Adaptive virtual impedance network type energy storage transient current limiting and voltage collaborative control method

Abstract

The application relates to the technical field of network-constructing energy storage control, and discloses a self-adaptive virtual impedance network-constructing energy storage transient current limiting and voltage collaborative control method, which collects real-time operation state data of a network-constructing energy storage system, pre-processes the operation state data, constructs a state vector and a control vector, inputs the state vector and the control vector into a transient control model, predicts the system state at the next moment and in a predicted time domain by using the transient control model, optimizes an optimal control amount under physical constraints, power constraints, SOC constraints and voltage and current limiting constraints by a prediction controller, dynamically adjusts a virtual resistor and a virtual inductor, wherein the weight and the control parameter are optimized by a particle swarm algorithm, and the optimal control amount is sent to the network-constructing energy storage system to perform switching from transient control to steady control; and the application significantly improves the transient control performance of the network-constructing energy storage.

Description

Technical Field

[0001] This invention relates to the field of grid-type energy storage control technology, specifically to an adaptive virtual impedance grid-type energy storage transient current limiting and voltage coordinated control method. Background Technology

[0002] As new power systems rapidly evolve towards higher proportions of renewable energy and higher proportions of power electronic equipment, the proportion of traditional synchronous generators is declining, system rotational inertia is decreasing, and anti-interference capabilities are weakening, posing a severe challenge to transient stability. Grid-based energy storage, as a voltage source device with independent grid-building capabilities, simulates the inertia and damping characteristics of traditional synchronous generators through control strategies such as virtual synchronous machines. It can provide voltage support, inertia support, and frequency and voltage regulation capabilities for the power grid, making it a key technical means to ensure the grid integration and consumption of high proportions of renewable energy. When transient faults (such as short circuits, sudden load changes, and sudden changes in renewable energy power) occur, grid-based energy storage systems will generate huge transient inrush currents, which will not only damage power electronic devices such as energy storage converters, but also cause a severe drop in the voltage at the common coupling point, leading to problems such as system power angle instability and power oscillation, and even causing the energy storage system to disconnect from the grid, threatening the safe operation of the power grid. Therefore, transient current limiting and voltage stability control are the core technical bottlenecks for the large-scale application of grid-based energy storage systems.

[0003] In existing technologies, the adjustment of adaptive virtual impedance often relies on a single parameter, resulting in low adjustment accuracy, slow response, and failure to consider the impact of power angle margin on control performance, making it difficult to achieve dynamic coordination between transient current limiting and voltage stability. Summary of the Invention

[0004] The purpose of this invention is to solve the above problems by designing an adaptive virtual impedance network-based energy storage transient current limiting and voltage coordinated control method.

[0005] This invention provides an adaptive virtual impedance network-based energy storage transient current limiting and voltage coordinated control method, which includes the following steps: Collect real-time operating status data of grid-type energy storage systems, and construct state vectors and control vectors after preprocessing the operating status data; The virtual impedance is embedded into the physical information neural network in the form of physical constraints to obtain the transient control model, and the state vector and control vector are input into the transient control model. A transient control model is used to predict the system state at the next moment and within the prediction time domain, and the prediction results are input into the predictive controller. The optimal control quantity is obtained by predicting the controller under physical constraints, power constraints, SOC constraints and voltage and current limiting constraints, and the virtual resistance and virtual inductance are dynamically adjusted. The weights and control parameters are optimized by particle swarm optimization algorithm. The optimal control quantity is sent to the grid-type energy storage system to perform the switch from transient control to steady-state control.

[0006] Optionally, in the first implementation of the present invention, the step of collecting real-time operating status data of the grid-type energy storage system, preprocessing the operating status data, and constructing a state vector and a control vector includes: Real-time operation status data of grid-type energy storage system is collected, including at least grid connection point voltage, output current, active power, reactive power, grid equivalent impedance, virtual synchronous machine power angle and energy storage state of charge. An adaptive detection algorithm based on variable density is used to process outliers in real-time operating status data to obtain processed valid data. The min-max standardization algorithm is used to standardize the processed effective data. Based on the standardized data, a state vector including system operating state parameters and a control vector including control parameters to be adjusted are constructed.

