Full-process optimization method and system for synthesizing ammonia through wind-solar hydrogen production

Abstract

The invention relates to the technical field of renewable energy source and process industrial intelligent control, and discloses a full-process optimization method and system for synthesizing ammonia through wind-solar hydrogen production, which can realize full-process collaboration and self-adaption to wind-solar fluctuation, and can guarantee system safety and economy in long-term operation. According to the scheme, the method comprises the steps that running state data of a wind-solar power generation system, a hydrogen production system, a hydrogen storage system and a synthetic ammonia system are collected in real time, and the collected data are preprocessed; establishing a device-level adaptive boundary model, and calculating a dynamic safe operation boundary of the key device; inputting wind-solar power prediction of a future time domain, performing rolling optimization based on the multi-objective optimization model according to constraint conditions, and generating load setting instructions of each system; and executing the load setting instruction, collecting real response data of the system, comparing a deviation between a real response and a predicted response, calculating a multi-dimensional deviation index, and when the deviation exceeds a set threshold value, performing feedback calibration on the equipment-level adaptive boundary model.

Description

Technical Field

[0001] This invention relates to the field of intelligent control technology for renewable energy and process industries, specifically to a method and system for optimizing the entire process of wind-solar hydrogen production and ammonia synthesis. Background Technology

[0002] Utilizing renewable energy sources such as wind and solar power to generate "green hydrogen" and further synthesize "green ammonia" has become a key path for the green and low-carbon transformation of the energy and chemical industry. Green ammonia is not only an important zero-carbon energy carrier, enabling long-term storage and cross-seasonal allocation of renewable energy, but also a basic chemical raw material with wide application value in agriculture, energy, and other fields. Its synthesis process is of great significance for promoting energy structure transformation and reducing carbon emissions.

[0003] However, the inherent volatility and intermittency of wind and solar power generation present a fundamental contradiction with the high requirements for reaction condition stability and material supply continuity in chemical processes such as ammonia synthesis. Ammonia synthesis, as a typical high-temperature and high-pressure process, demands strict stability of operating conditions for catalyst activity maintenance, reaction heat balance maintenance, and equipment mechanical stress control. Random fluctuations in wind and solar power output can easily disrupt this balance, leading to system operational risks. Currently, the operation and control of wind-solar hydrogen production for ammonia synthesis mainly faces the following technical bottlenecks:

[0004] (1) Isolated control lacks end-to-end coordination:

[0005] During operation, subsystems such as hydrogen production, hydrogen storage, and ammonia synthesis often employ independent control strategies, lacking a holistic perspective. When wind and solar power fluctuate drastically, existing systems cannot optimize load allocation based on the real-time status of each device (such as electrolyzer temperature, compressor fatigue level, storage tank pressure, and synthesis tower catalyst activity). This can easily lead to excessive stress on local equipment, frequent start-ups and shutdowns, affecting overall system efficiency and shortening equipment lifespan.

[0006] (2) Rigid boundary constraints limit system flexibility:

[0007] Existing control systems typically set fixed safe operating boundaries for key equipment (such as the minimum / maximum load rate of electrolyzers and the maximum speed of compressors). This "one-size-fits-all" boundary management approach is too conservative and cannot dynamically adjust operating thresholds based on the real-time health status of the equipment. This results in the inability to fully utilize the equipment's potential when wind and solar power is high, leading to a waste of renewable energy. When wind and solar power is low, there is a lack of flexible adjustment methods to adapt to power fluctuations, ultimately resulting in poor overall system flexibility and limited wind and solar power absorption rate.

[0008] (3) Model mismatch affects long-term operational economics:

[0009] Existing digital twin technologies are mostly used for offline simulation or condition monitoring, failing to achieve deep closed-loop integration with real-time control. Due to factors such as equipment performance degradation, catalyst deactivation, and scaling, model accuracy gradually decreases over time, making it impossible to provide a reliable basis for dynamic optimization control. This leads to discrepancies between optimized commands and actual system responses, resulting in decreased long-term operational economy.

[0010] In summary, existing technologies are insufficient to resolve the core contradiction between the volatility of wind and solar power and the continuity of chemical processes. There is an urgent need for an optimized technology that can achieve full-process coordination, adapt to wind and solar power fluctuations, and ensure system safety and economy in long-term operation, so as to promote the efficient and sustainable development of the wind and solar power hydrogen production and ammonia synthesis industry. Summary of the Invention

[0011] The technical problem to be solved by the present invention is to provide a whole-process optimization method and system for wind and solar hydrogen production and ammonia synthesis, which can achieve whole-process coordination, adapt to wind and solar fluctuations, and ensure system safety and economy in long-term operation.

[0012] The technical solution adopted by the present invention to solve the above-mentioned technical problems is as follows:

[0013] On the one hand, this invention provides a method for optimizing the entire process of wind and solar hydrogen production to synthesize ammonia, including the following steps:

[0014] S1. Real-time acquisition of operating status data from wind and solar power generation systems, hydrogen production systems, hydrogen storage systems, and ammonia synthesis systems, and preprocessing of the acquired data;

[0015] S2. Based on the data preprocessed in step S1, establish an equipment-level adaptive boundary model and calculate the dynamic safe operation boundary of key equipment;

[0016] S3. Input the wind and solar power forecast for the future time domain, and perform rolling optimization based on the multi-objective optimization model according to the constraints to generate load setting instructions for each system;

[0017] S4. Execute the load setting command, collect the actual response data of the system, compare the deviation between the actual response and the predicted response, calculate the multi-dimensional deviation index, and when the deviation exceeds the set threshold, perform feedback calibration on the device-level adaptive boundary model.

[0018] Furthermore, in step S1, the collected operational status data includes: real-time power and probabilistic prediction data of wind and solar power generation, operational status parameters of each device, and environmental condition data.

[0019] Furthermore, step S2 specifically includes:

[0020] S21. Based on the different operating characteristics of key equipment, establish corresponding fatigue damage accumulation models for key equipment; the key equipment includes an electrolyzer, a hydrogen compressor, and ammonia synthesis reactor;

[0021] S22. Set initial safety thresholds for equipment operating parameters, establish a database of coupling relationships between operating parameters and maximum allowable load of equipment, and determine key coefficients of the equipment-level adaptive boundary model;

[0022] S23. Based on the preprocessed data and combined with the fatigue damage accumulation model of key equipment, update the fatigue damage degree of the equipment in real time;

[0023] S24. Calculate the dynamic safety operation boundary model of key equipment for different equipment based on real-time fatigue damage degree and current operating condition parameters;

[0024] S25. Perform look-ahead optimization based on the dynamic safety boundary calculated in step S24 to determine the final dynamic safety operating boundary.

[0025] Furthermore, in step S24, the dynamic operating boundary models of the key equipment include: the safety boundary model of the compressor system, the safety boundary model of the electrolyzer system, and the safety boundary model of the ammonia synthesis reactor.

[0026] The compressor system safety boundary model includes dynamic speed limits and pressure fluctuation safety domains;

[0027] The safety boundary model of the electrolytic cell system includes an adaptive range for current density and a safe operating window for temperature.

[0028] The safety boundary model for the ammonia synthesis reactor includes dynamic limits on operating pressure and safety constraints on temperature gradient.

