Multi-VSG inertia and damping adaptive adjustment method and system based on data driving

Abstract

The invention discloses a multi-VSG inertia and damping adaptive adjustment method and system based on data driving. The method comprises the steps that for each VSG node in the multi-VSG grid-connected system, an intelligent agent is used for obtaining the optimal rotational inertia and the optimal damping coefficient of each VSG node at the current moment; the intelligent agent is used for selecting the action quantity according to the current state quantity at each time step, evaluating the effect of the action quantity through a reward function for feedback, and automatically adjusting the rotational inertia and the damping coefficient of each VSG node through multiple iteration feedback; and controlling the inverter through a VSG control algorithm according to the optimal rotational inertia and the optimal damping coefficient. According to the scheme provided by the invention, the frequency deviation and the frequency change rate are considered on the basis of ensuring the transient stability of the multiple VSGs, and the comprehensive optimization of the transient performance of the multiple VSGs is realized.

Description

Technical Field

[0001] The present invention relates to the technical field of virtual synchronous machine control, and in particular to a data-driven multi-VSG inertia and damping adaptive regulation method and system. Background Technique

[0002] With the steady increase in the installed capacity of new energy, the degree of power system electrification has gradually deepened. In traditional power systems, the main power generation units are synchronous generators (SGs) driven by non-renewable energy sources such as coal. These traditional synchronous generators can store mechanical energy through the rotor shaft, thus having a large inertia, which effectively maintains the speed of the synchronous generator in the face of disturbances and ensures the stability of the generator output. In contrast, power electronic devices lack the rotor shafts of traditional synchronous generators, resulting in the overall inertia loss of the new power system. When facing disturbances such as voltage and frequency dips, large voltage and frequency rates of change (Rate of Change of Frequency, RoCoF) may occur, and in severe cases, it may even endanger the safe and stable operation of the power system.

[0003] It is worth emphasizing that the inertia and damping of the new energy power grid-connected system mainly depend on the control strategy of the grid-connected inverter. Therefore, scholars at home and abroad have proposed a series of control schemes to enhance the inertia and damping of the system. Among them, virtual synchronous generator (VSG) control adds virtual "rotational inertia" and "damping" to the inverter by simulating the rotor swing equation of traditional synchronous generators, and is considered to be one of the most effective control schemes for enhancing the inertia of power systems and improving voltage and frequency responses under the new grid architecture. While inheriting the external characteristics of traditional SGs, VSG control also exhibits different characteristics from SGs. On the one hand, the rotational inertia and damping coefficient of traditional SGs are fixed and their numerical values are determined by physical entities, while the rotational inertia and damping coefficient of VSGs are virtualized through control strategies and can be flexibly adjusted within the allowable capacity range. On the other hand, the damping braking effect of the damping winding of traditional SGs is very small and its damping coefficient can be almost ignored, while the damping dissipation term of VSGs has a relatively large impact on the transient performance and cannot be ignored like SGs. Therefore, when the grid voltage drops or a short-circuit fault occurs in the transmission line, the transient stability problem of VSGs is relatively more complex than that of SGs. If the traditional transient analysis method for SGs is directly applied to VSGs, it may lead to deviations or overly conservative tendencies in the transient stability assessment results of VSGs, ultimately affecting the correctness of VSG parameter regulation and worsening the transient response of VSGs.

[0004] In summary, overcoming the conservatism of the existing VSG transient stability analysis method, adaptively reshaping the inertia and damping of the VSG system, improving the transient stability of the system, and optimizing the transient performance of the VSG system are of great significance for the stable operation of the new power system. Summary of the Invention

[0005] To solve the above technical problems, an embodiment of the present invention provides a data-driven multi-VSG inertia and damping adaptive adjustment method and system.

[0006] The technical solution of the embodiment of the present invention is implemented as follows:

[0007] An embodiment of the present invention provides a data-driven multi-VSG inertia and damping adaptive adjustment method, the method includes:

[0008] For each VSG node in the multi-VSG grid-connected system, an agent is used to obtain the optimal moment of inertia and the optimal damping coefficient of each VSG node at the current moment; the agent is used to select an action amount according to the current state amount at each time step, and evaluate the effect of the action amount through a reward function for feedback. Through multiple iterative feedbacks, the moment of inertia and damping coefficient of each VSG node are automatically adjusted; the state amount is used to characterize the operating state of the multi-VSG grid-connected system; the action amount is used to characterize the control action taken by the multi-VSG grid-connected system at each time step; the reward function is used to evaluate the performance of the multi-VSG grid-connected system;

[0009] According to the optimal moment of inertia and the optimal damping coefficient, the inverter is controlled through a VSG control algorithm.

