Micro-grid active power filter cluster cooperative management system for sensitive load

Abstract

The application discloses a sensitive load-oriented micro-grid active power filter cluster cooperative management system and belongs to the field of micro-grid power quality management. The system comprises: a sensitive load portrait module for extracting sensitive load tolerance indexes and constructing a demand portrait; a cluster hierarchical cooperation module for dividing load priorities and generating master-slave cooperation strategies according to the portrait; a dynamic topology reconfiguration module for reconfiguring a control model through a consistency algorithm based on topology changes and cooperation strategies; a virtual impedance adjustment module for optimizing the virtual impedance values of each APF according to the reconfigured model and performing harmonic spectrum shifting; and a harmonic energy dredging module for directing the remaining harmonic energy to non-sensitive loads or energy storage devices when resources are saturated. The application can adapt to differentiated needs of sensitive loads and dynamic topology changes of micro-grids, realize the cooperative management of micro-grid harmonic energy and preferentially guarantee the power quality of sensitive loads.

Description

Technical Field

[0001] This invention discloses a collaborative governance system for active power filter clusters in microgrids oriented towards sensitive loads, belonging to the field of power quality governance in microgrids. Background Technology

[0002] Existing active power filter (APF) technologies for microgrids primarily focus on current sharing control and reliability design for multi-unit parallel operation. For example, patent CN111030113A employs a centralized control method to achieve current sharing control in parallel APF systems. However, this type of solution relies on a dedicated control unit to handle calculations and fault diagnosis; if this unit fails, the entire system becomes unusable, resulting in low system reliability. Furthermore, this control method requires additional communication equipment and lines, complicating control, hindering expansion and maintenance, and failing to consider the varying power quality requirements of different sensitive loads within the microgrid.

[0003] In terms of distributed coordination and governance, existing technologies have introduced a hierarchical control architecture. For example, patent CN117117864A proposes a method for pre-evaluating the harmonic mitigation effect of grid-connected inverters based on virtual impedance control. However, this method mainly focuses on predicting the harmonic suppression effect at the inverter's grid connection point. It fails to establish a refined governance model for the differentiated needs of sensitive load nodes under dynamic changes in microgrid topology, and it also fails to solve the adaptive reconfiguration problem of the APF cluster collaborative control model.

[0004] For power quality management of sensitive loads, existing technologies mainly focus on the passive prevention and control of single power quality problems. For example, patent CN119134341A involves a harmonic energy recovery device that suppresses and recovers high-order harmonics through chopping control. However, the above-mentioned solutions do not form a complete closed-loop management system from sensitive load demand profiling and cluster hierarchical coordination to dynamic topology reconfiguration and harmonic energy diversion. It is difficult to dynamically reconfigure the coordination strategy to prioritize the power quality of critical sensitive loads when the APF cluster capacity is saturated or the microgrid topology changes.

[0005] Therefore, there is an urgent need in the market for a collaborative governance system of microgrid active power filter clusters for sensitive loads to solve the above problems. Summary of the Invention

[0006] The purpose of this invention is to provide a microgrid active power filter (APF) cluster collaborative governance system for sensitive loads, including a sensitive load profiling module, a cluster hierarchical collaboration module, a dynamic topology reconfiguration module, a virtual impedance adjustment module, and a harmonic energy diversion module. The system extracts the tolerance indicators of each sensitive load node and constructs a demand profile through the sensitive load profiling module. The cluster hierarchical collaboration module prioritizes loads based on the profiles and generates a hierarchical collaborative control strategy that includes master-slave relationships and preemption logic. The dynamic topology reconfiguration module combines real-time topology change information with the hierarchical collaboration strategy to adaptively reconfigure the APF cluster control model using a consensus algorithm. The virtual impedance adjustment module collaboratively optimizes the fundamental and harmonic virtual impedances of each APF based on the reconfigured model and actively performs harmonic spectrum shifting. The harmonic energy diversion module directs remaining harmonic energy to non-sensitive loads or energy storage devices when governance resources are saturated. This invention overcomes the limitations of existing technologies that cannot simultaneously address the differentiated needs of sensitive loads and the dynamic topology changes of microgrids. It constructs a collaborative governance system that includes demand perception, strategy hierarchical management, model reconstruction, impedance optimization, and harmonic energy management, achieving refined collaborative governance of harmonic energy and priority assurance of power quality for critical sensitive loads.

[0007] The objective of this invention can be achieved through the following technical solutions: A collaborative governance system for microgrid active power filter clusters for sensitive loads includes the following modules: The sensitive load profiling module is used to extract the power quality demand characteristics of sensitive loads based on the power quality monitoring data of each sensitive load node in the microgrid, and obtain a sensitive load demand profile. The cluster hierarchical coordination module is used to prioritize and generate coordinated control strategies for the microgrid active power filter cluster according to the sensitive load demand profile, thus forming a hierarchical coordinated control strategy. The dynamic topology reconfiguration module is used to construct the APF electrical coupling matrix based on real-time topology awareness information and perform dynamic reconfiguration of the APF cluster collaborative control model through a consensus algorithm. The virtual impedance adjustment module is used to optimize and adjust the fundamental impedance and virtual impedance of each harmonic of each APF online based on the virtual impedance control strategy. The harmonic energy diversion module is used to perform spectrum transfer on the harmonic energy of the sensitive load bus based on the real-time harmonic detection results and the virtual impedance adjustment command, and the harmonic spectrum adjustment meets the power quality standard limit of the microgrid, and generates harmonic energy management command.