[0029] Each boundary parameter is dynamically adjusted based on the real-time health status;

[0030] Dynamic speed limit:

[0031] ;

[0032] In the formula, The safe speed represents the maximum permissible safe speed that varies over time. All are weighting coefficients, among which, ; , For bearing temperature; Vibration power; Normalized fatigue damage degree; This refers to the standard vibration power value of the equipment under rated conditions. Rated speed, representing the standard operating speed of the compressor system during its design;

[0033] Pressure fluctuation safety domain:

[0034] ;

[0035] In the formula, Maximum permissible pressure fluctuation; A comprehensive health index; The critical health index; Rated pressure;

[0036] Current density adaptive range:

[0037] ;

[0038] In the formula, For safe current density; All are weighting coefficients, among which ; Real-time temperature; The optimal temperature; The reference current density; As for efficiency loss, among which , For the initial Faraday efficiency, For the current Faraday efficiency;

[0039] Temperature safety operation window:

[0040] ;

[0041] In the formula, Within the safe temperature range; The optimal temperature; This represents the maximum temperature deviation. A comprehensive health index; The critical health index;

[0042] Dynamic limits on operating pressure:

[0043] ;

[0044] In the formula, This refers to the reactor's maximum permissible operating pressure. Design pressure; Initial catalyst activity; This represents the current catalyst activity;

[0045] Temperature safety gradient constraints:

[0046] ;

[0047] In the formula, The maximum temperature gradient; For design lifespan.

[0048] Furthermore, in step S3, the multi-objective optimization model includes: a model of equipment lifespan, wind and solar energy consumption, operational economy, and process stability; a model of equipment lifespan loss; and a model of operating cost.

[0049] Furthermore, in step S3, the constraints include equipment operation constraints and process flow constraints; the equipment operation constraints include the dynamic safe operation boundary and gradient change rate limit; the process flow constraints include material balance, energy balance and gas purity requirements.

[0050] Furthermore, in step S3, the rolling optimization adopts a model predictive control framework; the model predictive control framework includes a state-space model of electrolysis power, stack temperature, hydrogen storage pressure, hydrogen quantity and fatigue damage, with current setting, compressor speed, valve opening degree and cooling power as control variables.

[0051] Furthermore, in step S4, the multi-dimensional deviation index includes instantaneous deviation and cumulative deviation: the instantaneous deviation is quantified using mean absolute error and root mean square error, and the cumulative deviation is quantified using integral absolute error and integral square error.

[0052] Furthermore, in step S4, when setting the threshold, the following is included: designing a multi-condition triggering mechanism, which includes quantitative judgment criteria and qualitative judgment criteria; the quantitative judgment criteria set differentiated thresholds based on error exceeding limits and trend deterioration.

[0053] On the other hand, the present invention also provides a complete process optimization system for wind and solar hydrogen production to synthesize ammonia, comprising:

[0054] Data sensing and edge processing module: used to collect real-time operating status data of wind and solar power generation system, hydrogen production system, hydrogen storage system and ammonia synthesis system, and to preprocess the collected data;

[0055] Dynamic digital twin module: Used to receive data processed by the data sensing and edge processing module, and calculate the dynamic safe operation boundary of key equipment based on the device-level adaptive boundary model established by the data;

[0056] Intelligent Decision-Making and Coordination Control Module: This module takes future wind and solar power forecasts as input, performs rolling optimization based on a multi-objective optimization model according to constraints, generates load setting instructions for each system, simulates the execution of these load setting instructions in the dynamic digital twin module's simulation scenario of the entire wind-solar-hydrogen-ammonia synthesis process, and collects the system's actual response data. By comparing the deviation between the actual response and the model's predicted response, a multi-dimensional deviation index is calculated. When the deviation exceeds a set threshold, the device-level adaptive boundary model is calibrated.

[0057] The beneficial effects of this invention are:

[0058] (1) Resolve the contradiction between the volatility of wind and solar power and the continuity of chemical industry, and improve the stability of system operation:

[0059] This invention addresses the core contradiction between the intermittency and volatility of wind and solar power generation and the high stability requirements of the ammonia synthesis process by constructing a collaborative mechanism of equipment-level adaptive boundary model and multi-objective rolling optimization. On one hand, the equipment-level adaptive boundary model does not use a fixed safety threshold, but dynamically adjusts the operating boundary based on the real-time health status of the equipment (such as fatigue damage and performance degradation), ensuring that the equipment load remains within a safe and stable operating range as wind and solar power changes. On the other hand, the rolling optimization based on model predictive control can plan load allocation in advance using wind and solar power forecast data, combined with the buffer adjustment of the hydrogen storage system, to prevent sudden changes in wind and solar power from directly impacting the high-temperature and high-pressure reaction process of ammonia synthesis. This ensures the stability of process parameters throughout the entire process and reduces problems such as catalyst deactivation and reaction imbalance caused by fluctuations in operating conditions.

[0060] (2) Extend the lifespan of critical equipment:

[0061] This invention introduces a fatigue damage accumulation model based on damage mechanics to achieve real-time quantification and dynamic control of equipment health status. Through the fatigue damage accumulation model, the fatigue degree of key equipment such as electrolytic cells, compressors, and synthesis towers can be tracked in real time. In multi-objective optimization, equipment life loss is given priority as one of the core objectives. When the equipment fatigue damage approaches the critical value, its load stress is automatically reduced. When the equipment health status is good, the adjustment potential is appropriately released to avoid the extreme situation of conservative operation that wastes life or excessive load that shortens life, ultimately extending the service life of the equipment by more than 30%.

[0062] (3) Enhance the capacity for wind and solar energy absorption:

[0063] This invention overcomes the limitations of traditional rigid boundary control by utilizing an equipment-level adaptive boundary model to break away from the "one-size-fits-all" fixed load restrictions. When wind and solar power is sufficient, and if the equipment's health permits, the operating boundary can be dynamically relaxed to fully accommodate surplus green electricity for hydrogen production. When wind and solar power is insufficient, the load can be flexibly adjusted based on the equipment's status, avoiding wind and solar curtailment caused by rigid load requirements. Furthermore, this invention incorporates wind and solar energy utilization into the core optimization objective using a multi-objective optimization model. Under the premise of ensuring process stability and equipment safety, wind and solar power is prioritized for allocation to the hydrogen production stage. Simultaneously, hydrogen produced from surplus green electricity is stored through a hydrogen storage system and used for ammonia synthesis when wind and solar power is insufficient, realizing the transfer and utilization of wind and solar energy. Ultimately, this invention improves wind and solar energy utilization capacity by 5% to 15%.

[0064] (4) Implement model self-learning iteration to ensure long-term operational reliability:

[0065] This invention constructs a closed-loop self-learning mechanism of instruction execution, deviation feedback, and model calibration, effectively solving the "model mismatch" problem caused by equipment performance degradation and catalyst deactivation in traditional digital twin models. During the entire process, the system collects real-time data on the actual response of the equipment after executing load commands and compares it with the model's predicted response. When the deviation exceeds a threshold, the system automatically calibrates the equipment-level adaptive boundary model and the full-process simulation model, updating the model parameters to match the actual performance changes of the equipment. Attached Figure Description

[0066] Figure 1 This is a flowchart of the optimized process for wind and solar hydrogen production to ammonia synthesis in an embodiment of the present invention.

