A wind-solar-hydrogen-ammonia-alcohol multi-energy system intelligent decision and dynamic optimization method and system

Abstract

The application discloses a kind of wind light hydrogen ammonia alcohol multi-purpose system wisdom decision and dynamic optimization method, comprising the following steps: step S1: real-time acquisition each data;Step S2: construct multi-objective optimization model, the weight factor between double target can be configured in predefinition range online;Step S3: using multi-objective genetic algorithm, based on the high-frequency optimization solution of the multi-objective optimization model to the rolling time window, generate collaborative control instruction;Step S4: the collaborative control instruction is issued to corresponding equipment execution.The application improves the economy and environmental protection of system by whole chain collaborative optimization, can still guarantee chemical process stable operation when wind light forecast is seriously distorted, avoids unplanned shutdown, solves the problem of renewable energy fluctuation restricts the reliability of chemical production.

Description

Technical Field

[0001] This invention relates to a smart decision-making and dynamic optimization method and system for a wind-solar-hydrogen-ammonia-ethanol multi-energy system, belonging to the field of industrial systems engineering and efficient utilization of renewable energy technology. Background Technology

[0002] Currently, in industrial production fields utilizing wind and solar power for hydrogen production, ammonia synthesis, and methanol synthesis, a unit-based independent control strategy is commonly adopted. Existing technologies typically design and operate subsystems such as wind and solar power generation, electrolytic hydrogen production, ammonia synthesis, methanol synthesis, and energy storage as independent units, with each unit only linked by simple start / stop signals or fixed power / flow commands.

[0003] This control method has significant drawbacks. First, the drastic fluctuations in wind and solar power directly lead to frequent start-ups and shutdowns or prolonged low-load operation of the electrolyzers, severely shortening equipment lifespan and increasing maintenance costs. To maintain stable process parameters such as temperature and pressure required for the synthesis sections (ammonia, methanol), the system typically needs to rely on external grid power purchases or self-supplied high-carbon energy for compensation, resulting in an increase in the overall carbon footprint of the system. Second, existing energy management strategies are mostly based on static optimization models with preset rules or single economic objectives, which cannot respond in real time to fluctuations in time-of-use electricity prices, changes in downstream product demand, and prediction errors in wind and solar power. The system lacks the ability to dynamically optimize and balance benefits among multiple energy conversion paths such as direct power sales, hydrogen production, ammonia production, and methanol production, leading to economic losses and insufficient renewable energy absorption. Furthermore, due to the lack of forward-looking collaborative optimization and rapid buffer adjustment mechanisms, existing systems struggle to ensure the stable operation of downstream chemical processes when there are significant deviations in wind and solar power predictions, easily triggering unplanned shutdowns and resulting in low operational reliability.

[0004] These technological bottlenecks severely restrict the economic viability, environmental friendliness, and operational reliability of large-scale integrated wind-solar-hydrogen-ammonia-methanol projects, hindering their commercialization. Therefore, there is an urgent need for an intelligent system and method that can connect the entire "electricity-hydrogen-ammonia-methanol-application" chain and achieve multi-objective real-time collaborative optimization decision-making. Summary of the Invention

[0005] The purpose of this invention is to provide a method and system for intelligent decision-making and dynamic optimization of a wind-solar-hydrogen-ammonia-ethanol multi-energy system. This invention aims to achieve coordinated control and multi-objective optimization of the entire chain, significantly improve the system's operational robustness under renewable energy fluctuations, and verify the feasibility of the technology through high-fidelity simulation.

[0006] To solve the above-mentioned technical problems, the present invention adopts the following technical solution: a smart decision-making and dynamic optimization method for a wind-solar-hydrogen-amine-ethanol multi-energy system, comprising the following steps:

[0007] Step S1: Real-time acquisition of wind and solar power forecast data, grid time-of-use electricity price signals, downstream product demand data, and operating status data of each unit within the system;

[0008] Step S2: Construct a multi-objective optimization model with the dual objectives of maximizing the daily operating revenue of the system and minimizing the carbon dioxide emissions of the system. The weighting factors between the dual objectives can be configured online within a predefined range.

[0009] Step S3: Using a multi-objective genetic algorithm, the multi-objective optimization model is solved at high frequency based on a rolling time window to generate collaborative control instructions; the collaborative control instructions include electrolyzer load rate instructions, energy storage system charging and discharging power instructions, synthesis section feed ratio instructions, and power trading strategy instructions;

[0010] Step S4: Send the collaborative control command to the corresponding equipment for execution, and use the hydrogen storage tank and lithium battery energy storage system to perform minute-level real-time power balancing to buffer the impact of wind and solar power fluctuations on downstream chemical processes;

[0011] The multi-objective genetic algorithm employs a hot-start mechanism, which uses the solution set of the previous optimization cycle as the initial population for the current cycle's algorithm iteration.