[0007] Optionally, in a second implementation of the present invention, the variable-density-based adaptive detection algorithm performs outlier processing on the real-time operating status data to obtain processed valid data, including: The collected real-time operating status data is organized into a continuous time-series data sample set. The sample set is then divided into local densities. The data distribution density within the neighborhood of each data sample is calculated, and the boundaries between high-density and low-density regions are defined. Data samples that are outside the high-density area and in the low-density area are identified as outlier samples. The outlier samples are located, and normal operating data at adjacent time points before and after the outlier samples are selected to complete and replace the data, and the processed valid data is output.

[0008] Optionally, in a third implementation of the present invention, the step of embedding the virtual impedance into a physical information neural network in the form of physical constraints to obtain a transient control model, and inputting the state vector and control vector into the transient control model, includes: The physical constraints of the virtual impedance are obtained, and the basic structure of the physical information neural network is built. The basic structure includes an input layer, a hidden layer and an output layer. The input layer is used to receive the state vector and the control vector. The hidden layer extracts features from the input data through the activation function. The output layer is used to output transient control parameters. The physical constraint expression of virtual impedance is embedded into the loss function of the physical information neural network. The physical information neural network with embedded constraints is initialized by setting the initial weights and biases of the network to obtain the transient control model. The state vector and control vector are then input into the transient control model.

[0009] Optionally, in the fourth implementation of the present invention, the constraints include at least virtual impedance limit constraints, grid impedance matching constraints, transient current suppression constraints, voltage sag compensation constraints, power angle stability margin constraints, power oscillation suppression constraints, and equipment safe operation constraints.

[0010] Optionally, in the fifth implementation of the present invention, the step of using a transient control model to predict the system state at the next moment and within the prediction time domain, and inputting the prediction result into the prediction controller, includes: The input state vector and control vector are fed into the input layer of the transient control model, and features are extracted through the hidden layer to obtain the state prediction result at the prediction time. Based on the state prediction results at the prediction time, the operating state data at the next time is first deduced, and then the system state data at all times in the entire prediction time domain are deduced sequentially. The prediction results at a single moment and the prediction results at consecutive moments in the time domain are arranged and integrated in chronological order to obtain the prediction result, which is then input into the prediction controller.

[0011] Optionally, in the sixth implementation of the present invention, the optimal control quantity is obtained by the predictive controller under physical constraints, power constraints, SOC constraints, and voltage and current limiting constraints, and the virtual resistance and virtual inductance are dynamically adjusted. Initialize the particle swarm, where each particle corresponds to a set of weights and control parameters; The fitness value of each particle is calculated, and the weight coefficients of the cost function, the upper and lower limits of the virtual impedance, and the controller parameters are iteratively optimized to obtain the globally optimal parameter combination. The global optimal parameter combination and prediction results are input into the predictive controller, which performs optimization calculations under the constraints of physical constraints, power balance constraints, energy storage state of charge operation constraints, and voltage and current safety limit constraints, and outputs the optimal virtual resistance and virtual inductance adjustment amounts that can be controlled in a coordinated manner.

[0012] Optionally, in the seventh implementation of the present invention, the step of calculating the fitness value of each particle and iteratively optimizing the cost function weight coefficients, virtual impedance upper and lower limits, and controller parameters to obtain the globally optimal parameter combination includes: Compare the current fitness value of each particle with the historical best fitness value, and update the best position and corresponding best fitness value of each particle. Iterate through the individual best positions of all particles, compare and update the global best position of the population with the corresponding global best fitness value; Calculate the current iteration weight, update the particle velocity based on the individual particle's optimal position and the population's global optimal position, and then update the particle position vector based on the velocity; Perform a local search operation on the updated particle position to generate new candidate positions in the neighborhood of the particle's current position and replace the particle position with the better fitness. Determine if the current iteration count has reached the maximum iteration count. If so, terminate the iteration and output the globally optimal position to obtain the globally optimal parameter combination.