[0067] Figure 2 This is a flowchart illustrating the dynamic safety operation boundary of key equipment in an embodiment of the present invention.

[0068] Figure 3 This is a flowchart illustrating the rolling optimization generation of load setting instructions in an embodiment of the present invention.

[0069] Figure 4 This is a flowchart illustrating the model self-learning process in an embodiment of the present invention.

[0070] Figure 5 This is a system framework diagram of the entire process optimization for wind and solar hydrogen production to synthesize ammonia in an embodiment of the present invention. Detailed Implementation

[0071] This invention aims to provide a method and system for optimizing the entire process of wind-solar hydrogen production and ammonia synthesis, enabling full-process coordination, adaptability to wind and solar fluctuations, and ensuring system safety and economy during long-term operation. Its core idea is to resolve the contradiction between wind and solar fluctuations and chemical continuity, and to achieve dynamic collaborative optimization of the entire process. This is achieved by constructing a technical system of real-time perception, dynamic modeling, rolling optimization, and closed-loop self-learning, breaking through the limitations of isolated control, rigid boundaries, and model mismatch in traditional wind-solar hydrogen production and ammonia synthesis systems.

[0072] In its specific implementation, the present invention achieves the above core idea through the following means:

[0073] (1) Traditional systems rely on fixed safety thresholds to control equipment, which cannot adapt to dynamic changes such as equipment performance degradation and fatigue damage, and are prone to wasting potential or overload risks. This invention abandons this model and introduces a fatigue damage accumulation model based on damage mechanics to quantify the health status of key equipment such as electrolytic cells, compressors, and synthesis towers in real time. Based on this, an equipment-level adaptive boundary model is constructed to dynamically adjust the equipment operating boundary, so that the equipment operation can adapt to the fluctuation of wind and solar power and always match its own health status, shifting from passive protection to active prevention, and laying a solid foundation for safe operation of the entire process.

[0074] (2) In response to the core contradiction between the intermittency of wind and solar power generation and the stability of the ammonia synthesis process, this invention constructs a predictive-planning-regulation collaborative mechanism: on the one hand, wind and solar power prediction data are introduced to perceive the future power change trend in advance; on the other hand, based on the model predictive control framework, equipment life, wind and solar power consumption, process stability, and operating costs are incorporated into the multi-objective optimization model, and load allocation instructions are generated in real time through rolling optimization. When wind and solar power is sufficient, green electricity production is prioritized and the hydrogen storage system is used for buffering; when power is scarce, the load is flexibly adjusted based on the equipment status to avoid directly impacting the high temperature and high pressure process of ammonia synthesis.

[0075] (3) Traditional digital twin models often become inaccurate due to factors such as equipment performance degradation and catalyst deactivation, leading to a disconnect between optimized instructions and actual responses. This invention constructs a closed-loop mechanism of instruction execution-deviation feedback-model calibration. By collecting real-time data on the actual response after the system executes load instructions, comparing and analyzing the model's predicted response, and calculating multi-dimensional deviation indicators, when the deviation exceeds a threshold, the equipment-level adaptive boundary model and the full-process simulation model are automatically calibrated, ensuring that the model always accurately matches the actual state of the system, shifting from static modeling to dynamic iteration, and guaranteeing the effectiveness and reliability of the optimization strategy during long-term operation.

[0076] The present invention will be further described below with reference to the accompanying drawings and embodiments.

[0077] This embodiment first provides a method for optimizing the entire process of wind and solar hydrogen production to synthesize ammonia, see [link to relevant documentation]. Figure 1 The implementation process includes the following steps:

[0078] S1. Real-time acquisition of operating status data from wind and solar power generation systems, hydrogen production systems, hydrogen storage systems, and ammonia synthesis systems, followed by data preprocessing.

[0079] In this step, a multi-source sensor network is deployed at key nodes throughout the process to establish a highly reliable and low-latency data acquisition and communication channel, and a data redundancy and secure transmission mechanism is configured. Then, the sensor network is used to collect system operation status data in real time, including: real-time power and probabilistic prediction data of wind and solar power generation (point prediction and interval prediction), operation status parameters of each device (electrical parameters, mechanical parameters, thermal parameters, etc.), and environmental condition data (ambient temperature, humidity, air pressure, etc.).

[0080] The specific implementation of this step is as follows:

[0081] The first step is to deploy a multi-source sensor network and establish data channels. During the system initialization phase, a multi-source sensor network covering the entire process of wind and solar power generation, electrolytic hydrogen production, hydrogen compression and storage, and ammonia synthesis is constructed. This network adopts a hierarchical distributed architecture design, spatially divided into four levels: site layer, process layer, equipment layer, and product layer. Each level is configured with sensors of different accuracies and sampling frequencies according to monitoring requirements.

[0082] The power monitoring unit employs a 0.2-grade accuracy sensor conforming to IEC 61869, ensuring that the power measurement error is controlled within ±0.2% of full scale. The environmental monitoring unit uses a Class A PT100 temperature sensor, whose calculated temperature uncertainty does not exceed ±0.15℃. For process parameter monitoring, pressure transmitters with corresponding ranges are configured for equipment of different pressure levels: 0-2.5MPa range with an accuracy of ±0.1%FS for the electrolytic cell area; 0-25MPa range with an accuracy of ±0.25%FS for the compressor unit; and 0-30MPa range with an accuracy of ±0.15%FS for the synthesis tower. For mechanical condition monitoring, triaxial vibration sensors covering a frequency range of 5-2000Hz are deployed, accurately capturing changes in the mechanical condition of the equipment.

[0083] The data communication architecture employs a combination of Industrial Ethernet and Time-Sensitive Networking (TSN) to ensure real-time and deterministic data transmission. A precision clock synchronization mechanism is deployed, based on the IEEE 1588 protocol, achieving a time synchronization accuracy of better than 100μs at each sampling point. To enhance system reliability, the communication network uses a redundant ring topology, ensuring that the failure of a single node does not affect overall data transmission.

[0084] In terms of data acquisition and preprocessing, this embodiment ensures the real-time performance, accuracy, and security of the data through a layered design and precise strategies, laying a solid foundation for subsequent dynamic boundary calculations and optimization control.

[0085] (1) Construction of a hierarchical distributed data acquisition network:

[0086] To address the differences in characteristics of various physical quantities such as power, vibration, process parameters, and environmental parameters, a differentiated sampling strategy is adopted, while a unified time-scale system is established to solve the problem of multi-source data synchronization.

[0087] Differentiated sampling rules: Power data is sampled at a high frequency of 1kHz, combined with a 10ms moving average filter to balance real-time performance and smoothness; Vibration data is sampled at a frequency of 5kHz to capture details, and spectral features are extracted through a 20ms window to meet the needs of equipment fault diagnosis; Process parameters are sampled at 100Hz and noise is eliminated by a 100ms digital filter; Environmental parameters are sampled at a low frequency of 1Hz, combined with 1s smoothing to ensure data stability.

[0088] Time synchronization mechanism: A unified time standard system is established based on the IEEE 1588 precision clock protocol to achieve millisecond-level time synchronization of sampling data from various sensors, avoiding data correlation failure due to timing deviations.

