Dynamic voltage recovery and flicker suppression method based on light-storage cooperation

Abstract

The invention discloses a dynamic voltage recovery and flicker suppression method based on light-storage cooperation, and relates to the technical field of power quality control of a power distribution network, and the method comprises the steps: obtaining a dynamic topological structure and a node impedance spectrum of the power distribution network in real time; calculating the vulnerability index of each node based on the data; generating a risk thermodynamic diagram according to the vulnerability index; proactive power scheduling is performed on the energy storage system based on a thermodynamic diagram, node impedance characteristics can be actively shaped and optimized in combination with impedance, the system comprises an intelligent monitoring unit, a central control platform, the energy storage system and a communication network, weak links of the system can be accurately recognized, the problems that a voltage disturbance propagation path is uncertain and the early warning adaptability is poor are solved, and the system reliability is improved. The perspectiveness and the accuracy of voltage disturbance suppression are improved, and the voltage stability and the power supply reliability of the power distribution network are enhanced.

Description

Technical Field

[0001] This invention relates to the field of power quality control technology for power distribution networks, specifically a dynamic voltage recovery and flicker suppression method based on photovoltaic-storage synergy. Background Technology

[0002] As the penetration rate of photovoltaic power generation in power distribution networks continues to increase, the intermittency and volatility of its output are increasingly impacting power quality, particularly causing voltage flicker and voltage sag. These voltage disturbances propagate along power lines, leading to flickering lighting, uneven motor speeds, and even shutdowns or damage to precision equipment, severely affecting power supply reliability and user experience. The combination of dynamic voltage restorers and energy storage systems has become a crucial technical solution to these problems.

[0003] In existing technologies, some patent applications compensate for voltage dips by detecting voltage sag events and controlling the energy storage system to release energy. However, this method is a typical post-event remedial strategy, acting only after a voltage drop occurs and failing to prevent the occurrence and propagation of disturbances. There is also a method and system for suppressing voltage fluctuations in a distribution network using distributed photovoltaic and energy storage collaboration, as described in patent application publication number CN120728623B. While this method considers the characteristics of photovoltaic fluctuations and utilizes historical data for some prediction, its core control strategy still relies on a pre-set, fixed grid model. This fails to address the dynamic changes in topology caused by line switching and load changes in actual distribution networks, limiting its adaptability and early warning accuracy in real-world operating environments.

[0004] In summary, existing technologies generally suffer from insufficient depth of perception of power grid operation status, failing to effectively address the core challenge of uncertain voltage disturbance propagation paths and poor adaptability of traditional early warning methods caused by the combined effects of dynamic changes in distribution network topology and distributed generation fluctuations. Early warning and control methods relying on fixed topology models struggle to accurately identify weak points in the structurally variable real-world distribution networks, making it impossible to suppress voltage flicker and sags at their source.

[0005] Therefore, there is an urgent need in this field for a new method that can sense the dynamic operating characteristics of the power grid in real time and thereby achieve forward-looking voltage disturbance suppression, so as to make up for the shortcomings of existing technologies in terms of active defense and adaptive capabilities. Summary of the Invention

[0006] The purpose of this invention is to overcome the shortcomings of existing technologies and provide a dynamic voltage recovery and flicker suppression method based on photovoltaic-storage synergy. By intelligently monitoring key data of the distribution network, dynamically updating the topology and analyzing node impedance, and combining proactive scheduling of energy storage system power output, the method can adapt to changes in distribution network topology and fluctuations in distributed power sources, accurately locate voltage disturbance risks, effectively suppress voltage flicker and sags, improve the voltage recovery capability and operational stability of the distribution network, and overcome the shortcomings of existing technologies that rely on fixed models.