[0013] Optionally, in the eighth implementation of the present invention, the step of sending the optimal control quantity to the grid-type energy storage system to perform the switching from transient control to steady-state control includes: Real-time monitoring of system fault indicators, output current amplitude, grid connection point voltage, power angle stability, and energy storage operation status to determine whether the fault has subsided and whether the system meets the steady-state recovery conditions; When the steady-state recovery condition is met, the virtual impedance parameter and control loop parameter of the transient operation will be adjusted from the current transient value to the preset steady-state operation value. The optimal control quantity after smooth transition is sent to the energy storage converter drive unit to complete the switch from transient control mode to steady-state control mode.

[0014] Optionally, in the ninth implementation of the present invention, the apparatus for implementing the adaptive virtual impedance network-type energy storage transient current limiting and voltage coordinated control method includes: The preprocessing module is used to collect real-time operating status data of the grid-type energy storage system, and after preprocessing the operating status data, construct state vectors and control vectors. The embedding module is used to embed virtual impedance into the physical information neural network in the form of physical constraints to obtain a transient control model, and input the state vector and control vector into the transient control model; The prediction module is used to predict the system state at the next moment and within the prediction time domain using a transient control model, and inputs the prediction results into the prediction controller. The optimization module is used to obtain the optimal control quantity by predicting the controller under physical constraints, power constraints, SOC constraints and voltage and current limiting constraints, and dynamically adjust the virtual resistance and virtual inductance. The weights and control parameters are optimized by the particle swarm optimization algorithm. The switching module is used to send the optimal control quantity to the grid-type energy storage system to perform the switching from transient control to steady-state control.

[0015] The technical solution provided by this invention involves collecting real-time operating status data of a grid-type energy storage system, preprocessing the operating status data, and constructing state vectors and control vectors. Virtual impedance is embedded into a physical information neural network in the form of physical constraints to obtain a transient control model, and the state vectors and control vectors are input into the transient control model. The transient control model is used to predict the system state at the next moment and within the predicted time domain, and the prediction results are input into a predictive controller. The predictive controller optimizes the optimal control quantity under physical constraints, power constraints, SOC constraints, and voltage and current limiting constraints, dynamically adjusting the virtual resistance and virtual inductance, wherein the weights and control parameters are optimized using a particle swarm optimization algorithm. The optimal control quantity is then sent to the grid-type energy storage system to perform the switch from transient control to steady-state control. This invention significantly improves the transient control performance of grid-type energy storage; dynamically adjusting the virtual impedance can quickly suppress fault inrush currents while stabilizing the grid connection point voltage; it features fast response, high accuracy, and strong robustness, effectively improving operational reliability. Attached Figure Description

[0016] Various other advantages and benefits will become apparent to those skilled in the art upon reading the following detailed description of preferred embodiments. The accompanying drawings are for illustrative purposes only and are not intended to limit the invention.

[0017] Figure 1 A schematic diagram of the first embodiment of the adaptive virtual impedance network-based energy storage transient current limiting and voltage coordinated control method provided by the present invention; Figure 2 A schematic diagram of the second embodiment of the adaptive virtual impedance network-based energy storage transient current limiting and voltage coordinated control method provided in this invention; Figure 3 This is a schematic diagram of the adaptive virtual impedance network-type energy storage transient current limiting and voltage coordinated control device provided in an embodiment of the present invention. Detailed Implementation

[0018] The terms “first,” “second,” “third,” “fourth,” etc. (if present) in the specification, claims, and accompanying drawings of this invention are used to distinguish similar objects and are not necessarily used to describe a particular order or sequence. It should be understood that such data can be interchanged where appropriate so that the embodiments described herein can be implemented in orders other than those illustrated or described herein. Furthermore, the terms “comprising” or “having,” and any variations thereof, are intended to cover a non-exclusive inclusion; for example, a process, method, apparatus, product, or device that comprises a series of steps or units is not necessarily limited to those steps or units explicitly listed, but may include other steps or units not explicitly listed or inherent to such processes, methods, products, or devices.