[0010] In one embodiment, the state amount is the angular frequency and the angular frequency change rate of the current VSG node and adjacent VSG nodes, and the calculation expression of the state amount is:

[0011]

[0012] where s ti represents the state amount, ω i represents the frequency of the i-th VSG, represents the frequency change rate of the i-th VSG, ω j represents the frequency of the j-th VSG, represents the frequency change rate of the j-th VSG, and Ω is the set composed of adjacent VSG nodes of the i-th VSG node.

[0013] In one embodiment, the action amount is the moment of inertia and the damping coefficient of the current VSG node, and the calculation expression of the action amount is:

[0014] a ti ={Ji D i}

[0015] Among them, a ti represents the action amount, J i represents the moment of inertia of the i-th VSG node, D i represents the damping coefficient of the i-th VSG node.

[0016] In one embodiment, the calculation expression of the reward function is:

[0017]

[0018] Among them, r 1i represents the penalty term related to the frequency deviation of VSG, r 2i represents the penalty term related to the rate of change of frequency of VSG, r 3i represents the penalty term for the frequency deviation between neighboring VSGs, r ti represents the reward function, ρ ω 、ρ ω0 、ρ dω 、ρ dω0 、ρ m are all penalty coefficients, Δω i represents the frequency deviation of the i-th VSG, represents the rate of change of frequency of the i-th VSG, ω i represents the frequency of the i-th VSG, ω j represents the frequency of the j-th VSG, and Ω is the set composed of the neighboring VSG nodes of the i-th VSG node.

[0019] In one embodiment, according to the optimal moment of inertia and the optimal damping coefficient, the inverter is controlled through the VSG control algorithm, including:

[0020] Measure the output current and output voltage of the inverter, and calculate the output power according to the output current and the output voltage;

[0021] According to the optimal moment of inertia and the optimal damping coefficient, obtain the voltage amplitude reference value and frequency through the VSG control algorithm;

[0022] Synthesize the output reference voltage according to the voltage amplitude reference value and the frequency; after passing the output reference voltage through a voltage-current double closed loop, control the inverter using a modulation signal through a pulse width modulation unit.

[0023] In one embodiment, calculating the output power according to the output current and the output voltage includes:

[0024]

[0025] where P i is the active power output of the i-th VSG, Q i is the reactive power output of the i-th VSG, V d is the d-axis component of the inverter output voltage in the dq coordinate system, V q is the q-axis component of the inverter output voltage in the dq coordinate system, I d is the d-axis component of the inverter output current in the dq coordinate system, I q is the q-axis component of the inverter output current in the dq coordinate system.

[0026] In one embodiment, according to the optimal moment of inertia and the optimal damping coefficient, obtaining a voltage amplitude reference value and a frequency through a VSG control algorithm includes:

[0027]

[0028] E = K(Q ref - Q i ) + E ref

[0029] where J i is the virtual moment of inertia of the i-th VSG node; D i is the virtual damping coefficient of the i-th VSG node; and P i respectively represent the active power reference value and the actual output active power of the i-th VSG node; ω i is the actual operating frequency of the i-th VSG node, ω j is the actual operating frequency of the j-th VSG node, ω ref is the rated frequency value; D k is the frequency mutual damping, δ is the phase difference between the common coupling point voltage vpcc and the grid voltage vg, which is defined as the power angle; E is the voltage amplitude reference value of the voltage-current double closed loop; E ref is the rated voltage amplitude; K is the reactive-voltage droop coefficient; Q ref and Q i are respectively the reactive power reference value and the reactive power actually output by the VSG, is the derivative of the VSG output frequency.

[0030] An embodiment of the present invention also provides a data-driven multi-VSG inertia and damping adaptive regulation system, including: a processor and a memory for storing a computer program that can run on the processor; wherein, when the processor is used to run the computer program, it executes the steps of the above-mentioned method.

[0031] An embodiment of the present invention also provides a storage medium, in which a computer program is stored. When the computer program is executed by a processor, the steps of the above-mentioned method are implemented.