[0008] Preferably, the requirement profile construction process of the sensitive load profile module includes: By using the voltage and current waveform data of each sensitive load node, voltage distortion, voltage sag and harmonic tolerance indices are extracted and processed for the sensitive load to obtain the initial tolerance index set of the sensitive load. By using historical monitoring data and machine learning algorithms, the sensitive load demand profile is iteratively optimized and dynamically updated to form an updated sensitive load demand profile that can characterize the evolution of power quality demand during different operating periods, different load rates, and different operating conditions.

[0009] Preferably, the demand profile feature dimensions of the sensitive load profile module include: Based on the equipment type, core production process requirements, and real-time operating status of sensitive loads, a joint construction of voltage distortion sensitivity threshold and threshold over-limit tolerance time is performed on sensitive loads to build a sensitive load demand profile that includes voltage distortion sensitivity threshold and over-limit tolerance time. Based on the historical power quality event records of sensitive loads, the equipment's factory withstand curve data, and the field measured waveforms, the sensitive loads are subjected to a joint construction of voltage sag withstand curves and harmonic spectrum sensitive intervals. This results in a sensitive load demand profile that includes voltage sag amplitude-duration withstand curves, sensitive limits for the voltage content of each harmonic, and sensitive frequency bands for interharmonics.

[0010] Preferably, the master-slave relationship generation process of the cluster hierarchical collaboration module includes: Sensitive loads are prioritized according to the voltage sag tolerance index, harmonic sensitivity index, and load importance label defined by the user side in the sensitive load demand profile. The weight of the compensation current reference value of each APF is adjusted according to the priority classification result to generate a refined classification result including special-grade guaranteed loads, first-grade sensitive loads, and second-grade ordinary sensitive loads. Using the priority classification results, the master control filter is directionally allocated to each APF and the subordinate APF is dynamically adapted and configured, resulting in a hierarchical collaborative control strategy that includes a master-slave APF correspondence table, the electrical boundary definition of the auxiliary governance responsibility area of ​​each subordinate APF, and the master-slave role dynamic switching logic.

[0011] Preferably, the collaborative control strategy generated by the cluster hierarchical collaboration module includes: Based on the hierarchical collaborative control strategy, a refined allocation scheme is generated for the reference values ​​of the fundamental reactive current and the compensation current of each harmonic for each APF, resulting in a refined allocation scheme that includes the compensation current amplitude, phase command and compensation dead zone setting that each APF should output under the fundamental and characteristic harmonics. Referring to the hierarchical collaborative control strategy, the collaborative action timing of each APF during the microgrid grid-connected mode and islanded mode switching process, as well as the priority preemption logic of the APF corresponding to the high-priority load to the low-priority APF, are generated to obtain a collaborative control strategy that includes grid-connected and islanded mode switching time markers, transient process collaborative control rules, preemption trigger threshold conditions, and recovery exit mechanism after preemption.

[0012] Preferably, the collaborative control model reconstruction process of the dynamic topology reconstruction module includes: Based on the master-slave relationship definition and responsibility area division in the hierarchical collaborative control strategy, and combined with the topology change information caused by microgrid feeder switching, distributed power grid connection or disconnection, the original collaborative control model of the APF cluster is reconstructed through the consensus algorithm to generate an updated collaborative control model that meets the hierarchical collaboration requirements and adapts to the changes in electrical distance between nodes under the new topology. Based on the APF device's own network connection or disconnection events, communication link interruption or recovery status, and communication latency fluctuation information, the APF cluster collaborative control model is reconstructed online using a consensus algorithm to form an updated collaborative control model that adapts to the new communication topology, the number of new access nodes, and the changed communication condition constraints.

[0013] Preferably, the predictive control model update process of the dynamic topology reconfiguration module includes: A distributed model predictive control architecture is adopted to perform rolling predictions of future multi-step control commands for each APF, and to exchange and verify the consistency of predicted command values ​​between adjacent APFs. The objective function of the predictive control includes priority weight coefficients set according to the hierarchical cooperative control strategy, so as to obtain the neighbor APF predictive control command set and cooperative constraint boundary required for local optimization of each APF. By utilizing real-time topology sensing information and communication status monitoring results, the prediction time domain length, priority weight coefficients, and optimization objective function constraint boundaries of each APF are dynamically tuned, thus constructing an updated collaborative control model that can still maintain a dynamic balance between model prediction accuracy and control response speed under conditions of communication topology abrupt changes or time delay fluctuations.