[0019] For ease of understanding, the specific process of the embodiments of the present invention is described below. Please refer to [link / reference]. Figure 1 A schematic diagram of the first embodiment of the adaptive virtual impedance network-based energy storage transient current limiting and voltage coordinated control method provided by this invention. The method specifically includes the following steps: Step 101: Collect real-time operating status data of the grid-type energy storage system, preprocess the operating status data, and construct the state vector and control vector; In this embodiment, a high-precision multi-channel synchronous acquisition device is used to continuously and in real time acquire the full-condition operation information of the grid-type energy storage system. According to the set high-frequency sampling sequence, the device synchronously acquires multiple key operating status data such as the instantaneous value of the three-phase voltage at the grid connection point, the three-phase current output of the converter, the real-time active power, the real-time reactive power, the grid equivalent impedance value, the power angle of the virtual synchronous machine, the state of charge of the energy storage battery, and the system frequency, ensuring that all acquired signals are strictly aligned in time and completely covered in dimensions. The collected real-time operating status data is organized into a continuous time-series data sample set. The sample set is divided into local densities, and the data distribution density within the neighborhood of each data sample is calculated. The boundaries of high-density and low-density regions are defined. Data samples that are outside the high-density region and in the low-density region are identified as outlier samples. The outlier samples are located, and the normal operating data at the time adjacent to the outlier sample are selected for data completion and replacement. The processed valid data is then output. The min-max standardization algorithm is used to uniformly dimensionless process the effective data after outlier processing. Parameters with different physical meanings, numerical ranges and dimensions, such as voltage, current, power, impedance, power angle and state of charge, are uniformly mapped to the preset standard numerical range through linear transformation, eliminating the calculation deviation caused by the difference in numerical magnitude between parameters. Then, according to the preset state space structure, the standardized operating data is combined to construct a state vector containing complete system operating information. At the same time, virtual impedance, controller gain and other optimization adjustment variables are encapsulated into control vectors.

[0020] Step 102: Embed the virtual impedance into the physical information neural network in the form of physical constraints to obtain the transient control model, and input the state vector and control vector into the transient control model; In this embodiment, based on the operational requirements of transient current limiting and voltage coordinated control in a grid-type energy storage system, a comprehensive set of physical constraints corresponding to the virtual impedance are extracted and determined. These constraints include upper and lower limits for virtual impedance, dynamic matching constraints for grid equivalent impedance, transient inrush current suppression constraints, grid connection point voltage dip compensation constraints, virtual synchronous machine power angle stability margin constraints, system power oscillation suppression constraints, and converter equipment safe operation constraints. This ensures that all constraints cover electrical characteristics, control objectives, equipment boundaries, and stability requirements. A basic network structure suitable for transient control is built according to the standard architecture of a physical information neural network, sequentially constructing an input layer, multiple hidden layers, and an output layer. The input layer receives and carries the pre-processed system state vector. Along with the control vector, the hidden layer uses a set activation function to perform nonlinear mapping and deep feature extraction on the input data. The output layer is used to output the calculation results related to transient control, such as virtual impedance and control parameters, to ensure that the network structure is adapted to multi-dimensional input and multi-objective control output. The various physical constraints of the determined virtual impedance are transformed into embeddable soft constraints and added to the loss function construction process of the physical information neural network. This forces the network to follow the physical laws of the power system during inference calculation. Then, the network after embedding constraints is initialized, and the initial weights, initial biases and basic inference parameters are set reasonably. Finally, a transient control model that integrates data-driven and physical rules is formed, and the state vector and control vector are input into the model to perform subsequent state prediction.