[0032] The method of this embodiment has the following beneficial effects:

[0033] Aiming at the comprehensive optimization problem of transient frequency deviation, frequency change rate and frequency coupled oscillation in a multi-machine VSG grid-connected system, the method of this embodiment adaptively adjusts and optimizes the moment of inertia and damping coefficient of multiple VSGs in a distributed network based on deep reinforcement learning, overcomes the reliability problems and security problems caused by the high dependence of centralized reinforcement learning on communication, takes into account the frequency deviation and frequency change rate while ensuring the transient stability of VSG, and realizes the comprehensive optimization of the transient performance of VSG. BRIEF DESCRIPTION OF THE DRAWINGS

[0034] Figure 1 It is a schematic flow chart of the method for adaptively adjusting the inertia and damping of multiple VSGs based on data driving according to an embodiment of the present invention;

[0035] Figure 2 It is a schematic diagram of the optimization process of the intelligent agent according to an embodiment of the present invention;

[0036] Figure 3 It is a schematic diagram of the distributed optimization scheme of the multi-VSG grid-connected system according to an embodiment of the present invention;

[0037] Figure 4 It is a schematic diagram of the control scheme according to an embodiment of the present invention;

[0038] Figure 5 It is an internal structure diagram of the computer device according to an embodiment of the present invention. DETAILED DESCRIPTION OF THE EMBODIMENTS

[0039] The present invention will be further described in detail below with reference to the drawings and embodiments.

[0040] An embodiment of the present invention provides a method for adaptively adjusting the inertia and damping of multiple VSGs based on data driving, as Figure 1 shown, the method includes:

[0041] Step 101: For each VSG node in the multi-VSG grid-connected system, use an agent to obtain the optimal moment of inertia and optimal damping coefficient of each VSG node at the current moment; the agent is used to select the action amount according to the current state quantity at each time step, and evaluate the effect of the action amount through the reward function for feedback. Through multiple iterative feedbacks, automatically adjust the moment of inertia and damping coefficient of each VSG node; the state quantity is used to characterize the operating state of the multi-VSG grid-connected system; the action amount is used to characterize the control action taken by the multi-VSG grid-connected system at each time step; the reward function is used to evaluate the performance of the multi-VSG grid-connected system.

[0042] Step 102: According to the optimal moment of inertia and the optimal damping coefficient, control the inverter through the VSG control algorithm.

[0043] The method of this embodiment aims at the comprehensive optimization problem of transient frequency deviation, frequency change rate and frequency coupling oscillation in the multi-machine VSG grid-connected system, and adaptively adjusts and optimizes the moment of inertia and damping coefficient of multiple VSGs in the distributed network based on deep reinforcement learning, overcoming the reliability problems and safety problems caused by the high dependence of centralized reinforcement learning on communication. On the basis of ensuring the transient stability of the VSG, it takes into account the frequency deviation and frequency change rate, and realizes the comprehensive optimization of the transient performance of the VSG.

[0044] The method of this embodiment first constructs the mathematical model and second-order transfer function of the virtual synchronous generator (VSG); secondly, establishes the transient space and action space of the multi-VSG grid-connected system, and evaluates the system state through the reward function mechanism; furthermore, proposes a control scheme and trains the neural network using the deep reinforcement learning algorithm; finally, based on the trained neural network, adjusts the moment of inertia J and damping coefficient D of the multi-VSG grid-connected system, and the simulation experiment verifies the effectiveness and operability of the method.

[0045] Specifically, the solution of this embodiment includes the following:

[0046] 1. Taking the rotor motion equation of the synchronous generator as the prototype, deduce the mathematical model of the virtual synchronous generator:

[0047] The active power control simulates the primary frequency regulation function of the synchronous generator governor:

[0048]

[0049] J is the virtual moment of inertia; D is the virtual damping coefficient; P ref and P respectively represent the active power reference value and the active power actually output by the VSG; ω and ω ref are the actual operating frequency and rated frequency value of the VSG; δ is the phase difference between the voltage vpcc at the point of common coupling and the grid voltage vg, which is defined as the power angle.

[0050] Simulate the excitation regulation function of a traditional synchronous generator, and the reactive power control adopts droop control to achieve voltage regulation:

[0051] E = K(Q ref - Q) + E ref

[0052] Among them, E is the reference value of the voltage amplitude of the double closed-loop of voltage and current; E ref is the rated voltage amplitude; K is the reactive power-voltage droop coefficient; Q ref and Q are the reference value of reactive power and the reactive power actually output by the VSG respectively.

