Dynamic identification method and system for GFM inverter dominant power distribution system and related device

Abstract

The invention provides a dynamic identification method and system for a GFM inverter dominant power distribution system and a related device. The method comprises the following steps: step 1, establishing a phasor domain mathematical model of a GFM inverter in a phasor domain; step 2, constructing a nonlinear least square optimization model based on the phasor domain mathematical model, obtaining a parameter estimation value corresponding to the GFM inverter, substituting the parameter estimation value into the phasor domain mathematical model, and solving to obtain an optimized output phasor; step 3, optimizing a system dynamic prediction model pre-constructed based on the NARX neural network by using the optimized output phasor obtained in the step 2 to obtain an optimized system dynamic prediction model; and step 4, based on the pre-constructed equivalent model of the GFM inverter in the phasor domain, utilizing the optimized system dynamic prediction model obtained in the step 3 to dynamically predict the dominant power distribution system of the GFM inverter, and through organic combination of physical optimization and data driving, high-precision and high-generalization system dynamic representation is realized.

Description

Technical Field

[0001] This invention belongs to the field of power system modeling and identification technology, specifically relating to a dynamic identification method, system and related devices for GFM inverter-dominated power distribution systems. Background Technology

[0002] With the increasing penetration of power sources based on power electronic inverters, such as photovoltaics and wind power, into the power grid, modern power systems are gradually evolving into a new type of power system dominated by power electronics and inverters. In particular, grid-forming (GFM) inverters, due to their voltage / frequency self-regulation capabilities and provision of virtual inertia, are considered a key solution for improving the stability of inverter-dominated systems. However, the large-scale integration of GFM inverters also brings challenges to system modeling and dynamic identification. First, GFM inverters exhibit complex nonlinear dynamic behavior and control coupling relationships. While traditional switching and averaging models can comprehensively capture system dynamics, computational costs increase rapidly with system scale. Furthermore, the topology or proprietary parameters of GFM inverters are often unavailable due to manufacturer secrecy or engineering limitations, resulting in limited model accuracy. Therefore, for GFM inverter-dominated distribution systems, a new method that balances physical interpretability, computational efficiency, and dynamic identification accuracy is urgently needed.

[0003] To address this technical issue, the paper "Artificial Intelligence Aided Black-Box Modeling of Three-Phase Single-Stage Photovoltaic Inverter Systems," published in IEEE Transactions on Industry Applications, vol. 61, no. 2, pp. 3317-3328, March-April 2025, proposes a black-box modeling method for photovoltaic inverter systems. This method, based on input-output measurement data, utilizes neural networks to model and dynamically identify photovoltaic inverters with hierarchical control structures, aiming to capture the nonlinear dynamic behavior of inverter-dominated systems in the time domain. However, due to its complete reliance on data-driven approaches and lack of physical mechanism constraints and interpretability, its generalization performance is poor under conditions of insufficient data or variable operating conditions. Furthermore, it remains limited to computationally expensive switching and averaging models, thus restricting the practical application value of this method in distribution systems dominated by high proportions of GFM inverters. Summary of the Invention

[0004] The purpose of this invention is to provide a dynamic identification method, system and related device for GFM inverter-dominated power distribution systems, which solves the defects of existing modeling and dynamic identification methods for grid-connected inverter-dominated power distribution systems, which have poor generalization performance when there is insufficient data or variable operating conditions.

[0005] To achieve the above objectives, the technical solution adopted by the present invention is as follows: In a first aspect, the present invention provides a dynamic identification method for a GFM inverter-dominated power distribution system, comprising the following steps: Step 1: Establish the phasor domain mathematical model of the GFM inverter in the phasor domain; Step 2: Construct a nonlinear least squares optimization model based on the phasor domain mathematical model, obtain the parameter estimates corresponding to the GFM inverter, substitute the parameter estimates into the phasor domain mathematical model, and solve to obtain the optimized output phasor. Step 3: Use the optimized output phasor obtained in Step 2 to optimize the pre-built system dynamic prediction model based on the NARX neural network, and obtain the optimized system dynamic prediction model. Step 4: Based on the pre-built equivalent model of the GFM inverter in the phasor domain, use the optimized system dynamic prediction model obtained in Step 3 to perform dynamic prediction on the GFM inverter-dominated power distribution system.

