A pumped storage participates in grid frequency modulation control method

Abstract

This invention discloses a method for pumped-storage hydropower participating in grid frequency regulation control. The invention relates to the field of grid dispatch automation technology and includes the following steps: S1: Receiving a total frequency regulation power command issued by the grid dispatch terminal; S2: Obtaining upper and lower reservoir water level data through the power station monitoring system and calculating the real-time operating head of the power station. This method for pumped-storage hydropower participating in grid frequency regulation control effectively solves the adaptability problem of power allocation under dynamic head conditions by combining feedforward optimization of a digital twin model with feedback verification from local sensors of the generating unit. This method can automatically generate a power allocation scheme that balances operational economy and equipment safety when responding to grid frequency regulation commands, avoiding unit failures or entry into unstable operating zones due to exceeding actual capacity, thus improving the reliability of frequency regulation response.

Description

Technical Field

[0001] This invention relates to the field of power grid dispatch automation technology, specifically to a method for pumped storage to participate in power grid frequency regulation control. Background Technology

[0002] In existing technologies, control methods for pumped-storage power plants participating in grid frequency regulation typically focus on improving command response speed, optimizing mode switching processes, or coordinating frequency regulation tasks at different levels. These methods often treat the power plant as a single power unit or employ fixed allocation strategies to distribute total power commands to multiple units. However, the operating head of the power plant is not a constant value but a dynamic parameter that fluctuates continuously with changes in reservoir water level; this crucial physical constraint is often overlooked. The dynamic changes in the operating head directly determine the optimal efficiency range and the actual power output limit of each turbine-generator unit.

[0003] Current control strategies fail to incorporate real-time head change information, leading to power commands issued under non-rated head conditions that may force units to operate in inefficient zones, causing unnecessary energy losses. More seriously, commands may even exceed the safe operating boundaries of the units under the current head, causing frequency regulation tasks to fail and threatening equipment safety. Furthermore, existing multi-unit power allocation strategies are often too rigid, failing to dynamically optimize based on real-time head and individual unit differences, thus failing to achieve optimal overall frequency regulation economy and reliability for the power plant. Therefore, the urgent technical challenge is how to achieve adaptive and collaborative optimization of multi-unit frequency regulation power allocation that takes into account dynamically changing head constraints. Summary of the Invention

[0004] To achieve the above objectives, the present invention provides the following technical solution: a method for pumped storage hydropower to participate in grid frequency regulation control, comprising the following steps:

[0005] S1: Receive the total frequency regulation power command issued by the power grid dispatch terminal;

[0006] S2: Obtain water level data of the upper and lower reservoirs through the power station monitoring system, and calculate the real-time operating head of the power station;

[0007] S3: Input the total frequency regulation power command and the real-time operating head into the power station's digital twin model, and generate multiple power allocation plans through the digital twin model;

[0008] S4: Perform efficiency optimization screening and safety boundary screening on the multiple power allocation schemes to select the optimal power allocation scheme;

[0009] S5: The optimal power allocation plan is sent to each unit controller, and each unit controller performs a feasibility check on the received power allocation based on the actual head data measured locally.

[0010] S6: Each unit controller uploads the verified power recommendation value to the power plant central controller;

[0011] S7: The power plant central controller coordinates and calculates based on the power recommendation values ​​uploaded by each unit to generate the final power allocation instruction;

[0012] S8: Issue the final power allocation command to each unit for execution.

[0013] Preferably, the efficiency optimization screening refers to: calculating the total water consumption of the power station corresponding to each power allocation plan, and screening out a predetermined number of plans with the minimum total water consumption.

[0014] Preferably, the safety boundary screening refers to: detecting whether the allocated power of each unit in each power allocation plan exceeds the power operating range of the unit under the current head, and eliminating plans that exceed the limits.

[0015] Preferably, the feasibility verification includes: each unit controller comparing the received allocated power with the locally calculated maximum allowable power; when the allocated power exceeds the maximum allowable power, adjusting the power recommendation value of the unit to the maximum allowable power.

[0016] Preferably, the locally calculated maximum allowable power is obtained based on the actual head data measured by the unit's inlet water pressure sensor.

[0017] Preferably, the coordination calculation refers to: when there is a deviation between the sum of the power recommendation values ​​uploaded by each unit and the total frequency regulation power command, the power is redistributed according to the proportion of the available power capacity of each unit.