[0053] Under a high-voltage power grid, the line impedance ignores the resistance and is approximately purely inductive.

[0054] The second-order transfer function model of the VSG system is:

[0055]

[0056] Among them, ω n is the natural frequency and ζ is the damping ratio, U grid is the grid voltage, and X is the line impedance.

[0057] The moment of inertia J is a parameter that simulates the inertia of the rotor of a synchronous generator in the VSG system, and it reflects the inertia of the system to frequency changes: a larger J can smooth the frequency fluctuations and improve the frequency stability of the system, and a smaller J can improve the dynamic response speed of the system, but may lead to increased frequency fluctuations.

[0058] The damping coefficient D is a parameter that simulates the damping characteristics of a synchronous generator in the VSG system, and it reflects the ability of the system to suppress oscillations: a larger D can effectively suppress the oscillations of the system and improve the stability of the system, and a smaller D can improve the dynamic response speed of the system, but may lead to system oscillations.

[0059] J and D jointly determine the dynamic response characteristics of the VSG system. In practical applications, the values of J and D need to be optimized according to system requirements and operating conditions to achieve good dynamic performance and stability.

[0060] 2. Establish the transient space and action space of the multi-VSG grid-connected system, and evaluate the system state through the reward function mechanism

[0061] In this embodiment, the optimization process is implemented through a deep reinforcement learning algorithm. Specifically, the system constructs the environment and policy of reinforcement learning through the state variables, action variables, non-linear state space equation and reward function defined above, so as to realize the adaptive adjustment of the moment of inertia (J) and damping coefficient (D) of the multi-VSG system.

[0062] For a deep reinforcement learning task, the top priorities are to define three elements: the state space, the action space, and the reward function.

[0063] See Figure 2 , after a group of actions is output to the environment, the environment generates a new set of state variables according to the action parameters. This set of state variables is used as the input of the agent. Through the constructed reward function and the current input state, the current state is judged, and then the optimization of the previous set of action variables is judged, and the action output of the next set is adjusted. The purpose is to maximize the cumulative reward, so as to achieve the optimal action. Among them, an agent refers to a subject that learns and takes actions by interacting with the environment (Environment).

[0064] State variables: State variables are the current state information of the system, which is used to describe the dynamic behavior of the system. In this embodiment, the frequency value ω and the rate of change of angular frequency RoCoF that can characterize the frequency characteristics of the VSG are selected as state variables. Each VSG node cooperates with local information and neighbor communication. In this embodiment, the state variables obtained by each VSG node are the angular frequencies and the rates of change of angular frequency of the local VSG and two neighbor VSGs:

[0065]

[0066] where Ω is the set composed of adjacent VSGs of the target VSG unit.

[0067] Each agent obtains state variables through local information and neighbor communication. The selection of state variables can characterize the frequency characteristics of the VSG and help the agent understand the current operating state of the system.

[0068] Action variables: Action variables are the control actions that the system can take at each time step. In the present invention, the moment of inertia and the damping coefficient are defined as action variables:

[0069] a ti ={J i D i}

[0070] The state variable s ti is mapped to the action variable a through the control policy u(.) ti , and the functional relationship of the mapping can be expressed as the formula:

[0071] a ti =u(s ti )

[0072] where x tiRepresent all state variables in the entire system, including the output current \(i_f\), output voltage \(v_o\), angular frequency \(\omega\), output power \(P\), etc. of the VSG; \(d\) ti Represent uncertain disturbances or variables, such as changes in the active power reference value, sudden increases and decreases in load, grid voltage dips, etc.

[0073] The agent selects an appropriate action amount according to the current state variables and optimizes the dynamic response of the system by adjusting these parameters.

[0074] Agent: An agent is a subject that learns and takes actions by interacting with the environment. In the present invention, the agent is each VSG node, which selects an action amount according to the current state variables and evaluates the effect of the action through a reward function. The goal of the agent is to find the optimal moment of inertia and damping coefficient through continuous learning and adjustment to optimize the transient performance of the system.