[0014] Preferably, the virtual impedance optimization adjustment process of the virtual impedance adjustment module includes: Referring to the updated collaborative control model, the virtual resistance and virtual reactance values ​​of each APF are collaboratively optimized at the fundamental frequency to generate the optimal setting value of the fundamental virtual impedance of each APF. Using the updated collaborative control model, the virtual harmonic resistance and virtual harmonic reactance values ​​of each APF are decoupled and optimized at each characteristic harmonic frequency. Based on the dynamic adjustment of virtual impedance, the harmonic spectrum of the sensitive load bus is actively shifted from the sensitive frequency band to the non-sensitive frequency band, resulting in a virtual impedance adjustment command that includes the fundamental virtual impedance setting value, the virtual impedance setting values ​​of each harmonic, and the spectrum shifting collaborative parameters.

[0015] Preferably, the energy dissipation guiding process of the harmonic energy conduction module includes: Based on the real-time detection results that the total capacity of the APF cluster is insufficient to completely offset all harmonics and the virtual impedance adjustment has reached its upper limit, the remaining harmonic energy that cannot be completely offset is directed and injected into the non-sensitive load branch connected to the microgrid. The harmonic energy is consumed by the natural damping characteristics of the non-sensitive load, and an energy diversion control command containing the preferred result of the injected branch and the amplitude and phase of the injected current is generated.

[0016] Preferably, the energy storage and guiding process of the harmonic energy diversion module includes: Based on the monitoring results that the harmonic energy of the sensitive load bus exceeds the overall governance capacity of the APF cluster and the non-sensitive load branch has insufficient acceptance capacity, the remaining harmonic energy is directed to the energy storage device configured in the microgrid or the specially set energy dissipation circuit, forming an energy diversion control command that includes the charging and discharging power command of the energy storage device and the conduction angle control parameter of the energy dissipation circuit.

[0017] The beneficial effects of this invention are: This invention constructs a profile of sensitive load demand and establishes a cluster hierarchical coordination mechanism, enabling active power filters in microgrids to implement differentiated compensation control based on the power quality requirements of different sensitive loads, thereby prioritizing the power quality of critical load nodes when governance resources are limited.

[0018] This invention, through the collaborative design of a dynamic topology reconfiguration module and a virtual impedance adjustment module, enables the system to update the APF cluster collaborative control model based on changes in microgrid feeder switching, distributed power source access, and communication conditions, thereby improving the system's adaptability under topology change conditions.

[0019] This invention sets up a harmonic energy diversion module to guide the remaining harmonic energy in a directional manner when the APF cluster compensation capability is insufficient, so that the harmonic energy can be absorbed in non-sensitive load branches or energy storage devices, thereby reducing the accumulation of harmonic energy at the sensitive load bus. Attached Figure Description

[0020] Figure 1This is a schematic diagram of the structure of the microgrid active power filter cluster collaborative management system for sensitive loads according to the present invention. Detailed Implementation

[0021] The technical solutions of the present invention will be clearly and completely described below with reference to the embodiments of the present invention. Obviously, the described embodiments are only some embodiments of the present invention, and not all embodiments. Based on the embodiments of the present invention, all other embodiments obtained by those of ordinary skill in the art without creative effort are within the scope of protection of the present invention.

[0022] Example: Figure 1 As shown, the collaborative governance system for microgrid active power filter clusters oriented towards sensitive loads includes the following modules: The sensitive load profiling module is used to extract the power quality demand characteristics of sensitive loads based on the power quality monitoring data of each sensitive load node in the microgrid, and obtain a sensitive load demand profile. The cluster hierarchical coordination module is used to prioritize and generate coordinated control strategies for the microgrid active power filter cluster according to the sensitive load demand profile, thus forming a hierarchical coordinated control strategy. The dynamic topology reconfiguration module is used to construct the APF electrical coupling matrix based on real-time topology awareness information and perform dynamic reconfiguration of the APF cluster collaborative control model through a consensus algorithm. The virtual impedance adjustment module is used to optimize and adjust the fundamental impedance and virtual impedance of each harmonic of each APF online based on the virtual impedance control strategy. The harmonic energy diversion module is used to perform spectrum transfer on the harmonic energy of the sensitive load bus based on the real-time harmonic detection results and the virtual impedance adjustment command, and the harmonic spectrum adjustment meets the power quality standard limit of the microgrid, and generates harmonic energy management command.

[0023] In this embodiment, power quality demand characteristics are extracted from the sensitive loads based on the power quality monitoring data of each sensitive load node in the microgrid to obtain a demand profile of the sensitive loads. The specific implementation method is as follows: Power quality monitoring devices are installed at each sensitive load node of the microgrid to collect the instantaneous values ​​of the three-phase voltage and current at that node in real time. The sampling period is set to [value missing]. For the collected instantaneous voltage values and instantaneous current value Perform a Fast Fourier Transform to obtain the voltage amplitude of each harmonic. and current amplitude Where h represents the harmonic order, with the fundamental frequency h=1, and typical analysis up to h=50. Based on the above spectrum analysis results, the total harmonic distortion rate of the voltage at sensitive load nodes is calculated. and the voltage content of each harmonic : , In the formula, Let be the fundamental voltage RMS value, and H be the highest harmonic order considered. Simultaneously, voltage sag events are recorded by detecting the depth and duration of the voltage RMS drop. The residual voltage amplitude of the j-th voltage sag event is defined as . The duration is By statistically analyzing historical sag events, a dataset of sag tolerance characteristics for sensitive loads is constructed. For the voltage distortion sensitivity threshold, a limit vector of the harmonic voltage content under steady-state operating conditions is set. When any monitored value exceeds the corresponding limit and the duration exceeds the tolerance time, When this occurs, it is determined to be a voltage distortion exceeding the limit event.

