Load shedding control method, system and equipment for wind power system and storage medium

Abstract

The invention discloses a wind power system load shedding control method, system and device and a storage medium, and the method comprises the following steps: building a wind power system frequency prediction model, and predicting the frequency change and frequency deviation according to the frequency prediction model; determining a load shedding scheme according to the frequency change, wherein the load shedding scheme comprises a load shedding amount; the stable supporting capacity of the wind turbine generator is obtained; and designing a load shedding control strategy based on graph deep reinforcement learning, wherein the load shedding control strategy determines a final load shedding action according to the load shedding scheme and the stable supporting capability of the wind turbine generator. The load shedding timeliness, self-adaptability and robustness of the wind power system are improved, and the load shedding control efficiency of the wind power system is improved.

Description

Technical Field

[0001] This invention belongs to the field of wind power control and relates to a method, system, equipment and storage medium for load shedding control of wind power systems. Background Technology

[0002] With the continuous increase in wind power penetration, the nonlinearity and uncertainty of power systems are enhanced, and stability mechanisms are becoming more complex, posing new requirements for their safe and stable operation. Load shedding control is the last line of defense for ensuring the safe and stable operation of wind power systems. When the wind power penetration rate is too high, the system may not be able to provide sufficient response in a short period of time, making it difficult to effectively implement load shedding control and ultimately leading to a wider range of power outages. In addition, load shedding control failure may lead to wind curtailment and reduce wind power utilization. Therefore, load shedding control is crucial in wind power systems. Existing technologies are difficult to adapt to and extend to new topology changes in power networks and wind power integration scenarios, which greatly affects the reliability of control strategies. Traditional solutions face challenges under high wind power penetration, mainly in the following aspects: 1. It is difficult to quantify safety conditions and stability control capabilities; 2. The control strategy fails to consider the support capabilities of wind turbine units; 3. The control strategy has poor performance and poor adaptability under topology changes. Summary of the Invention

[0003] The purpose of this invention is to overcome the shortcomings of the prior art and provide a method, system, device and storage medium for load shedding control of wind power systems, which improves the timeliness, adaptability and robustness of load shedding in wind power systems and improves the efficiency of load shedding control in wind power systems.

[0004] To achieve the above objectives, the present invention employs the following technical solution: A method for load shedding control in a wind power system includes the following processes: Establish a frequency prediction model for wind power systems, and predict frequency changes and frequency deviations based on the frequency prediction model; The load shedding scheme is determined based on frequency changes, and the load shedding scheme includes the load shedding amount. To obtain stable support capabilities for wind turbine units; A load shedding control strategy is designed based on graph deep reinforcement learning. The load shedding control strategy determines the final load shedding action based on the load shedding scheme and the stability support capability of the wind turbine.

[0005] Preferably, the steps for establishing a frequency prediction model for a wind power system include: constructing a dynamic frequency response model of the power system and simulating the dynamic frequency characteristics of the system through a transfer function.

[0006] Preferably, the initial frequency change rate, steady-state frequency deviation, and maximum frequency deviation are calculated based on the power system dynamic frequency response model.

[0007] Preferably, the steps for determining the load shedding scheme include: changing the frequency threshold based on the frequency deviation to initially determine the load shedding scheme; and determining the load shedding amount online through the frequency deviation.

[0008] Preferably, the steps for obtaining the stability support capability of wind turbine units include: establishing a dynamic frequency safety domain for the wind power system and quantifying the frequency support capability; deriving the constraints of the maximum frequency deviation and RoCoF through a frequency response model to form a frequency safety domain.

[0009] Preferably, the steps for designing a load shedding control strategy based on graph deep reinforcement learning include: designing states, actions, and reward functions. The states are mainly based on the predicted rate of change of frequency, steady-state deviation, maximum deviation, and load of each node. The actions are represented by binary vectors to indicate whether to shed loads. The reward function is determined based on the frequency deviation and load priority weights.

[0010] Preferably, graph deep reinforcement learning training includes sample acquisition and convolutional neural network (CNN) training.

[0011] A wind power system load shedding control system, comprising: The frequency prediction module is used to build a frequency prediction model for wind power systems and predict frequency changes and frequency deviations based on the frequency prediction model. The load shedding scheme determination module is used to determine the load shedding scheme based on frequency changes. The load shedding scheme includes the load shedding amount. The support acquisition module is used to acquire the stable support capability of the wind turbine. The control strategy module is used to design load shedding control strategies based on graph deep reinforcement learning. The load shedding control strategy determines the final load shedding action based on the load shedding scheme and the stability support capability of the wind turbine.