[0012] The aforementioned method uses a rolling time window that is optimized based on forecast data for the next 4 hours and operates on a 5-minute cycle.

[0013] The aforementioned method uses a multi-objective genetic algorithm that is an improved NSGA-II algorithm, the improvement of which includes at least one of the following:

[0014] (a) Adaptive crossover and mutation probabilities are used, and their values ​​are dynamically adjusted based on the number of iterations and the degree of aggregation of individuals at the Pareto front.

[0015] (b) An improved constraint dominance relationship handling rule is adopted, which comprehensively considers the merits of the objective function and the degree of constraint violation when comparing two infeasible solutions;

[0016] (c) In the elite selection strategy, the smoothness of control instructions is taken into consideration.

[0017] The aforementioned method, after step S4, further includes:

[0018] Step S5: Simulate and verify the intelligent decision-making and dynamic optimization process using a high-fidelity digital twin platform; the digital twin platform integrates multi-physics modeling tools for electrical, chemical, and control systems, and exchanges data in real time via the OPC UA protocol.

[0019] In the aforementioned method, the sum of the weights of the economic objective and the carbon emission objective in the objective function of the multi-objective optimization model is 1, and their respective weight configuration ranges are 0.2-0.8.

[0020] A smart decision-making and dynamic optimization system for a wind-solar-hydrogen-amine-ethanol multi-energy system includes:

[0021] The data acquisition and communication module is used to collect real-time wind and solar power forecast data, grid time-of-use electricity price signals, downstream product demand data, and operating status data of each unit in the system.

[0022] A multi-objective rolling optimization decision module is used to construct a multi-objective optimization model with the dual objectives of maximizing the daily operating revenue of the system and minimizing the carbon dioxide emissions of the system. A multi-objective genetic algorithm is used to perform high-frequency optimization based on a rolling time window to generate collaborative control commands. These collaborative control commands include electrolyzer load rate commands, energy storage system charging and discharging power commands, synthesis section feed ratio commands, and power trading strategy commands. The improved multi-objective genetic algorithm employs a hot-start mechanism.

[0023] The multi-timescale coordinated control module is used to send the coordinated control commands to the corresponding devices for execution, and to control the hydrogen storage tank and lithium battery energy storage system to perform minute-level real-time power balancing, so as to buffer the impact of wind and solar power fluctuations on downstream chemical processes.

[0024] The aforementioned system also includes:

[0025] The digital twin verification module integrates multiphysics modeling tools for electrical, chemical, and control systems. It exchanges data with the multi-objective rolling optimization decision module and the multi-timescale coordinated control module via the OPC UA protocol to perform high-fidelity simulation verification of the system decision-making and control processes.

[0026] In the aforementioned system, the multi-objective rolling optimization decision module uses a rolling time window that is based on predicted data for the next 4 hours and performs rolling optimization in 5-minute cycles.

[0027] In the aforementioned system, the multi-objective genetic algorithm is an improved NSGA-II algorithm, and its improvements include at least one of the following:

[0028] (a) Adaptive crossover and mutation probabilities are used;

[0029] (b) Adopt improved rules for handling constraint dominance relationships;

[0030] (c) Incorporate consideration of the smoothness of control instructions into the elite selection strategy.

[0031] In the aforementioned system, the objective function of the multi-objective optimization model has an allocation range of 0.2-0.8 for the weights of the economic objective and the carbon emission objective.

[0032] Compared with the prior art, the present invention has at least the following beneficial effects:

[0033] (1) This invention achieves coordinated dynamic matching of the entire chain of electricity-hydrogen-ammonia-alcohol through a multi-objective rolling optimization mechanism. Under the same wind and solar resources and equipment configuration, compared with the traditional unit independent control strategy, this invention can increase the total operating revenue of the system by about 17.3% and reduce CO2 emissions by about 28.1%, fundamentally solving the core contradiction that the traditional strategy cannot take into account both economic and environmental benefits.

[0034] (2) This invention allows the system to smoothly switch between different operating modes, such as "economic priority" and "low-carbon priority," without any hardware changes, by adjusting the weighting factors of the economic and carbon emission targets in the multi-objective optimization model online. Experiments show that by simply adjusting the weights, an adjustment range of approximately 9.1% for operating revenue and approximately 36% for CO2 emissions can be achieved, greatly enhancing the system's adaptability to carbon policies, environmental regulations, and corporate operating orientations in different regions.