[0018] Preferably, the method further includes: after executing the final power allocation command, collecting the actual operating parameters of each unit and feeding these parameters back to the digital twin model to update the model's operating parameter library.

[0019] Preferably, the actual operating parameters include at least one of the following: unit output, efficiency, and vibration value.

[0020] Preferably, the digital twin model of the power station stores the efficiency characteristic curves and vibration characteristic curves of each unit under different water heads.

[0021] Preferably, each unit controller acquires local head data through a pressure sensor installed in the unit's inlet pipe.

[0022] This invention provides a method for pumped-storage hydropower to participate in grid frequency regulation control. It has the following beneficial effects:

[0023] This method for pumped storage participating in grid frequency regulation control effectively solves the adaptability problem of power allocation under dynamic head conditions by combining feedforward optimization of a digital twin model with feedback verification from local unit sensors. When responding to grid frequency regulation commands, this method can automatically generate a power allocation scheme that balances operational economy and equipment safety, preventing unit failures or entry into unstable operating zones due to exceeding actual capacity, thus improving the reliability of frequency regulation response.

[0024] This method for pumped-storage hydropower participating in grid frequency regulation control, through coordinated computing mechanisms and continuous learning optimization functions, enables the system to bridge the discrepancy between global commands and local execution capabilities. It also continuously self-corrects using historical operating data, making the control strategy increasingly closer to the actual operating characteristics of the power station. This not only enhances the accuracy and adaptability of frequency regulation control under different head conditions but also provides high-level automated decision support for pumped-storage power stations participating in grid frequency regulation, comprehensively improving the intelligence level and economic efficiency of power station operation. Attached Figure Description

[0025] Figure 1 This is a schematic diagram of the system module interaction of a pumped storage hydropower participating in grid frequency regulation control method according to the present invention;

[0026] Figure 2 This is a flowchart illustrating a method for pumped storage to participate in grid frequency regulation control according to the present invention. Detailed Implementation

[0027] The technical solutions of the embodiments of the present invention will be clearly and completely described below with reference to the accompanying drawings. Obviously, the described embodiments are only some embodiments of the present invention, and not all embodiments. Based on the embodiments of the present invention, all other embodiments obtained by those skilled in the art without creative effort are within the scope of protection of the present invention.

[0028] Please see Figure 1 and Figure 2 This invention provides a technical solution: a method for pumped storage to participate in grid frequency regulation control, comprising the following steps:

[0029] S1: Receive the total frequency regulation power command issued by the power grid dispatch terminal;

[0030] S2: Obtain water level data of the upper and lower reservoirs through the power station monitoring system, and calculate the real-time operating head of the power station;

[0031] S3: Input the total frequency regulation power command and real-time operating head into the power station's digital twin model, and generate multiple power allocation plans through simulation using the digital twin model;

[0032] S4: Perform efficiency optimization screening and safety boundary screening on multiple power allocation plans to select the optimal power allocation plan;

[0033] S5: The optimal power allocation plan is sent to the controllers of each unit. Each unit controller verifies the feasibility of the received power allocation based on the actual head data measured locally.

[0034] S6: Each unit controller uploads the verified power recommendation value to the power plant central controller;

[0035] S7: The power plant central controller coordinates and calculates based on the power recommendation values ​​uploaded by each unit to generate the final power allocation instruction;

[0036] S8: Issue the final power allocation command to each unit for execution.

[0037] It should be further explained that, in the specific implementation process, firstly, the power plant's central control system receives the total frequency regulation power command issued from the power grid dispatching terminal. Subsequently, the system collects water level data in real time through water level sensors deployed in the upper and lower reservoirs, and calculates the current real-time operating head of the power plant based on these two water level values. This operating head data and the received total frequency regulation power command are synchronously input into the pre-constructed digital twin model of the power plant.

[0038] Based on the current head conditions, the digital twin model automatically simulates and generates multiple different power allocation plans, each specifying the exact power allocation values ​​for each generating unit within the power station. The model then performs a two-step screening of these plans: first, an efficiency optimization screening is conducted, calculating the potential total water consumption of the power station after each plan is implemented, and selecting several plans with lower total water consumption levels; second, a safety boundary screening is performed, verifying whether the power values ​​allocated to each generating unit in each plan are within the upper and lower limits of the power that the unit can safely operate under the current head, eliminating any plan that causes the generating unit's power to exceed these limits. After these two screening steps, the plan with the best overall performance is selected as the optimal power allocation plan.

