Primary frequency modulation method and system for photovoltaic power station

Abstract

The embodiment of the invention provides a photovoltaic power station primary frequency modulation method and system, and belongs to the field of photovoltaic power station frequency modulation. The method comprises the following steps: collecting multi-source frequency modulation data of a photovoltaic power station, and preprocessing the multi-source frequency modulation data of the photovoltaic power station; according to the preprocessed multi-source frequency modulation data of the photovoltaic power station, performing frequency deviation grading on power grid frequency fluctuation of a grid-connected point of the photovoltaic power station; dynamically distributing frequency modulation contribution weights of the equipment by using an asymmetric game algorithm based on the frequency deviation classification; generating a topology instruction based on the frequency deviation grading result and the equipment frequency modulation contribution weight; and controlling equipment to execute primary frequency modulation based on the frequency modulation contribution weight and the topology instruction. According to the invention, the primary frequency modulation performance of the photovoltaic power station in a high-proportion renewable energy power grid is obviously improved, and economical efficiency and equipment safety are both considered while the frequency stability is ensured.

Description

Technical Field

[0001] The present invention relates to the technical field of frequency modulation of photovoltaic power stations, and particularly relates to a primary frequency modulation method and system for a photovoltaic power station. Background Art

[0002] With the continuous increase in the penetration rate of renewable energy in the power system, the large-scale grid connection of photovoltaic power stations has posed new challenges to the grid frequency stability. The primary frequency modulation of traditional power systems mainly relies on the inertial response and droop control mechanism of synchronous generator sets. This mechanism quickly responds to grid frequency changes through the storage and release of the kinetic energy of the generator rotor. However, photovoltaic power stations are connected to the grid through power electronic interfaces, and their power generation units lack rotational inertia and cannot provide natural inertial support like traditional synchronous generators. This characteristic makes photovoltaic power stations unable to directly participate in frequency regulation during grid frequency fluctuations and may even exacerbate the frequency stability problems of the system. Currently, the participation of photovoltaic power stations in primary frequency modulation mainly relies on the active power control strategy of inverters. Common technical routes include fixed droop coefficient control, virtual synchronous generator (VSG) technology, and dynamic regulation strategies based on frequency-power curves. Although the fixed droop coefficient control is simple to implement, its static characteristics are difficult to adapt to the dynamic operation requirements of the grid; the VSG technology provides virtual inertia by simulating the operating characteristics of synchronous generators, but its control parameters often need to be manually tuned and its adaptability under different operating conditions is limited; although the dynamic regulation strategy based on the frequency-power curve can achieve a certain degree of adaptive control, there are still defects of regulation lag when dealing with rapid and large-amplitude frequency fluctuations. In addition, with the expansion of the scale of photovoltaic power stations and the enhancement of their distributed characteristics, the existing centralized control architecture also faces severe challenges in multi-source collaborative optimization and it is difficult to achieve the optimal cooperation of each frequency modulation unit. Especially in the grid environment with a high proportion of renewable energy access, due to the intermittency and volatility of photovoltaic power output, the random disturbances of the grid frequency are more frequent and the amplitude is larger, which makes it more difficult to balance the response speed, regulation accuracy, and economy of traditional frequency modulation strategies. Existing methods often can only be optimized for specific scenarios, lack universality for complex and variable operating conditions, and are difficult to meet the higher requirements of modern power systems for frequency regulation performance. Summary of the Invention

[0003] The purpose of the embodiments of the present invention is to provide a primary frequency modulation method and system for a photovoltaic power station, which is used to improve the primary frequency modulation performance of the photovoltaic power station in a power grid with a high proportion of renewable energy, taking into account economy and equipment safety while ensuring frequency stability.

[0004] To achieve the above object, an embodiment of the present invention provides a primary frequency modulation method for a photovoltaic power station, including: collecting multi-source frequency modulation data of the photovoltaic power station and preprocessing the multi-source frequency modulation data of the photovoltaic power station; performing frequency deviation classification on the grid frequency fluctuation at the grid connection point of the photovoltaic power station according to the preprocessed multi-source frequency modulation data of the photovoltaic power station; dynamically allocating the frequency modulation contribution weights of devices using an asymmetric game algorithm based on the frequency deviation classification; generating a topology instruction based on the result of the frequency deviation classification and the frequency modulation contribution weights of the devices; and controlling the devices to perform primary frequency modulation based on the frequency modulation contribution weights and the topology instruction.