[0007] To solve the above-mentioned technical problems, the present invention provides the following technical solution: On one hand, a dynamic voltage recovery and flicker suppression method based on photoelectric-storage synergy, comprising the following steps:

[0008] Step 1: Real-time acquisition of the dynamic topology and node impedance spectrum of the distribution network;

[0009] Step 2: Calculate the vulnerability index of each node based on the dynamic topology and node impedance spectrum;

[0010] Step 3: Generate a risk heat map of the distribution network based on the vulnerability index;

[0011] Step 4: Based on the risk heat map, perform proactive power scheduling on the energy storage system connected to the distribution network to suppress voltage disturbances;

[0012] Among them, the dynamic topology structure collects switch status signals in real time by deploying intelligent monitoring units at key nodes of the distribution network and updates the topology connection relationship using graph theory algorithms. The node impedance spectrum is obtained by injecting multi-frequency test signals into the distribution network and measuring the voltage and current response to calculate the impedance amplitude and impedance angle at each frequency. The vulnerability index is calculated by integrating impedance ratio, electrical distance, load sensitivity and voltage deviation through mathematical formulas. The risk heat map maps the vulnerability index into color values ​​for real-time visualization through a graphical interface. Proactive power dispatch identifies high-risk nodes based on the risk heat map and sends power control commands to nearby energy storage systems.

[0013] Furthermore, the real-time acquisition of the dynamic topology of the distribution network in step one specifically includes:

[0014] Intelligent monitoring units are deployed at substation outlets, main branch boxes, and grid connection points of distributed photovoltaic power generation systems in the distribution network. These units are configured to collect and upload the opening and closing status of all controllable switching devices at key nodes in real time. These controllable switching devices include circuit breakers and load switches. A central control platform receives status signals from all intelligent monitoring units. Based on the adjacency matrix method in graph theory, the central control platform abstracts the distribution network as a set of nodes and edges. It dynamically adjusts the element values ​​in the adjacency matrix according to the switch status, thereby reconstructing the distribution network topology in real time. The update frequency of the dynamic topology is synchronized with switch status change events; that is, when any switch status change is detected, a topology update calculation is immediately triggered to maintain consistency between the topology and the actual power grid. The intelligent monitoring units include remote terminal units or feeder terminal units. Status signals are transmitted to the central control platform via a communication network. The topology update calculation includes verifying node connectivity and updating the electrical distance matrix.

[0015] Furthermore, the real-time acquisition of the nodal impedance spectrum in step one specifically includes:

[0016] The inverters of distributed power sources connected in the distribution network or a dedicated signal injection device inject a set of predefined multi-frequency sinusoidal test signals into the power grid. The amplitude of the test signals is lower than a set threshold to ensure grid safety. Voltage and current waveforms of multiple nodes in the distribution network are acquired simultaneously. The voltage and current components at the test signal frequencies are extracted using signal processing algorithms. For each node and each test frequency, the impedance value is calculated according to Ohm's law. The impedance amplitude is the ratio of the voltage component amplitude to the current component amplitude, and the impedance angle is the difference between the voltage component phase and the current component phase. The impedance amplitudes and impedance angles at all frequencies are combined to form the impedance spectrum of the node. The frequency range of the multi-frequency test signals covers the typical voltage disturbance frequency band caused by load fluctuations and changes in the output of distributed power sources. The signal processing algorithms include Fast Fourier Transform or Wavelet Analysis. The impedance spectrum is stored in a database for subsequent analysis.

[0017] Furthermore, the calculation of the vulnerability index of each node in step two specifically includes: collecting the following real-time parameters for each node in the distribution network:

[0018] High-frequency impedance, low-frequency impedance, electrical distance from the disturbance source, local load sensitivity, and current voltage deviation are considered. High-frequency impedance is obtained through impedance spectrum analysis of the average impedance amplitude in the high-frequency range. Low-frequency impedance is obtained through impedance spectrum analysis of the impedance amplitude at the power frequency. Electrical distance is calculated as the sum of the impedance amplitudes of all branches along the path from the node to the nearest known disturbance source. Local load sensitivity is defined as the ratio of the change in active power at the node to the change in voltage, derived through real-time measurement or load characteristic curves. Current voltage deviation is calculated as the absolute value of the difference between the real-time voltage measurement at the node and the rated voltage. The vulnerability index is calculated using the following mathematical formula:

[0019]

[0020] Among them, VI i Z is the vulnerability index of node i. high,i Z is the impedance magnitude of node i in the high-frequency range. low,i D is the impedance magnitude of node i at the power frequency. i S is the electrical distance between node i and the disturbance source. i It is the local load sensitivity of node i, ΔV i Let w1, w2, w3, and w4 be the current voltage deviation of node i, and w1, w2, w3, and w4 be weighting coefficients that satisfy w1 + w2 + w3 + w4 = 1.