[0035] (3) By integrating high-frequency rolling optimization decision-making with minute-level real-time energy storage regulation, the system can effectively cope with the drastic fluctuations and prediction errors of renewable energy. Even under extreme conditions where the wind and solar power prediction error is as high as 5.1%, the system can still ensure the stability of key process parameters in the downstream ammonia / methanol synthesis section and completely avoid unplanned shutdowns, significantly improving the reliability and safety of the entire system.

[0036] (4) The optimized collaborative control strategy of this invention significantly reduces the frequent start-ups and shutdowns of the electrolyzer caused by fluctuations in wind and solar power. Experimental data shows that the number of start-ups and shutdowns of the electrolyzer can be reduced by 82%, effectively extending the service life of key equipment such as the electrolyzer and reducing system maintenance costs. Attached Figure Description

[0037] Figure 1 This is a flowchart of the intelligent decision-making and dynamic optimization method of the present invention;

[0038] Figure 2 This is a schematic diagram of the system structure of the present invention.

[0039] The present invention will be further described below with reference to the accompanying drawings and specific embodiments. Detailed Implementation

[0040] Embodiment 1 of the present invention:

[0041] A method for intelligent decision-making and dynamic optimization of a wind-solar-hydrogen-amine-ethanol multi-energy system includes the following steps:

[0042] Step S1: Real-time acquisition of wind and solar power forecast data, grid time-of-use electricity price signals, downstream product demand data, and operating status data of each unit within the system.

[0043] Step S2: Construct a multi-objective optimization model with the dual objectives of maximizing the daily operating revenue of the system and minimizing the carbon dioxide emissions of the system. The weighting factors between the dual objectives can be configured online within a predefined range.

[0044] Step S3: Using a multi-objective genetic algorithm, the multi-objective optimization model is solved at high frequency based on a rolling time window to generate collaborative control instructions; the collaborative control instructions include electrolyzer load rate instructions, energy storage system charging and discharging power instructions, synthesis section feed ratio instructions, and power trading strategy instructions.

[0045] Step S4: Send the collaborative control command to the corresponding equipment for execution, and use the hydrogen storage tank and lithium battery energy storage system to perform minute-level real-time power balancing to buffer the impact of wind and solar power fluctuations on downstream chemical processes;

[0046] The multi-objective genetic algorithm employs a hot-start mechanism, which uses the solution set of the previous optimization cycle as the initial population for the current cycle's algorithm iteration.

[0047] Step S5: Simulate and verify the intelligent decision-making and dynamic optimization process using a high-fidelity digital twin platform; the digital twin platform integrates multi-physics modeling tools for electrical, chemical, and control systems, and exchanges data in real time via the OPC UA protocol.

[0048] Specifically, the multi-objective genetic algorithm is an improved NSGA-II algorithm. Based on the classic NSGA-II framework, it makes the following key improvements to address the high-dimensional, nonlinear, and strongly constrained optimization characteristics of the wind-solar-hydrogen-amine-ethanol coupled system, thereby improving convergence speed, solution set diversity, and computational efficiency to meet real-time optimization requirements. These improvements include at least one of the following:

[0049] (a) Adaptive Crossover and Mutation Operators: To avoid premature convergence and maintain population diversity, this improved algorithm designs adaptive crossover probability ($P_c$) and mutation probability ($P_m$). Their values ​​are no longer fixed but dynamically adjusted with the number of generations and the degree of clustering of individuals. Specifically, larger $P_c$ and $P_m$ are used in the early stages of evolution to promote global exploration, and gradually decreased in the later stages to facilitate local development; for individuals clustered locally in the Pareto Front, their $P_m$ is automatically increased to encourage them to diffuse to other sparse regions.

[0050] Its adjustment strategy is defined by the following formula:

[0051] $P_c(g)=P_{c,\text{max}}-(P_{c,\text{max}}-P_{c,\text{min}})\cdot\frac{g}{G_{\te xt{max}}}}$

[0052] $P_m(g)=P_{m,\text{max}}-(P_{m,\text{max}}-P_{m,\text{min}})\cdot\frac{g}{G_{\te xt{max}}}}$

[0053] Where $g$ is the current iteration number, and $G_{\text{max}}$ is the maximum iteration number.

[0054] $P_{c,\text{max}}$,$P_{c,\text{min}}$,$P_{m,\text{max}}$,$P_{m,\text{min}}$ are probability boundary values ​​preset based on experience.