[0039] The optimal contingency plan is distributed to the local controllers of each unit. Each local controller uses the actual local head data measured in real time by the pressure sensors connected to it and installed on the unit's inlet water pipeline to verify the feasibility of the received allocated power value. The core of the verification is to calculate the maximum allowable power that the unit can safely output at this moment based on the measured local head. If the distributed value does not exceed this limit, it is recommended to execute the original value; if the distributed value exceeds this limit, the power recommendation value for the unit is adjusted to the calculated maximum allowable power to ensure the safe operation of the unit.

[0040] The power recommendations for all generating units are uploaded back to the power plant's central controller. The central controller aggregates these recommendations and calculates their sum. If the sum matches the total frequency regulation power command issued by the grid, these recommendations are directly adopted to form the final command; if there is a discrepancy, a coordination calculation program is initiated to redistribute the total power value proportionally based on the current actual available power capacity of each generating unit, generating a final power allocation command set that satisfies both the total demand and the actual capacity of each generating unit.

[0041] Ultimately, the instruction set was distributed to each generating unit for execution, thus completing this grid frequency regulation task. Throughout the process, the digital twin model and local sensors constituted a dual guarantee mechanism, ensuring the economy, safety, and reliability of power allocation under dynamic head conditions.

[0042] Efficiency optimization screening refers to calculating the total water consumption of the power plant corresponding to each power allocation plan and selecting the plan with the minimum total water consumption. It should be further explained that, in the specific implementation process, during the efficiency optimization screening step, the power plant's digital twin model evaluates each of the generated power allocation plans. The model calls upon its internally stored data on the unit water consumption rate characteristics of each unit under different water heads, and combines this with the current real-time operating water head to calculate the total water consumption required by all units in the entire station to achieve their allocated power for each plan. After completing the total water consumption calculation for all plans, the model sorts these plans in ascending order of total water consumption value.

[0043] Subsequently, the system selects a predetermined subset of plans with the lowest total water consumption from the ranking results and outputs them as candidates for the subsequent safety boundary screening stage. This screening mechanism ensures that the finally selected power allocation scheme can maximize water conservation and improve the economic efficiency of power plant operation while meeting the grid frequency regulation requirements.

[0044] Safety boundary screening refers to detecting whether the allocated power of each unit in each power allocation plan exceeds the unit's operating power range under the current head, and eliminating plans that exceed these limits. It should be further explained that, in the specific implementation process, during the safety boundary screening step, the digital twin model performs a plan-by-plan and unit-by-unit operational safety assessment on the candidate plans selected through efficiency optimization screening. For each plan to be inspected, the model, based on the current real-time operating head, calls upon its embedded safe operation database for each unit. This database defines the upper and lower power limits for each unit to operate stably under different heads, forming a dynamic safe operation corridor.

[0045] The model compares the power value allocated to each unit in the contingency plan with the upper and lower power limits of that unit obtained based on the current head. If the allocated power value of any unit is found to be higher than its upper power limit or lower than its lower power limit, the contingency plan is determined to have a safety risk, and it is then marked and removed from the candidate sequence.

[0046] This screening process ensures that all power allocation plans entering the final decision-making stage have command values ​​within the actual physical capabilities of each unit under the current head conditions. This avoids potential unit instability or equipment protection actions caused by commands exceeding limits, thus guaranteeing the safety and reliability of the frequency regulation task execution process.

[0047] Feasibility verification includes: each unit controller comparing the received allocated power with the locally calculated maximum allowable power. When the allocated power exceeds the maximum allowable power, the recommended power value for that unit is adjusted to the maximum allowable power. It should be further noted that in the specific implementation process, during the feasibility verification step, each unit controller immediately initiates its local verification procedure after receiving the optimal power allocation plan from the central controller. The controller first reads the local head data collected in real time by the pressure sensor directly connected to it and installed on the unit's inlet pipe. This data reflects the immediate operating conditions of the flow channel where the unit is located. Based on this measured head, the controller calls upon its internally stored unit performance curves to calculate the maximum allowable power value at which the unit can operate safely and stably under the current head.

[0048] Subsequently, the controller compares the received planned power allocation with the maximum allowable power value calculated by the machine. If the planned power allocation value does not exceed the maximum allowable power, the controller accepts the instruction and prepares to upload it as the machine's power suggestion value; if the planned power allocation value exceeds the maximum allowable power, the controller will no longer use the issued instruction value, but will set the machine's power suggestion value to the calculated maximum allowable power value.