[0006] Preferably, the phasor domain mathematical model of the GFM inverter includes the apparent power expression, average active power expression, average reactive power expression, voltage amplitude and phase angle expression, frequency expression, reference voltage expression, and output current expression of the GFM inverter.

[0007] Preferably, the expression for the nonlinear least squares optimization model is:

[0008] in, To optimize output; The values ​​are the inverter-side measurements obtained through the phasor domain mathematical model of the GFM inverter; p represents the set of unknown system parameters to be identified.

[0009] Preferably, the optimized output obtained in step 2 is used to optimize the pre-built system dynamic prediction model based on the NARX neural network. The specific method is as follows: A dynamic prediction model for the system was constructed based on the NARX neural network. The optimized output data obtained in step 2, the historical input data of the GFM inverter, the historical output data of the GFM inverter, and the instantaneous input data of the GFM inverter are combined to form the input vector of the neural network. The system dynamic prediction model is optimized using the generated input vector to obtain the optimized system dynamic prediction model.

[0010] Preferably, the expression for the system dynamic prediction model is:

[0011] in, y opt-NN To optimize and enhance the output of the neural network; u This is the input vector of the neural network; λ Represents the hyperparameters of a neural network; N (·) represents the NARX neural network structure.

[0012] Preferably, the pre-constructed equivalent model of the GFM inverter in the phasor domain is specifically implemented as follows: Based on the Norton equivalent principle, the optimized system dynamic prediction model obtained in step 3 is integrated with the controlled current source into the main circuit to construct a "phasor-neural network" hybrid architecture, thereby obtaining the equivalent model of the GFM inverter in the phasor domain.

[0013] Secondly, the present invention provides a dynamic identification system for a GFM inverter-dominated power distribution system, based on the method described, comprising: The mathematical model building unit is used to build the phasor domain mathematical model of the GFM inverter in the phasor domain. The phasor acquisition unit is used to construct a nonlinear least squares optimization model based on the phasor domain mathematical model, obtain the parameter estimates corresponding to the GFM inverter, substitute the parameter estimates into the phasor domain mathematical model, and solve for the optimized output phasor. The model optimization unit is used to optimize the pre-built system dynamic prediction model based on the NARX neural network using the obtained optimized output phasor, so as to obtain the optimized system dynamic prediction model. The dynamic prediction unit is used to perform dynamic prediction of the power distribution system dominated by the GFM inverter based on the pre-built equivalent model of the GFM inverter in the phasor domain and the obtained optimized system dynamic prediction model.

[0014] Thirdly, the present invention provides an electronic device including a processor and a memory, wherein the memory stores computer instructions, and when the computer instructions are executed by the processor, the electronic device performs the method described thereon.

[0015] Fourthly, the present invention provides a computer program product that includes computer-executable instructions, which, when executed, implement the method described herein.

[0016] Fifthly, the present invention provides a computer-readable storage medium storing computer-executable instructions that, when executed by a processor, implement the method described herein.

[0017] Compared with the prior art, the beneficial effects of the present invention are: This invention provides a dynamic identification method for GFM inverter-dominated power distribution systems. By integrating physical optimization and data learning, it achieves high-precision, robust, and physically interpretable phasor domain modeling and dynamic identification of GFM inverter-dominated power distribution systems. Specifically, it includes: (1) Balancing accuracy and efficiency in modeling: Modeling is performed in the phasor domain, filtering high-frequency dynamics and focusing on the fundamental frequency response, balancing model accuracy and computational efficiency. A coupled model of neural networks and controllable current source models is used to equivalently characterize the GFM inverter, forming a "phasor" model of the GFM inverter-dominated power distribution system. – The hybrid architecture of "neural network" ensures the feasibility of simulation and real-time identification of power distribution systems with a high proportion of GFM inverters.