[0012] A computer device includes a memory, a processor, and a computer program stored in the memory and executable on the processor, wherein the processor executes the computer program to implement the steps of the wind power system load shedding control method.

[0013] A computer-readable storage medium storing a computer program that, when executed by a processor, implements the steps of the wind power system load shedding control method.

[0014] Compared with the prior art, the present invention has the following beneficial effects: This invention proposes a wind power system load shedding control method based on frequency prediction and graph deep reinforcement learning. This method not only innovates in control strategy but also incorporates frequency prediction. By shedding load proportionally according to frequency deviation, the load shedding amount is calculated online based on the frequency prediction results, resulting in stronger adaptability compared to traditional staged load shedding schemes. By considering the support capacity of wind turbines, the accuracy and stability of the control strategy are enhanced. The graph deep reinforcement learning algorithm improves the prediction accuracy of the load shedding strategy. This method improves the speed of load shedding control while ensuring control accuracy, avoiding the delay and poor control stability problems of existing control strategies. Attached Figure Description

[0015] Figure 1 This is a flowchart of the wind power system load shedding control method according to an embodiment of the present invention; Figure 2 This is a schematic diagram of the dynamic model of the system under disturbance according to an embodiment of the present invention; Figure 3 This is a schematic diagram of the wind turbine load reduction control curve according to an embodiment of the present invention; Figure 4 This is a schematic diagram illustrating the load shedding control effect of an embodiment of the present invention. Detailed Implementation

[0016] The technical solution of the present invention will now be clearly and completely described with reference to the accompanying drawings. Obviously, the described embodiments are only some, not all, of the embodiments of the present invention. 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.

[0017] like Figure 1 As shown, this embodiment illustrates the wind power system load shedding control method based on frequency prediction and graph deep reinforcement learning. The method includes the following steps: S1. Establish a frequency prediction model for wind power systems: Frequency is a crucial indicator for measuring the balance between active power generation and load. When generation losses lead to insufficient active power, the frequency drops. Systems with a high proportion of wind power are more susceptible to frequency deviations. When a wind power system is disturbed, generation and load are rebalanced by shedding an appropriate amount of load. Accurate frequency prediction can provide guidance for load shedding control.

[0018] First, a dynamic frequency response model of the power system is constructed. The dynamic frequency characteristics of the system are simulated by the transfer function, and indicators such as steady-state frequency deviation and maximum frequency deviation are calculated.

[0019] A. Frequency response transfer function: Figure 2For the dynamic model of the system under disturbance, the transfer function after neglecting the frequency dead zone is: In the formula , , , These are the model coefficients. To accumulate step power disturbance, The generator set's gradeability. , These are the backup startup delay and the maximum duration, respectively.

[0020] B. Calculation of dynamic frequency index: 1. Initial frequency change rate:

[0021] 2. Steady-state frequency deviation:

[0022] 3. Maximum frequency deviation (minutes) > , = , < (Derived in three cases).

[0023] S2. Determine the load shedding scheme: This invention determines the load shedding scheme by changing the frequency threshold based on the steady-state frequency deviation and the maximum frequency deviation.

[0024] The load shearing amount is determined online through the frequency deviation: ,in This is the load shedding ratio factor. During normal operation, the frequency deviation should be limited to a small range (e.g., ±0.2Hz). UFLS requires a frequency threshold to avoid erroneous shedding. The actual load shedding amount in the improved scheme is: The higher the threshold, the worse the frequency performance. Ks needs to be increased to cope with high thresholds.

[0025] S3. Study the stable support capability of wind turbine units: Wind turbines typically use converters to control the rotor's kinetic energy to autonomously respond to frequency changes, providing power support to the power grid. Characterizing the power support capability of wind turbines is crucial, requiring the establishment of a dynamic frequency security domain for wind power systems and the quantification of frequency support capabilities.