[0024] To quantify the overall sensitivity of sensitive loads to different power quality issues, a sensitivity level assessment model is constructed. Taking into account the load's equipment type, production process requirements, and historical operating data, a comprehensive sensitivity coefficient is defined for the i-th sensitive load node. : In the formula, For the weighting coefficients, satisfying It can be determined based on load characteristics through the analytic hierarchy process and expert experience; It is a voltage sag sensitivity function, which is jointly determined by the sag tolerance curve and the probability distribution of the grid sag amplitude. This is a voltage distortion sensitivity function, which is related to the load's tolerance to total distortion; Let be the harmonic spectrum sensitivity function, characterizing the distribution of load sensitivity to each harmonic. Specifically, the harmonic spectrum sensitivity function... The specific form of expression is: In the formula, Let be the sensitivity weighting coefficient of the i-th sensitive load to the h-th harmonic, satisfying This coefficient can be obtained based on the operating frequency characteristics of the sensitive components inside the load or through on-site measurement. This is the measured value of the voltage content of the current h-th harmonic; This is the withstand limit value for the voltage content of the h-th harmonic. The larger this ratio, the closer the current harmonic is to the load withstand limit, and the higher the sensitivity.

[0025] Machine learning algorithms are used to dynamically update the demand profile of sensitive loads. An online sequential extreme learning machine algorithm is employed, using newly acquired power quality monitoring data as input and load anomalies or outages as feedback labels to dynamically adjust the tolerance boundary parameters of sensitive loads. The updated voltage sag tolerance curve parameter vector for the i-th sensitive load at time t is defined. for: ,in, This is the parameter vector of the tolerance curve at the previous moment, containing the coordinates of multiple key points on the transient amplitude-duration curve; The learning rate controls the step size for parameter updates; Let L be the gradient of the loss function with respect to the parameters. The loss function is defined as the cross-entropy between the predicted load state and the actual load state. The power quality feature vector input at time t includes the current voltage sag depth, duration, and harmonic content, etc. This label represents the actual operating status of the load at time t. Through the aforementioned online learning mechanism, the sensitive load demand profile can adaptively reflect the evolution of power quality requirements during different operating periods, load rates, and switching between different operating conditions, providing a precise governance basis for the subsequent cluster hierarchical coordination module.

[0026] In this embodiment, based on the sensitive load demand profile, priority division and collaborative control strategy generation are implemented for the microgrid active power filter cluster, forming a hierarchical collaborative control strategy. The specific implementation method is as follows: Based on multi-dimensional indicators in the sensitive load demand profile, a quantitative evaluation model for sensitive load priority is constructed. Assume there are N sensitive load nodes and M active power filters in the microgrid. For the i-th sensitive load node, its comprehensive priority coefficient is defined. for: In the formula, The voltage sag sensitivity index for the i-th sensitive load is obtained by normalizing the sag tolerance curve parameters output by the preceding sensitive load profiling module. The total voltage distortion sensitivity index is based on the load pair. Calculation of tolerance threshold; As a comprehensive index of harmonic spectrum sensitivity, it is composed of Calculation, where Let be the sensitivity weight of the i-th load to the h-th harmonic. This is the tolerance limit of the load to the h-th harmonic; The load importance label is defined for the user side, with a value ranging from 0 to 1. The larger the value, the higher the importance. For the weighting coefficients, satisfying This can be determined through the analytic hierarchy process or expert experience. The overall priority coefficient is calculated based on this. Sensitive loads are classified as special-grade guaranteed loads ( Level 1 sensitive load () ) and secondary ordinary sensitive loads ( ),in and This is the preset priority threshold.

[0027] Based on the priority classification results, the directional allocation of the master control filter and the dynamic adaptation configuration of the subordinate APFs are performed on each APF. The electrical distance between the j-th APF and the i-th sensitive load node is defined. for: in, and These are the equivalent resistance and equivalent reactance between the j-th APF installation point and the i-th sensitive load node, respectively; and This is the reference impedance value used for normalization. For high-level protection loads, the APF with the smallest electrical distance is selected as the main control filter, denoted as . For Level 1 sensitive loads, optimization matching is performed based on load priority and remaining APF capacity. A master-slave allocation optimization objective function is constructed. as follows: In the formula, Let j be a binary decision variable, representing whether the j-th APF is assigned to the i-th sensitive load as a master or slave APF, satisfying the following condition: That is, each APF can serve multiple loads simultaneously, but capacity constraints must be considered; This represents the maximum electrical distance within the microgrid. Let J be the rated compensation capacity of the j-th APF; The required compensation capacity allocated to the j-th APF is determined by the total demand of the sensitive loads it serves; The weighting coefficients balance electrical distance and capacity utilization. By solving this optimization problem, we obtain the master-slave APF correspondence table and the electrical boundary definitions of the auxiliary governance responsibility areas of each subordinate APF.