[0018] (2) Dynamic identification balances physical interpretability and flexibility: An optimization model based on nonlinear least squares is constructed to initially estimate the key parameters of the GFM inverter, providing a physically consistent initialization for the neural network. In addition, a nonlinear autoregressive exogenous (NARX) neural network structure is used to introduce historical input-output data and optimized output, enhancing the ability to capture the nonlinear behavior of the power distribution system dominated by the GFM inverter.

[0019] In summary, this invention significantly improves the dynamic identification accuracy of GFM inverter-dominated distribution networks by deeply integrating the optimization model with neural networks, maintaining stable performance even under complex disturbance conditions. This method not only introduces prior physical information to enhance the interpretability of the modeling results but also effectively captures unmodeled nonlinear dynamic characteristics, thus balancing physical rationality with data-driven flexibility. Simultaneously, phasor domain modeling significantly reduces computational complexity, ensuring the feasibility of large-scale power grid simulation and real-time identification. It possesses good robustness and scalability, thus demonstrating significant practical value and broad application prospects in power system stability analysis, control optimization, and engineering applications. Attached Figure Description

[0020] Figure 1 A schematic diagram of the "phasor-neural network" hybrid framework of a GFM inverter system; Figure 2 A schematic diagram illustrating an optimized and enhanced neural network modeling method; Figure 3 This is a schematic diagram of voltage amplitude disturbance; Figure 4 This is a schematic diagram of phase angle disturbance; Figure 5 This is a schematic diagram of frequency disturbance; Figure 6 This is a schematic diagram of high voltage ride-through. Figure 7 This is a schematic diagram of low voltage ride-through. Detailed Implementation

[0021] In the following description, specific details such as particular system architectures and techniques are set forth for illustrative purposes and not for limitation, in order to provide a thorough understanding of the embodiments of this application. However, those skilled in the art will understand that this application may also be implemented in other embodiments without these specific details. In other instances, detailed descriptions of well-known systems, apparatuses, circuits, and methods have been omitted so as not to obscure the description of this application with unnecessary detail.

[0022] It should be understood that, when used in this application specification and the appended claims, the term "comprising" indicates the presence of the described features, integrals, steps, operations, elements and / or components, but does not exclude the presence or addition of one or more other features, integrals, steps, operations, elements, components and / or a collection thereof.

[0023] It should also be understood that the term “and / or” as used in this application specification and the appended claims means any combination of one or more of the associated listed items and all possible combinations, and includes such combinations.

[0024] As used in this application specification and the appended claims, the term "if" may be interpreted, depending on the context, as "when," "once," "in response to determination," or "in response to detection." Similarly, the phrase "if determined" or "if detected [the described condition or event]" may be interpreted, depending on the context, as meaning "once determined," "in response to determination," "once detected [the described condition or event]," or "in response to detection [the described condition or event]."

[0025] Furthermore, in the description of this application and the appended claims, the terms "first," "second," "third," etc., are used only to distinguish descriptions and should not be construed as indicating or implying relative importance.

[0026] References to "one embodiment" or "some embodiments" as described in this specification mean that one or more embodiments of this application include a specific feature, structure, or characteristic described in connection with that embodiment. Therefore, the phrases "in one embodiment," "in some embodiments," "in other embodiments," "in still other embodiments," etc., appearing in different parts of this specification do not necessarily refer to the same embodiment, but rather mean "one or more, but not all, embodiments," unless otherwise specifically emphasized. The terms "comprising," "including," "having," and variations thereof mean "including but not limited to," unless otherwise specifically emphasized.