[0005] Optionally, the performing frequency deviation classification on the grid frequency fluctuation at the grid connection point of the photovoltaic power station includes: setting a first frequency deviation threshold and a second frequency deviation threshold; when the absolute value of the frequency deviation value ≤ the first frequency deviation threshold, determining that the response level is level one; when the first frequency deviation threshold < the absolute value of the frequency deviation value < the second frequency deviation threshold, determining that the response level is level two; and when the absolute value of the frequency deviation value ≥ the second frequency deviation threshold, determining that the response level is level three.

[0006] Optionally, dynamically allocating the frequency modulation contribution weights of devices using an asymmetric game algorithm based on the frequency deviation classification includes: when the frequency deviation first exceeds the dead zone threshold, triggering the asymmetric game algorithm, and the asymmetric game algorithm includes: defining an objective function; calculating the initial weights of each device based on the frequency deviation classification; taking the derivative of the objective function and iteratively updating the frequency modulation contribution weights of the devices using the gradient descent method.

[0007] Optionally, the defining the objective function includes: defining a device utility function and optimizing the objective function by introducing the Lagrange multiplier method into the device utility function; and the calculating the initial weights of each device based on the frequency deviation classification includes: determining the frequency deviation response level, selecting a corresponding response level correction coefficient according to the frequency deviation response level, calculating the initial weights of each device, and verifying whether the sum of the initial weights meets the conditions.

[0008] Optionally, optimizing the objective function by introducing the Lagrange multiplier method into the device utility function includes that the value of the Lagrange multiplier in the Lagrange multiplier method is set corresponding to the frequency deviation response level.

[0009] Optionally, generating a topology instruction based on the result of the frequency deviation classification and the frequency modulation contribution weights of the devices includes: using the preprocessed multi-source frequency modulation data of the photovoltaic power station to define the state input of the DRL model; designing the action space of the DRL model, where the action space includes physical topology switching and virtual cluster division; designing the reward function of the DRL model, training the DRL model using the training set, and verifying the DRL model using the verification set; and generating a topology instruction using the trained DRL model.

[0010] Optionally, the reward function of the designed DRL model includes: determining an optimization target, where the optimization target includes frequency restoration, device safety, and economy; designing reward function reward terms, where the reward function reward terms include a frequency restoration reward term, a device safety penalty term, and an economy penalty term, and obtaining a comprehensive reward value after fusing the reward function reward terms.

[0011] Optionally, controlling the device to perform primary frequency modulation based on the frequency modulation contribution weight and the topology instruction includes: sending the topology instruction to the inverter so that the inverter dynamically adjusts the output power according to the device frequency modulation contribution weight and the frequency deviation.

[0012] Optionally, controlling the device to perform primary frequency modulation based on the frequency modulation contribution weight and the topology instruction further includes: when the response level is level one, only adjusting the output power of the inverter; when the response level is level two, first adjusting the output power of the inverter and then controlling the flexible output of the energy storage according to the frequency change rate and the state of charge of the energy storage; when the response level is level three, first adjusting the output power of the inverter and then controlling the forced output of the energy storage according to the frequency change rate and the state of charge of the energy storage.

[0013] On the other hand, the present invention provides a primary frequency modulation system for a photovoltaic power station, which is used to implement the primary frequency modulation method for a photovoltaic power station. The system includes a control module, and the control module includes a memory, a processor, and a computer program stored on the memory and executable on the processor. The processor executes the computer program to implement the primary frequency modulation method for a photovoltaic power station.

[0014] Through the above technical solution, by integrating frequency deviation classification, asymmetric game algorithm, and deep reinforcement learning (DRL), the system solves the key problems of primary frequency modulation in a photovoltaic power station. Through a three-level frequency deviation classification mechanism (setting double thresholds), refined control is achieved, and adjustment strategies are intelligently matched for different disturbance amplitudes; the asymmetric game algorithm is used to dynamically optimize the frequency modulation weights of each device, fully considering the heterogeneous characteristics of photovoltaic and energy storage, significantly improving resource utilization; based on the DRL model, a multi-objective optimization framework is designed to integrate frequency restoration, device safety, and economy indicators, and the optimal topology instruction is generated through intelligent decision-making; the primary frequency modulation performance of the photovoltaic power station in a high-proportion renewable energy power grid is significantly improved, taking into account economy and device safety while ensuring frequency stability.