[0028] For microgrid switching between grid-connected and islanded modes, a cooperative control strategy is generated, including coordinated action timing and priority preemption logic. The reference value of the compensation current for the j-th APF at time t is defined. for: In the formula, The set of sensitive load nodes served by the j-th APF; Let be the compensation contribution coefficient of the j-th APF to the i-th load, satisfying... ; It is used to compensate for the current component and is used for conventional harmonic suppression; The priority preemption trigger flag is set when a high-priority load experiences a power quality violation. Otherwise, it is 0; The priority preemption compensation current component is transferred from the APF corresponding to the preempted low-priority load. Mode switching timing control is implemented by a state machine: when a grid-connected to islanded event is detected, the system first enters a pre-synchronization phase, coordinating the output phase adjustments of each APF; then it enters a transient support phase, activating priority preemption logic to prioritize the high-priority load; finally, it enters a steady-state operation phase, operating according to the updated collaborative control strategy. The preemption trigger threshold is set as follows: when the total distortion rate of the high-priority load bus voltage... Or the voltage content of a certain harmonic And the duration exceeds When this occurs, priority preemption is triggered, and the APF compensation capacity corresponding to adjacent low-priority loads is reallocated to high-priority loads. The recovery exit mechanism after preemption occurs when the power quality of the high-priority load returns to normal and remains stable. After a period of time, the preempted APF resources will be gradually released, and the system will return to the normal collaborative control mode.

[0029] In this embodiment, an APF electrical coupling matrix is ​​constructed based on real-time topology-aware information, and a consensus algorithm is used to dynamically reconstruct the APF cluster collaborative control model. The specific implementation method is as follows: The dynamic topology reconfiguration module establishes a real-time sensing mechanism for microgrid topology changes. Assume there are M active power filters in the microgrid, and define the electrical coupling matrix at time t. , of which elements The electrical coupling strength between the m-th APF and the n-th APF is calculated in real time based on the feeder impedance parameters and switch status. , This represents the equivalent complex impedance magnitude between the two nodes. The electrical coupling matrix is ​​used when events such as feeder switching, distributed generation grid connection, or grid disconnection occur. A change has occurred. Define the topology change detection quantity. The relative rate of change of the Frobenius norm of the electrical coupling matrix between the current time step and the previous time step: In the formula, Denotes the Frobenius norm of a matrix; The sampling period is [value]. When [condition]... Exceeding the preset threshold When a significant change in the microgrid topology is detected, the collaborative control model reconfiguration process is triggered.

[0030] The APF cluster collaborative control model is reconstructed based on a consensus algorithm. Upon detecting a topology change, the module reads the master-slave relationship and responsibility area division information from the hierarchical collaborative control strategy and sets the initial compensation command vector for each APF. Where H is the total number of harmonic orders to be compensated. To achieve coordinated allocation of global compensation, a dynamic average consensus algorithm is used to iteratively update the compensation instructions of each APF. The instruction update formula for the m-th APF in the (k+1)th iteration is: In the formula, The set of neighboring APFs that have a real-time communication connection with the m-th APF; Let be the consistency weight coefficient, satisfying The weights are typically set as Metropolis weights based on the quality of the communication link. This is the traction coefficient, with a value range of (0,1), used to maintain the constraint relationship with the initial setpoint. After iterative convergence, the steady-state compensation command for each APF is obtained. These are the core parameters of the updated cooperative control model that adapts to the new topology.

[0031] To address dynamic events such as APF device grid connection or disconnection and communication link interruption, distributed model predictive control is employed to update the cooperative control model online, ensuring the system's robustness under changing communication conditions. The prediction time domain length of the m-th APF at time t is defined as... The control time domain length is Its optimization objective function for In the formula, For time t with respect to time t The compensation instruction prediction vector; This refers to the steady-state command after the consensus algorithm converges; It is a diagonal positive definite weight matrix used to penalize the deviation between the predicted output and the steady-state target; The communication delay from the neighbor APFn to the current APFm; The received neighbor prediction value after taking time delay into account; The coordination deviation weight matrix has its diagonal elements dynamically adjusted according to the priority weights set in the hierarchical coordination control strategy. High-priority loads have larger coordination weights to ensure their output consistency. The dynamic topology reconfiguration module monitors the communication status in real time and automatically updates the neighbor set when it detects APF merging, disconnection, or changes in communication links. and weight matrix And resolve the above optimization problem, taking the instruction at the first time step. As the actual output, it is continuously optimized in the next time step to achieve adaptive updates of the collaborative control model under conditions of communication topology changes or time delay fluctuations. Through the above mechanism, the dynamic topology reconfiguration module realizes a complete closed loop from topology perception, model reconstruction to prediction updates, ensuring that the APF cluster collaborative control model always matches the real-time operating status of the microgrid.