[0021] The high-frequency range is defined as 100Hz to 1000Hz, and the power frequency is defined as 50Hz. The weighting coefficients are obtained through training with historical data.

[0022] Furthermore, the process of determining the weighting coefficients w1, w2, w3, and w4 includes:

[0023] Time-series data is extracted from the historical operation database of the power distribution network. Each data record includes the impedance ratio, electrical distance, load sensitivity, voltage deviation, and whether a voltage disturbance event was recorded at that moment. The data is divided into training and test sets. A classification model is trained using the training set data. The input features are impedance ratio, the reciprocal of electrical distance, load sensitivity, and voltage deviation. The output is the probability of voltage disturbance occurrence. The classification model used is a random forest or a neural network. During training, the model parameters are adjusted through backpropagation or decision tree generation algorithms. After training, the model's performance on the test set is evaluated, and the average importance score of each feature is calculated. The feature importance scores are normalized so that the sum of the four scores is 1. These normalized scores are set as weight coefficients. The weight coefficients are embedded in the control system for real-time calculation of the vulnerability index. The classification model is trained monthly to ensure adaptability.

[0024] Furthermore, the generation of the risk heat map of the distribution network in step three specifically includes:

[0025] The central control platform maintains a fusion database of a power distribution network geographic information system model and an electrical topology model. It uses a graphics rendering engine to read the current topology from the fusion database and displays a simplified single-line diagram of the power distribution network on a large monitoring screen. Nodes are represented by circles, and branches by lines. For each node, the platform reads the latest calculated vulnerability index from its real-time data buffer and designs a color mapping function. This function normalizes the vulnerability index to the zero-to-one interval and then obtains the corresponding color in the color space through linear interpolation, where zero corresponds to green and one corresponds to red. The color of each node's circle is set as the output value of the color mapping function. The graphics rendering engine redraws the entire single-line diagram at a rate of at least one frame per second to ensure the real-time nature of the risk heatmap. The system provides interactive functionality, allowing users to click on nodes to view detailed vulnerability index values ​​and component decompositions. The color mapping is implemented using the HSL color model, and the node size scales with the vulnerability index value.

[0026] Furthermore, the proactive power scheduling of the energy storage system in step four specifically includes:

[0027] From the risk heatmap, all nodes with vulnerability indices exceeding a preset threshold are extracted to form a high-risk node list. For each node in the list, Dijkstra's algorithm is used to calculate the shortest electrical path to all energy storage systems in the dynamic topology. The energy storage system with the shortest path is selected as the control target. The power imbalance of the node is evaluated, and the amount of power compensation that needs to be injected or absorbed is calculated. The power compensation amount is estimated using the following formula:

[0028] ΔP=sign(V i -V nom )·|VI i |·K p

[0029] Where sign is the sign function, V i It is the real-time voltage of the node, V nom It is the rated voltage, VI i It is a vulnerability index, K p It is the power gain coefficient;

[0030] A control command is generated, which includes the identifier of the energy storage system, the power command value, and the duration of the action. The control command is sent to the local controller of the energy storage system via the IEC61850 or Modbus protocol. The energy storage system adjusts the output power of its converter according to the command to achieve forward power support. The power gain coefficient is adjusted according to the system stability.

[0031] Furthermore, the method also includes an active impedance shaping step:

[0032] Nodes with consistently high vulnerability indices in the risk heatmap are identified, and it is checked whether these nodes are connected to distributed power sources with controllable inverters. If so, an impedance reshaping command is sent to the controller of the distributed power source. Impedance reshaping is achieved by adding a virtual impedance calculation module to the inverter's control system. The virtual impedance calculation module calculates in real time the virtual voltage drop that should be compensated in the output current based on the command parameters. This virtual voltage drop is proportional to the output current, and the proportionality coefficient is the complex impedance Z. v =R v +jωL v By setting R v A negative value achieves negative resistance characteristics within a specific frequency range, thereby offsetting the positive resistance of the line and reducing the overall impedance. The parameter R of the virtual impedance is... v and L v Dynamic adjustments are made based on the node impedance spectrum analysis results. To target the most sensitive disturbance frequencies, impedance shaping and proactive power scheduling work together to improve voltage stability. The virtual impedance element is inserted into the inverter's dual closed-loop control structure.