[0015] Other features and advantages of the embodiments of the present invention will be described in detail in the subsequent specific implementation part. Description of the Drawings

[0016] The accompanying drawings are used to provide a further understanding of the embodiments of the present invention and form a part of the specification. Together with the following specific embodiments, they are used to explain the embodiments of the present invention, but do not constitute a limitation to the embodiments of the present invention. In the accompanying drawings: Figure 1 is a flowchart of a primary frequency regulation method for a photovoltaic power station.

[0017] Figure 2 is a flowchart for generating topology instructions. Specific Embodiments

[0018] The following will Figure 1 - be Figure 2 described in detail the specific embodiments of the embodiments of the present invention. It should be understood that the specific embodiments described herein are only used to illustrate and explain the embodiments of the present invention, and are not used to limit the embodiments of the present invention.

[0019] It should be noted that in the technical solution of this application, the acquisition, transmission, storage, use, processing, etc. of data all comply with the relevant regulations of national laws and regulations. In the embodiments of this application, some industry-existing solutions such as certain software, components, models, etc. may be mentioned. They should be regarded as exemplary. The purpose is only to illustrate the feasibility in the implementation of the technical solution of this application, but it does not mean that the applicant has already or necessarily used this solution.

[0020] The inventors of this application found in the process of implementing the present invention that most of the existing technologies are fixed-threshold grading methods, such as two-segment frequency deviation division, which cannot adapt to the dynamic response requirements of different disturbance intensities, resulting in over-regulation of resources during slight frequency fluctuations or insufficient response during severe disturbances; traditional game algorithms mostly adopt symmetric Nash equilibrium in weight allocation and do not consider the heterogeneous characteristics of devices such as photovoltaic and energy storage, such as response speed and capacity limitation, resulting in a mismatch between the frequency regulation contribution degree and the actual ability. Moreover, the traditional topology instruction generation depends on preset rules and lacks the ability of autonomous learning and optimization of the real-time state of the power grid.

[0021] Embodiment 1 Referring to Figures 1 - 2 , this is the first embodiment of the present invention. This embodiment provides a primary frequency regulation method for a photovoltaic power station, including: S100: Collect multi-source frequency regulation data of the photovoltaic power station and preprocess the multi-source frequency regulation data of the photovoltaic power station.

[0022] Specifically, frequency deviation data, including frequency magnitude and phase information, is obtained in real time from the phasor measurement unit (PMU) in the power grid; operating data of each device in the photovoltaic power station, such as output power, current, voltage, etc., is collected, through which the working status and power generation status of the equipment can be known; the charge state of the energy storage system, that is, the current power percentage of the energy storage battery, and the health status of the energy storage battery, including parameters such as the number of battery cycles and internal resistance changes, are obtained; light sensors and temperature sensors are used to collect light intensity and ambient temperature data in the area where the photovoltaic power station is located.

[0023] Furthermore, the collected multi-source frequency modulation data of photovoltaic power stations are preprocessed to remove noise points and abnormal values in the collected raw data, such as eliminating frequency deviation values or power values that are obviously beyond the reasonable range; the data from different data sources are integrated to form a complete data set, such as integrating the frequency deviation, equipment power, energy storage status and environmental data at the same time point for subsequent analysis; the data of different dimensions and numerical ranges are standardized to make them comparable and consistent; for the small amount of missing data that may appear in the collection process, appropriate methods are used to fill them, such as interpolation or estimation methods based on adjacent data points, to ensure the integrity of the data set.

[0024] Preferably, a complete grid status perception system is built by integrating multi-dimensional data such as PMU, equipment operating parameters, energy storage status and environmental sensors, providing a comprehensive data basis for subsequent analysis. Sensor errors and environmental interference are eliminated through outlier removal and standardization processing to improve data confidence, especially the accuracy of frequency deviation data; the time series of multi-source data collected can ensure the effectiveness of causal relationship analysis between parameters after alignment, avoiding misjudgment caused by delays.

[0025] S200: Classifying frequency deviation of the power grid frequency fluctuation of the photovoltaic power station grid connection point according to the pre-processed multi-source frequency modulation data of the photovoltaic power station.

[0026] Specifically, a first frequency deviation threshold and a second frequency deviation threshold are set; the first frequency deviation threshold is 0.1 Hz, and the second frequency deviation threshold is 0.2 Hz.

[0027] Furthermore, when the absolute value of the frequency deviation value is ≤ the first frequency deviation threshold, the response level is determined to be level one; when the first frequency deviation threshold is < the absolute value of the frequency deviation value < the second frequency deviation threshold, the response level is determined to be level two; when the absolute value of the frequency deviation value is ≥ the second frequency deviation threshold, the response level is determined to be level three.