[0089] (2) Multidimensional data signal processing:

[0090] A combination of main algorithm and auxiliary mechanisms is used to perform data noise reduction and anomaly repair, ensuring data quality.

[0091] Core filtering algorithm: The core data processing adopts an adaptive Kalman filter algorithm, which establishes a multivariable state-space model to make optimal estimates of key parameters such as temperature, pressure, and voltage, and accurately filters out random interference; at the same time, it dynamically calculates the Kalman gain to adapt to the changing characteristics of parameters.

[0092] Anomaly detection and repair: Combining a dynamic threshold mechanism and the 3σ criterion, abnormal data such as sensor malfunctions and signal mutations are identified through sliding window statistical analysis; for detected abnormal values, linear interpolation or historical similar operating conditions are used to repair them, ensuring the integrity of the dataset.

[0093] (3) Feature engineering and data augmentation:

[0094] Effective information is extracted from both the time and frequency domains to enrich data representation and provide support for subsequent equipment status analysis and model building.

[0095] Temporal feature extraction: For the collected raw data, calculate temporal statistics such as mean, variance, peak value, and kurtosis to capture the overall trend and instantaneous characteristics of parameter changes.

[0096] Frequency domain feature analysis: By using Fourier transform to convert non-stationary data such as vibration to the frequency domain, extract indicators such as characteristic frequency band energy and spectral peak values, and uncover potential fault signals of equipment.

[0097] Simultaneously perform data augmentation processing to improve the generalization ability of subsequent model training by reasonably expanding the effective data dimensions.

[0098] This embodiment balances computational efficiency, data security, and system stability by constructing a hierarchical edge computing architecture and a full-process quality control mechanism.

[0099] Hierarchical edge computing architecture: High-frequency sampled data is processed directly at the edge gateway to reduce transmission latency; tasks such as device model calculation and feature fusion are undertaken by the edge server; and computationally intensive tasks such as complex multi-objective optimization are completed through cloud collaboration to achieve reasonable allocation of computing resources.

[0100] Timing and quality control: Establish a strict timing control system to keep the delay of the entire process from data acquisition to signal processing, feature extraction, and boundary update within a preset range; monitor the calculation accuracy in real time through relative error indicators, track system performance parameters, and ensure the reliability of data processing results.

[0101] Security strategy safeguards: A security scheme combining encrypted transmission and hierarchical authorization is adopted to prevent data leakage and unauthorized operations; when performance abnormalities are detected, the system automatically triggers alarms and activates emergency plans to ensure continuous and stable operation.

[0102] S2. Based on the data preprocessed in step S1, establish a device-level adaptive boundary model and calculate the dynamic safe operation boundary of key equipment:

[0103] In this step, the equipment-level adaptive boundary model is the core of achieving a balance between safety and flexibility. By dynamically adjusting the safe operating range in real time by associating the equipment's fatigue state with operating parameters, it ensures equipment safety while fully releasing its adjustment capabilities. See also Figure 2 The implementation process for establishing an equipment-level adaptive boundary model and calculating the dynamic safe operating boundary of critical equipment is as follows:

[0104] (1) Based on the different working characteristics of key equipment, establish a fatigue damage accumulation model for key equipment:

[0105] For rotating equipment (such as compressors), a fatigue life prediction model based on vibration characteristics is adopted. This model not only considers the traditional Miner linear accumulation theory, but also introduces a continuous damage factor α to characterize the additional damage to the equipment under unstable operating conditions. Vibration feature extraction adopts the frequency band energy weighting method. By analyzing the root mean square value of vibration acceleration in each characteristic frequency band, the equivalent stress σ_eq is calculated, providing accurate input for fatigue damage assessment.

[0106] For electrochemical equipment (such as electrolyzers), performance degradation models primarily consider the effects of current density and temperature on equipment lifespan. The Faraday efficiency degradation function is in exponential form, where the degradation coefficient β is determined through accelerated life testing, with a typical value of [value missing]. The temperature effect correction term uses a quadratic function to ensure that the equipment operates near its optimal temperature range.

[0107] In one exemplary implementation, the compressor fatigue life prediction model is as follows:

[0108] Fatigue damage calculation based on vibration characteristics:

[0109] ;

[0110] In the formula, Let i be the number of cycles at stress level i. The fatigue life corresponding to the stress level. The continuous damage factor is 0.001 to 0.005. This is the equivalent stress.

[0111] Vibration feature extraction:

[0112] ;

[0113] In the formula, Let j be the weighting coefficient for frequency band j. Let J be the root mean square of the vibration acceleration in frequency band j. The characteristic frequency is denoted as .

[0114] In one exemplary implementation, the electrolyzer performance degradation model includes an efficiency degradation function and a temperature effect correction, wherein the efficiency degradation function is as follows:

[0115] ;

[0116] In the formula, For the initial Faraday efficiency, The attenuation coefficient is... For operating current density

[0117] Temperature effect correction:

[0118] .

[0119] (2) Set the initial safety threshold for the equipment operating parameters, establish a database of the coupling relationship between operating parameters and the maximum allowable load of the equipment, and determine the key coefficients of the equipment-level adaptive boundary model.

[0120] (3) Based on the data after prediction processing in step S1, and combined with the equipment fatigue damage accumulation model in step S21, update the equipment fatigue damage degree in real time.

[0121] (4) Calculate the dynamic safety operation boundary model of key equipment based on real-time fatigue damage degree and current working condition parameters; establish a multi-parameter coupled safety factor model for the dynamic safety boundary calculation system.

[0122] Taking a compressor as an example, the calculation of its safe speed range comprehensively considers three key parameters: fatigue damage state, bearing temperature, and vibration power, using weighting coefficients. The degree of influence of each factor is quantified. The safe current density range of the electrolyzer is dynamically adjusted according to the degree of deviation of the real-time temperature from the optimal value and the efficiency degradation, ensuring that the equipment can fully realize its potential while avoiding excessive damage.

[0123] The dynamic operating boundary models for key equipment include safety boundary models for the compressor system, electrolyzer system, and hydrogen synthesis reactor. The compressor system safety boundary model includes dynamic speed limits and pressure fluctuation safety domains. The electrolyzer system safety boundary model includes an adaptive current density range and a safe operating temperature window. The hydrogen synthesis reactor safety boundary model includes dynamic operating pressure limits and temperature gradient safety constraints.

[0124] Dynamic speed limit:

[0125] ;

[0126] In the formula, All are weighting coefficients, among which , , For bearing temperature, For vibration power, Normalized fatigue damage degree; This refers to the standard vibration power value of the equipment under rated conditions. Rated speed, representing the standard operating speed of the compressor system during its design; The safe speed represents the maximum permissible safe speed that varies over time.

[0127] Pressure fluctuation safety domain:

[0128] ;

[0129] In the formula, To assess overall health, This is the critical health index. For rated pressure, This represents the maximum permissible pressure fluctuation.

[0130] Current density adaptive range:

[0131] ;

[0132] In the formula, All are weighting coefficients, among which , For real-time temperature, For optimal temperature, For safe current density, As a reference current density, As for efficiency loss, among which , For the initial Faraday efficiency, This represents the current Faraday efficiency.