[0033] On the other hand, a dynamic voltage recovery and flicker suppression system based on photoelectric-storage synergy, the system comprising:

[0034] Multiple intelligent monitoring units are installed in substations, branch boxes, and distributed power access points of the distribution network. Each intelligent monitoring unit includes a status input module, an electrical quantity measurement module, and a communication module. The central control platform consists of a server cluster and runs topology management software, impedance analysis software, vulnerability calculation software, visualization software, and control decision software. At least one energy storage system includes a battery bank, a supercapacitor bank, and a bidirectional converter, which is connected to the distribution network through a circuit breaker. The communication network adopts wired Ethernet or wireless cellular network to provide data transmission between the intelligent monitoring units and the central control platform, and between the central control platform and the energy storage system. The intelligent monitoring units upload switch status and voltage and current data in real time. The central control platform processes the data and generates control commands, which are then sent to the energy storage system for execution via the communication network. The central control platform also includes a database to store historical data and real-time parameters.

[0035] Furthermore, the central control platform also includes an adaptive learning module for continuously optimizing the vulnerability index calculation model. The adaptive learning module initiates the optimization process at fixed time intervals. The optimization process includes collecting all historical data within the past time window. The data includes the impedance ratio, electrical distance, load sensitivity, voltage deviation, and whether voltage disturbances have occurred for each node. After cleaning and feature engineering the data, a new classifier is trained using the random forest algorithm. The input features are the impedance ratio, the reciprocal of the electrical distance, the load sensitivity, and the voltage deviation. The output is the disturbance probability. After training, the average Gini importance score of each feature in the random forest is calculated. The score is normalized to obtain new weight coefficients. The difference between the new weight coefficients and the old weight coefficients is compared. If the difference exceeds a threshold, the vulnerability index calculation formula is updated with the new weight coefficients. At the same time, the old model is retained as a backup to back up when the new model fails to be validated. The adaptive learning module also generates an optimization report to record the history of weight coefficient changes. The time window is set to seven days, and the threshold is set to 10%.

[0036] Compared with existing technologies, this dynamic voltage recovery and flicker suppression method based on optical-storage synergy has the following advantages:

[0037] I. This invention utilizes intelligent monitoring units to collect real-time switch status and electrical quantity data of key nodes in the distribution network. It updates the dynamic topology using graph theory algorithms, calculates node impedance spectra by combining multi-frequency test signals, and then calculates node vulnerability indices and generates risk heat maps based on the dynamic topology and node impedance spectra. Finally, it implements proactive power scheduling for the energy storage system based on the heat maps. This invention can adapt to dynamic changes in the distribution network topology and operational status changes caused by distributed power source fluctuations in real time, accurately locate weak nodes and potential voltage disturbance risks in the system, and solve the problems of uncertain voltage disturbance propagation paths and poor adaptability of early warning methods caused by existing technologies that rely on fixed grid models. It improves the distribution network's proactive suppression capability against voltage disturbances and effectively reduces the probability and propagation range of voltage flicker and sag.

[0038] Second, this invention adds an active impedance shaping step, dynamically adjusting the virtual impedance parameters of the distributed power inverter based on the node impedance spectrum analysis results, thereby optimizing the impedance characteristics of highly vulnerable nodes. At the same time, by using the adaptive learning module of the central control platform to periodically train the classification model and update the weight coefficients for calculating the vulnerability index, it can further improve the voltage stability of the distribution network and enhance the system's adaptability to complex operating conditions such as different load fluctuations and changes in distributed power output, thus significantly improving the power supply reliability of the distribution network and the user's electricity experience.

[0039] Other advantages, objectives and features of the invention will be set forth in part in the description which follows, and in part will be apparent to those skilled in the art from the following examination or study, or may be learned from the practice of the invention. Attached Figure Description

[0040] To more clearly illustrate the technical solutions in the embodiments of the present invention or the prior art, the accompanying drawings used in the description of the embodiments or the prior art will be briefly introduced below. Obviously, the drawings described below are merely some embodiments of the present invention. For those skilled in the art, other drawings can be obtained based on these drawings without any creative effort.