[0021] Step 103: Use a transient control model to predict the system state at the next moment and within the prediction time domain, and input the prediction results into the predictive controller; Step 104: The optimal control quantity is obtained by predicting the controller under physical constraints, power constraints, SOC constraints and voltage and current limiting constraints, and the virtual resistance and virtual inductance are dynamically adjusted. The weights and control parameters are optimized by particle swarm optimization algorithm. In this embodiment, the population initialization operation is completed according to the operation rules of the improved particle swarm algorithm. The number of particles, the number of iterations, the search interval and the velocity boundary are set according to the control target and parameter range. The cost function weight coefficient, the virtual impedance upper and lower limits and the key adjustment parameters of the controller are uniformly encoded into the position vector corresponding to each particle. Initial particle positions and initial particle velocities that meet the constraints are randomly generated to construct a complete initial iterative population. Each particle's fitness value in the current iteration cycle is compared with the best fitness value recorded in its own historical iterations. An update judgment is performed based on the comparison results. If the current fitness is better, the historical best record is replaced. The individual best position and the matching individual best fitness value of the particle are updated simultaneously to refresh the individual best information of the particle. A global traversal and horizontal comparison are performed on the optimal position and optimal fitness value of each individual corresponding to all particles in the population. The information of the individual with the best fitness in the current population is selected. Based on this, the global optimal position of the population is updated, and the corresponding global optimal fitness value is updated simultaneously, providing a unified global guiding benchmark for the search direction of subsequent particles. Based on the adaptive inertia weight calculation rules set by the improved particle swarm algorithm, and combined with the ratio of the current iteration number to the total iteration number, the dynamic inertia weight used in this iteration is determined. The optimal position of the individual particle and the optimal position of the population are used as guiding terms. The velocity update rules are substituted to complete the calculation and update of the particle velocity. Then, the particle position vector is adjusted based on the updated velocity. For each particle position after velocity and position updates, a local fine-grained search operation is performed. Several new candidate positions are generated within a preset neighborhood of the particle's current position according to a set step size. The fitness value of each candidate position is calculated, and the original position of the particle is replaced with a candidate position with better fitness, thereby improving the local search capability and avoiding the algorithm from getting stuck in a local optimum. The algorithm continuously counts the number of iterations completed by the current algorithm and compares it with the preset maximum number of iterations. If the maximum number of iterations has been reached, the entire iterative optimization process is terminated immediately, and the final determined global optimal position of the population is output. The parameters of the optimal position are then analyzed and decomposed to obtain the global optimal parameter combination that meets the requirements of multi-objective optimization.

[0022] The globally optimal parameter combination obtained by iterative optimization and the system state prediction results output by the prediction module are synchronously input into the distributed model predictive controller. Under the common constraints of virtual impedance physical constraints, system power balance constraints, energy storage state of charge safety operation constraints, and voltage and current safety limiting constraints, rolling time-domain optimization calculation is performed. Through constraint judgment and optimal objective solution, the optimal virtual resistance adjustment and optimal virtual inductance adjustment that can simultaneously achieve transient current limiting and voltage coordinated support are output.

[0023] Step 105: Send the optimal control quantity to the grid-type energy storage system to perform the switch from transient control to steady-state control.

[0024] In this embodiment, by continuously collecting and analyzing the system's fault judgment indicators, converter output current amplitude, grid connection point voltage RMS value, virtual synchronous machine power angle stability, and energy storage battery state of charge, the system judges item by item whether the fault has completely subsided, whether the system electrical quantities have returned to the safe range, and whether the recovery conditions for exiting transient control and entering steady-state operation are met, according to the preset steady-state judgment threshold and stable duration. When the monitoring and judgment link confirms that the system has met the steady-state recovery conditions, the virtual impedance parameters, controller parameters, power loop parameters, etc., which are in operation during the transient control phase, are slowly, continuously, and without impact adjusted from the current transient operating values ​​to the preset steady-state operating rated values ​​according to the exponential smoothing gradual transition rule, avoiding parameter abrupt changes that may cause system oscillations or secondary disturbances. The optimal virtual impedance adjustment amount and control reference amount, etc., which have completed the smooth transition and meet the steady-state operation requirements, are encapsulated and time-calibrated, and sent to the energy storage converter drive execution unit according to the communication protocol to drive the converter to switch the operating mode, ultimately realizing the smooth and reliable switch of the system from the transient current limiting and voltage support control mode to the conventional steady-state operation control mode.

[0025] Please see Figure 2 A schematic diagram of the second embodiment of the adaptive virtual impedance network-based energy storage transient current limiting and voltage coordinated control method provided in this invention. The method includes: Step 201: Pass the input state vector and control vector into the input layer of the transient control model, extract features through the hidden layer, and obtain the state prediction result at the prediction time. In this embodiment, the state vector and control vector are completely fed into the input layer of the transient control model according to the data format and input order specified by the model. With the help of the multi-layer hidden layers of the physical information neural network, the multi-dimensional operating information contained in the vector, such as grid connection point voltage, output current, active and reactive power, grid equivalent impedance, virtual synchronous machine power angle, and energy storage state of charge, is deeply mined and fused. Combined with the embedded virtual impedance physical constraints and system dynamics laws, feature mapping and state reasoning are completed, and the accurate system state prediction result at the current prediction time is output.