[0032] In this embodiment, the fundamental impedance and virtual impedance of each APF are optimized and adjusted online based on a virtual impedance control strategy. The specific implementation method is as follows: Collaborative optimization of the virtual impedance at the fundamental frequency is performed to achieve a reasonable distribution of reactive power among the APFs and stable support for the voltage of sensitive load buses. Assume there are M APFs in the microgrid, and the virtual impedance of the m-th APF at the fundamental frequency is determined by the virtual resistance. and virtual reactance Composition, denoted as Define the objective function for fundamental virtual impedance optimization. for: In the formula, Let m be the fundamental reactive power vector actually output by the m-th APF; The reactive power reference value of this APF is given by the hierarchical collaborative control strategy; It is a positive definite weight matrix used to adjust the penalty degree for reactive power distribution errors of different APFs; These are the weighting coefficients; The measured value of the voltage vector at the point of common coupling; Let be the voltage reference vector. Solve the objective function using a particle swarm optimization algorithm to obtain the result... Minimum optimal setting value of fundamental virtual impedance for each APF and .

[0033] Decoupling optimization is performed on the virtual impedance at each characteristic harmonic frequency to suppress harmonic voltage distortion of the sensitive load bus and achieve coordinated distribution of harmonic current among the APFs. Let the highest harmonic order to be mitigated be K, and the virtual impedance of the m-th APF to the k-th harmonic be... ,in Define the optimization objective function for the k-th harmonic. for: In the formula, The vector of the kth harmonic current actually output by the m-th APF; The reference value for the kth harmonic compensation current of the APF is given by the updated cooperative control model; This is the weight matrix corresponding to the k-th harmonic; This is the harmonic voltage suppression weighting coefficient; Let k be the k-th harmonic voltage vector of the sensitive load bus. The gradient descent method is used to determine the k-th harmonic voltage vector. Perform iterative optimization to obtain the result. Minimum virtual impedance settings for each harmonic of each APF and .

[0034] Dynamic adjustment based on virtual impedance enables active shifting control of the harmonic spectrum of sensitive load buses. A harmonic spectrum shifting coefficient vector is defined. ,in This represents the intensity of the k-th harmonic shift from the sensitive frequency band to the non-sensitive frequency band. Spectral shifting is achieved by reshaping the harmonic virtual impedance frequency characteristics of each APF, and its control law is: In the formula, The virtual impedance value ultimately used for the kth harmonic of the m-th APF; This is the relocation adjustment coefficient; For the preset frequency band sensitivity markers, if the k-th harmonic belongs to the sensitive frequency band, then... ,otherwise ; This is the real-time monitoring value of the kth harmonic voltage content of the sensitive load busbar; This represents the tolerance limit for the harmonic voltage content. When the harmonic voltage in the sensitive frequency band exceeds the limit, the corresponding harmonic virtual impedance value is increased, forcing the harmonic current to flow to the low-impedance path corresponding to the non-sensitive frequency band, thereby transferring the harmonic energy from the sensitive frequency band to the non-sensitive frequency band, achieving spectrum shifting. Combining the above optimization results of the fundamental and harmonic virtual impedances with the spectrum shifting control parameters, a spectrum shifting control is generated that includes... , , , and The virtual impedance adjustment command is sent to each APF for execution.

[0035] In this embodiment, based on the real-time harmonic detection results and the virtual impedance adjustment command, spectrum transfer is performed on the harmonic energy of the sensitive load bus, and the harmonic spectrum adjustment meets the microgrid power quality standard limits, generating a harmonic energy management command. The specific implementation method is as follows: The harmonic energy mitigation module monitors the capacity margin and virtual impedance adjustment status of the APF cluster in real time, and quantitatively assesses the remaining harmonic energy that cannot be offset. Assume there are M APFs in the microgrid, and the rated compensation capacity of the m-th APF is... The currently used compensation capacity is Then the remaining capacity of the m-th APF is Simultaneously, a virtual impedance adjustment saturation index is defined. This represents the degree of virtual impedance adjustment of the m-th APF to the k-th harmonic, with a value range of [0,1]. The time indicates that the adjustment has reached its upper limit. Therefore, the remaining harmonic energy that the system cannot completely cancel at time t is... for: In the formula, This represents the measured value of the kth harmonic current vector of the sensitive load busbar. The system reference impedance; Let m be the compensation capacity allocation coefficient of the m-th APF for the k-th harmonic, satisfying K represents the highest harmonic order. The first term represents the total harmonic energy demand of the sensitive load bus, and the second term represents the remaining compensation capacity that the APF cluster can currently provide. The difference between the two is the remaining harmonic energy that needs to be diverted.

[0036] when Furthermore, when the virtual impedance adjustment has reached its upper limit, the module evaluates the ability of non-sensitive load branches within the microgrid to absorb harmonic energy. Assume there are L non-sensitive load branches in total, and the equivalent fundamental impedance of the nth branch is... The harmonic impedance can be approximated as Define the upper limit of the acceptance capacity of the nth branch for the kth harmonic current. Determined by both branch heat capacity and power quality limits: In the formula This represents the upper limit of the heat capacity of the nth branch. This is the rated fundamental voltage of the branch. The fundamental current of the branch; This represents the k-th harmonic current of the branch. This is the limit value for the kth harmonic voltage of the branch busbar; Let be the impedance magnitude of the k-th harmonic. Considering all harmonic orders, the total harmonic reception capacity of the n-th branch is... This is the square root of the sum of the squares of the harmonic receiving capacities. The optimal injection branch is selected by comparing the receiving capacities of each branch with the current harmonic distribution. And determine the amplitude and phase of the injected current.