[0133] Temperature safety operation window:

[0134] ;

[0135] In the formula, Within a safe temperature range, For optimal temperature, For the maximum temperature deviation, To assess overall health, This is the critical health index.

[0136] Dynamic limits on operating pressure:

[0137] ;

[0138] In the formula, The maximum allowable operating pressure of the reactor. To design pressure, Initial catalyst activity, This represents the current catalyst activity.

[0139] Temperature safety gradient constraints:

[0140] ;

[0141] In the formula, For the maximum temperature gradient, For design lifespan.

[0142] (5) Perform forward optimization based on the calculated dynamic safety boundary to determine the final dynamic safety operation boundary.

[0143] The aforementioned forward optimization is based on the forward adjustment of the predicted boundary, which involves adjusting the boundary in advance through short-term wind and solar power prediction and dynamically adjusting the gradient limit response through load change prediction.

[0144] The short-term power prediction model:

[0145] ;

[0146] Boundary adjustment in advance:

[0147] ;

[0148] Load Change Prediction Response – Gradient Limit Dynamic Adjustment:

[0149] .

[0150] S3. Input the wind and solar power forecast for the future time domain, and based on the constraints, perform rolling optimization using a multi-objective optimization model to generate load setting instructions for each system:

[0151] The core of this step is to solve the problem of optimally allocating load to adapt to wind and solar power fluctuations while ensuring equipment safety and process stability. This is achieved through model predictive control and multi-objective optimization to achieve full-process coordination, balancing safety, efficiency, and economy. For the specific process of rolling optimization, please refer to [link to details]. Figure 3 The implementation process includes the following:

[0152] (1) Construct a hierarchical optimization framework to achieve multi-timescale coordinated control at the second, minute, and hour levels:

[0153] The second-level optimization layer focuses on handling instantaneous fluctuations in wind and solar power, ensuring equipment operation within dynamic safety boundaries through rapid load allocation; the minute-level optimization layer performs load planning based on short-term power forecasts, coordinating the operating status of various subsystems; and the hour-level optimization layer formulates day-ahead operating plans, taking into account equipment maintenance cycles and energy storage. Spatially, a three-tiered coordination mechanism is established from the equipment level to the system level and then to the entire process level to achieve global optimization goals.

[0154] The hierarchical optimization framework design includes time-scale division and spatial-scale coordination. The time-scale division includes: a second-level optimization layer (1 to 10 seconds): handling instantaneous fluctuations in wind and solar power; performing rapid load distribution adjustments; and ensuring equipment operation within safe boundaries. A minute-level optimization layer (1 to 10 minutes): performing load planning based on short-term power forecasts; coordinating the operating status of various subsystems; and optimizing system operating economy. An hourly optimization layer (1 to 24 hours): developing day-ahead operating plans; considering equipment maintenance cycles; and optimizing energy storage charging and discharging strategies.

[0155] The spatial scale coordination includes: equipment level: optimization of individual equipment operation; system level: coordination within subsystems; and full-process level: global optimization across systems.

[0156] (2) Construct a multi-objective optimization function that comprehensively considers equipment lifespan, wind and solar energy integration, operational economy, and process stability:

[0157] A weighted summation method is used to balance the conflicts among various objectives, with equipment lifespan having the highest weight, reflecting a design philosophy centered on equipment health. The equipment lifespan loss model is based on real-time fatigue damage data, and the operating cost model includes energy consumption and maintenance costs. The optimization objective is to find the optimal load allocation scheme while satisfying all constraints.

[0158] First, a comprehensive objective function is constructed, including the main objective items, equipment lifespan, wind and solar energy integration, operational economy, process stability, equipment lifespan loss model, and operating cost model.

[0159] The stability model for the economic operation of the equipment's lifespan and its ability to integrate wind and solar power is as follows:

[0160] ;

[0161] In the formula, Weighted by equipment lifespan. ; To reduce the weight of wind and solar energy absorption, ; Weighted by operating costs, ; For process stability weighting, .

[0162] The equipment lifespan loss model is as follows:

[0163] ;

[0164] The operating cost model is as follows:

[0165] .

[0166] (3) Establish a complete system of constraints, including equipment operation constraints and process flow constraints:

[0167] Equipment operation constraints encompass dynamic safety boundary limits and gradient rate of change limits, with the rate of change constraint dynamically adjusted based on the equipment's health status. Process constraints include stringent material balance, energy balance, and gas purity requirements to ensure stable operation throughout the entire process. All constraints employ a combination of hard and soft constraints to enhance the feasibility of the optimization problem while ensuring safety.

[0168] The dynamic safety boundary constraint is:

[0169] ;

[0170] The gradient rate of change constraint is:

[0171] ;

[0172] Material balance constraints are:

[0173] ;

[0174] The energy balance constraint is:

[0175] ;

[0176] The gas purity constraint is:

[0177] ;

[0178] (4) Rolling optimization is achieved using a Model Predictive Control (MPC) framework:

[0179] A state-space model was established, incorporating key parameters such as electrolysis power, stack temperature, hydrogen storage pressure, hydrogen quantity, and fatigue damage, with current setting, compressor speed, valve opening, and cooling power as control variables. An optimization window of 50 minutes for prediction and 25 minutes for control was designed, and an interior-point solver was used for efficient computation. A hot-start strategy was employed to improve solution efficiency, ensuring that optimization calculations were completed within 30 seconds.

[0180] The MPC controller includes a predictive model, a state vector, and a control vector.

[0181] The prediction model is as follows:

[0182] ;

[0183] The state vector is:

[0184] ;

[0185] The control vector is:

[0186] ;

[0187] The rolling optimization problem is:

[0188] ;

[0189] In the formula, For prediction of the time domain (50 minutes); To control the time domain, (25 minutes); This is the weight matrix. ; To control the weights, .

[0190] The real-time optimization algorithm includes interior-point solver configuration and hotspot initiation strategy; the hotspot initiation strategy includes using the solution of the previous cycle as the initial point; updating the initial value based on sensitivity analysis; and adaptively adjusting the step size parameter.

[0191] (5) Construction of hybrid optimization strategy:

[0192] To address the issues of insufficient global search, slow local convergence, and long computation time in complex multi-objective optimization using traditional single algorithms, this embodiment integrates the advantages of traditional optimization algorithms and intelligent algorithms to construct an efficient hybrid optimization strategy:

[0193] A collaborative mechanism of global search + local fine-tuning is adopted: the genetic algorithm is responsible for traversing the global solution space to quickly lock the approximate range of the optimal solution and avoid getting trapped in local optima; then the particle swarm optimization algorithm is used to perform a fine search in this region to improve the accuracy of the solution.

[0194] Introducing a neural network surrogate model: To address the problem of complex and computationally time-consuming objective functions (such as equipment lifespan loss and operating costs) in multi-objective optimization, machine learning methods are used to approximate the objective function, replacing the complex calculations of traditional mechanistic models and significantly shortening the solution time for a single optimization.

[0195] A reinforcement learning module is added: Based on the optimization experience of the system's long-term operation (such as the optimal decision under different wind and light fluctuation scenarios), continuous learning is carried out to dynamically optimize algorithm parameters (such as the crossover rate of the genetic algorithm and the inertia weight of the particle swarm), improve the adaptability of the strategy to complex working conditions, and enhance the intelligence level of the system.