[0075] Reward function: The reward function is used to evaluate the performance of the system and guide the optimization direction of the reinforcement learning algorithm. In the present invention, considering the requirements of the training objective for the frequency deviation and the rate of change of frequency, the reward function is designed as follows:

[0076]

[0077] where \(r\) 1i Represents the penalty term related to the frequency deviation of the VSG. Since the training objective of the deep reinforcement learning in this embodiment is to reduce the frequency deviation value of the VSG to not exceed 0.5 Hz, when the frequency deviation is less than 0.5 Hz, a smaller penalty coefficient \(\rho\) ω Is multiplied by the frequency deviation amount as the penalty term, and when the frequency deviation exceeds 0.5 Hz, a relatively large penalty coefficient \(\rho\) ω0 Is selected, hoping to avoid this situation through training.

[0078] \(r\) 2i Represents the penalty term related to the rate of change of frequency of the VSG. Since another training objective of the deep reinforcement learning in this embodiment is to reduce the rate of change of frequency of the VSG to not exceed 2 Hz / s, when the rate of change of frequency is less than 2 Hz / s, a smaller penalty coefficient \(\rho\) dω Is multiplied by the absolute value of the rate of change of frequency as the penalty term, and when the rate of change of frequency exceeds 2 Hz / s, a relatively large penalty coefficient \(\rho\) dω0 Is selected, hoping to avoid this situation through training.

[0079] \(r\) 3i Represents the penalty term for the frequency deviation between neighboring VSGs, and the penalty coefficient \(\rho\) m .

[0080] The above three penalties are normalized to obtain the reward function of the deep reinforcement learning algorithm in this section, denoted as r ti .

[0081] Through the reward function, the agent can evaluate the effect of the current action and adjust the policy to maximize the cumulative reward.

[0082] Environment: The environment is the object with which the agent interacts. It generates new state quantities based on the agent's actions and returns rewards. In this embodiment, the environment is a multi-VSG grid-connected system. The agent affects the frequency response of the system by adjusting the moment of inertia and damping coefficient. The environment then generates new frequencies and rates of change of frequency based on these adjustments and returns the corresponding rewards.

[0083] Through the above definitions, the reinforcement learning algorithm can select the action quantity (a ti ) according to the current state quantity (s ti ) at each time step, and the reward function is used to evaluate the effect of this action and feedback it to the algorithm to adjust the policy. Through continuous iteration and optimization, the optimal value that maximizes the cumulative reward accumulated by the reward function is found, and the reinforcement learning algorithm finally finds the optimal moment of inertia (J) and damping coefficient (D), so as to achieve the comprehensive optimization of frequency deviation and rate of change of frequency in the multi-VSG grid-connected system and improve the transient stability of the system.

[0084] The schematic diagram of the distributed optimization scheme for the multi-VSG grid-connected system is shown in Figure 3 as follows.

[0085] In summary, the process of achieving optimal parameters in this embodiment is as follows:

[0086] State observation: The agent first observes the current state quantities (frequency and rate of change of frequency) to understand the current operating state of the system.

[0087] Action selection: The agent selects the action quantity (moment of inertia and damping coefficient) according to the current state quantity and maps it to specific parameter adjustments through the control strategy.

[0088] Environmental feedback: The environment generates new state quantities based on the agent's actions and calculates the reward function to evaluate the effect of the current action.

[0089] Policy optimization: The agent adjusts the policy according to the feedback of the reward function, with the goal of maximizing the cumulative reward. Through continuous iteration and optimization, the agent gradually finds the optimal moment of inertia and damping coefficient.

[0090] Optimal parameters: After multiple iterations and optimizations, the agent finally finds the optimal moment of inertia and damping coefficient, so that the frequency deviation and rate of change of frequency of the system are comprehensively optimized, improving the transient stability of the system.

[0091] 3. Propose a control scheme

[0092] See the schematic diagram of the control scheme in Figure 4 as shown below.

[0093] The following are the calculation steps of the control scheme:

[0094] Step 1) Measure the output current and voltage of the inverter and calculate the corresponding output power P i , Q i ;

[0095] The output power P i , Q i is calculated as follows:

[0096]

[0097] where V d is the d-axis component of the output voltage of the inverter in the dq coordinate system, V q is the q-axis component of the output voltage of the inverter in the dq coordinate system, I d is the d-axis component of the output current of the inverter in the dq coordinate system, I q is the q-axis component of the output current of the inverter in the dq coordinate system.