[0055] (b) Improvements to System Constraint Handling: For various complex constraints such as electrolyzer start-up / shutdown constraints, energy storage charging / discharging power constraints, and chemical process safety operation constraints, the comparison rules for constraint domination have been improved. For two infeasible solutions (solutions that violate constraints), the solution with the smaller constraint violation is no longer simply considered better. Instead, the merits of their objective functions and the degree of constraint violation are comprehensively considered, granting them a certain degree of competitiveness. This guides the search to efficiently transition from the infeasible region to the feasible region, preventing the population from falling into local optima when the feasible region is discontinuous.

[0056] (c) Application-oriented elite retention strategy: In each generation of elite selection, in addition to non-dominated ranking and crowding distance, considerations of the practicality and stability of the system operation strategy are introduced. For example, for solutions with excessively drastic changes in control commands, even if their Pareto ranking is high, their priority will be appropriately reduced, and solutions with smoother control commands and more suitable for engineering implementation will be selected first, thereby ensuring that the optimization results can be directly used for system control and avoiding frequent equipment operations.

[0057] Specifically, the rolling time window is based on the predicted data for the next 4 hours and is optimized in 5-minute cycles. To adapt to the 5-minute rolling optimization, this invention uses the solution set of the previous optimization cycle as the initial population for the algorithm iteration in the current cycle. Since the system state and wind and solar prediction data of adjacent cycles are continuous, this "hot start" mechanism can significantly reduce the number of iterations required for algorithm convergence, thereby obtaining high-quality solutions within a limited computation time (<5 minutes) and meeting the timeliness requirements of real-time control.

[0058] Specifically, in the objective function of the multi-objective optimization model, the sum of the weights of the economic objective and the carbon emission objective is 1, and their respective weight configuration ranges from 0.2 to 0.8. By adjusting a single parameter, the system's operating strategy can seamlessly switch between different modes such as "economic priority" and "environmental priority," thus flexibly adapting to the carbon tax policies, environmental regulations, and enterprise operational orientations of different regions without changing the hardware structure. The adjustment ranges for operating revenue and CO2 emissions can reach 9.1% and 36%, respectively.

[0059] This invention integrates MATLAB / Simulink, Aspen, and Python, and employs the OPC UA protocol for real-time data exchange to construct a high-fidelity digital twin simulation platform covering multiple physical fields such as electrical, chemical, and control. It provides a fully reproducible and operable verification environment for the aforementioned intelligent decision-making methods, with all parameters, algorithms, and results clearly defined, ensuring the feasibility and reliability of the technical solution.

[0060] Embodiment 2 of the present invention:

[0061] A smart decision-making and dynamic optimization system for a wind-solar-hydrogen-amine-ethanol multi-energy system includes:

[0062] The data acquisition and communication module is used to collect real-time wind and solar power forecast data, grid time-of-use electricity price signals, downstream product demand data, and operating status data of each unit in the system.

[0063] A multi-objective rolling optimization decision module is used to construct a multi-objective optimization model with the dual objectives of maximizing the daily operating revenue of the system and minimizing the carbon dioxide emissions of the system. A multi-objective genetic algorithm is used to perform high-frequency optimization based on a rolling time window to generate collaborative control commands. These collaborative control commands include electrolyzer load rate commands, energy storage system charging and discharging power commands, synthesis section feed ratio commands, and power trading strategy commands. The improved multi-objective genetic algorithm employs a hot-start mechanism.

[0064] The multi-timescale coordinated control module is used to send the coordinated control commands to the corresponding devices for execution, and to control the hydrogen storage tank and lithium battery energy storage system to perform minute-level real-time power balancing, so as to buffer the impact of wind and solar power fluctuations on downstream chemical processes.

[0065] The digital twin verification module integrates multiphysics modeling tools for electrical, chemical and control fields, and exchanges data with the multi-objective rolling optimization decision module and the multi-timescale coordinated control module through the OPC UA protocol, for high-fidelity simulation verification of the system decision and control process;

[0066] The modules mentioned above interact with each other through industrial standard protocols to jointly achieve synergistic optimization and stable control of the entire chain of electricity-hydrogen-ammonia-alcohol.

[0067] Specifically, in the multi-objective rolling optimization decision module, the rolling time window is based on the predicted data for the next 4 hours and is rolled with a period of 5 minutes.

[0068] Specifically, the multi-objective genetic algorithm is an improved NSGA-II algorithm, and its improvements include at least one of the following:

[0069] Adaptive crossover and mutation probabilities are employed.