[0041] Figure 1 This is a step diagram of the present invention;

[0042] Figure 2 This is a system architecture diagram of the present invention;

[0043] Figure 3 This is a flowchart illustrating the operation of the present invention. Detailed Implementation

[0044] To further illustrate the technical means and effects of the present invention in achieving its intended purpose, the following detailed description of the specific implementation methods, structures, features, and effects of the present invention, in conjunction with the accompanying drawings and preferred embodiments, is provided below.

[0045] Example 1

[0046] like Figure 1 and Figure 2 As shown, this embodiment discloses a specific implementation of a dynamic voltage recovery and flicker suppression method based on photovoltaic-storage synergy, relating to the field of power quality control technology for distribution networks. Addressing the problems of existing technologies relying on post-event remediation and poor adaptability of fixed grid models, this embodiment acquires the dynamic topology and node impedance spectrum of the distribution network in real time, calculates the node vulnerability index and generates a risk heatmap, and combines proactive power scheduling and active impedance shaping to achieve proactive suppression of voltage disturbances. This embodiment can effectively improve the adaptability of the distribution network to dynamic topology changes and distributed power source fluctuations, enhancing voltage stability and power supply reliability.

[0047] This embodiment is applied to a 10kV distribution network in a certain area, which is connected to multiple distributed photovoltaic power generation systems and energy storage systems. The energy storage systems are connected to the distribution network through bidirectional converters. The distribution network includes key nodes such as substation outlets, main branch boxes, and grid connection points of distributed photovoltaic power generation systems. In daily operation, there are risks of voltage disturbances caused by load fluctuations and changes in the output of distributed power sources. This method is needed to achieve dynamic voltage recovery and flicker suppression.

[0048] The system implementation process is as follows:

[0049] The dynamic voltage recovery and flicker suppression system based on photovoltaic-storage synergy adopted in this embodiment includes an intelligent monitoring unit, a central control platform, an energy storage system, and a communication network. The functions and connections of each part are as follows:

[0050] Intelligent monitoring unit: Deployed at substation outlets, main branch boxes, and grid connection points of distributed photovoltaic power generation systems in the distribution network. Each unit includes a status input module, an electrical quantity measurement module, and a communication module. The status input module is used to collect the opening and closing status of controllable switching equipment; the electrical quantity measurement module is used to collect voltage and current waveforms of nodes; the communication module is used to transmit the collected status signals and electrical quantity data to the central control platform. This unit can be implemented using a remote terminal unit or a feeder terminal unit.

[0051] Central control platform: Composed of server clusters, it runs topology management software, impedance analysis software, vulnerability calculation software, visualization software, control decision software, and adaptive learning module. It also maintains a fusion database of the distribution network geographic information system model and electrical topology model for data storage, processing, analysis, and control command generation.

[0052] Energy storage system: It includes at least a battery bank, a supercapacitor bank and a bidirectional converter. It is connected to the distribution network through a circuit breaker and can adjust the output power of the converter according to the instructions issued by the central control platform to achieve power injection or absorption.

[0053] Communication network: Wired Ethernet or wireless cellular network is used to build data transmission channels between the intelligent monitoring unit and the central control platform, and between the central control platform and the energy storage system, to ensure real-time transmission of signals and commands.

[0054] The implementation process is as follows:

[0055] Step 1: Real-time acquisition of the dynamic topology and node impedance spectrum of the distribution network.

[0056] Intelligent monitoring units are deployed at substation outlets, main branch boxes, and grid connection points of distributed photovoltaic power generation systems in the distribution network. The intelligent monitoring units adopt remote terminal units or feeder terminal units, which are configured to collect the opening and closing status of all controllable switching equipment at key nodes in real time. The controllable switching equipment includes circuit breakers and load switches.

[0057] The status signals collected by the intelligent monitoring unit are transmitted to the central control platform through a communication network, which can be either wired Ethernet or wireless cellular network. The central control platform uses the adjacency matrix method from graph theory to abstract the distribution network as a set of nodes and edges: nodes correspond to key locations in the distribution network such as substation outlets, main branch boxes, and photovoltaic grid connection points, while edges correspond to the lines connecting each node.