[0098] Step 2) Through the VSG control algorithm:

[0099]

[0100] J i is the virtual moment of inertia of the i-th VSG; D i is the virtual damping coefficient of the i-th VSG; and P i respectively represent the active power reference value and the actual output active power of the i-th VSG; ω i , ω j and ω ref are the actual operating frequency and the rated frequency value of the i-th and j-th VSGs; D k is the frequency mutual damping, and δ is the phase difference between the voltage vpcc at the point of common coupling and the grid voltage vg, which is defined as the power angle.

[0101] E = K(Q ref - Q) + E ref

[0102] where E is the voltage amplitude reference value of the voltage-current double closed loop; E ref is the rated voltage amplitude; K is the reactive power-voltage droop coefficient; Q ref and Q are the reactive power reference value and the reactive power actually output by the VSG respectively.

[0103] It should be particularly noted that J here i and D i are given by the intelligent agent in the previous text.

[0104] In step 3), after the reference voltage amplitude and frequency obtained from the previous step are synthesized into the output reference voltage, through the voltage-current double closed-loop, the inverter is controlled by the modulation signal through the pulse width modulation (PWM) unit.

[0105] This embodiment proposes a data-driven multi-VSG inertia and damping adaptive online learning and adjustment method, which can overcome the reliability problems and safety problems caused by the high dependence of centralized reinforcement learning on communication, solve the disadvantages that traditional adaptive VSG control cannot take into account both frequency deviation and frequency change rate, take into account both frequency deviation and frequency change rate on the basis of ensuring VSG transient stability, and realize the comprehensive optimization of VSG transient performance.

[0106] To implement the method of this embodiment of the present invention, this embodiment of the present invention also provides a data-driven multi-VSG inertia and damping adaptive adjustment system, including: a processor and a memory for storing a computer program that can run on the processor; wherein, when the processor is used to run the computer program, it executes the steps of the above-mentioned method.

[0107] The above system provided in this embodiment and the above method embodiment belong to the same concept, and the specific implementation process is detailed in the method embodiment and will not be repeated here.

[0108] To implement the method of this embodiment of the present invention, this embodiment of the present invention also provides a computer program product, the computer program product includes computer instructions, and the computer instructions are stored in a computer-readable storage medium. The processor of the computer device reads the computer instructions from the computer-readable storage medium, and the processor executes the computer instructions, so that the computer device executes the steps of the above method.

[0109] Based on the hardware implementation of the above program module, and to implement the method of this embodiment of the present invention, this embodiment of the present invention also provides an electronic device (computer device). Specifically, in one embodiment, the computer device may be a terminal, and its internal structure diagram may be as Figure 5As shown. The computer device includes a processor A01, a network interface A02, a display screen A04, an input device A05, and a memory (not shown in the figure) connected through a system bus. Among them, the processor A01 of the computer device is used to provide computing and control capabilities. The memory of the computer device includes an internal memory A03 and a non-volatile storage medium A06. The non-volatile storage medium A06 stores an operating system B01 and a computer program B02. The internal memory A03 provides an environment for the operation of the operating system B01 and the computer program B02 in the non-volatile storage medium A06. The network interface A02 of the computer device is used to communicate with an external terminal through a network connection. When the computer program is executed by the processor A01, it implements the method of any one of the above embodiments. The display screen A04 of the computer device can be a liquid crystal display screen or an electronic ink display screen. The input device A05 of the computer device can be a touch layer covered on the display screen, or a button, a trackball, or a touchpad provided on the computer device housing, or an external keyboard, touchpad, or mouse, etc.

[0110] Those skilled in the art can understand that Figure 5 the structure shown in is only a block diagram of some structures related to the solution of the present application, and does not constitute a limitation on the computer device to which the solution of the present application is applied. The specific computer device may include more or fewer components than those shown in the figure, or combine some components, or have different component arrangements.

[0111] The device provided by the embodiment of the present invention 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 method of any one of the above embodiments.

[0112] Those skilled in the art should understand that the embodiments of the present application can be provided as a method, a system, or a computer program product. 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 storage, CD-ROM, optical storage, etc.) containing computer-usable program code.