[0026] When wind energy drives the wind turbine blades to rotate, the active power output of the wind turbine is approximately equal to the mechanical power of the wind turbine. ,in For wind speed, Where is the blade radius. air density, The wind energy utilization coefficient is the ratio of the blade tip speed to the blade tip speed. and pitch angle Related: Variable-speed wind turbines simulate the electromechanical characteristics of synchronous generators through integrated inertia control, autonomously responding to frequency changes via rotor energy release / absorption. The active power regulation during frequency changes is: In the formula , These are the droop control and virtual inertia control coefficients, respectively. For frequency, Real-time frequency and initial frequency deviation, t is RoCoF. Assuming that the turbine parameters are the same in the same wind farm and they are all operating in FAR mode, the derivative of equation (4) can be obtained for the wind farm. Autonomous Adjustment Speed ​​(ARS): The sum of all power supply ARS is called the Comprehensive Autonomous Speed ​​(SARS). When grid power imbalance is severe, FAR alone cannot prevent frequency overruns. Switching the wind turbine to constant power control and adjusting the active power reference value can quickly regulate output. The adjustment range of CPCS is determined by the rotor speed. If CPCS still cannot prevent frequency overruns, then GT / LS must be enabled. Figure 2 The power operating point movement trajectory is shown under different control methods. CPCS has a faster power regulation speed than FAR, while GT has the fastest regulation speed. However, both CPCS and GT will reduce the system SARS, so they need to be optimized together to reduce the control input.

[0027] Under power disturbance, the system frequency response equation is: ,in For the initial disturbance, The system's equivalent inertia time constant, For GT / LS quantities, For CPCS power variation, This represents the continuous adjustment variable for FAR. The system's equivalent inertia varies with CPCS and GT: The constraints of Maximum Frequency Deviation (MFD) and RoCoF can be derived through the frequency response model, forming a frequency safety domain. Within this domain, the combination of power surges and SARS ensures that the frequency does not exceed the limit, and the two exhibit an inverse relationship. The wind turbine load shedding control curve is shown below. Figure 3 As shown.

[0028] S4. Design of load shedding control strategy based on graph deep reinforcement learning: The parameter design for graph deep reinforcement learning mainly includes the design of states, actions, and reward functions. States primarily consist of the predicted rate of change, steady-state deviation, maximum deviation, and load at each node. In the Markov decision-making process of graph deep reinforcement learning, the support capacity of the wind turbine is used as a node feature in the state space, allowing the agent to learn its association with load shedding actions. Actions are represented by binary vectors. Indicates whether to remove the first The load is [number]. The reward function can be expressed as: , of which Load priority weighting: if the strategy fully utilizes wind power support within the safety domain, it will be prioritized for application. If the load is cut again, a higher reward will be given. Graph deep reinforcement learning training includes sample acquisition and CNN training. Sample acquisition is achieved through... - Greedy strategy (linear decay) generates actions and stores them. To the experience replay pool. CNN training is achieved by fitting the Q function. Randomly select batches of samples for forward / backward propagation. The load shedding control effect is as follows: Figure 4 As shown.

[0029] The following are embodiments of the apparatus of the present invention, which can be used to execute embodiments of the method of the present invention. For details not omitted in the apparatus embodiments, please refer to the embodiments of the method of the present invention.

[0030] In another embodiment of the present invention, a wind power system load shedding control system is provided. This wind power system load shedding control system can be used to implement the above-mentioned wind power system load shedding control method. Specifically, the wind power system load shedding control system includes a frequency prediction module, a load shedding scheme determination module, a support acquisition module, and a control strategy module.

[0031] The frequency prediction module is used to establish a frequency prediction model for wind power systems and to predict frequency changes and frequency deviations based on the frequency prediction model.

[0032] The load shedding scheme determination module is used to determine the load shedding scheme based on frequency changes. The load shedding scheme includes the load shedding amount.

[0033] The support acquisition module is used to acquire the stable support capability of the wind turbine.

[0034] The control strategy module is used to design load shedding control strategies based on graph deep reinforcement learning. The load shedding control strategy determines the final load shedding action based on the load shedding scheme and the stability support capability of the wind turbine.