[0058] When the central control platform detects a change in the state of any controllable switchgear, it immediately triggers a topology update calculation: The values ​​of elements in the adjacency matrix are dynamically adjusted based on the switch state—when the switch is closed, the corresponding edge's adjacency matrix element value is set to 1, indicating connectivity between nodes; when the switch is open, the corresponding edge's adjacency matrix element value is set to 0, indicating disconnection between nodes. During the topology update calculation, node connectivity is simultaneously verified and the electrical distance matrix is ​​updated to maintain consistency between the dynamic topology and the actual operating state of the distribution network.

[0059] The central control platform controls the inverters of distributed power sources connected to the distribution network or a dedicated signal injection device to inject a set of predefined multi-frequency sinusoidal test signals into the distribution network. The amplitude of the test signals is lower than the set safety threshold, and the frequency range covers the typical voltage disturbance frequency band caused by load fluctuations and changes in the output of distributed power sources.

[0060] Voltage and current waveforms from multiple nodes in the distribution network are simultaneously acquired. Signal processing algorithms such as Fast Fourier Transform or wavelet analysis are used to extract the voltage and current components at the test signal frequencies. For each node and each test frequency, impedance parameters are calculated according to Ohm's law.

[0061] Impedance amplitude: the ratio of the voltage component amplitude to the current component amplitude at that frequency;

[0062] Impedance angle: the difference between the phase of the voltage component and the phase of the current component at this frequency.

[0063] The impedance magnitude and impedance angle of the same node at all test frequencies are combined to form the impedance spectrum of the node, and the impedance spectrum is stored in a database for subsequent vulnerability index calculation.

[0064] Step 2: Calculate the vulnerability index of each node.

[0065] For each node in the distribution network, the following parameters are collected in real time:

[0066] High-frequency impedance: The average impedance amplitude in the high-frequency range of 100Hz to 1000Hz is obtained through impedance spectrum analysis;

[0067] Low-frequency impedance: The impedance amplitude at the 50Hz power frequency is obtained through impedance spectrum analysis;

[0068] Electrical distance: Calculate the sum of the magnitudes of the impedances of all branches along the path from this node to the nearest known disturbance source;

[0069] Local load sensitivity: defined as the ratio of the change in active power of the load at a node to the change in voltage, derived through real-time measurement or load characteristic curve;

[0070] Current voltage deviation: Calculated as the absolute value of the difference between the real-time voltage measurement of the node and the rated voltage.

[0071] The vulnerability index of a node is calculated using the following formula:

[0072]

[0073] In the formula:

[0074] VI i : Vulnerability index of node i. The larger the value, the higher the node's sensitivity to voltage disturbances and the weaker its disturbance resistance.

[0075] Z high,i : The high-frequency impedance of node i, i.e., the average impedance amplitude in the range of 100Hz to 1000Hz;

[0076] Z low,i : The low-frequency impedance of node i, i.e., the impedance amplitude at the 50Hz power frequency;

[0077] Impedance ratio reflects the difference in impedance characteristics of a node at different frequencies. The larger the ratio, the more sensitive the node is to high-frequency disturbances.

[0078] D i : The electrical distance between node i and the nearest disturbance source. The smaller the electrical distance, the higher the probability that the node is affected by the disturbance.

[0079] The reciprocal of the electrical distance is used to convert the effect of electrical distance on vulnerability into a positive correlation;

[0080] S i The local load sensitivity of node i is as follows: the higher the sensitivity, the more significant the impact of load power changes on voltage, and the more susceptible the node is to voltage disturbances.

[0081] ΔV i The current voltage deviation of node i; the larger the deviation, the more unstable the current voltage state of the node.

[0082] w1, w2, w3, w4: Weighting coefficients, satisfying w1+w2+w3+w4=1, used to characterize the degree of influence of each parameter on the vulnerability index, and their values ​​are determined through training with historical data.

[0083] The process of determining the weight coefficient w1+w2+w3+w4=1 is as follows:

[0084] Data extraction: Time series data are extracted from the historical operation database of the distribution network. Each data record includes the impedance ratio, electrical distance, load sensitivity, voltage deviation of the node at a specific moment, and whether a voltage disturbance event was recorded at that moment.

[0085] Data partitioning: The extracted historical data is divided into training and testing sets for training and performance validation of the classification model;

[0086] Model training: Random forest or neural network is used as the classification model, with impedance ratio, reciprocal of electrical distance, load sensitivity, and voltage deviation as input features, and the probability of voltage disturbance as the output. The model parameters are adjusted through backpropagation or decision tree generation algorithm to complete the model training.