[0113] This application is described with reference to the flowcharts and / or block diagrams of methods, apparatus (systems), and computer program products according to embodiments of the application. It should be understood that each flow and / or block in the flowchart and / or block diagram, and the combination of flows and / or blocks in the flowchart and / or block diagram, can be realized by computer program instructions. These computer program instructions can be provided to the processor of a general-purpose computer, special-purpose computer, embedded processor, or other programmable data processing device to generate a machine, such that the instructions executed by the processor of the computer or other programmable data processing device generate means for realizing the functions specified in one or more of the flows Figure 1 one or more of the flows and / or blocks Figure 1 or means for realizing the functions specified in one or more of the blocks.

[0114] These computer program instructions can also be stored in a computer-readable memory that can direct a computer or other programmable data processing device to work in a specific manner, such that the instructions stored in the computer-readable memory generate a manufactured article including instruction means that realizes the functions specified in one or more of the flows Figure 1 one or more of the flows and / or blocks Figure 1 or means for realizing the functions specified in one or more of the blocks.

[0115] These computer program instructions can also be loaded onto a computer or other programmable data processing device, such that a series of operation steps are executed on the computer or other programmable device to generate a computer-implemented process, so that the instructions executed on the computer or other programmable device provide steps for realizing the functions specified in one or more of the flows Figure 1 one or more of the flows and / or blocks Figure 1 or means for realizing the functions specified in one or more of the blocks.

[0116] In a typical configuration, a computing device includes one or more processors (CPUs), an input / output interface, a network interface, and a memory.

[0117] 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). The memory is an example of computer-readable media.

[0118] A computer-readable medium includes permanent and non-permanent, removable and non-removable media that can store information by any method or technology. 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-transitory medium that can be used to store information that can be accessed by a computing device. As defined herein, a computer-readable medium does not include transitory computer-readable media such as modulated data signals and carrier waves.

[0119] It can be understood that the memory in the embodiments of the present invention can be a volatile memory or a non-volatile memory, or can include both volatile and non-volatile memories. Among them, the non-volatile memory can be a read-only memory (ROM, Read-Only Memory), a programmable read-only memory (PROM, Programmable Read-Only Memory), an erasable programmable read-only memory (EPROM, Erasable Programmable Read-Only Memory), an electrically erasable programmable read-only memory (EEPROM, Electrically Erasable Programmable Read-Only Memory), a ferromagnetic random access memory (FRAM, ferromagnetic random access memory), a flash memory (Flash Memory), a magnetic surface memory, an optical disc, or a compact disc read-only memory (CD-ROM, Compact Disc Read-Only Memory); the magnetic surface memory can be a disk memory or a tape memory. The volatile memory can be a random access memory (RAM, RandomAccess Memory), which is used as an external cache. By way of example but not limitation, many forms of RAM are available, such as a static random access memory (SRAM, Static Random Access Memory), a synchronous static random access memory (SSRAM, Synchronous Static Random Access Memory), a dynamic random access memory (DRAM, Dynamic Random Access Memory), a synchronous dynamic random access memory (SDRAM, SynchronousDynamic Random Access Memory), a double data rate synchronous dynamic random access memory (DDRSDRAM, Double Data Rate Synchronous Dynamic Random Access Memory), an enhanced synchronous dynamic random access memory (ESDRAM, Enhanced Synchronous Dynamic Random Access Memory), a sync link dynamic random access memory (SLDRAM, SyncLink Dynamic Random Access Memory), a direct rambus random access memory (DRRAM, Direct Rambus Random Access Memory).The memories described in the embodiments of the present invention are intended to include, but are not limited to, these and any other suitable types of memories.

[0120] It should also be noted that the term "comprising", "including" or any other variant thereof is intended to cover non-exclusive inclusion, such that a process, method, commodity or device comprising a series of elements includes not only those elements but also other elements not expressly listed, or 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.

[0121] The above are only the 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 data-driven multi-VSG inertia and damping adaptive regulation method, characterized in that The method includes: For each VSG node in the multi-VSG grid-connected system, an agent is used to obtain the optimal moment of inertia and the optimal damping coefficient of each VSG node at the current moment; the agent is used to select the action quantity according to the current state quantity at each time step, and evaluate the effect of the action quantity through a reward function for feedback. Through multiple iterative feedbacks, the moment of inertia and the damping coefficient of each VSG node are automatically adjusted; the state quantity is used to characterize the operating state of the multi-VSG grid-connected system; the action quantity is used to characterize the control action taken by the multi-VSG grid-connected system at each time step; the reward function is used to evaluate the performance of the multi-VSG grid-connected system; According to the optimal moment of inertia and the optimal damping coefficient, the inverter is controlled through a VSG control algorithm.