[0035] In another embodiment of the present invention, a terminal device is provided, the terminal device including a processor and a memory, the memory being used to store a computer program, the computer program including program instructions, and the processor being used to execute the program instructions stored in the computer storage medium. The processor may be a Central Processing Unit (CPU), or other general-purpose processors, digital signal processors (DSPs), application-specific integrated circuits (ASICs), field-programmable gate arrays (FPGAs), or other programmable logic devices, discrete gate or transistor logic devices, discrete hardware components, etc. It is the computing and control core of the terminal, suitable for implementing one or more instructions, specifically suitable for loading and executing one or more instructions to achieve corresponding method flows or corresponding functions. The processor described in this embodiment can be used in the operation of a wind power system load shedding control method, including: establishing a wind power system frequency prediction model; predicting frequency changes and frequency deviations based on the frequency prediction model; determining a load shedding scheme based on frequency changes, the load shedding scheme including the load shedding amount; obtaining the stable support capability of the wind turbine; and designing a load shedding control strategy based on graph deep reinforcement learning, the load shedding control strategy determining the final load shedding action based on the load shedding scheme and the stable support capability of the wind turbine.

[0036] In another embodiment, the present invention also provides a computer-readable storage medium (Memory), which is a memory device in a terminal device for storing programs and data. It is understood that the computer-readable storage medium here may include both the built-in storage medium in the terminal device and extended storage media supported by the terminal device. The computer-readable storage medium provides storage space that stores the terminal's operating system. Furthermore, the storage space also stores one or more instructions suitable for loading and execution by a processor, which may be one or more computer programs (including program code). It should be noted that the computer-readable storage medium here may be high-speed RAM or non-volatile memory, such as at least one disk storage device.

[0037] One or more instructions stored in a computer-readable storage medium can be loaded and executed by a processor to implement the corresponding steps of the wind power system load shedding control method in the above embodiments. One or more instructions in the computer-readable storage medium are loaded and executed by the processor to perform the following steps: establishing a frequency prediction model for the wind power system, predicting frequency changes and frequency deviations based on the frequency prediction model; determining a load shedding scheme based on the frequency changes, the load shedding scheme including the load shedding amount; obtaining the stability support capability of the wind turbine; designing a load shedding control strategy based on graph deep reinforcement learning, the load shedding control strategy determining the final load shedding action based on the load shedding scheme and the stability support capability of the wind turbine.

[0038] Those skilled in the art will understand that embodiments of this application can be provided as methods, systems, or computer program products. Therefore, this application can take the form of a completely hardware embodiment, a completely software embodiment, or an embodiment combining software and hardware aspects. Furthermore, this application can take the form of a computer program product embodied on one or more computer-usable storage media (including but not limited to disk storage, CD-ROM, optical storage, etc.) containing computer-usable program code.

[0039] This application is described with reference to flowchart illustrations and / or block diagrams of methods, apparatus (systems), and computer program products according to embodiments of this application. It will be understood that each block of the flowchart illustrations and / or block diagrams, and combinations of blocks in the flowchart illustrations and / or block diagrams, can be implemented by computer program instructions. These computer program instructions can be provided to a processor of a general-purpose computer, special-purpose computer, embedded processor, or other programmable data processing apparatus to produce a machine, such that the instructions, which execute via the processor of the computer or other programmable data processing apparatus, generate instructions for implementing the flowchart... Figure 1 One or more processes and / or boxes Figure 1 A device that provides the functions specified in one or more boxes.

[0040] These computer program instructions may also be stored in a computer-readable storage medium that can direct a computer or other programmable data processing device to function in a particular manner, such that the instructions stored in the computer-readable storage medium produce an article of manufacture including instruction means, which are implemented in a process Figure 1 One or more processes and / or boxes Figure 1 The function specified in one or more boxes.

[0041] These computer program instructions may also be loaded onto a computer or other programmable data processing equipment to cause a series of operational steps to be performed on the computer or other programmable equipment to produce a computer-implemented process, thereby providing instructions that execute on the computer or other programmable equipment for implementing the process. Figure 1 One or more processes and / or boxes Figure 1 The steps of the function specified in one or more boxes.

[0042] The sequence numbers of the embodiments in this application are for descriptive purposes only and do not represent the superiority or inferiority of the embodiments.

[0043] In the above embodiments of this application, the descriptions of each embodiment have different focuses. For parts not described in detail in a certain embodiment, please refer to the relevant descriptions of other embodiments.

[0044] In the several embodiments provided in this application, it should be understood that the disclosed technical content can be implemented in other ways. The device embodiments described above are merely illustrative; for example, the division of units can be a logical functional division, and in actual implementation, there may be other division methods. For instance, multiple units or components may be combined or integrated into another system, or some features may be ignored or not executed. Furthermore, the displayed or discussed mutual coupling, direct coupling, or communication connection may be through some interfaces; the indirect coupling or communication connection between units or modules may be electrical or other forms.