[0028] Further, read the frequency deviation value collected by the PMU and preprocessed, where the frequency deviation value represents the deviation between the actual frequency and the rated frequency of the current power grid.

[0029] Further, divide the response levels according to the frequency deviation value, specifically as follows: When the absolute value of the frequency deviation value ≤ the first frequency deviation threshold of 0.1 Hz, the response level is determined to be level one.

[0030] When the first frequency deviation threshold of 0.1 Hz < the absolute value of the frequency deviation value < the second frequency deviation threshold of 0.2 Hz, the response level is determined to be level two.

[0031] When the absolute value of the frequency deviation value ≥ the second frequency deviation threshold of 0.2 Hz, the response level is determined to be level three.

[0032] Preferably, based on the high-precision measurement of the PMU, real-time grading can be achieved. Compared with the fixed threshold strategy, the frequency recovery time can be shortened by about 15%; through the three-level threshold division, a differentiated response system is established. When the response level is level one, only fine-tune the frequency deviation. When the response level is level two, quickly suppress the frequency fluctuation while avoiding excessive equipment loss. When the response level is level three, trigger coercive measures, thereby improving the response efficiency; the frequency deviation value at the response level of level three is close to the power grid safety critical value, which can leave a buffer margin for emergency control.

[0033] S300: Based on the frequency deviation grading, use the asymmetric game algorithm to dynamically allocate the frequency modulation contribution weights of the equipment.

[0034] Further, when the frequency deviation first exceeds the dead zone threshold of ±0.05 Hz, trigger the asymmetric game algorithm, and the asymmetric game algorithm includes: defining the objective function; calculating the initial weights of each device based on the frequency deviation grading; taking the derivative of the objective function and using the gradient descent method to iteratively update the frequency modulation contribution weights of the equipment.

[0035] Specifically, according to the preprocessed multi-source frequency modulation data of the photovoltaic power station, use the asymmetric game algorithm to solve the optimal weight allocation value based on the frequency deviation grading. Defining the objective function includes: defining the device utility function, introducing the Lagrange multiplier method to optimize the objective function in the device utility function, and the value of the Lagrange multiplier in the Lagrange multiplier method is set according to the corresponding frequency deviation response level.

[0036] Further, define the objective function, and the utility function of each inverter is as follows:

[0037] Introduce the Lagrange multiplier, and the objective function becomes:

[0038] Among them, represents the optimal weight allocation value of device i in the Nash equilibrium game, that is, the frequency modulation contribution weight of the device, represents the utility function of device i, which measures the benefits and losses of device i during the frequency modulation process, represents the adjustable power of device i, represents the loss cost coefficient, represents the device loss cost, represents the Lagrange multiplier, which is related to the response level, represents the frequency modulation weight of the j-th device, N represents the total number of devices, and j represents the index.

[0039] It should be noted that when the response level is level one, = 0.1; when the response level is level two, = 0.2; when the response level is level three, = 0.3.

[0040] Furthermore, calculating the initial weights of each device based on the frequency deviation classification includes: determining the frequency deviation response level, selecting the corresponding response level correction coefficient according to the frequency deviation response level, calculating the initial weights of each device, and verifying whether the sum of the initial weights meets the condition. The condition to be met is that the sum of the initial weights of each device should be 1.

[0041] Specifically, calculating the initial weights of each device based on the frequency deviation classification, the calculation formula is as follows:

[0042] Among them, represents the initial weight of device i (affected by the response level ), represents the rated power of device i, represents the sum of the rated powers of all devices, represents the response level correction coefficient.

[0043] It should be noted that when the response level is level one, = 1.0; when the response level is level two, = 1.2; when the response level is level three, = 1.5.

[0044] Taking the partial derivative of the objective function, the update direction of required by the gradient descent method is obtained:

[0045] Among them, represents the unit loss coefficient of the device.

[0046] Iterate using the gradient descent method, and the update for each time is as follows:

[0047] where, represents the newly calculated weight at present, represents the weight calculated in the previous time. When using the gradient descent method for the first iteration, = , represents the learning rate.

[0048] Furthermore, when the weight update amplitude is less than the set threshold or the maximum number of iterations is reached, stop the iteration. At this time, the weight is the weight of the device frequency modulation contribution obtained by the asymmetric game algorithm.