[0196] (6) Full-process performance monitoring and dynamic adaptation:

[0197] To ensure the effectiveness and real-time nature of the optimization strategy, a comprehensive performance monitoring and dynamic adjustment system should be established to form a closed loop of optimization execution, effect evaluation, and strategy iteration.

[0198] Real-time evaluation of key performance indicators: Real-time monitoring of core indicators such as wind and solar energy absorption rate, equipment life extension rate, unit hydrogen production energy consumption, and synthetic ammonia production stability, to comprehensively quantify the implementation effect of optimization strategies.

[0199] Optimize weight dynamic adjustment: Based on the indicator evaluation results, automatically adapt the weight coefficients of the multi-objective optimization model. If an indicator (such as wind and solar energy absorption rate) deviates from the preset target, immediately increase its corresponding weight to guide the optimization strategy toward that target and avoid neglecting one aspect for another.

[0200] Computational performance optimization and assurance: A parallel computing architecture is adopted to allocate multi-objective optimization computing tasks. Combined with algorithm acceleration technology, the solution time of a single rolling optimization is controlled within the control cycle to ensure the real-time performance of optimization instructions. At the same time, parameters such as computing resource utilization and solution convergence speed are monitored in real time. When a computing anomaly occurs, a degradation plan is automatically activated to ensure stable system operation.

[0201] Key performance indicator (KPI), wind and solar energy absorption rate:

[0202] ;

[0203] Equipment life extension rate:

[0204] ;

[0205] Energy consumption per unit of hydrogen production:

[0206] ;

[0207] The real-time adjustment mechanism includes:

[0208] Weight adaptive adjustment:

[0209] ;

[0210] Constraint relaxation strategy:

[0211] ;

[0212] In terms of computer performance optimization, a parallel computer architecture is adopted, utilizing the sparsity of the Jacobian matrix; Hessian matrix approximation is performed; and real-time linear algebra library optimization is implemented.

[0213] S4. Execute the load setting command, collect the actual response data of the system, compare the deviation between the actual response and the predicted response, calculate the multi-dimensional deviation index, and when the deviation exceeds the set threshold, perform feedback calibration on the device-level adaptive boundary model:

[0214] In this step, closed-loop self-learning ensures that the model always matches the actual system, resolving the optimization failure problem caused by model mismatch. The specific implementation process is as follows:

[0215] (1) Optimize instruction issuance and execution control:

[0216] A three-tiered instruction distribution system is established, comprising a central controller, regional actuators, and local controllers. The central controller receives optimization instructions and performs safety checks; regional actuators parse instructions according to subsystems; and local controllers directly drive the actuators and provide status feedback. The system employs a multi-level priority management mechanism: safety protection instructions are executed immediately (response time <100ms), optimization and adjustment instructions are executed quickly (response time <1s), and setpoint adjustment instructions are executed smoothly (response time <5s).

[0217] The actuators employ a precise control strategy. Electrolytic cell current density control is achieved through a PID algorithm, equipped with a feedforward compensation mechanism, achieving a control accuracy of ±0.5%, and ensuring overshoot is less than 3% during the dynamic response process. Compressor speed regulation combines frequency conversion drive and operating condition compensation, dynamically adjusting the output based on real-time temperature and pressure parameters. Valve positioning utilizes an intelligent control algorithm, incorporating dual compensation for pressure change rate and flow deviation to achieve precise flow regulation.

[0218] The precise control of the electrolytic cell density is achieved through a power converter control model.

[0219] ;

[0220] In the formula, , , All are weighting coefficients , , .

[0221] The compressor speed is precisely controlled using a variable frequency drive control algorithm.

[0222] ;

[0223] The compensation function is:

[0224] ;

[0225] The valve opening degree is precisely adjusted through intelligent valve positioning control.

[0226] ;

[0227] In the formula, , is the pressure change compensation coefficient; , is the flow deviation compensation coefficient.

[0228] (2) Construct a multi-dimensional monitoring system to track command execution accuracy and equipment health status in real time:

[0229] Command tracking error is strictly controlled within 2%, and system response time is ensured to be no more than 3 seconds. Actuator health monitoring is achieved through motor current harmonic analysis, with health indicators required to be maintained above 0.9. A four-level safety interlock protection system is established, from parameter over-limit alarms to emergency shutdown protection, ensuring system safety layer by layer.

[0230] The safety verification logic includes checking the rationality of commands and verifying device status. When the change in control commands exceeds a set threshold or the device health status falls below a safe limit, the system automatically rejects the command or enters protection mode. This mechanism effectively prevents misoperation and device overload, providing a reliable guarantee for stable system operation.

[0231] (3) Collect real system response data:

[0232] A high-performance data acquisition system is deployed, employing a tiered sampling strategy. Fast variables (vibration, current) are acquired at a frequency of 1 kHz, medium-speed variables (pressure, temperature) at a frequency of 100 Hz, and slow variables (concentration, liquid level) at a frequency of 1 Hz. The system achieves millisecond-level time synchronization based on the IEEE 1588 protocol, ensuring the accuracy of data timing.

[0233] The data quality assurance system includes signal integrity checks and sensor health diagnostics. Signal validity is required to reach over 98%, and sensor health status indicators must be greater than 0.95. Feature data extraction focuses on dynamic response characteristics, including key parameters such as rise time, settling time, and overshoot. Real-time calculation of system energy efficiency indicators and equipment operating performance provides quantitative evidence for performance evaluation.

[0234] (4) Data transmission and processing:

[0235] A hybrid communication network architecture is constructed, with an industrial Ethernet backbone achieving high-speed transmission of 1000Mbps, a fieldbus network ensuring real-time performance, and a wireless sensor network covering long-distance monitoring points. Data transmission adopts a multi-protocol parallel mechanism: real-time data is transmitted via the OPC UA protocol, historical data is compressed and transmitted using the MQTT protocol, and alarm data is transmitted with priority.

[0236] The edge computing layer implements a data preprocessing process, including outlier filtering and data alignment. An intelligent compression algorithm achieves an 8:1 compression ratio, with information loss controlled to within 3%, while retaining key feature data intact. This processing mechanism effectively balances data transmission efficiency and information integrity.

[0237] (5) Establish a real-time control performance evaluation system to quantify the control effect by tracking and adjusting performance indicators:

[0238] System stability monitoring employs Lyapunov exponent analysis to ensure the system remains in a stable operating state. The fault diagnosis system, based on residual analysis and trend early warning, promptly detects anomalies.

[0239] The intelligent diagnostic system employs a Bayesian classification algorithm to accurately identify fault modes through feature parameters. The system possesses self-healing capabilities, automatically adjusting control parameters to maintain optimal operating conditions when performance degradation is detected. This closed-loop evaluation mechanism ensures the long-term reliability and stability of the system.

[0240] For the implementation process of model self-learning and updating, please refer to [link / reference]. Figure 4 :

[0241] 1. Compare predicted and actual values ​​in depth:

[0242] A multi-dimensional deviation analysis system is established to comprehensively evaluate the model's prediction accuracy through time series analysis and frequency domain feature comparison. The system calculates point-by-point deviations and sliding window statistics in real time to identify model performance degradation trends. Power spectral density analysis technology is employed to ensure the model's prediction accuracy under dynamic operating conditions.