[0026] Step 202: Based on the state prediction results at the prediction time, first deduce the operating state data at the next time, and then deduce the system state data at all times in the entire prediction time domain in a sequential manner. In this embodiment, the single-step state prediction result obtained at the current moment is used as the initial input condition. The physical constraints and state recursion rules followed by the transient control model are strictly followed to gradually deduce and calculate the system operating state data at the next moment. Then, the deduction result at the next moment is used as the initial state input for a new round. The same reasoning and calculation process is repeated and deduced sequentially to generate complete system state data corresponding to all discrete moments in the entire prediction time domain.

[0027] Step 203: Arrange and integrate the single-time prediction results and the prediction results of consecutive times in the time domain in chronological order to obtain the prediction results, and input the prediction results into the prediction controller.

[0028] In this embodiment, the prediction results at a single moment and the prediction state data generated at multiple consecutive moments in the entire prediction time domain are uniformly aligned in time sequence, amplitude limit normalization and data format integration according to the actual time sequence to form a continuous, complete and standardized system state prediction sequence. The processed prediction results are then input to the distributed model prediction controller according to the communication interface standard.

[0029] Figure 3 This is a schematic diagram of the structure of an adaptive virtual impedance grid-based energy storage transient current limiting and voltage coordinated control device 300 provided in an embodiment of the present invention. The adaptive virtual impedance grid-based energy storage transient current limiting and voltage coordinated control device 300 can vary significantly due to different configurations or performance characteristics. It may include one or more central processing units (CPUs) 310 (e.g., one or more processors) and a memory 320, and one or more storage media 330 (e.g., one or more mass storage devices) storing application programs 333 or data 332. The memory 320 and storage media 330 can be temporary or persistent storage. The program stored in the storage media 330 may include one or more modules (not shown in the diagram), and each module may include a series of instruction operations on the adaptive virtual impedance grid-based energy storage transient current limiting and voltage coordinated control device 300. Furthermore, the processor 310 can be configured to communicate with the storage medium 330 and execute a series of instruction operations in the storage medium 330 on the adaptive virtual impedance network type energy storage transient current limiting and voltage coordinated control device 300 to implement the method provided in the above embodiment.

[0030] The adaptive virtual impedance network-based energy storage transient current limiting and voltage coordinated control device 300 may also include one or more power supplies 340, one or more wired or wireless network interfaces 350, one or more input / output interfaces 360, and / or one or more operating systems 331, such as Windows Server, Mac OS X, Unix, Linux, FreeBSD, etc. Those skilled in the art will understand that... Figure 3 The adaptive virtual impedance network-type energy storage transient current limiting and voltage coordinated control device structure shown does not constitute a limitation on the computer device provided by the present invention. It may include more or fewer components than shown, or combine certain components, or have different component arrangements.

[0031] The present invention also provides a computer-readable storage medium, which can be a non-volatile computer-readable storage medium or a volatile computer-readable storage medium. The computer-readable storage medium stores instructions that, when executed on a computer, cause the computer to perform the various steps of the adaptive virtual impedance network-type energy storage transient current limiting and voltage coordinated control method provided in the above embodiments.

[0032] Those skilled in the art will clearly understand that, for the sake of convenience and brevity, the specific working process of the above-described equipment or apparatus or unit can be referred to the corresponding process in the foregoing method embodiments, and will not be repeated here.

[0033] If the integrated unit is implemented as a software functional unit and sold or used as an independent product, it can be stored in a computer-readable storage medium. Based on this understanding, the technical solution of the present invention, in essence, or the part that contributes to the prior art, or all or part of the technical solution, can be embodied in the form of a software product. This computer software product is stored in a storage medium and includes several instructions to cause a computer device (which may be a personal computer, server, or network device, etc.) to execute all or part of the steps of the methods described in the various embodiments of the present invention. The aforementioned storage medium includes various media capable of storing program code, such as USB flash drives, portable hard drives, read-only memory (ROM), random access memory (RAM), magnetic disks, or optical disks.