[0027] Example 1 This embodiment provides a method for modeling and dynamic identification of a grid-connected inverter-dominated power distribution system in the phasor domain, including the following steps: Step 1: Establish the phasor domain mathematical model of the GFM inverter in the phasor domain; Step 2: Construct a nonlinear least squares optimization model based on the phasor domain mathematical model, obtain the parameter estimates corresponding to the GFM inverter, substitute the parameter estimates into the phasor domain mathematical model to solve the optimized output phasor, and the optimized output provides physical prior features for subsequent NARX neural network training. Step 3: Use the optimized output obtained in Step 2 to optimize the pre-built system dynamic prediction model based on the NARX neural network to obtain the optimized system dynamic prediction model. Step 4: Based on the Norton equivalence principle, construct a hybrid architecture of "phasor-neural network" to represent the GFM inverter. Step 5: Based on the equivalent model of the GFM inverter obtained in Step 4, the optimized system dynamic prediction model is used to perform dynamic prediction on the GFM inverter.

[0028] Example 2 This invention relates to a dynamic identification method for GFM inverter-dominated power distribution systems, aiming to address the shortcomings of existing modeling methods in terms of parameter sensitivity, dynamic response accuracy, and generalization ability.

[0029] The power distribution system includes a GFM inverter, line impedance, AC load, and a MATLAB / Simulink simulation modeling platform.

[0030] Specifically, the following steps are included: Step 1: Phasor Domain Modeling (1) Derive the phasor domain mathematical model of the GFM inverter, which includes the apparent power expression, average active power expression, average reactive power expression, voltage amplitude and phase angle expression, frequency expression, reference voltage expression and output current expression of the GFM inverter, wherein: The apparent power of a GFM inverter is defined as: (1) in, V o The inverter output voltage phasor; I o * The real and imaginary parts of the apparent power correspond to the active power output by the inverter, respectively. P o and reactive power Q o ; j It represents the imaginary unit.

[0031] Average active power P and average reactive power Q The output can be obtained through a low-pass filter: (2a) (2b) In the formula, ω c This is the cutoff frequency of the low-pass filter; s It is the Laplace operator.

[0032] In order to adjust the frequency of the GFM inverter ω and voltage amplitude V It can adopt widely used P–f and Q–V Drooping control relationship: (3a) (3b) In the formula, ω 0 and These represent the rated voltage frequency and amplitude, respectively. P set and Q set These represent the active power and reactive power setpoints, respectively. m p and n q This is the droop coefficient.

[0033] Voltage phase angle δ Represented as: (4) The reference voltage for a GFM inverter is expressed as follows: (5) The output current of the GFM inverter is calculated as follows: (6) In the formula, Z eq This is the equivalent impedance between the inverter and the point of common coupling (PCC).

[0034] Step 2: Parameter Identification and Optimization Output Solution (1) Construct a nonlinear least squares optimization problem: Based on the phasor domain model obtained in step 1, a nonlinear least squares optimization problem is constructed, and the optimization model can be expressed as: (7) in, To optimize output; The variable 'p' represents the inverter-side measurements (e.g., current or voltage phasors) obtained from the phasor domain mathematical model of the GFM inverter; 'p' represents the set of unknown system parameters to be identified. Function f (·) contains algebraic or differential equations describing the dynamics of a physical system.

[0035] (2) Solve the optimization problem to obtain the parameter estimates. The optimization process aims to determine the parameter set p such that the model system behavior and the observed system behavior are optimally consistent. Parameter identification is achieved by solving a nonlinear least squares problem, as follows: (8) in, y meas It is the measurement output of the physical system.

[0036] Using the phasor domain mathematical model of the GFM inverter as a constraint condition for a nonlinear least squares problem, and solving it using the interior point method (8), the optimal estimate of the system parameters can be obtained, and the corresponding output... y opt As physical information characteristics for the next stage.

[0037] (3) Substitute the parameter estimates into the phasor domain mathematical model of the GFM inverter to solve the optimized output phasor. The optimized output will provide physical prior features for subsequent neural network training.

[0038] Step 3: Optimize and enhance neural network modeling (1) Constructing the NARX neural network The NARX model extends the linear ARX framework by integrating nonlinear and linear functions into a unified structure, enabling it to effectively identify complex system dynamics. The NARX neural network model integrates feedback loops, using past output values ​​as inputs to provide a memory mechanism for the network, allowing historical values ​​to influence future predictions. This architecture makes hybrid NARX neural network models particularly well-suited for capturing the dynamic characteristics of inverter-dominated systems, especially in predicting time-series data (e.g., voltage and current signals) and modeling highly nonlinear dynamic behaviors (e.g., coupling between control outputs).

