in the next window is predicted based on the Kalman filter. The Kalman filter is an algorithm that optimally estimates the system state using the state equation and observation equation of the system. Assume the state equation of the new energy power generation system is , and the observation equation is , where is the state of the system at time , , , are coefficient matrices, is the control input, , are noises. The state estimate value is continuously updated through the Kalman filter to predict the deviation of new energy output in the next window.

[0039] Fuzzy logic rules are used to adjust the power allocation priority of the hydrogen production equipment. The input variables of the fuzzy logic rules include the new energy output deviation rate, hydrogen production efficiency coefficient and power grid peak shaving demand intensity. The new energy output deviation rate , where is the actual new energy output, is the predicted new energy output. The hydrogen production efficiency coefficient is the value calculated by the dynamic sub-model of hydrogen production efficiency. The power grid peak shaving demand intensity can be determined according to factors such as the change rate of the power grid load. The output variable is the membership function of the power adjustment amplitude of the hydrogen production equipment. For example, when the new energy output deviation rate is large and positive (indicating overproduction), the hydrogen production efficiency coefficient is high, and the power grid peak shaving demand intensity is low, the fuzzy logic rules determine that the power allocation priority of the hydrogen production equipment should be increased significantly, and the power of the hydrogen production equipment should be increased to absorb the excess new energy power. By establishing a series of fuzzy logic rules, the dynamic and intelligent adjustment of the power allocation priority of the hydrogen production equipment is realized, ensuring the effectiveness of the multi-energy collaborative scheduling scheme.

[0040] Example 5: This embodiment focuses on illustrating the calculation method of the actual new energy consumption rate and the update process of model parameters in the data feedback mechanism to improve the accuracy and adaptability of the new energy consumption measurement method. In the data feedback mechanism, the calculation formula for the actual new energy consumption rate is:

[0041] Among them, is the grid-connected power of new energy within time , which can be accurately obtained through grid metering equipment. is the power consumed by the hydrogen production equipment, which is recorded by the power metering device of the hydrogen production equipment. is the total power generation of new energy, which is obtained by statistically analyzing the metering data of new energy power generation equipment. For example, within a certain hour, the grid-connected power of new energy is , the power consumed by the hydrogen production equipment is , and the total power generation of new energy is , then the actual new energy consumption rate for that hour .

[0042] The parameter update process adopts an online reinforcement learning algorithm. Using the deviation between the actual consumption rate and the predicted value as the reward signal, dynamically optimize the network weights of the new energy power generation power prediction sub-model and the weight coefficients of the hydrogen production efficiency dynamic sub-model. The online reinforcement learning algorithm continuously interacts with the environment and adjusts its own strategy according to the reward signal. When the actual consumption rate is higher than the predicted value, a positive reward is given, and the algorithm will adjust the network weights of the prediction sub-model and the weight coefficients of the hydrogen production efficiency dynamic sub-model in the direction of making the predicted value closer to the actual value; conversely, when the actual consumption rate is lower than the predicted value, a negative reward is given, prompting the algorithm to optimize the model parameters. For example, using the Deep Q-Network (DQN) algorithm, taking the deviation of the actual consumption rate as the basis for Q-value update, and continuously iteratively training to enable the model to better adapt to the actual situation and improve the accuracy of new energy consumption measurement.

[0043] It should be noted that in this article, relational terms such as first and second are only used to distinguish one entity or operation from another entity or operation, and do not necessarily require or imply any actual relationship or order between these entities or operations. Moreover, the terms "including", "comprising" or any other variant thereof are intended to cover non-exclusive inclusion, so that a process, method, article or device including a series of elements not only includes those elements, but also includes other elements not expressly listed, or also includes elements inherent to such process, method, article or device.

[0044] Although embodiments of the present invention have been shown and described, it will be understood by those of ordinary skill in the art that various changes, modifications, substitutions and variations can be made to these embodiments without departing from the principles and spirit of the present invention, and the scope of the present invention is defined by the appended claims and their equivalents.