2. The data-driven multi-VSG inertia and damping adaptive regulation method according to claim 1, wherein The state quantity is the angular frequency and the angular frequency change rate of the current VSG node and adjacent VSG nodes, and the calculation expression of the state quantity is: Among them, s ti represents the state quantity, ω i represents the frequency of the i-th VSG, represents the frequency change rate of the i-th VSG, ω j represents the frequency of the j-th VSG, represents the frequency change rate of the j-th VSG, and Ω is the set composed of the adjacent VSG nodes of the i-th VSG node.

3. The data-driven multi-VSG inertia and damping adaptive regulation method according to claim 1, wherein The action quantity is the moment of inertia and the damping coefficient of the current VSG node, and the calculation expression of the action quantity is: a ti = {J i D i} Among them, a ti represents the amount of action, J i represents the moment of inertia of the i-th VSG node, D i represents the damping coefficient of the i-th VSG node.

4. The data-driven multi-VSG inertia and damping adaptive regulation method according to claim 1, characterized in that The calculation expression of the reward function is: Among them, r 1i represents the penalty term related to the frequency deviation of the VSG, r 2i represents the penalty term related to the frequency change rate of the VSG, r 3i represents the penalty term for the frequency deviation between neighboring VSGs, r ti represents the reward function, ρ ω 、ρ ω0 、ρ dω 、ρ dω0 、ρ m 、ρ i are all penalty coefficients, Δω represents the frequency deviation of the i-th VSG, i ω represents the frequency of the i-th VSG, ω j represents the frequency of the j-th VSG, and Ω is the set composed of neighboring VSG nodes of the i-th VSG node.

5. The data-driven multi-VSG inertia and damping adaptive regulation method according to claim 1, wherein According to the optimal moment of inertia and the optimal damping coefficient, controlling the inverter through a VSG control algorithm includes: Measuring the output current and output voltage of the inverter, and calculating the output power according to the output current and the output voltage; According to the optimal moment of inertia and the optimal damping coefficient, obtaining the voltage amplitude reference value and the frequency through a VSG control algorithm; Synthesizing an output reference voltage according to the voltage amplitude reference value and the frequency; after passing the output reference voltage through a voltage-current double closed loop, controlling the inverter by using a modulation signal through a pulse width modulation unit.

6. The data-driven multi-VSG inertia and damping adaptive regulation method according to claim 5, characterized in that, Calculating the output power according to the output current and the output voltage includes: Among them, P i is the active power output of the i-th VSG, Q i is the reactive power output of the i-th VSG, V d is the d-axis component of the output voltage of the inverter in the dq coordinate system, V q is the q-axis component of the output voltage of the inverter in the dq coordinate system, I d is the d-axis component of the output current of the inverter in the dq coordinate system, I q is the q-axis component of the output current of the inverter in the dq coordinate system.

7. The data-driven multi-VSG inertia and damping adaptive regulation method according to claim 5, characterized in that According to the optimal moment of inertia and the optimal damping coefficient, obtaining the voltage amplitude reference value and the frequency through a VSG control algorithm includes: E = K(Q ref -Q i ) + E ref Among them, J i is the virtual moment of inertia of the i-th VSG node; D i is the virtual damping coefficient of the i-th VSG node; and P i represent the active power reference value and the actually output active power of the i-th VSG node respectively; ω i is the actual operating frequency of the i-th VSG node, ω j is the actual operating frequency of the j-th VSG node, ω ref is the rated frequency value; D k is the frequency cross-damping, δ is the phase difference between the voltage vpcc at the point of common coupling and the grid voltage vg, which is defined as the power angle; E is the reference voltage amplitude of the voltage-current double closed-loop; E ref is the rated voltage amplitude; K is the reactive power-voltage droop coefficient; Q ref and Q i are the reactive power reference value and the actually output reactive power of the VSG respectively, is the derivative of the VSG output frequency.

8. A data-driven multi-VSG inertia and damping adaptive regulation system, characterized in that, Includes: A processor and a memory for storing a computer program that can run on the processor; wherein, when the processor is used to run the computer program, it executes the steps of the method according to any one of claims 1 to 7.

9. A storage medium, in which a computer program is stored, characterized in that, When the computer program is executed by the processor, it implements the steps of the method according to any one of claims 1 to 7.

Images