[0039] The NARX neural network extends the linear ARX model by introducing a nonlinear mapping function and a feedback mechanism. Its general structure can be represented as: (10) in, W and b Represents the weight matrix and bias vector. Φ (·) represents the regression quantity formed by past outputs and exogenous inputs. d represents the time delay, and tanh(·) represents the selected hyperbolic tangent activation function.

[0040] (2) Couple “optimized output + historical input - output data” as neural network input. To further capture nonlinear and unmodeled dynamic characteristics, the optimized output obtained in step 2 is... y opt The historical inputs and outputs of the GFM inverter are combined to form the input vector of the neural network. This relationship can be expressed as: (9) in, y opt-NN To optimize and enhance the output of the neural network; u This includes historical output data of the GFM inverter, instantaneous input data of the GFM inverter, and historical input data of the GFM inverter. λ Represents the hyperparameters of a neural network (number of network layers, number of neurons, time, and delay steps); N (·) represents the NARX neural network structure.

[0041] (3) Data preprocessing, training the neural network and adjusting the hyperparameters to minimize the prediction error; Before training, the collected data is normalized to improve convergence speed and mitigate the impact of outliers. The dataset is then divided into training and validation subsets. During training, network structure parameters (e.g., activation function, number of layers, number of neurons, time, and latency steps) are adjusted based on dynamic characteristics until the prediction error is minimized.

[0042] Step 4: Equivalent Modeling of GFM Inverter like Figure 1 As shown, based on the Norton equivalence principle, the optimized and enhanced NARX neural network obtained in step 3 is integrated with the controlled current source into the main circuit to form the "phasor" of the GFM inverter. – The hybrid architecture of "neural networks" enables the equivalent representation of GFM inverters. This representation effectively captures all the unmodeled dynamic and nonlinear characteristics of GFM inverters while maintaining physical interpretability and providing flexibility to adapt to system uncertainties and unmodeled dynamic behaviors, ensuring the engineering feasibility of the method.

[0043] Step 5: Model Validation and Performance Evaluation (1) Establish a test system and set up test scenarios: To verify the effectiveness of the proposed optimized augmented neural network modeling method, a test system was built in MATLAB / Simulink based on the equivalent phasor domain model of the GFM inverter obtained in step 4. The system consists of four GFM inverters, line impedance, AC load, and AC power grid.

[0044] In this example, the input and output vectors consist of the measured inverter output voltage and current, respectively. The dataset is divided into training and validation subsets and preprocessed with normalization. Furthermore, offline training is performed on the normalized dataset to improve convergence speed and reduce sensitivity to outliers. The optimized model identifies the droop coefficient as... m p =0.051 and n q =0.049, and then substitute the identified droop coefficient into the phasor domain mathematical equation (3) to solve the optimized output current phasor. This optimized output current provides physical information for subsequent neural network enhancement.

[0045] Then, an optimized and enhanced NARX neural network is constructed, consisting of two hidden layers (each containing eight neurons) and a two-step time delay. For example... Figure 2 As shown, the input vector includes the measured voltage at PCC. V pcc and current output optimized based on physical model I o ’ The corresponding current phasor at PCC I pcc As output, the following perturbation scenarios are considered for a comprehensive evaluation of the proposed model: (1) Voltage amplitude disturbance: The grid voltage amplitude increases by 2%, 4%, 6%, 8% and 10% respectively, and then returns to the nominal value.

[0046] (2) Phase angle disturbance: The grid voltage phase angle is disturbed by 10°, 15°, 20°, 25° and 30° respectively, and then restored to the nominal value.

[0047] (3) Frequency disturbance: The grid frequency varies by 0.2Hz, 0.4Hz, 0.6Hz, 0.8Hz and 1Hz respectively around the nominal 50 Hz.