Claims

1. A new energy consumption measurement method considering the hydrogen production process, characterized in that, The method includes: Constructing a hydrogen production coupling model, including establishing a new energy power generation prediction sub-model and a hydrogen production efficiency dynamic sub-model; the new energy power generation prediction sub-model outputs a predicted value of new energy power generation based on meteorological data and historical output data; the hydrogen production efficiency dynamic sub-model integrates electrolyzer temperature, electrolysis pressure, and catalyst activity parameters and outputs the real-time efficiency coefficient of the hydrogen production process; Designing a multi-energy collaborative optimization framework, including inputting the predicted value of new energy output, grid load demand, and hydrogen production efficiency coefficient into the collaborative optimization framework, and generating a multi-energy collaborative scheduling plan through a dynamic game theory algorithm; the dynamic game theory algorithm includes defining grid operators, new energy power stations, and hydrogen production equipment as game participants and solving the optimal absorption strategy based on Nash equilibrium; Executing a dynamic correction strategy, including adjusting the collaborative scheduling plan through rolling horizon optimization according to real-time new energy output fluctuations and the operating status of hydrogen production equipment, and outputting a corrected energy allocation instruction; Establishing a data feedback mechanism, including collecting the actual new energy absorption rate and the operating data of hydrogen production equipment, and updating the parameters of the new energy power generation prediction sub-model and the hydrogen production efficiency dynamic sub-model.

2. The method according to claim 1, characterized in that, The construction of the new energy power generation prediction sub-model includes: using a time series network based on the attention mechanism to process the wind speed, light intensity, and temperature sequences in meteorological data; when inputting historical output data, filling in missing values through spatio-temporal interpolation and using wavelet transform to extract the frequency domain characteristics of the output curve.

3. The method according to claim 1, characterized in that, The calculation formula of the hydrogen production efficiency dynamic sub-model is: ； Among them, is the hydrogen production efficiency coefficient with respect to time , is the real-time temperature of the electrolyzer is the optimal temperature threshold is the electrolysis pressure is the reference pressure is the starting time of the catalyst activity decay , , , are weight coefficients 4. The method according to claim 1, characterized in that, The dynamic game theory algorithm includes defining the utility functions of each game participant: the grid operator aims to minimize the absorption cost, the new energy power station aims to minimize the curtailment rate of wind and light, and the hydrogen production equipment aims to maximize the efficiency; Solving the Pareto optimal solution of the three-party game through mixed integer programming.

5. The method according to claim 1, characterized in that The execution process of the rolling horizon optimization includes: taking a 15-minute time window, predicting the new energy output deviation in the next window based on the Kalman filter, and adjusting the power allocation priority of the hydrogen production equipment using fuzzy logic rules.

6. The method according to claim 1, wherein In the data feedback mechanism, the calculation formula of the actual new energy absorption rate is: ； Among them, is the time the grid-connected electricity of new energy within it, is the electricity consumed by hydrogen production equipment, is the total power generation of new energy.

7. The method according to claim 2, characterized in that, The spatio-temporal interpolation filling uses Kriging interpolation method, combines the historical output data of neighboring power stations and the meteorological spatial correlation, and generates an estimated value of new energy output in the missing period.

8. The method according to claim 1, wherein The multi-energy collaborative scheduling plan includes defining the flexible operation range of the hydrogen production equipment: when the new energy output is excessive, the high-efficiency mode of the hydrogen production equipment is preferentially started; when the grid load demand increases, the hydrogen production equipment is switched to the low-power standby mode.

9. The method according to claim 5, characterized in that The input variables of the fuzzy logic rules include the new energy output deviation rate, the hydrogen production efficiency coefficient, and the grid peak shaving demand intensity, and the output variable is the membership function of the power adjustment amplitude of the hydrogen production equipment.

10. The method according to claim 1, characterized in that, The parameter update process uses an online reinforcement learning algorithm, takes the deviation between the actual absorption rate and the predicted value as the reward signal, and dynamically optimizes the network weights of the new energy power generation prediction sub-model and the weight coefficients of the hydrogen production efficiency dynamic sub-model.

Images