[0070] An improved constraint dominance relationship handling rule is adopted;

[0071] Incorporate consideration of the smoothness of control commands into the elite selection strategy.

[0072] Specifically, in the objective function of the multi-objective optimization model, the weights of the economic objective and the carbon emission objective are configured in the range of 0.2-0.8.

[0073] The present invention will be further explained below with reference to specific experimental cases and comparative data.

[0074] Embodiment 3 of the present invention: Multi-objective optimization benchmark test based on actual wind and solar data.

[0075] The experiment used measured wind and solar power data from a base in my country in August 2023, with a total installed capacity of 300MW (200MW wind power + 100MW solar power). The system was configured with four 1000Nm turbines. 3 / h alkaline electrolyzer (Huadian Huazhen type, minimum load rate 20%, cold start time 45min, unit energy consumption 4.3kWh / Nm³) 3The ammonia synthesis section uses an iron-based catalyst (Fe3O4:Al2O3:K2O = 68:28:4wt%), with an operating pressure of 15.2MPa and a temperature control of 455℃; the methanol synthesis section uses a Cu / ZnO / Al2O3 catalyst (Cu:Zn:Al = 45:45:10wt%), with an operating pressure of 7.8MPa and a temperature control of 235℃; it is equipped with a 20MWh lithium battery energy storage (efficiency 92%) and a 2000kg hydrogen storage tank (4.5MPa).

[0076] Downstream demand is set at a constant load: 500 kg / h hydrogen, 20 t / h liquid ammonia, and 15 t / h methanol. The power grid adopts time-of-use pricing: RMB 1.2 / kWh during peak hours (10:00-15:00, 18:00-22:00) and RMB 0.7 / kWh during normal hours.

[0077] (7:00-10:00, 15:00-18:00, 22:00-23:00), off-peak price 0.3 yuan / kWh (23:00-7:00).

[0078] The experiment constructed a digital twin platform using MATLAB / Simulink and Aspen, and exchanged data using the OPC UA protocol. Specific implementation details include:

[0079] 1) A rigorous thermodynamic model was established in Aspen, with the Temkin-Pyzhev kinetic equation used for ammonia synthesis and the Langmuir-Hinshelwood kinetic model used for methanol synthesis;

[0080] 2) Deploy the LSTM wind and solar power prediction module in Simulink (prediction error 8.2%);

[0081] 3) Configure a multi-objective optimization core and adopt an improved NSGA-II algorithm (population size 100, iterations 200 times). The objective function is to maximize daily operating revenue and minimize CO2 emissions. The decision variables include electrolyzer load rate, energy storage power, feed ratio of synthesis section and power trading strategy.

[0082] 4) Perform rolling optimization with a 5-minute cycle and generate control commands based on a 4-hour prediction window; 5) Perform continuous simulation for 720 hours (30 days).

[0083] The experimental results are shown in the table below:

[0084] Performance indicators This invention system Traditional control strategies Improvement effect Total operating revenue (ten thousand yuan) 1243.5 1060.2 +17.3％ CO2 emissions (tons) 321.8 447.5 -28.1％ Electrolytic cell start-up and shutdown times 18 101 -82％ Ammonia production volatility ±2.1％ ±12.5％ -83％ Wind and solar curtailment rate 1.5％ 5.8％ -74％

[0085] This embodiment verifies the significant advantages of the present invention in terms of economic benefits, environmental benefits, and equipment maintenance through specific parameters and comparative data.

[0086] Example 4 of the present invention: Sensitivity analysis of optimized target weights.

[0087] This embodiment, based on all parameters and configurations of Embodiment 3, uniquely changes the weight allocation between the economic objective and the carbon emission objective in the multi-objective optimization algorithm. The weight of the economic objective (maximizing operating revenue) is adjusted from 0.5 to 0.2, and the weight of the carbon emission objective (minimizing CO2 emissions) is increased from 0.5 to 0.8. The optimization algorithm still uses the improved NSGA-III algorithm, with a population size of 100, 200 iterations, rolling optimization with a 5-minute cycle, control commands generated based on a 4-hour prediction window, and continuous simulation for 720 hours (30 days).