[0048] (4) High Voltage Ride-through (HVRT): The grid voltage rises from 1 p.u. to 1.5 p.u. and then decreases to 1.4 p.u., 1.3 p.u. and 1.2 p.u. respectively.

[0049] (5) Low voltage ride-through (LVRT): The grid voltage drops from 1 p.u. to 0 p.u., gradually recovers to 0.8 p.u., and finally returns to 1 p.u.

[0050] In all test scenarios, the optimized augmented neural network exhibited high fidelity in reproducing the dynamic response of the PCC current phasor.

[0051] (2) NRMSE is used as the evaluation indicator: Based on NRMSE, the system's dynamic fit can be calculated as follows: (11) in Y Represents the measurement output vector; Ỹ This represents the output vector estimated by the model. The higher the goodness of fit, the higher the model accuracy.

[0052] like Figure 3-7 As shown, the output of the proposed model was compared with test data under conditions of voltage amplitude variation, phase angle disturbance, frequency deviation, HVRT, and LVRT. The results confirm that the proposed model can accurately capture transient and steady-state behavior. The goodness of fit of the optimized enhanced neural network model in the five scenarios is summarized as follows: Voltage amplitude variation: current amplitude 98.8%, phase angle 97.2%; Phase angle disturbance: current amplitude 95.6%, phase angle 96.2%; Frequency deviation: current amplitude 95.8%, phase angle 97.3%; HVRT condition: current amplitude 94.1%, phase angle 93.8%; LVRT condition: current amplitude 97.3%, phase angle 96.1%.

[0053] The proposed optimized and enhanced neural network model exhibits high fitting accuracy (both current amplitude and phase fitting accuracy exceed 90%) under various scenarios, including short-circuit faults, voltage amplitude variations, phase disturbances, frequency fluctuations, and high / low voltage ride-through. The results demonstrate that the proposed method effectively improves the ability to capture dynamic responses and the model's generalization performance while maintaining physical interpretability.

[0054] Example 3 This embodiment provides a dynamic identification system for a GFM inverter-dominated power distribution system, comprising: The mathematical model building unit is used to build the phasor domain mathematical model of the GFM inverter in the phasor domain. The phasor acquisition unit is used to construct a nonlinear least squares optimization model based on the phasor domain mathematical model, obtain the parameter estimates corresponding to the GFM inverter, substitute the parameter estimates into the phasor domain mathematical model, and solve for the optimized output phasor. The model optimization unit is used to optimize the pre-built system dynamic prediction model based on the NARX neural network using the obtained optimized output phasor, so as to obtain the optimized system dynamic prediction model. The dynamic prediction unit is used to perform dynamic prediction of the power distribution system dominated by the GFM inverter based on the pre-built equivalent model of the GFM inverter in the phasor domain and the obtained optimized system dynamic prediction model.

[0055] Example 4 This embodiment also provides a computing device. The computing device includes a bus, a processor, a memory, and a communication interface. The processor, memory, and communication interface communicate with each other via the bus. The computing device can be a server or a terminal device. It should be understood that this application does not limit the number of processors and memory in the computing device.

[0056] A bus can be a Peripheral Component Interconnect (PCI) bus or an Extended Industry Standard Architecture (EISA) bus, etc. Buses can be categorized as address buses, data buses, control buses, etc. For ease of representation, a bus can include a path for transmitting information between various components of a computing device (e.g., memory, processor, communication interfaces).

[0057] The processor may include any one or more of the following: central processing unit (CPU), graphics processing unit (GPU), tensor processing unit (TPU), application specific integrated circuit (ASIC), field-programmable gate array (FPGA), microprocessor (MP), or digital signal processor (DSP).

[0058] The memory may include volatile memory, such as random access memory (RAM). The processor may also include non-volatile memory, such as read-only memory (ROM), flash memory, hard disk drive (HDD), or solid state drive (SSD).

[0059] The memory stores executable program code, which the processor executes to implement the functions of the aforementioned units, thereby achieving, for example, the method described in Embodiment 1. That is, the memory may store instructions for the methods and functions relating to the computing device in any of the above embodiments.