[0049] This localized verification mechanism forms the second line of defense for system security. It effectively compensates for the local deviations that may exist in the central model based on global head prediction, ensuring that the final execution instructions issued to each unit match its real, instantaneous physical operating environment, thereby guaranteeing equipment safety and the reliability of frequency regulation response.

[0050] The locally calculated maximum allowable power is derived from the actual head data measured by the unit's inlet water pressure sensor. It's important to further clarify that, in the specific implementation process, the core basis for calculating the local maximum allowable power comes from the real-time measurement data provided by the unit's inlet water pressure sensor. This sensor is directly installed on the pressure steel pipe before the turbine inlet and can sensitively capture and transmit instantaneous changes in water pressure within the pipe. The control system continuously reads this pressure signal and, based on fluid dynamics principles, converts it into a precise head value representing the unit's current actual operating environment.

[0051] This head data, based on direct physical measurements, can more accurately characterize the actual operating conditions of a specific unit compared to the power plant-level average head calculation, especially reflecting the head differences caused by hydraulic losses in the water intake pipeline or local flow regime changes.

[0052] Each unit controller uses this real-time, local head data to query or calculate the maximum safe output value in the performance envelope of the unit under the current conditions.

[0053] This mechanism ensures that the power limit setting is based on the most direct and reliable field sensor information, providing a solid and accurate basis for subsequent power command feasibility verification. It effectively avoids the risk of unit overload or operation in the vibration zone that may be caused by head measurement deviation, and is a key foundation for achieving high-reliability frequency regulation control.

[0054] Coordination calculation refers to the redistribution of power according to the available power capacity of each unit when the sum of the power recommendations uploaded by each unit deviates from the total frequency regulation power command. It should be further explained that in the specific implementation process, during the coordination calculation step, the central controller collects the actual power recommendations reported by each unit and calculates their sum. If this sum equals the total frequency regulation power command issued by the grid, the recommendations are directly adopted to form the final command. If a deviation is detected between the sum and the total command, an arbitration and redistribution procedure is initiated. The controller first calculates the specific value of the deviation, and then queries the maximum allowable power value actually reported by each unit after completing local verification, calculated based on its local head, to use as the current actual available power capacity of each unit.

[0055] The controller allocates the existing power deviation value to each unit according to the proportion of each unit's available power capacity to the total available capacity. For units that need to increase power, the allocation is increased based on their recommended value, but the total increase must not exceed the maximum allowable power reported by the unit; for units that need to reduce power, the reduction is made based on their recommended value.

[0056] In this way, a new final power allocation instruction set is generated that fully responds to the total grid demand and strictly follows the instantaneous operating capacity constraints of each unit, thereby ensuring that the frequency regulation task can be executed safely and completely.

[0057] The method also includes: after executing the final power allocation command, collecting the actual operating parameters of each unit and feeding these parameters back to the digital twin model to update the model's operating parameter library. It should be further explained that, in the specific implementation process, during the learning and optimization step, after the final power allocation command set is issued and executed by each unit, the system initiates a continuous data acquisition and model optimization closed loop. The power plant monitoring system continuously collects the actual operating parameters of each unit during the execution of frequency regulation tasks. These parameters include at least the actual output, operating efficiency, and vibration monitoring values ​​of key components. The massive amount of actual operating data collected is transmitted to the data processing center of the power plant's digital twin model. The model does not simply store this data, but rather performs in-depth comparison and analysis with the predicted data used in the previous simulation to generate the power allocation plan and the expected operating state.

[0058] Through this comparison, the model can identify any systematic deviations or evolutionary trends between its internally stored efficiency characteristic curves, vibration characteristic curves, and even the definition of safe operating boundaries and the actual situation. Based on these analysis results, the model initiates a self-learning and adaptive update process to fine-tune and correct the relevant curves and boundary parameters in its kernel database, enabling these key model parameters to approximate the real, dynamically changing operating characteristics of the power plant unit.

[0059] This process enables the prediction accuracy and simulation reliability of the digital twin model to continuously improve with the accumulation of system operating time, forming a smart optimization cycle that gets smarter the more it is used. This provides increasingly accurate basis for subsequent frequency regulation control decisions, fundamentally improving the overall economy and long-term operational safety of the power plant's frequency regulation response.