[0037] When the capacity of non-sensitive load branches is still insufficient, the module will direct the remaining harmonic energy to the energy storage device or a specially designed energy-consuming circuit. Assume the microgrid is equipped with an energy storage device, and its current state of charge is... Maximum charging and discharging power is It also includes an energy-dissipating circuit, the power of which can be controlled by the thyristor conduction angle. Continuous regulation. Command for the energy storage device to absorb harmonic power. With the conduction angle of the energy dissipation circuit Determined by the following formulas: , In the formula, This is the energy storage priority coefficient, with a value range of [0,1], which is dynamically adjusted based on the energy storage health status and economics. This represents the maximum energy consumption power of the energy-consuming circuit. The arcsine function is used to convert the power command into the thyristor conduction angle, ensuring that the power absorbed by the energy dissipation circuit matches the remaining energy. The charging and discharging power command of the energy storage device and the conduction angle control parameters of the energy dissipation circuit are calculated using the above formula. Together with the optimal injection branch result and the amplitude and phase of the injection current, these constitute the harmonic energy management command, which is then sent to each execution unit. This enables flexible channeling of harmonic energy from the sensitive load bus when the APF cluster capacity is saturated or the virtual impedance adjustment capability is insufficient.

[0038] It should be noted that, in this document, relational terms such as "first" and "second" are used only to distinguish one entity or operation from another, and do not necessarily require or imply any such actual relationship or order between these entities or operations. Furthermore, the terms "comprising," "including," or any other variations thereof are intended to cover non-exclusive inclusion, such that a process, method, article, or apparatus that comprises a list of elements includes not only those elements but also other elements not expressly listed, or elements inherent to such process, method, article, or apparatus.

[0039] Although embodiments of the invention have been shown and described, it will be understood by those skilled in the art that various changes, modifications, substitutions and alterations can be made to these embodiments without departing from the principles and spirit of the invention, the scope of which is defined by the appended claims and their equivalents.

Claims

1. A collaborative governance system for microgrid active power filter clusters oriented towards sensitive loads, characterized in that, Includes the following modules: The sensitive load profiling module is used to extract the power quality demand characteristics of sensitive loads based on the power quality monitoring data of each sensitive load node in the microgrid, and obtain a sensitive load demand profile. The cluster hierarchical coordination module is used to prioritize and generate coordinated control strategies for the microgrid active power filter cluster according to the sensitive load demand profile, thus forming a hierarchical coordinated control strategy. The dynamic topology reconfiguration module is used to construct the APF electrical coupling matrix based on real-time topology awareness information and perform dynamic reconfiguration of the APF cluster collaborative control model through a consensus algorithm. The virtual impedance adjustment module is used to optimize and adjust the fundamental impedance and virtual impedance of each harmonic of each APF online based on the virtual impedance control strategy. The harmonic energy diversion module is used to perform spectrum transfer on the harmonic energy of the sensitive load bus based on the real-time harmonic detection results and the virtual impedance adjustment command, and the harmonic spectrum adjustment meets the power quality standard limit of the microgrid, and generates harmonic energy management command.

2. The microgrid active power filter cluster collaborative management system for sensitive loads according to claim 1, characterized in that, The requirement profile construction process for the sensitive load profile module includes: By using the voltage and current waveform data of each sensitive load node, voltage distortion, voltage sag and harmonic tolerance indices are extracted and processed for the sensitive load to obtain the initial tolerance index set of the sensitive load. By using historical monitoring data and machine learning algorithms, the sensitive load demand profile is iteratively optimized and dynamically updated to form an updated sensitive load demand profile that can characterize the evolution of power quality demand during different operating periods, different load rates, and different operating conditions.

3. The microgrid active power filter cluster collaborative management system for sensitive loads according to claim 1, characterized in that, The requirements profile feature dimensions of the sensitive load profile module include: Based on the equipment type, core production process requirements, and real-time operating status of sensitive loads, a joint construction of voltage distortion sensitivity threshold and threshold over-limit tolerance time is performed on sensitive loads to build a sensitive load demand profile that includes voltage distortion sensitivity threshold and over-limit tolerance time. Based on the historical power quality event records of sensitive loads, the equipment's factory withstand curve data, and the field measured waveforms, the sensitive loads are subjected to a joint construction of voltage sag withstand curves and harmonic spectrum sensitive intervals. This results in a sensitive load demand profile that includes voltage sag amplitude-duration withstand curves, sensitive limits for the voltage content of each harmonic, and sensitive frequency bands for interharmonics.

4. The microgrid active power filter cluster collaborative management system for sensitive loads according to claim 1, characterized in that, The process of generating master-slave relationships in the cluster hierarchical collaboration module includes: Sensitive loads are prioritized according to the voltage sag tolerance index, harmonic sensitivity index, and load importance label defined by the user side in the sensitive load demand profile. The weight of the compensation current reference value of each APF is adjusted according to the priority classification result to generate a refined classification result including special-grade guaranteed loads, first-grade sensitive loads, and second-grade ordinary sensitive loads. Using the priority classification results, the master control filter is directionally allocated to each APF and the subordinate APF is dynamically adapted and configured, resulting in a hierarchical collaborative control strategy that includes a master-slave APF correspondence table, the electrical boundary definition of the auxiliary governance responsibility area of ​​each subordinate APF, and the master-slave role dynamic switching logic.