[0034] The foregoing has shown and described the basic principles, main features, and advantages of the present invention. Those skilled in the art should understand that the present invention is not limited to the above embodiments. The embodiments and descriptions in the specification are merely preferred examples and are not intended to limit the invention. Various changes and modifications can be made to the invention without departing from its spirit and scope, and all such changes and modifications fall within the scope of the present invention as claimed. The scope of protection of the present invention is defined by the appended claims and their equivalents.

Claims

1. An adaptive virtual impedance network-based energy storage transient current limiting and voltage coordinated control method, characterized in that, The method includes the following steps: Collect real-time operating status data of grid-type energy storage systems, and construct state vectors and control vectors after preprocessing the operating status data; The virtual impedance is embedded into the physical information neural network in the form of physical constraints to obtain the transient control model, and the state vector and control vector are input into the transient control model. A transient control model is used to predict the system state at the next moment and within the prediction time domain, and the prediction results are input into the predictive controller. The optimal control quantity is obtained by predicting the controller under physical constraints, power constraints, SOC constraints and voltage and current limiting constraints, and the virtual resistance and virtual inductance are dynamically adjusted. The weights and control parameters are optimized by particle swarm optimization algorithm. The optimal control quantity is sent to the grid-type energy storage system to perform the switch from transient control to steady-state control.

2. The adaptive virtual impedance network-based energy storage transient current limiting and voltage coordinated control method as described in claim 1, characterized in that, The system collects real-time operating status data of the grid-type energy storage system. After preprocessing the operating status data, it constructs state vectors and control vectors, including: Real-time operation status data of grid-type energy storage system is collected, including at least grid connection point voltage, output current, active power, reactive power, grid equivalent impedance, virtual synchronous machine power angle and energy storage state of charge. An adaptive detection algorithm based on variable density is used to process outliers in real-time operating status data to obtain processed valid data. The min-max standardization algorithm is used to standardize the processed effective data. Based on the standardized data, a state vector including system operating state parameters and a control vector including control parameters to be adjusted are constructed.

3. The adaptive virtual impedance network-based energy storage transient current limiting and voltage coordinated control method as described in claim 2, characterized in that, The variable-density-based adaptive detection algorithm performs outlier processing on real-time operating status data to obtain processed valid data, including: The collected real-time operating status data is organized into a continuous time-series data sample set. The sample set is then divided into local densities. The data distribution density within the neighborhood of each data sample is calculated, and the boundaries between high-density and low-density regions are defined. Data samples that are outside the high-density area and in the low-density area are identified as outlier samples. The outlier samples are located, and normal operating data at adjacent time points before and after the outlier samples are selected to complete and replace the data, and the processed valid data is output.

4. The adaptive virtual impedance network-based energy storage transient current limiting and voltage coordinated control method as described in claim 1, characterized in that, The step of embedding virtual impedance into a physical information neural network in the form of physical constraints to obtain a transient control model, and inputting the state vector and control vector into the transient control model, includes: The physical constraints of the virtual impedance are obtained, and the basic structure of the physical information neural network is built. The basic structure includes an input layer, a hidden layer and an output layer. The input layer is used to receive the state vector and the control vector. The hidden layer extracts features from the input data through the activation function. The output layer is used to output transient control parameters. The physical constraint expression of virtual impedance is embedded into the loss function of the physical information neural network. The physical information neural network with embedded constraints is initialized by setting the initial weights and biases of the network to obtain the transient control model. The state vector and control vector are then input into the transient control model.

5. The adaptive virtual impedance network-based energy storage transient current limiting and voltage coordinated control method as described in claim 4, characterized in that, The constraints include at least virtual impedance limit constraints, grid impedance matching constraints, transient current suppression constraints, voltage sag compensation constraints, power angle stability margin constraints, power oscillation suppression constraints, and equipment safe operation constraints.