[0088] Simulation results show that changes in the optimization objective weights directly and significantly affect the system's operating strategy and output. The system prioritizes energy flow to achieve deep emission reduction, specifically by using more electricity for hydrogen electrolysis and storage during off-peak electricity prices, rather than selling it; and by maintaining the synthesis unit's operation rather than purchasing electricity from the grid when wind and solar power are sufficient, even when the purchased electricity price is normal, in order to reduce the grid's carbon footprint. The operating results are compared with those of Example 1 and Comparative Example 1 in the table below:

[0089]

[0090] Data analysis shows that, compared to the balanced strategy in Example 3, the low-carbon priority strategy in this example further reduces CO2 emissions by 11% (321.8 tons → 286.4 tons), a 36% reduction compared to Comparative Example 3. Although the total operating revenue is lower than that in Example 3 (a decrease of 6.97%), it is still significantly higher than the traditional strategy in Comparative Example 1, with a 9.1% increase in revenue. This result fully demonstrates that the system and method described in this invention possess high flexibility and configurability. By adjusting a single parameter (target weight), the system's operating strategy can smoothly switch between different modes such as "economic priority" and "environmental priority," thereby flexibly adapting to carbon tax policies, environmental regulations, or different operational orientations of enterprises in different regions. This is a disruptive advantage that traditional rigid control strategies simply cannot achieve.

[0091] Embodiment 5 of the present invention: Robustness test of wind and solar power prediction error.

[0092] This embodiment, based on the system and parameters of Embodiment 3, specifically tests the adaptability of the intelligent decision-making system of this invention under extreme conditions of significant deviations in wind and solar power prediction. The experiment employs artificial interference: a Gaussian white noise sequence with a mean of 0 and a standard deviation of 30% is superimposed in real-time onto the wind and solar power prediction values ​​output by the LSTM prediction module, thereby worsening the prediction error from the normal 8.2% to approximately 35.1%, simulating a severe scenario of extremely inaccurate forecasts. The target weights of the optimization algorithm remain 0.5 each for economic efficiency and carbon emissions, the rolling optimization cycle remains 5 minutes, and the continuous simulation lasts for 720 hours.

[0093] Experimental results show that despite severe prediction distortion, the system of this invention exhibits excellent robustness thanks to its high-frequency rolling optimization mechanism and the rapid response capability of the energy storage system. The optimizer rapidly replans the energy flow based on the latest measured data, utilizing the hydrogen storage tank and lithium battery for charge-discharge adjustments on a minute-scale timescale, effectively absorbing most power fluctuations and ensuring stable operation of the downstream synthesis section. Key process parameters are strictly controlled within safe ranges: temperature fluctuations in the ammonia synthesis loop are +4.5℃ / -4.1℃ (setpoint 455℃), and temperature fluctuations in the methanol synthesis loop are +2.9℃ / -2.7℃ (setpoint 235℃). The final product yield volatility is extremely low: ±3.5% for ammonia synthesis and ±4.2% for methanol.

[0094] In contrast, the traditional control strategy in Comparative Example 1 completely failed in this extreme scenario. Due to its lack of forward-looking optimization and rapid coordinated control capabilities, the system experienced three emergency shutdowns caused by severe imbalances in the hydrogen-nitrogen ratio and exceeding reaction temperature limits, resulting in a total downtime of 11.2 hours. Ammonia production fluctuated by ±19.3%, and methanol production fluctuated by ±21.5%, causing significant material and energy losses, compromising both economic efficiency and safety. Specific performance comparisons are shown in the table below:

[0095] Performance indicators The system of this invention (prediction error 35.1%) Comparative Example 1 (Prediction Error 35.1%) Ammonia production volatility ±3.5％ ±19.3％ Methanol production volatility ±4.2％ ±21.5％ Number of times the synthesis reactor is shut down due to interlocking 0 3 Cumulative unplanned downtime (hours) 0 11.2 Maximum temperature fluctuation during ammonia synthesis (°C) +4.5 / -4.1 +22.7 / -18.9

[0096] This embodiment powerfully verifies the core advantages of the intelligent decision-making and dynamic optimization method of the present invention: it can still ensure the safe, stable and efficient operation of the entire system when facing the inherent high uncertainty of renewable energy and the limitations of prediction technology. This is a key reliability guarantee necessary for the commercial success of the wind-solar-hydrogen-ammonia-methanol integrated project.

[0097] Comparative Example 1: Traditional Unit-Based Control Strategy

[0098] This comparative example uses the independent control strategy widely used in the industry as the benchmark, whose control logic involves independent operation of each subsystem and a lack of collaborative optimization. In practice, the wind and solar power generation system operates in a "self-consumption, surplus power fed into the grid" mode, prioritizing the supply of wind and solar power to the plant's loads, feeding excess power into the grid, and purchasing insufficient power from the grid. The electrolyzer system employs simple start-stop control: when the total wind and solar power exceeds the minimum operating power of four electrolyzers (4 * 20% * 1000 Nm³),... 3 / h*4.3kWh / Nm 3 When the power output reaches 3.44MW, all electrolyzers are started and operated at a fixed load rate (80%); otherwise, all are shut down. The ammonia and methanol synthesis sections always operate at full load and constant conditions. The required reaction heat and compression work are supplied by the plant's power grid, and fluctuations are balanced entirely through power exchange with the external power grid. The energy storage system (20MWh lithium battery) is only used to smooth power fluctuations in the grid to meet grid guidelines and does not participate in the plant's energy optimization scheduling.