[0087] Performance evaluation: Input the test set data into the trained classification model to evaluate the model's predictive performance and ensure the model's effectiveness in predicting voltage disturbances;

[0088] Weight Calculation: Calculate the average importance score of the input features in the classification model, normalize the importance scores of the four features, and the normalized scores are the weight coefficients w1, w2, w3, and w4.

[0089] Model update: The determined weight coefficients are fixed into the control system for real-time calculation of the vulnerability index; the classification model is trained once a month to adapt to changes in the operating status of the distribution network and ensure the accuracy of the weight coefficients.

[0090] Step 3: Generate a risk heat map of the power distribution network

[0091] The central control platform maintains a fusion database of the distribution network geographic information system model and electrical topology model. The fusion database stores the geographical location information of the distribution network, the electrical parameters of nodes and lines, and real-time operation data.

[0092] The current dynamic topology is read from the fusion database using a graphics rendering engine, and a simplified single-line diagram of the power distribution network is drawn on the monitoring screen: nodes are represented by circles, and lines are represented by lines connecting the nodes.

[0093] The latest calculated vulnerability index is read from the real-time data buffer of each node, and a color mapping function is designed: the vulnerability index is normalized to the range of 0 to 1, and the corresponding color is obtained in the HSL color space through linear interpolation - a normalized value of 0 corresponds to green, and a normalized value of 1 corresponds to red.

[0094] The color of each node circle in the single-line graph is set to the output value of the color mapping function, and the size of the node circle is adjusted according to the vulnerability index to intuitively reflect the node risk level.

[0095] The graphics rendering engine redraws the entire single-line graph at a rate of at least one frame per second, ensuring the real-time nature of the risk heatmap. The system provides interactive functionality: users can click on nodes in the single-line graph to view the vulnerability index value of that node, the specific values ​​of each parameter, and their contribution percentage to the vulnerability index, making it easier for operations and maintenance personnel to understand the sources of node risks.

[0096] Step 4: Perform proactive power scheduling on the energy storage system

[0097] All nodes with vulnerability indices exceeding a preset threshold are extracted from the risk heatmap to form a high-risk node list. For each node in the high-risk node list, Dijkstra's algorithm is used to calculate the shortest electrical path from the node to all energy storage systems in the dynamic topology, and the energy storage system with the shortest path is selected as the control target for that node.

[0098] Assess the power imbalance at high-risk nodes and estimate the amount of power compensation that needs to be injected or absorbed using the following formula:

[0099] ΔP=sign(V i -V nom )·|VI i |·K p

[0100] In the formula:

[0101] ΔP: Power compensation amount. A positive value indicates that the energy storage system needs to inject power into the distribution network, while a negative value indicates that the energy storage system needs to absorb power from the distribution network.

[0102] sign: sign function, when V i >V nom At that time, sign(V) i -V nom ) = 1; when V i <V nom At that time, sign(V) i -V nom ) = -1;

[0103] V i Real-time voltage of high-risk nodes;

[0104] V nom The rated voltage of the power distribution network;

[0105] |VI i |: The absolute value of the vulnerability index of high-risk nodes, which represents the degree of risk of the node. The larger the value, the greater the power required for compensation.

[0106] K p Power gain coefficient, adjusted according to the stability of the power distribution network system.

[0107] The central control platform generates control commands, which include the energy storage system's identifier, power command value, and duration of action. These control commands are then sent to the energy storage system's local controller via the IEC 61850 or Modbus protocol.

[0108] After receiving the control command, the local controller of the energy storage system adjusts the output power of the bidirectional converter: when ΔP is positive, the converter controls the energy storage system to release energy and inject power into the distribution network; when ΔP is negative, the converter controls the energy storage system to absorb energy and absorb power from the distribution network, thereby providing proactive power support for high-risk nodes and suppressing the occurrence and propagation of voltage disturbances.

[0109] Impedance active shaping steps:

[0110] Identify nodes with consistently high vulnerability indices in the risk heatmap and check whether these nodes are connected to distributed power sources with controllable inverters.