[0049] Preferably, use the asymmetric game algorithm to achieve Nash equilibrium through gradient descent iteration, so that high-rated power devices obtain higher weights, adjust the device loss cost through the loss cost coefficient. Compared with the traditional equal-weight strategy, it can extend the device life; associate the Lagrange multiplier with the response level, and the Lagrange multiplier increases with the increase of the response level to achieve elastic constraints on the frequency modulation weight distribution, strengthen the convergence speed and output during emergencies, and reduce losses under normal conditions to balance efficiency and device safety; when the response level is level three, force acceleration of convergence to meet emergency response requirements; the response level correction coefficient in the initial weight also increases with the increase of the response level, thereby amplifying the initial weight of high-rated power devices, so that the weight ratio of key devices, such as large-capacity inverters, at the response level of three increases, achieving the purpose of preferentially using high-regulation-capability resources for regulation when the response level reaches three; the gradient formula can reflect the device loss characteristics, protect high-loss devices, extend the cycle life in frequent frequency modulation scenarios, and reduce maintenance costs. Achieve Pareto optimality of weight distribution through game theory, avoid local optimality, improve the fairness of frequency modulation contribution, and equalize device utilization.

[0050] S400: Generate a topology instruction based on the result of the frequency deviation classification and the device frequency modulation contribution weight.

[0051] Furthermore, use the preprocessed multi-source frequency modulation data of the photovoltaic power station to define the state input of the DRL model; design the action space of the DRL model, and the action space includes physical topology switching and virtual cluster partitioning; design the reward function of the DRL model, and use the training set to train the DRL model and the validation set to verify the DRL model; use the trained DRL model to generate a topology instruction.

[0052] Specifically, based on the preprocessed multi-source frequency regulation data of the photovoltaic power station, the weight distribution entropy, line impedance matrix, energy storage SOC (State of Charge), and frequency deviation value are used as the state inputs of deep reinforcement learning (DRL), and all state variables are standardized to ensure consistent numerical ranges. Physical topology switching and virtual cluster partitioning are selected as the action space. Physical topology switching includes the selection of star, ring, tree, and hybrid topology structures to obtain topology switching instructions, such as switching from star to hybrid. Different topology structures affect the power flow and stability of the power grid. Virtual cluster partitioning divides the inverters and energy storage systems of the power station into 2 to 4 virtual synchronous units to obtain a cluster partitioning scheme, such as Cluster 1 includes Inverters 1 - 3, and Cluster 2 includes Inverters 4 - 6.

[0053] Further, the reward function for designing the DRL model includes: determining the optimization objectives, where the optimization objectives include frequency recovery, equipment safety, and economy; designing the reward function reward terms, where the reward function reward terms include a frequency recovery reward term, an equipment safety penalty term, and an economy penalty term, and the comprehensive reward value is obtained after fusing the reward function reward terms.

[0054] Further, design the reward function. The design logic of the reward function is that the faster the frequency recovery, the higher the reward; according to indicators such as frequency regulation response time, regulation accuracy, and duration, train the model to optimize the topology decision to maximize the frequency regulation performance reward and ensure a fast, accurate, and persistent response to frequency deviation. The worse the equipment safety, the higher the penalty value; combined with abnormal event data such as equipment overload and communication interruption, train the model to avoid decisions that may cause equipment failures during topology optimization, and regard equipment safety as an important part of the reward function to ensure the reliability of power station operation. The worse the economy, the higher the penalty value. Considering economic factors such as the energy storage cycle life and equipment loss, train the DRL model to reduce unnecessary energy storage use and equipment wear by optimizing the topology structure and lower the long-term operation cost of the power station; the comprehensive reward value is obtained after weighted fusion of the frequency recovery reward term, the equipment safety penalty term, and the economy penalty term.

[0055] Further, use historical frequency event data as the training set, set key parameters such as the learning rate and discount factor, and use the training set to train the DRL model. Use the validation set to verify the DRL model. The validation set is independent historical data or extreme scenarios generated by simulation to obtain a trained DRL model; input the state into the trained DRL model, the DRL model outputs the optimal action, convert the optimal action to generate a topology instruction, and recalculate the equipment frequency regulation contribution weights based on the new topology.

[0056] Preferably, a 10-second cooling period is started after the topology switch, continuous operation of the relay / circuit breaker is prohibited, and the number of topology switches is limited in combination with the historical operation times.