[0099] After running under this strategy for 720 hours (30 days), the key performance indicators are compared with those of the present invention, as shown in the table below:

[0100] Performance indicators Comparative Example 1 (Traditional Sub-unit Control) The system of the present invention (Example 3) Performance gap Total operating revenue (ten thousand yuan) 1060.2 1243.5 -17.3％ <![CDATA[CO2 emissions (tons)]]> 447.5 321.8 +39.1％ Electrolytic cell start-up and shutdown times 101 18 +461％ Electricity purchased from the grid (MWh) 12520 9280 +34.9％ Electricity sold to the grid (MWh) 15850 13220 -19.9％ Ammonia production volatility ±12.5％ ±2.1％ +495％

[0101] Experimental results show that traditional unit-based control strategies, due to their inability to coordinate and optimize the various links of "source-grid-load-storage," result in poor system operating economy, high carbon footprint, and severe equipment wear and tear. Their core flaw lies in:

[0102] 1) Inefficient energy management prevents the company from selling more electricity when prices are high and producing more hydrogen when prices are low, thus missing arbitrage opportunities;

[0103] 2) Rigid operation mode makes the high-energy-consuming synthesis section a stable "grid load", consuming a large amount of high-carbon grid electricity and increasing carbon emissions;

[0104] 3) Frequent start-ups and shutdowns of the electrolytic cell (more than 3 times per day on average) severely shorten equipment lifespan and increase maintenance costs;

[0105] 4) Lacking a global optimization perspective, energy storage resources are only used to smooth grid connection power, failing to create maximum value for the entire system. This result, from the opposite perspective, fully demonstrates the necessity and significant superiority of the intelligent decision-making and dynamic optimization method proposed in this invention.

[0106] Comparative Example 2: Single-objective optimization based solely on economics

[0107] This comparative example is used to verify the negative consequences of pursuing only economic single-objective optimization while ignoring environmental impact. Its experimental setup is completely identical to Example 3, including the same wind and solar data, equipment parameters, grid electricity prices, and downstream demand. The only change is that the multi-objective optimization algorithm is replaced with a single-objective optimization, i.e., maximizing the system's daily operating revenue is the sole objective function, completely ignoring CO2 emissions. The optimization algorithm uses the same improved NSGA-II framework, but retains only the economic objective, with a population size of 100, 200 iterations, and a rolling optimization cycle of 5 minutes.

[0108] Simulation results show that this single-objective optimization strategy drives the system to operate in an extremely profit-driven manner: during peak electricity prices (1.2 yuan / kWh), it minimizes on-site power consumption and sells all wind and solar power; during off-peak electricity prices (0.3 yuan / kWh), it purchases large amounts of electricity from the grid for water electrolysis to produce hydrogen and maintain the synthesis section, completely disregarding the high carbon content of grid electricity. After 720 hours of simulation operation, the key performance indicators are compared with those of Example 3 and Comparative Example 1 in the following table:

[0109]

[0110] Data analysis shows that while simple economic optimization yielded the highest operating revenue (12.802 million yuan, 3% higher than Example 3), its environmental cost was extremely high. CO2 emissions reached 498.6 tons, 54.9% higher than Example 3, and even 11.4% higher than the traditional control strategy in Comparative Example 1. This is mainly because this strategy purchased large amounts of cheap but high-carbon-intensity grid electricity during off-peak electricity prices, significantly increasing the system's carbon footprint. This result strongly demonstrates that, in the context of energy transition, single-objective economic optimization is no longer feasible. The multi-objective optimization method adopted in this invention can significantly reduce environmental impact while maintaining high economic returns, making it a necessary technical approach to achieve sustainable development and highlighting the comprehensive superiority and necessity of the strategy proposed in this invention.

[0111] The following is a data table comparing the data from each experiment.