[0060] The communication interface uses transceiver modules such as, but not limited to, network interface cards and transceivers to enable communication between computing devices and other devices or communication networks.

[0061] Example 5 This embodiment also provides a computer-readable storage medium storing computer instructions that, when executed by a processor, cause the processor to perform the methods and functions of the computing device involved in any of the above embodiments.

[0062] Generally, the various embodiments of this disclosure can be implemented in hardware or dedicated circuitry, software, logic, or any combination thereof. Some aspects can be implemented in hardware, while others can be implemented in firmware or software, which can be executed by a controller, microprocessor, or other computing device. Although various aspects of the embodiments of this disclosure are shown and described as block diagrams, flowcharts, or represented using some other illustration, it should be understood that the blocks, apparatuses, systems, techniques, or methods described herein can be implemented as, as non-limiting examples, in hardware, software, firmware, dedicated circuitry or logic, general-purpose hardware or controllers or other computing devices, or some combination thereof.

[0063] Example 6 This embodiment provides at least one computer program product tangibly stored on a non-transitory computer-readable storage medium. The computer program product includes computer-executable instructions, such as instructions included in program modules, which execute in a device on a target real or virtual processor to perform the processes / methods as described above with reference to the accompanying drawings. Typically, program modules include routines, programs, libraries, objects, classes, components, data structures, etc., that perform specific tasks or implement specific abstract data types. In various embodiments, the functionality of program modules can be combined or divided among program modules as needed. The machine-executable instructions for the program modules can execute within a local or distributed device. In a distributed device, the program modules can reside in both local and remote storage media.

[0064] Computer program code used to implement the methods of this disclosure may be written in one or more programming languages. This computer program code may be provided to a processor of a general-purpose computer, a special-purpose computer, or other programmable data processing apparatus, such that when executed by the computer or other programmable data processing apparatus, the program code causes the functions / operations specified in the flowcharts and / or block diagrams to be performed. The program code may be executed entirely on a computer, partially on a computer, as a stand-alone software package, partially on a computer and partially on a remote computer, or entirely on a remote computer or server.

[0065] In the context of this disclosure, computer program code or related data may be carried on any suitable carrier to enable a device, apparatus, or processor to perform the various processes and operations described above. Examples of carriers include signals, computer-readable media, and so on. Examples of signals may include electrical, optical, radio, sound, or other forms of propagation signals, such as carrier waves, infrared signals, etc.

[0066] Computer-readable media can be any tangible medium that contains or stores programs for or relating to an instruction execution system, apparatus, or device, or a data storage device such as a data center containing one or more available media. Computer-readable media can be computer-readable signal media or computer-readable storage media. Computer-readable media can be, but is not limited to, electronic, magnetic, optical, electromagnetic, infrared, or semiconductor systems, apparatus, or devices, or any suitable combination thereof. More detailed examples of computer-readable storage media include electrical connections with one or more wires, portable computer disks, hard disks, random access memory (RAM), read-only memory (ROM), erasable programmable read-only memory (EPROM or flash memory), optical storage devices, magnetic storage devices, or any suitable combination thereof.

[0067] The above-described embodiments are only used to illustrate the technical solutions of this application, and are not intended to limit them. Although this application has been described in detail with reference to the foregoing embodiments, those skilled in the art should understand that modifications can still be made to the technical solutions described in the foregoing embodiments, or equivalent substitutions can be made to some of the technical features. Such modifications or substitutions do not cause the essence of the corresponding technical solutions to deviate from the spirit and scope of the technical solutions of the embodiments of this application, and should all be included within the protection scope of this application.