[0243] 2. Construct a model bias system:

[0244] A multi-level evaluation system incorporating both instantaneous and cumulative biases is constructed. Mean absolute error (MAE) and root mean square error (RMSE) are used to quantify instantaneous prediction accuracy, while integral absolute error (IAE) and integral squared error (ISE) assess long-term cumulative bias. Statistical significance tests are conducted to ensure the scientific validity and reliability of the bias analysis.

[0245] 3. Design calibration triggers intelligent judgment:

[0246] A multi-condition triggering mechanism is designed, combining quantitative and qualitative judgment criteria. Quantitative triggering is based on dual judgments of error exceeding limits and trend deterioration, with differentiated thresholds set for different parameters. Qualitative triggering considers special operating conditions such as changes in operating modes and seasonal variations to ensure the timeliness and effectiveness of calibration. Fuzzy logic and Bayesian decision theory are employed to achieve intelligent triggering judgment.

[0247] First, the quantitative trigger condition is triggered when the error exceeds the limit:

[0248] ;

[0249] The threshold settings include a temperature model: Pressure model: ; Flow model:

[0250] The deterioration of the trend was triggered by:

[0251] ;

[0252] 4. Perform precise calibration of model parameters:

[0253] A parameter identification strategy combining recursive least squares and Kalman filtering is adopted. The recursive algorithm is equipped with a forgetting factor mechanism to ensure the timeliness of parameter estimation; Kalman filtering effectively handles system noise and improves the accuracy of parameter estimation. Simultaneously, machine learning-assisted calibration is introduced, achieving global parameter optimization through neural networks and genetic algorithms.

[0254] 5. Digital twin model parameter update:

[0255] Implement a gradual and safe update strategy, controlling the magnitude of parameter updates through a smoothing factor to avoid model mutations. Establish a comprehensive version control management system to record parameter changes and performance metrics for each update. In multi-model collaborative updates, consider the coupling relationships between parameters to ensure overall model consistency. Strictly adhere to physical constraints and maintain fundamental principles such as energy conservation to ensure the rationality and reliability of the updated model.

[0256] This complete model self-learning and updating system ensures that the digital twin model can continuously track changes in the actual system, providing an accurate model basis for optimized control and realizing continuous self-improvement and performance enhancement of the system.

[0257] The above describes the overall process optimization method for wind-solar hydrogen production and ammonia synthesis. To provide corresponding system support, this embodiment also provides an overall process optimization system for wind-solar hydrogen production and ammonia synthesis, the framework of which can be found in [reference needed]. Figure 5 It includes the following modules:

[0258] The data sensing and edge processing module includes a multi-type sensor network deployed on wind and solar power generation units, electrolyzers, compressors, hydrogen storage tanks, and ammonia synthesis units, as well as an edge computing gateway for data preprocessing; it is used to collect real-time operating status data of wind and solar power generation systems, hydrogen production systems, hydrogen storage systems, and ammonia synthesis systems, and to preprocess the collected data.

[0259] The dynamic digital twin module includes an equipment adaptive boundary model and a full-process simulation model. It receives data processed by the data sensing and edge processing module and calculates the dynamic safe operating boundaries of key equipment based on the equipment-level adaptive boundary model established by the data. The equipment-level adaptive boundary model calculates the upper and / or lower limits of dynamically changing loads for the electrolyzer, compressor, and ammonia synthesis reactor by coupling the equipment fatigue damage model, real-time operating conditions, and short-term predictions. The full-process simulation model combines mechanism modeling with data-driven approaches to simulate the entire process of wind and solar hydrogen production and ammonia synthesis.

[0260] The full-process simulation model combines mechanistic modeling with data-driven approaches, as detailed below:

[0261] First, the dynamic model of the electrolytic hydrogen production system is established based on electrochemical principles:

[0262] The mathematical model of the alkaline electrolyzer includes voltage-current characteristic equations and thermal balance equations. The voltage-current characteristic equations comprehensively consider the effects of reversible voltage, ohmic overpotential, and activation overpotential. The parameters r1, r2, s, t1, t2, and t³ are determined by fitting experimental data. The PEM electrolyzer supplements the proton transport equation, describing the concentration distribution and transport process within the proton exchange membrane.

[0263] The refined model of the ammonia synthesis system focuses on constructing a reactor dynamic model and a catalyst activity model. The reactor dynamic model includes material and energy balance equations, with the reaction rate employing a modified Temkin-Pyzhev equation to account for the impact of internal diffusion limitations on the actual reaction rate. The catalyst activity model adopts an exponential decay form, with the deactivation rate constant k_deact correlated with the poison concentration, enabling accurate prediction of catalyst lifetime.

[0264] The dynamic model of the storage and transportation system establishes the pressure-temperature coupling equation and thermodynamic balance equation of the hydrogen storage tank, taking into account the pressure changes caused by gas entry and exit and the heat exchange process with the environment, providing a basis for the assessment of the system's buffering capacity.

[0265] The voltage-current characteristic is:

[0266] ;

[0267] In the formula, It is a reversible voltage, ranging from 1.23V to 1.48V; All parameters are ohmic resistance parameters; This is the overpotential parameter.

[0268] Heat balance equation:

[0269] ;

[0270] Proton transport equation:

[0271] ;

[0272] Material balance:

[0273] ;

[0274] Energy balance:

[0275] ;

[0276] Activity decay function:

[0277] ;

[0278] Pressure-temperature coupling:

[0279] ;

[0280] Thermodynamic equilibrium:

[0281] ;

[0282] Second, model parameter identification and verification:

[0283] Model parameter identification employs a strategy combining maximum likelihood estimation and least squares. For linear or linearizable model parts, least squares is used for parameter estimation; for more nonlinear model parts, maximum likelihood estimation is used to improve the statistical properties of parameter estimation.

[0284] Model validation establishes a multi-indicator evaluation system, including statistical indicators such as the coefficient of determination and mean relative error (MRE). It requires that the coefficient of determination between the model's predicted values ​​and the actual measured values ​​for each key variable be greater than 0.95, and the MRE be less than 5%, ensuring that the model has sufficient predictive accuracy and engineering applicability.

[0285] The intelligent decision-making and coordination control module includes a dynamic margin coordinator and a model self-learning module. The dynamic margin coordinator is used to input the wind and solar power prediction in the future time domain, perform rolling optimization based on the multi-objective optimization model according to the constraints, generate load setting instructions for each system, execute the load setting instructions, and collect the actual response data of the system. The model self-learning module is used to compare the deviation between the actual response and the model prediction response, calculate multi-dimensional deviation indicators, set thresholds, and perform feedback calibration on the equipment-level adaptive boundary model when the deviation exceeds the threshold.

[0286] In addition, the system also includes a human-machine interface for displaying the system's real-time status, dynamic safety boundaries, optimization instructions, and equipment health assessment reports, as well as actuators, including a hydrogen production system, a hydrogen storage system, and an ammonia synthesis system.

[0287] Although embodiments of the present invention have been described, it will be understood by those skilled in the art that various changes, modifications, substitutions and alterations can be made to these embodiments without departing from the principles and spirit of the present invention, and all such changes and alterations shall not depart from the protection scope of the present invention.