[0060] Actual operating parameters include at least one of the following: unit output, efficiency, and vibration value. It should be further noted that during implementation, the actual operating parameters involved in data acquisition and feedback are carefully selected to comprehensively reflect the unit's operating status. These parameters mainly include the unit's actual output, used to directly compare its compliance with commands; the unit's operating efficiency, calculated by monitoring data such as inlet and outlet water pressure, flow rate, and rotational speed, used to assess its economic performance under current head and load; and vibration monitoring values ​​of key mechanical components such as bearings and main shafts, used to determine operational stability and potential risks. These parameters are not used independently but are integrated into a multi-dimensional operating status assessment dataset.

[0061] The system continuously acquires this data through a sensor network deployed throughout the unit, and after each frequency modulation task cycle, it transmits the summarized data packets to the digital twin model. Upon receiving the data, the model does not simply archive it, but initiates a deep analysis process. This process compares the actual collected efficiency values ​​with the predicted values ​​of the original efficiency characteristic curves in the model under that operating condition, and matches and verifies the actual vibration spectrum with the original vibration safety boundary database in the model.

[0062] By identifying systematic deviations or trend changes, the model can autonomously calibrate and update the characteristic curves and safety thresholds of its core, such as fine-tuning the optimal efficiency point under a specific head or converging the allowable range of safe vibrations. This allows the virtual mapping of the digital twin to continuously closely approximate the real state of the physical entity, providing a more accurate basis for subsequent predictions and decisions.

[0063] The power plant's digital twin model stores efficiency and vibration characteristic curves for each unit under different water heads. It's important to further clarify that, in practical implementation, the core feature of the power plant's digital twin model lies in its embedded unit characteristic database. This database does not store static design parameters but rather contains dynamic characteristic curves formed through continuous learning and updating using historical operating data. Among these, the efficiency characteristic curve records the correspondence between the output power and operating efficiency of each unit under different operating water heads. This curve reflects the movement of the unit's optimal efficiency point as the water head changes, providing crucial input for early efficiency optimization screening.

[0064] The vibration characteristic curve defines the mapping relationship between the vibration level of the key mechanical components of each unit and its operating status under different head and load conditions. This curve is used to identify the safe operating area and potential vibration risk area of ​​the unit and is the core basis for performing safety boundary screening.

[0065] These characteristic curves together form the knowledge base for digital twin models to perform simulation, prediction, and optimization decisions. This enables them to accurately simulate the operating behavior of power plants in complex environments with dynamic changes in water head, thereby generating economical and safe power allocation plans. This effectively solves the problem of reduced frequency regulation performance caused by neglecting the dynamic influence of water head in traditional control methods.

[0066] Each unit controller acquires local head data via pressure sensors installed on the unit's inlet pipe. It should be further noted that, in practice, each unit controller acquires local head data through pressure sensors directly installed at specific locations on each unit's inlet pipe. These sensors monitor the fluid pressure within the pipeline in real time and transmit continuous analog signals to the unit controller's data acquisition module.

[0067] The controller's built-in signal processing unit filters and standardizes the raw pressure signal to eliminate measurement noise and transient fluctuations, resulting in a stable pressure measurement value. Subsequently, based on the principles of hydrostatics, the controller converts this pressure value into a corresponding head value, which represents the actual energy potential of the water flow when it reaches the turbine of the unit.

[0068] This localized measurement method ensures the real-time nature and accuracy of head data, effectively avoiding the deviations that may occur when relying solely on the average head of the power station. It provides the most direct and reliable field data support for subsequent local verification of power commands and is the foundation for the effective operation of the entire distributed security verification mechanism.

[0069] It should be further explained that, during the implementation process, the power plant's central control system continuously monitors and maintains communication with the power grid dispatch center. Upon receiving the total frequency regulation power command issued by the power grid dispatch center, it immediately initiates the control process. The system first collects real-time reservoir water level data through level sensors or gauges deployed in the upper and lower reservoirs, and calculates the current real-time operating head of the power plant based on the elevation difference between these two water levels. This head value is a key parameter characterizing the potential energy of the water flow.

[0070] Subsequently, the total frequency regulation power command and the calculated real-time operating head are synchronously input into a pre-built and continuously updated digital twin model of the power plant. This digital twin model is a dynamic mapping of the physical power plant in virtual space, and it stores characteristic curves of each unit obtained through continuous self-learning based on historical operating data. These curves mainly include efficiency characteristic curves and vibration characteristic curves. The efficiency characteristic curves describe the relationship between unit output power and operating efficiency under different heads; the vibration characteristic curves define the vibration safety thresholds of key mechanical components of the unit under different combinations of heads and loads.