5. The microgrid active power filter cluster collaborative management system for sensitive loads according to claim 1, characterized in that, The collaborative control strategy generated by the cluster hierarchical collaboration module includes: Based on the hierarchical collaborative control strategy, a refined allocation scheme is generated for the reference values ​​of the fundamental reactive current and the compensation current of each harmonic for each APF, resulting in a refined allocation scheme that includes the compensation current amplitude, phase command and compensation dead zone setting that each APF should output under the fundamental and characteristic harmonics. Referring to the hierarchical collaborative control strategy, the collaborative action timing of each APF during the microgrid grid-connected mode and islanded mode switching process, as well as the priority preemption logic of the APF corresponding to the high-priority load to the low-priority APF, are generated to obtain a collaborative control strategy that includes grid-connected and islanded mode switching time markers, transient process collaborative control rules, preemption trigger threshold conditions, and recovery exit mechanism after preemption.

6. The microgrid active power filter cluster collaborative management system for sensitive loads according to claim 1, characterized in that, The collaborative control model reconfiguration process of the dynamic topology reconfiguration module includes: Based on the master-slave relationship definition and responsibility area division in the hierarchical collaborative control strategy, and combined with the topology change information caused by microgrid feeder switching, distributed power grid connection or disconnection, the original collaborative control model of the APF cluster is reconstructed through the consensus algorithm to generate an updated collaborative control model that meets the hierarchical collaboration requirements and adapts to the changes in electrical distance between nodes under the new topology. Based on the APF device's own network connection or disconnection events, communication link interruption or recovery status, and communication latency fluctuation information, the APF cluster collaborative control model is reconstructed online using a consensus algorithm to form an updated collaborative control model that adapts to the new communication topology, the number of new access nodes, and the changed communication condition constraints.

7. The microgrid active power filter cluster collaborative management system for sensitive loads according to claim 1, characterized in that, The predictive control model update process of the dynamic topology reconfiguration module includes: A distributed model predictive control architecture is adopted to perform rolling predictions of future multi-step control commands for each APF, and to exchange and verify the consistency of predicted command values ​​between adjacent APFs. The objective function of the predictive control includes priority weight coefficients set according to the hierarchical cooperative control strategy, so as to obtain the neighbor APF predictive control command set and cooperative constraint boundary required for local optimization of each APF. By utilizing real-time topology sensing information and communication status monitoring results, the prediction time domain length, priority weight coefficients, and optimization objective function constraint boundaries of each APF are dynamically tuned, thus constructing an updated collaborative control model that can still maintain a dynamic balance between model prediction accuracy and control response speed under conditions of communication topology abrupt changes or time delay fluctuations.

8. The microgrid active power filter cluster collaborative management system for sensitive loads according to claim 1, characterized in that, The virtual impedance optimization adjustment process of the virtual impedance adjustment module includes: Referring to the updated collaborative control model, the virtual resistance and virtual reactance values ​​of each APF are collaboratively optimized at the fundamental frequency to generate the optimal setting value of the fundamental virtual impedance of each APF. Using the updated collaborative control model, the virtual harmonic resistance and virtual harmonic reactance values ​​of each APF are decoupled and optimized at each characteristic harmonic frequency. Based on the dynamic adjustment of virtual impedance, the harmonic spectrum of the sensitive load bus is actively shifted from the sensitive frequency band to the non-sensitive frequency band, resulting in a virtual impedance adjustment command that includes the fundamental virtual impedance setting value, the virtual impedance setting values ​​of each harmonic, and the spectrum shifting collaborative parameters.

9. The microgrid active power filter cluster collaborative management system for sensitive loads according to claim 1, characterized in that, The energy dissipation guidance process of the harmonic energy conduction module includes: Based on the real-time detection results that the total capacity of the APF cluster is insufficient to completely offset all harmonics and the virtual impedance adjustment has reached its upper limit, the remaining harmonic energy that cannot be completely offset is directed and injected into the non-sensitive load branch connected to the microgrid. The harmonic energy is consumed by the natural damping characteristics of the non-sensitive load, and an energy diversion control command containing the preferred result of the injected branch and the amplitude and phase of the injected current is generated.

10. The microgrid active power filter cluster collaborative management system for sensitive loads according to claim 1, characterized in that, The energy storage and guiding process of the harmonic energy conduction module includes: Based on the monitoring results that the harmonic energy of the sensitive load bus exceeds the overall governance capacity of the APF cluster and the non-sensitive load branch has insufficient acceptance capacity, the remaining harmonic energy is directed to the energy storage device configured in the microgrid or the specially set energy dissipation circuit, forming an energy diversion control command that includes the charging and discharging power command of the energy storage device and the conduction angle control parameter of the energy dissipation circuit.

Images