6. The adaptive virtual impedance network-based energy storage transient current limiting and voltage coordinated control method as described in claim 1, characterized in that, The process of using a transient control model to predict the system state at the next moment and within the prediction time domain, and inputting the prediction results into the prediction controller, includes: The input state vector and control vector are fed into the input layer of the transient control model, and features are extracted through the hidden layer to obtain the state prediction result at the prediction time. Based on the state prediction results at the prediction time, the operating state data at the next time is first deduced, and then the system state data at all times in the entire prediction time domain are deduced sequentially. The prediction results at a single moment and the prediction results at consecutive moments in the time domain are arranged and integrated in chronological order to obtain the prediction results, which are then input into the prediction controller.

7. The adaptive virtual impedance network-based energy storage transient current limiting and voltage coordinated control method as described in claim 1, characterized in that, The optimal control quantity is obtained by optimizing the predictive controller under physical constraints, power constraints, SOC constraints, and voltage and current limiting constraints, and the virtual resistance and virtual inductance are dynamically adjusted. Initialize the particle swarm, where each particle corresponds to a set of weights and control parameters; The fitness value of each particle is calculated, and the weight coefficients of the cost function, the upper and lower limits of the virtual impedance, and the controller parameters are iteratively optimized to obtain the globally optimal parameter combination. The global optimal parameter combination and prediction results are input into the predictive controller, which performs optimization calculations under the constraints of physical constraints, power balance constraints, energy storage state of charge operation constraints, and voltage and current safety limit constraints, and outputs the optimal virtual resistance and virtual inductance adjustment amounts that can be controlled in a coordinated manner.

8. The adaptive virtual impedance network-based energy storage transient current limiting and voltage coordinated control method as described in claim 7, characterized in that, The process of calculating the fitness value of each particle, iteratively optimizing the cost function weight coefficients, virtual impedance upper and lower limits, and controller parameters to obtain the globally optimal parameter combination includes: Compare the current fitness value of each particle with the historical best fitness value, and update the best position and corresponding best fitness value of each particle. Iterate through the individual best positions of all particles, compare and update the global best position of the population with the corresponding global best fitness value; Calculate the current iteration weight, update the particle velocity based on the individual particle's optimal position and the population's global optimal position, and then update the particle position vector based on the velocity; Perform a local search operation on the updated particle position to generate new candidate positions in the neighborhood of the particle's current position and replace the particle position with the better fitness. Determine if the current iteration count has reached the maximum iteration count. If so, terminate the iteration and output the globally optimal position to obtain the globally optimal parameter combination.

9. The adaptive virtual impedance network-based energy storage transient current limiting and voltage coordinated control method as described in claim 1, characterized in that, The step of distributing the optimal control quantity to the grid-type energy storage system to perform the switch from transient control to steady-state control includes: Real-time monitoring of system fault indicators, output current amplitude, grid connection point voltage, power angle stability, and energy storage operation status to determine whether the fault has subsided and whether the system meets the steady-state recovery conditions; When the steady-state recovery condition is met, the virtual impedance parameter and control loop parameter of the transient operation will be adjusted from the current transient value to the preset steady-state operation value. The optimal control quantity after smooth transition is sent to the energy storage converter drive unit to complete the switch from transient control mode to steady-state control mode.

10. An apparatus for implementing the adaptive virtual impedance network-based energy storage transient current limiting and voltage coordinated control method as described in claim 1, characterized in that, The device includes: The preprocessing module is used to collect real-time operating status data of the grid-type energy storage system, and after preprocessing the operating status data, construct state vectors and control vectors. The embedding module is used to embed virtual impedance into the physical information neural network in the form of physical constraints to obtain a transient control model, and input the state vector and control vector into the transient control model; The prediction module is used to predict the system state at the next moment and within the prediction time domain using a transient control model, and inputs the prediction results into the prediction controller. The optimization module is used to obtain the optimal control quantity by predicting the controller under physical constraints, power constraints, SOC constraints and voltage and current limiting constraints, and dynamically adjust the virtual resistance and virtual inductance. The weights and control parameters are optimized by the particle swarm optimization algorithm. The switching module is used to send the optimal control quantity to the grid-type energy storage system to perform the switching from transient control to steady-state control.

Images