Claims

1. A dynamic identification method for GFM inverter master leading power distribution system, characterized in that, The method comprises the following steps: Step 1, a phasor domain mathematical model of the GFM inverter is established in the phasor domain; Step 2, a nonlinear least squares optimization model is constructed based on the phasor domain mathematical model, parameter estimation values corresponding to the GFM inverter are obtained, the parameter estimation values are substituted into the phasor domain mathematical model, and an optimized output phasor is solved; Step 3, the optimized output phasor obtained in Step 2 is used to optimize a system dynamic prediction model pre-constructed based on a NARX neural network, and an optimized system dynamic prediction model is obtained; Step 4, based on an equivalent model of the GFM inverter in the phasor domain pre-constructed, the optimized system dynamic prediction model obtained in Step 3 is used to dynamically predict a GFM inverter dominant power distribution system.

2. A dynamic identification method for GFM inverter master leading power distribution system according to claim 1, characterized in that, The phasor domain mathematical model of the GFM inverter comprises apparent power expressions, average active power expressions, average reactive power expressions, voltage amplitude and phase angle expressions, frequency expressions, reference voltage expressions and output current expressions of the GFM inverter.

3. A dynamic identification method for GFM inverter master leading power distribution system according to claim 1, characterized in that, An expression of the nonlinear least squares optimization model is: wherein, to optimize the output; are inverter-side measurements obtained by a phasor domain mathematical model of the GFM inverter; p represents a set of unknown system parameters to be identified.

4. A dynamic identification method for GFM inverter master leading power distribution system according to claim 1, characterized in that, The optimized output obtained in Step 2 is used to optimize the system dynamic prediction model pre-constructed based on the NARX neural network, and the specific method is: A system dynamic prediction model is constructed based on the NARX neural network; Optimized output data obtained in Step 2, historical input data of the GFM inverter, historical output data of the GFM inverter and instantaneous input data of the GFM inverter are combined to form an input vector of the neural network; The formed input vector is used to optimize the system dynamic prediction model, and an optimized system dynamic prediction model is obtained.

5. A dynamic identification method for GFM inverter master leading power distribution system according to claim 1 or 4, characterized in that, An expression of the system dynamic prediction model is: wherein, y opt-NN to optimize the output of the enhanced neural network; u input vector to the neural network; λ denotes a neural network hyperparameter; N (·) denotes a NARX neural network structure.

6. A dynamic identification method for GFM inverter master leading power distribution system according to claim 1, characterized in that, The equivalent model of the GFM inverter in the phasor domain pre-constructed, and the specific method is: Based on the Norton equivalent principle, the optimized system dynamic prediction model obtained in Step 3 is integrated into a main circuit together with a controlled current source, a phasor-neural network hybrid architecture is constructed, and an equivalent model of the GFM inverter in the phasor domain is obtained.

7. A dynamic identification system for a GFM inverter master lead power distribution system, characterized by, The method of claim 1 comprises: A mathematical model establishment unit is configured to establish a phasor domain mathematical model of the GFM inverter in the phasor domain; A phasor acquisition unit is configured to construct a nonlinear least squares optimization model based on the phasor domain mathematical model, obtain parameter estimation values corresponding to the GFM inverter, substitute the parameter estimation values into the phasor domain mathematical model, and solve an optimized output phasor; A model optimization unit is configured to use the optimized output phasor to optimize a system dynamic prediction model pre-constructed based on a NARX neural network, and obtain an optimized system dynamic prediction model; A dynamic prediction unit is configured to use the optimized system dynamic prediction model to dynamically predict a GFM inverter dominant power distribution system based on an equivalent model of the GFM inverter in the phasor domain pre-constructed.

8. An electronic device, comprising: The electronic device comprises a processor and a memory, and the memory stores computer instructions, when the computer instructions are executed by the processor, the electronic device executes the method in any one of claims 1 to 6. The electronic device comprises a processor and a memory, and the memory stores computer instructions, when the computer instructions are executed by the processor, the electronic device executes the method in any one of claims 1 to 6.

9. A computer program product, characterised in that, The computer program product has computer executable instructions embodied thereon, the computer executable instructions, when executed, implement the method of any one of claims 1 to 6.

10. A computer-readable storage medium, characterized in that, The computer readable storage medium has stored thereon computer executable instructions, the computer executable instructions, when executed by a processor, implement the method of any one of claims 1 to 6.

Images