[0057] Preferably, the reward function integrates three elements: frequency restoration, equipment safety, and economy, and is represented by a comprehensive reward value. The Pareto optimal solution is found during the training of the DRL model, and the DRL model is guided by the Pareto optimal solution to generate topology instructions. Physical topology switching, such as switching from star to hybrid, can make the power flow distribution more balanced and reduce the node voltage fluctuation; virtual cluster partitioning (2 - 4 groups) can improve the coordination of VSG control; at the same time, a cooling period mechanism (10-second latch) is set to limit the topology switching frequency, which can prevent the failure rate of equipment such as relays from increasing due to frequent operation.

[0058] S500: Based on the frequency modulation contribution weight and the topology instruction, control the device to perform primary frequency modulation.

[0059] Furthermore, the topology instruction is sent to the inverter so that the inverter dynamically adjusts the output power according to the device frequency modulation contribution weight and the frequency deviation.

[0060] Specifically, according to the hierarchical architecture of master station → regional controller → inverter, the frequency modulation instructions are transmitted step by step. The master station integrates the device frequency modulation contribution weight, the topology instruction, and the response level to generate a global control signal. The regional controller disassembles the instruction into device-level instructions (such as the inverter power setting value, the energy storage charge and discharge power), and the inverter dynamically adjusts the output power according to the device frequency modulation contribution weight and the frequency deviation. The power adjustment formula of the inverter is as follows:

[0061] Among them, represents the power adjustment amount of the i-th inverter; represents the proportional gain coefficient, which is used to adjust the sensitivity of the inverter power response, and the value range is 1.2 - 2.0, and it is dynamically adjusted according to the frequency deviation; represents the grid frequency deviation, represents the newly calculated weight at present, represents the rated power of device i.

[0062] Furthermore, when the response level is level one, only the output power of the inverter is adjusted; when the response level is level two, the output power of the inverter is adjusted first, and then the flexible output of the energy storage is controlled according to the frequency change rate and the state of charge of the energy storage; when the response level is level three, the output power of the inverter is adjusted first, and then the forced output of the energy storage is controlled according to the frequency change rate and the state of charge of the energy storage.

[0063] Further, when the response level is secondary, on the basis of adjusting the output power of the inverter to cope with the frequency deviation, the energy storage system outputs power proportionally according to the frequency change rate and its own SOC (State of Charge). The calculation formula for the output power of the energy storage system is as follows:

[0064] Among them, represents the output power of the energy storage system, represents the grid frequency deviation, and SOC represents the state of charge of the energy storage system.

[0065] Further, when the response level is tertiary, on the basis of adjusting the output power of the inverter to cope with the frequency deviation, when SOC≥20%, the energy storage system forcibly outputs at 10% of the total system power.

[0066]

[0067] Among them, represents the output power of the energy storage system, represents the total system power capacity, which refers to the sum of the rated powers of all devices in the cluster.

[0068] When SOC<20%, take the smaller value of 10% of the total system power and the power calculated proportionally according to the remaining SOC capacity.

[0069]

[0070] Among them, k represents the proportionality coefficient used to control the power output decline rate of the energy storage system at low SOC.

[0071] Preferably, a three-level architecture of master station → regional controller → inverter is used to achieve fast response. Compared with centralized control communication, the load is reduced. When the response level is secondary, differential control is introduced. By capturing the frequency change rate, the system can sense the accelerated deviation of the frequency in advance, such as a rapid decline or rise, and actively intervene before the frequency deviation reaches the tertiary response level; differential control only requires the energy storage to output at high efficiency when the frequency changes violently, and reduces the output during the flat stage to avoid excessive losses caused by the energy storage running at full power continuously; SOC dynamic regulation can effectively improve the utilization rate of the energy storage, and the three-level response forced output mechanism can ensure the grid inertia support in extreme cases; and when the SOC is low, that is, when SOC<20%, it can effectively limit the deep discharge of the energy storage, increase the cycle times of the energy storage device, and reduce the maintenance cost of the energy storage device.

[0072] The present invention also provides a primary frequency regulation system for a photovoltaic power station, which is used to implement the primary frequency regulation method for the photovoltaic power station. The system includes a control module, and the control module includes a memory, a processor, and a computer program stored on the memory and executable on the processor. The processor executes the computer program to implement the primary frequency regulation method for the photovoltaic power station.

[0073] An embodiment of the present invention provides a storage medium, on which a program is stored, and when the program is executed by a processor, it implements the primary frequency regulation method for the photovoltaic power station.