Claims

1. A full-process optimization method for hydrogen production by wind and solar energy and ammonia synthesis, characterized in that, The method comprises the following steps: S1. Real-time acquisition of the operating state data of the wind-solar power generation system, hydrogen production system, hydrogen storage system and ammonia synthesis system, and preprocessing of the acquired data; S2. Based on the preprocessed data of step S1, an equipment-level adaptive boundary model is established, and the dynamic safe operation boundary of the key equipment is calculated; S3. Input the wind-solar power prediction in the future time domain, and based on the constraint condition, a rolling optimization is carried out based on a multi-objective optimization model to generate load setting instructions for each system; S4. Execute the load setting instructions, collect the real response data of the system, compare the deviation between the real response and the predicted response, calculate the multi-dimensional deviation index, and when the deviation exceeds the set threshold, feedback and calibrate the equipment-level adaptive boundary model.

2. The whole process optimization method of wind-solar hydrogen production and ammonia synthesis according to claim 1, characterized in that: In step S1, the acquired operating state data includes real-time power and probabilistic prediction data of wind-solar power generation, operating state parameters of each equipment, and environmental condition data.

3. The whole process optimization method of wind-solar hydrogen production and ammonia synthesis according to claim 1, characterized in that: Step S2 specifically includes: S21. According to the different working characteristics of the key equipment, the corresponding key equipment fatigue damage accumulation model is established; the key equipment includes electrolytic cell, hydrogen compressor and ammonia synthesis reactor; S22. Set the initial safety threshold of the equipment operation parameter, establish the coupling relationship database of the working condition parameter and the maximum allowable load of the equipment, and determine the key coefficient of the equipment-level adaptive boundary model; S23. According to the preprocessed data, the real-time fatigue damage degree of the equipment is updated in combination with the key equipment fatigue damage accumulation model; S24. According to the real-time fatigue damage degree and the current working condition parameter, the key equipment dynamic safe operation boundary model is calculated for different equipment; S25. According to the dynamic safety boundary calculated in step S24, the forward-looking optimization is carried out to determine the final dynamic safe operation boundary.

4. The whole process optimization method of wind-solar hydrogen production and ammonia synthesis according to claim 3, characterized in that: In step S24, the key equipment dynamic operation boundary model includes: compressor system safety boundary model, electrolytic cell system safety boundary model, and ammonia synthesis reactor safety boundary model: The compressor system safety boundary model includes dynamic speed limit and pressure fluctuation safety domain; The electrolytic cell system safety boundary model includes current density adaptive range and temperature safety operation window; The ammonia synthesis reactor safety boundary model includes operating pressure dynamic limit and temperature gradient safety constraint; Each boundary parameter is dynamically adjusted according to the real-time health status; Dynamic speed limit: ； wherein is the maximum safe speed allowed as a function of time; are weight coefficients, wherein ; , is the bearing temperature; is the vibration power; is the normalized fatigue damage degree; is the standard vibration power value of the equipment in the rated state; is the rated speed, which represents the standard operating speed at the time of compressor system design; Pressure fluctuation safety domain: ； In the formula, is the maximum allowable pressure fluctuation; is the comprehensive health index; is the critical health index; is the rated pressure; Current density adaptive range: ； wherein is a safety current density; are weight coefficients, wherein ; is a real-time temperature; is an optimal temperature; is a reference current density; is an efficiency loss, wherein , is an initial Faraday efficiency, is a current Faraday efficiency; Temperature safety operation window: ； wherein, is a safe temperature range; is an optimal temperature; is a maximum temperature deviation; is a comprehensive health index; is a critical health index; Operating pressure dynamic limit: ； wherein Pmax is the maximum allowable operating pressure of the reactor; Pd is the design pressure; P0 is the initial catalyst activity; P is the current catalyst activity; Temperature safety gradient constraint: ； In the formula, is the maximum temperature gradient; is the design life.

5. The whole process optimization method of wind-solar hydrogen production and ammonia synthesis according to claim 1, characterized in that: In step S3, the multi-objective optimization model includes: equipment life wind-solar consumption operation economic stability model, equipment life loss model and operation cost model.

6. The method of claim 5, wherein the constraints in step S3 include equipment operation constraints and process flow constraints; the equipment operation constraints include the dynamic safe operation boundary and the gradient change rate limit; and the process flow constraints include material balance, energy balance, and gas purity requirements.

7. The method of claim 6, wherein the rolling optimization in step S3 uses a model predictive control framework; and the model predictive control framework includes state space models of electrolysis power, stack temperature, hydrogen storage pressure, hydrogen amount, and fatigue damage, with current setpoint, compressor speed, valve opening, and cooling power as control variables.

8. The method of claim 1, wherein the multi-dimensional deviation index in step S4 includes instantaneous deviation and cumulative deviation; the instantaneous deviation is quantified by mean absolute error and root mean square error; and the cumulative deviation is quantified by integral absolute error and integral square error.

9. The method of claim 8, wherein the threshold setting in step S4 includes designing a multi-condition triggering mechanism, which includes quantitative and qualitative judgment criteria; and the quantitative judgment criteria set differentiated thresholds based on error overrun and trend deterioration.

10. The method of claim 9, wherein the multi-condition triggering mechanism includes: a data sensing and edge processing module for real-time acquisition of operation state data of the wind-solar power generation system, the hydrogen production system, the hydrogen storage system, and the ammonia synthesis system, and pre-processing of the acquired data; a dynamic digital twin module for receiving the data processed by the data sensing and edge processing module, and calculating the dynamic safe operation boundary of the key equipment according to the equipment-level adaptive boundary model established based on the data; an intelligent decision-making and coordinated control module for inputting wind-solar power prediction in the future time domain, performing rolling optimization based on a multi-objective optimization model according to the constraints, generating load setpoint instructions for each system, simulating the execution of the load setpoint instructions in the dynamic digital twin module of the wind-solar hydrogen synthesis ammonia process simulation scene, and collecting real response data of the system; 10. A full-process optimization system for wind-solar hydrogen synthesis ammonia, characterized in that, comparing the real response and the model prediction response, calculating the multi-dimensional deviation index, and performing feedback calibration on the equipment-level adaptive boundary model when the deviation exceeds the set threshold.

11. A system for optimizing the whole process of wind-solar hydrogen synthesis ammonia, comprising: a data sensing and edge processing module for real-time acquisition of operation state data of the wind-solar power generation system, the hydrogen production system, the hydrogen storage system, and the ammonia synthesis system, and pre-processing of the acquired data; a dynamic digital twin module for receiving the data processed by the data sensing and edge processing module, and calculating the dynamic safe operation boundary of the key equipment according to the equipment-level adaptive boundary model established based on the data; an intelligent decision-making and coordinated control module for inputting wind-solar power prediction in the future time domain, performing rolling optimization based on a multi-objective optimization model according to the constraints, generating load setpoint instructions for each system, simulating the execution of the load setpoint instructions in the dynamic digital twin module of the wind-solar hydrogen synthesis ammonia process simulation scene, and collecting real response data of the system; comparing the real response and the model prediction response, calculating the multi-dimensional deviation index, and performing feedback calibration on the equipment-level adaptive boundary model when the deviation exceeds the set threshold.

Images