[0071] The digital twin model, based on the input current water head, simulates and generates multiple possible power allocation plans within the station. Each plan specifies the detailed power allocation values ​​for each generating unit within the station. The model then screens these plans for execution order. First, an efficiency optimization screening is performed, which involves calling efficiency characteristic curves to calculate the total water consumption required for all generating units to complete their allocated power for each plan. Based on the total water consumption, these plans are ranked, and several plans with lower total water consumption levels are selected for the next round of screening.

[0072] Next, a safety boundary screening is conducted. The model uses vibration characteristic curves and unit power limit data to verify whether the power value allocated to any unit in each candidate plan is within the range defined by the upper and lower limits of the power that the unit can safely operate under the current head. Simultaneously, it assesses whether the plan would cause the unit to enter an abnormal vibration zone. Any plan posing a risk of exceeding these limits is immediately eliminated. After these two rounds of screening, the plan with the best overall performance is selected as the optimal power allocation plan.

[0073] The central controller distributes this optimal contingency plan to the local controllers of each unit. Upon receiving the allocation command, each local controller does not execute it directly, but instead initiates a local feasibility verification procedure. The core of this procedure relies on a pressure sensor directly connected to it and installed on the unit's inlet pressure pipe. This sensor senses the fluid pressure within the pipe in real time. The controller, through its built-in signal processing unit, filters and standardizes the raw pressure signal, converting it into a local head value representing the actual energy potential at the unit's inlet, based on the principles of hydrostatics.

[0074] The controller then uses this measured head value to query or calculate the maximum permissible power value that allows the unit to operate safely and stably under the current head. Next, the controller compares the received power allocation command with this maximum permissible power value. If the command value does not exceed this limit, the command is accepted as a suggested value; if the command value exceeds this limit, for safety reasons, the suggested power value of the unit is adjusted to the calculated maximum permissible power value.

[0075] Each generating unit controller uploads its power recommendations back to the central controller of the power plant. The central controller aggregates all recommendations and calculates their sum. This sum is then compared to the total frequency regulation power command issued by the grid. If they match perfectly, the recommendations are adopted to form the final power allocation command. If a deviation is detected, a coordination calculation procedure is initiated. The controller first calculates the deviation, then queries the maximum permissible power value reported by each generating unit, using this as the available power capacity for each unit. Subsequently, the controller redistributes the existing power deviations proportionally to each generating unit based on the proportion of each unit's available power capacity to the total capacity.

[0076] For generating units that need to increase output, the allocation is increased based on their recommended values, but the total increase is ensured not to exceed their maximum allowable power. For generating units that need to reduce output, the allocation is reduced based on their recommended values. Through this coordination process, a final power allocation instruction set is generated that fully meets the total grid demand while strictly adhering to the instantaneous operating capacity constraints of each generating unit.

[0077] Ultimately, the instruction set was distributed to each generating unit for execution, completing the grid frequency regulation task. After task execution, the system entered the learning and optimization phase. The power plant monitoring system collected actual operating parameters of each unit during the execution process, such as actual output, calculated efficiency, and vibration values, and fed these actual data packets back to the digital twin model. The model performed a deep comparison and analysis of the actual data with the data used in previous predictions and the expected state, identifying whether there were systematic deviations in its internal characteristic curves or safety boundaries. Based on this analysis, the model initiated a self-learning process, fine-tuning and correcting the efficiency characteristic curves and vibration characteristic curves in its kernel database, enabling it to better approximate the actual operating characteristics of the physical units, thereby continuously improving the accuracy of subsequent predictions and decisions.

[0078] This method combines feedforward optimization of the power plant-level digital twin model with feedback verification of the unit-level local sensors, and is supplemented by a central coordination arbitration and continuous learning evolution mechanism. Together, these methods ensure that pumped storage power plants can achieve the optimal allocation of frequency regulation power between economy, safety and reliability when facing dynamically changing head conditions.

[0079] By combining feedforward optimization using a digital twin model with feedback verification from local unit sensors, the adaptability problem of power allocation under dynamic head conditions is effectively solved. This method can automatically generate a power allocation scheme that balances operational economy and equipment safety when responding to grid frequency regulation commands, preventing the unit from failing to execute due to exceeding its actual capacity or entering an unstable operating range, thus improving the reliability of frequency regulation response.