[0074] An embodiment of the present invention provides a processor, which is used to run a program. When the program runs, it executes the primary frequency regulation method for the photovoltaic power station.

[0075] An embodiment of the present invention provides a device, which includes a processor, a memory, and a program stored on the memory and executable on the processor. When the processor executes the program, it implements the primary frequency regulation method for the photovoltaic power station. The device herein can be a server, a PC, a PAD, a mobile phone, etc.

[0076] The present application also provides a computer program product, which is suitable for executing the primary frequency regulation method for the photovoltaic power station when executed on a data processing device.

[0077] Those skilled in the art should understand that the embodiments of the present application can provide methods, systems, or computer program products. Therefore, the present application can take the form of a complete hardware embodiment, a complete software embodiment, or an embodiment combining software and hardware aspects. Moreover, the present application can take the form of a computer program product implemented on one or more computer-usable storage media (including but not limited to disk memories, CD-ROMs, optical memories, etc.) containing computer-usable program codes.

[0078] The present application is described with reference to the flowcharts and / or block diagrams of methods, devices (systems), and computer program products according to the embodiments of the present application. It should be understood that each process and / or block in the flowcharts and / or block diagrams, and the combination of processes and / or blocks in the flowcharts and / or block diagrams, can be implemented by computer program instructions. These computer program instructions can be provided to the processors of general-purpose computers, special-purpose computers, embedded processors, or other programmable data processing devices to generate a machine, so that the instructions executed by the processors of the computer or other programmable data processing devices generate means for implementing the functions specified in one Figure 1 one process or multiple processes and / or blocks Figure 1 one block or multiple blocks.

[0079] These computer program instructions can also be stored in a computer-readable memory that can direct a computer or other programmable data processing apparatus to operate in a particular manner, such that the instructions stored in the computer-readable memory produce an article of manufacture including instruction means that implement the functions specified in the flowchart Figure 1 one or more flowcharts and / or block diagrams Figure 1 specified in one or more blocks or blocks.

[0080] These computer program instructions can also be loaded onto a computer or other programmable data processing apparatus to cause a series of operational steps to be performed on the computer or other programmable apparatus to produce a computer-implemented process, so that the instructions executed on the computer or other programmable apparatus provide steps for implementing the functions specified in the flowchart Figure 1 one or more flowcharts and / or block diagrams Figure 1 specified in one or more blocks or blocks.

[0081] In a typical configuration, a computing device includes one or more processors (CPUs), an input / output interface, a network interface, and memory.

[0082] The memory may include non-permanent memory in the form of computer-readable media, random access memory (RAM) and / or non-volatile memory such as read-only memory (ROM) or flash memory (flash RAM). Memory is an example of computer-readable media.

[0083] Computer-readable media includes both permanent and non-permanent, removable and non-removable media implemented by any method or technology for storing information. The information can be computer-readable instructions, data structures, program modules, or other data. Examples of computer storage media include, but are not limited to, phase change memory (PRAM), static random access memory (SRAM), dynamic random access memory (DRAM), other types of random access memory (RAM), read-only memory (ROM), electrically erasable programmable read-only memory (EEPROM), flash memory or other memory technologies, compact disc read-only memory (CD-ROM), digital versatile disc (DVD) or other optical storage, magnetic cassettes, magnetic tape disk storage or other magnetic storage devices, or any other non-transmission media that can be used to store information that can be accessed by a computing device. As defined herein, computer-readable media does not include transitory media such as modulated data signals and carrier waves.

[0084] It should also be noted that the terms "comprise", "include" or any other variants thereof are intended to cover non-exclusive inclusion, so that a process, method, commodity or device comprising a series of elements not only includes those elements, but also includes other elements not expressly listed, or further includes elements inherent to such process, method, commodity or device. Without further limitation, an element defined by the statement "comprising an..." does not exclude the presence of additional identical elements in the process, method, commodity or device comprising the element.

[0085] The above are only embodiments of the present application and are not intended to limit the present application. For those skilled in the art, various changes and modifications can be made to the present application. Any modification, equivalent replacement, improvement, etc. made within the spirit and principle of the present application shall be included within the scope of the claims of the present application.