[0045] The units described as separate components may or may not be physically separate. The components shown as units may or may not be physical units; that is, they may be located in one place or distributed across multiple units. Some or all of the units can be selected to achieve the purpose of this embodiment according to actual needs.

[0046] The above description is only a preferred embodiment of this application. It should be noted that for those skilled in the art, several improvements and modifications can be made without departing from the principle of this application, and these improvements and modifications should also be considered within the scope of protection of this application.

[0047] It should be understood that the above description is for illustrative purposes and not for limitation. Many embodiments and applications beyond the provided examples will be apparent to those skilled in the art upon reading the above description. Therefore, the scope of this patent should not be determined by reference to the above description, but rather by reference to the foregoing claims and the full scope of their equivalents. For purposes of completeness, all articles and references, including patent applications and publications, are incorporated herein by reference. The omission of any aspect of the subject matter disclosed herein in the foregoing claims is not intended as a waiver of that subject matter, nor should it be construed as an indication that the applicant has not considered that subject matter as part of the disclosed inventive subject matter.

Claims

1. A load shedding control method for a wind power system, characterized in that, Includes the following processes: Establish a frequency prediction model for wind power systems, and predict frequency changes and frequency deviations based on the frequency prediction model; The load shedding scheme is determined based on frequency changes, and the load shedding scheme includes the load shedding amount. To obtain stable support capabilities for wind turbine units; A load shedding control strategy is designed based on graph deep reinforcement learning. The load shedding control strategy determines the final load shedding action based on the load shedding scheme and the stability support capability of the wind turbine.

2. The wind power system load shedding control method according to claim 1, characterized in that, The steps to establish a frequency prediction model for a wind power system include: constructing a dynamic frequency response model of the power system and simulating the dynamic frequency characteristics of the system through a transfer function.

3. The wind power system load shedding control method according to claim 2, characterized in that, The initial frequency change rate, steady-state frequency deviation, and maximum frequency deviation are calculated based on the dynamic frequency response model of the power system.

4. The wind power system load shedding control method according to claim 1, characterized in that, The steps to determine the load shedding scheme include: adjusting the frequency threshold based on the frequency deviation to initially determine the load shedding scheme; and determining the load shedding amount online through the frequency deviation.

5. The wind power system load shedding control method according to claim 1, characterized in that, The steps to obtain the stability support capability of wind turbine units include: establishing the dynamic frequency safety domain of the wind power system and quantifying the frequency support capability; deriving the maximum frequency deviation and RoCoF constraint conditions through the frequency response model to form the frequency safety domain.

6. The wind power system load shedding control method according to claim 1, characterized in that, The steps for designing a load shedding control strategy based on graph deep reinforcement learning include: designing states, actions, and reward functions. The states are mainly based on the predicted rate of change of frequency, steady-state deviation, maximum deviation, and load of each node. The actions are represented by binary vectors to indicate whether to shed loads. The reward function is determined based on the frequency deviation and load priority weights.

7. The wind power system load shedding control method according to claim 6, characterized in that, Graph deep reinforcement learning training includes sample collection and training of convolutional neural networks (CNNs).

8. A load shedding control system for a wind power system, characterized in that, include: The frequency prediction module is used to build a frequency prediction model for wind power systems and predict frequency changes and frequency deviations based on the frequency prediction model. The load shedding scheme determination module is used to determine the load shedding scheme based on frequency changes. The load shedding scheme includes the load shedding amount. The support acquisition module is used to acquire the stable support capability of the wind turbine. The control strategy module is used to design load shedding control strategies based on graph deep reinforcement learning. The load shedding control strategy determines the final load shedding action based on the load shedding scheme and the stability support capability of the wind turbine.

9. A computer device comprising a memory, a processor, and a computer program stored in the memory and executable on the processor, characterized in that, When the processor executes the computer program, it implements the steps of the wind power system load shedding control method as described in any one of claims 1 to 7.

10. A computer-readable storage medium storing a computer program, characterized in that, When the computer program is executed by the processor, it implements the steps of the wind power system load shedding control method as described in any one of claims 1 to 7.

Images