[0112] Table 1: Summary of Performance Comparison of Various Solutions in the Baseline Scenario

[0113]

[0114]

[0115] Table 2: Comparison of Robustness under Extreme Disturbance Scenarios

[0116]

Claims

1. A method for intelligent decision-making and dynamic optimization of a wind-solar-hydrogen-amine-ethanol multi-energy system, characterized in that, Includes the following steps: Step S1: Real-time acquisition of wind and solar power forecast data, grid time-of-use electricity price signals, downstream product demand data, and operating status data of each unit within the system; Step S2: Construct a multi-objective optimization model with the dual objectives of maximizing the daily operating revenue of the system and minimizing the carbon dioxide emissions of the system. The weighting factors between the dual objectives can be configured online within a predefined range. Step S3: Using a multi-objective genetic algorithm, the multi-objective optimization model is solved at high frequency based on a rolling time window to generate collaborative control instructions; the collaborative control instructions include electrolyzer load rate instructions, energy storage system charging and discharging power instructions, synthesis section feed ratio instructions, and power trading strategy instructions; Step S4: Send the collaborative control command to the corresponding equipment for execution, and use the hydrogen storage tank and lithium battery energy storage system to perform minute-level real-time power balancing to buffer the impact of wind and solar power fluctuations on downstream chemical processes; The multi-objective genetic algorithm employs a hot-start mechanism, which uses the solution set of the previous optimization cycle as the initial population for the current cycle's algorithm iteration.

2. The method according to claim 1, characterized in that, The rolling time window is optimized by rolling data based on the forecast data for the next 4 hours, with a cycle of 5 minutes.

3. The method according to claim 1 or 2, characterized in that, The multi-objective genetic algorithm is an improved NSGA-II algorithm, and its improvements include at least one of the following: (a) Adaptive crossover and mutation probabilities are used, and their values ​​are dynamically adjusted based on the number of iterations and the degree of aggregation of individuals at the Pareto front. (b) An improved constraint dominance relationship handling rule is adopted, which comprehensively considers the merits of the objective function and the degree of constraint violation when comparing two infeasible solutions; (c) In the elite selection strategy, the smoothness of control instructions is taken into consideration.

4. The method according to claim 1, characterized in that, Following step S4, the following is also included: Step S5: Simulate and verify the intelligent decision-making and dynamic optimization process using a high-fidelity digital twin platform; the digital twin platform integrates multi-physics modeling tools for electrical, chemical, and control systems, and exchanges data in real time via the OPC UA protocol.

5. The method according to claim 1, characterized in that, In the objective function of the multi-objective optimization model, the sum of the weights of the economic objective and the carbon emission objective is 1, and their respective weight configuration ranges are 0.2-0.

8.

6. A smart decision-making and dynamic optimization system for a wind-solar-hydrogen-amine-ethanol multi-energy system, characterized in that, include: The data acquisition and communication module is used to collect real-time wind and solar power forecast data, grid time-of-use electricity price signals, downstream product demand data, and operating status data of each unit in the system. A multi-objective rolling optimization decision module is used to construct a multi-objective optimization model with the dual objectives of maximizing the daily operating revenue of the system and minimizing the carbon dioxide emissions of the system. A multi-objective genetic algorithm is used to perform high-frequency optimization based on a rolling time window to generate collaborative control commands. These collaborative control commands include electrolyzer load rate commands, energy storage system charging and discharging power commands, synthesis section feed ratio commands, and power trading strategy commands. The improved multi-objective genetic algorithm employs a hot-start mechanism. The multi-timescale coordinated control module is used to send the coordinated control commands to the corresponding devices for execution, and to control the hydrogen storage tank and lithium battery energy storage system to perform minute-level real-time power balancing, so as to buffer the impact of wind and solar power fluctuations on downstream chemical processes.

7. The system according to claim 6, characterized in that, Also includes: The digital twin verification module integrates multiphysics modeling tools for electrical, chemical, and control systems. It exchanges data with the multi-objective rolling optimization decision module and the multi-timescale coordinated control module via the OPC UA protocol to perform high-fidelity simulation verification of the system decision-making and control processes.

8. The system according to claim 6, characterized in that, In the multi-objective rolling optimization decision module, the rolling time window is based on the predicted data for the next 4 hours and is rolled with a period of 5 minutes.

9. The system according to claim 6, characterized in that, The multi-objective genetic algorithm is an improved NSGA-II algorithm, and its improvements include at least one of the following: (a) Adaptive crossover and mutation probabilities are used; (b) Adopt improved rules for handling constraint dominance relationships; (c) Incorporate consideration of the smoothness of control instructions into the elite selection strategy.

10. The system according to claim 6, characterized in that, In the objective function of the multi-objective optimization model, the weights of the economic objective and the carbon emission objective are configured within the range of 0.2-0.8.

Images