Claims

1. A primary frequency regulation method for a photovoltaic power station, characterized in that, Including: Collecting multi-source frequency regulation data of a photovoltaic power station and preprocessing the multi-source frequency regulation data of the photovoltaic power station; Based on the preprocessed multi-source frequency regulation data of the photovoltaic power station, performing frequency deviation classification on the grid frequency fluctuation at the grid connection point of the photovoltaic power station; Based on the frequency deviation classification, using an asymmetric game algorithm to dynamically allocate the frequency regulation contribution weights of devices; Based on the result of the frequency deviation classification and the frequency regulation contribution weights of the devices, generating a topology instruction; Based on the frequency regulation contribution weights and the topology instruction, controlling the devices to perform primary frequency regulation.

2. The primary frequency regulation method for a photovoltaic power station according to claim 1, characterized in that The performing frequency deviation classification on the grid frequency fluctuation at the grid connection point of the photovoltaic power station includes: Setting a first frequency deviation threshold and a second frequency deviation threshold; When the absolute value of the frequency deviation value ≤ the first frequency deviation threshold, determining that the response level is level one; When the first frequency deviation threshold < the absolute value of the frequency deviation value < the second frequency deviation threshold, determining that the response level is level two; When the absolute value of the frequency deviation value ≥ the second frequency deviation threshold, determining that the response level is level three.

3. The primary frequency regulation method for a photovoltaic power station according to claim 1, characterized in that The using an asymmetric game algorithm to dynamically allocate the frequency regulation contribution weights of devices based on the frequency deviation classification includes: When the frequency deviation first exceeds the dead zone threshold, triggering the asymmetric game algorithm, and the asymmetric game algorithm includes: Defining an objective function; Calculating the initial weights of each device based on the frequency deviation classification; Taking the derivative of the objective function and using the gradient descent method to iteratively update the frequency regulation contribution weights of the devices.

4. The primary frequency regulation method for a photovoltaic power station according to claim 3, wherein The defining an objective function includes: defining a device utility function and introducing the Lagrange multiplier method in the device utility function to optimize the objective function; The calculating the initial weights of each device based on the frequency deviation classification includes: determining the frequency deviation response level, selecting a corresponding response level correction coefficient according to the frequency deviation response level, calculating the initial weights of each device, and verifying whether the sum of the initial weights meets the conditions.

5. The primary frequency regulation method of the photovoltaic power station according to claim 4, wherein The introducing the Lagrange multiplier method in the device utility function to optimize the objective function includes that the value of the Lagrange multiplier in the Lagrange multiplier method is set according to the frequency deviation response level.

6. The primary frequency regulation method of a photovoltaic power station according to claim 1, characterized in that, The generating a topology instruction based on the result of the frequency deviation classification and the frequency regulation contribution weights of the devices includes: Using the preprocessed multi-source frequency regulation data of the photovoltaic power station to define the state input of the DRL model; Designing the action space of the DRL model, and the action space includes physical topology switching and virtual cluster partitioning; Designing the reward function of the DRL model, training the DRL model using the training set, and verifying the DRL model using the verification set; Using the trained DRL model to generate a topology instruction.

7. The primary frequency regulation method for a photovoltaic power station according to claim 6, characterized in that The designing the reward function of the DRL model includes: determining the optimization objectives, and the optimization objectives include frequency restoration, device safety, and economy; designing the reward items of the reward function, and the reward items of the reward function include frequency restoration reward items, device safety penalty items, and economy penalty items, and obtaining a comprehensive reward value after fusing the reward items of the reward function.

8. The primary frequency regulation method for a photovoltaic power station according to claim 1, wherein Based on the frequency modulation contribution weight and the topology instruction, controlling the device to perform primary frequency modulation includes: sending the topology instruction to the inverter so that the inverter dynamically adjusts the output power according to the device frequency modulation contribution weight and the frequency deviation.

9. The primary frequency regulation method of a photovoltaic power station according to claim 1, wherein Based on the frequency modulation contribution weight and the topology instruction, controlling the device to perform primary frequency modulation further includes: When the response level is level one, only adjust the output power of the inverter; When the response level is level two, first adjust the output power of the inverter and then control the flexible output of the energy storage according to the frequency change rate and the state of charge of the energy storage; When the response level is level three, first adjust the output power of the inverter, and then control the forced output of the energy storage according to the frequency change rate and the state of charge of the energy storage.

10. A primary frequency regulation system for a photovoltaic power station, characterized in that, The system includes a control module, the control module includes a memory, a processor, and a computer program stored on the memory and executable on the processor, and the processor executes the computer program to implement the primary frequency modulation method of the photovoltaic power station according to any one of claims 1-9.

Images