[0080] By coordinating computational mechanisms and continuous learning optimization functions, the system can bridge the gap between global commands and local execution capabilities, and continuously self-correct using historical operating data, making the control strategy increasingly closer to the actual operating characteristics of the power station. This not only enhances the accuracy and adaptability of frequency regulation control under different head conditions, but also provides high-level automated decision support for pumped storage power stations to participate in grid frequency regulation, thereby improving the overall intelligence level and economic efficiency of power station operation.

[0081] It should be noted that, in this document, relational terms such as "first" and "second" are used only to distinguish one entity or operation from another, and do not necessarily require or imply any such actual relationship or order between these entities or operations. Furthermore, the terms "comprising," "including," or any other variations thereof are intended to cover non-exclusive inclusion, such that a process, method, article, or apparatus that comprises a list of elements includes not only those elements but also other elements not expressly listed, or elements inherent to such a process, method, article, or apparatus. Without further limitations, an element defined by the phrase "comprising one..." does not exclude the presence of other identical elements in the process, method, article, or apparatus that includes said element.

[0082] Although embodiments of the invention have been shown and 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 invention, the scope of which is defined by the appended claims and their equivalents.

Claims

1. A method for pumped storage hydropower participating in grid frequency regulation control, characterized in that, Includes the following steps: S1: Receive the total frequency regulation power command issued by the power grid dispatch terminal; S2: Obtain water level data of the upper and lower reservoirs through the power station monitoring system, and calculate the real-time operating head of the power station; S3: Input the total frequency regulation power command and the real-time operating head into the power station's digital twin model, and generate multiple power allocation plans through the digital twin model; S4: Perform efficiency optimization screening and safety boundary screening on the multiple power allocation schemes to select the optimal power allocation scheme; S5: The optimal power allocation plan is sent to each unit controller, and each unit controller performs a feasibility verification on the received power allocation based on the actual head data measured locally. S6: Each unit controller uploads the verified power recommendation value to the power plant central controller; S7: The power plant central controller coordinates and calculates based on the power recommendation values ​​uploaded by each unit to generate the final power allocation command; S8: Issue the final power allocation command to each unit for execution.

2. The method for pumped storage participating in grid frequency regulation control according to claim 1, characterized in that: The efficiency optimization screening refers to: calculating the total water consumption of the power station corresponding to each power allocation plan, and screening out the plan with the minimum total water consumption.

3. The method for pumped storage participating in grid frequency regulation control according to claim 1, characterized in that: The safety boundary screening refers to detecting whether the allocated power of each unit in each power allocation plan exceeds the power operating range of that unit under the current head, and eliminating plans that exceed the limits.

4. The method for pumped storage participating in grid frequency regulation control according to claim 1, characterized in that: The feasibility verification includes: each unit controller comparing the received allocated power with the locally calculated maximum allowable power; when the allocated power exceeds the maximum allowable power, the power recommendation value of the unit is adjusted to the maximum allowable power.

5. A method for pumped storage participating in grid frequency regulation control according to claim 4, characterized in that: The maximum allowable power calculated locally is obtained based on the actual head data measured by the unit's inlet water pressure sensor.

6. The method for pumped storage participating in grid frequency regulation control according to claim 1, characterized in that: The coordination calculation refers to the redistribution of power according to the proportion of available power capacity of each unit when the sum of the power recommendation values ​​uploaded by each unit deviates from the total frequency regulation power command.

7. The method for pumped storage participating in grid frequency regulation control according to claim 1, characterized in that: The method further includes: after executing the final power allocation command, collecting the actual operating parameters of each unit and feeding these parameters back to the digital twin model to update the model's operating parameter library.

8. A method for pumped storage participating in grid frequency regulation control according to claim 7, characterized in that: The actual operating parameters include at least one of the following: unit output, efficiency, and vibration value.

9. A method for pumped storage participating in grid frequency regulation control according to claim 1, characterized in that: The digital twin model of the power station stores the efficiency characteristic curves and vibration characteristic curves of each unit under different water heads.

10. A method for pumped storage participating in grid frequency regulation control according to claim 1, characterized in that: Each unit controller obtains local head data through pressure sensors installed on the unit's inlet pipe.

Images