[0111] If a node is connected to a distributed power source with a controllable inverter, the central control platform sends an impedance reshaping command to the controller of that distributed power source. Impedance reshaping is achieved by adding a virtual impedance calculation module to the inverter's control system: the virtual impedance calculation module calculates in real time, based on the command parameters, the virtual voltage drop that should be compensated in the inverter's output current. This virtual voltage drop is proportional to the output current, and the proportionality coefficient is the complex virtual impedance Z. v =R v +jωL v In the formula:

[0112] R v Virtual resistance;

[0113] L v Virtual inductance;

[0114] ω: angular frequency, consistent with the operating frequency of the power distribution network;

[0115] j: Imaginary unit.

[0116] By setting R v A negative value indicates negative resistance characteristics within a specific frequency range, which can offset the positive resistance of distribution network lines and reduce the overall impedance of nodes; the parameter R of the virtual impedance... v and L v The system is dynamically adjusted based on the results of nodal impedance spectrum analysis to ensure that it works at the perturbation frequencies most sensitive to the nodes.

[0117] The virtual impedance link is inserted into the inverter's dual closed-loop control structure and works in conjunction with proactive power dispatch: proactive power dispatch quickly compensates for power imbalance through the energy storage system, while virtual impedance fundamentally reduces disturbance sensitivity by adjusting the node impedance characteristics, thus jointly improving the voltage stability of the distribution network.

[0118] In summary, this embodiment achieves accurate perception of the distribution network's operating status by acquiring the dynamic topology and node impedance spectrum of the distribution network in real time; it clearly identifies weak nodes in the system based on the vulnerability index calculation and risk heat map visualization using multi-parameter fusion; and it achieves proactive suppression of voltage disturbances by combining proactive power scheduling and active impedance shaping in coordinated control.

[0119] Example 2

[0120] like Figure 3 As shown in this embodiment, the specific steps of a dynamic voltage recovery and flicker suppression method based on photovoltaic-storage synergy are described in detail. This method identifies vulnerable nodes by real-time sensing of the distribution network's operating status and coordinates the energy storage system and distributed power sources to proactively suppress voltage disturbances. The specific workflow is as follows:

[0121] 1. The intelligent monitoring unit collects switch status signals and electrical quantity data of key nodes in the power distribution network in real time, including voltage waveforms and current waveforms.

[0122] 2. The central control platform receives the collected data and dynamically updates the distribution network topology based on graph theory algorithms to ensure that the topology is consistent with the actual power grid.

[0123] 3. The central control platform controls the distributed power inverter or dedicated signal injection device to inject multi-frequency test signals into the power grid, and the signal amplitude is set within the safety threshold.

[0124] 4. Simultaneously acquire voltage and current waveforms from multiple nodes, and use signal processing algorithms to extract voltage and current components at the test frequency.

[0125] 5. Calculate the impedance amplitude and impedance angle of each node at different frequencies according to Ohm's law, form the node impedance spectrum and store it.

[0126] 6. For each node, calculate high-frequency impedance, low-frequency impedance, electrical distance, load sensitivity, and voltage deviation parameters in real time.

[0127] 7. Using the weight coefficients obtained from training, the above parameters are combined to calculate the vulnerability index of each node.

[0128] 8. The central control platform generates a heat map of distribution network risks based on the vulnerability index, and visualizes the risk level of nodes in real time through color mapping.

[0129] 9. Identify high-risk nodes with vulnerability indices exceeding a preset threshold from the risk heatmap and create a list.

[0130] 10. For each high-risk node, use the shortest path algorithm to calculate the electrical distance to the energy storage system and select the nearest energy storage system as the control target.

[0131] 11. Assess the power imbalance at high-risk nodes and estimate the required power compensation.

[0132] 12. Generate a control command that includes the energy storage system identifier, power command value, and duration.

[0133] 13. Control commands are sent to the local controller of the energy storage system via communication protocols.

[0134] 14. The energy storage system adjusts the output power of the bidirectional converter to achieve power injection or absorption in order to support the node voltage.

[0135] 15. For nodes with consistently high vulnerability indices, check whether they are connected to distributed power sources with controllable inverters.

[0136] 16. If a controllable distributed power source exists, send an impedance reshaping command and adjust the inverter output characteristics through the virtual impedance calculation module.

[0137] 17. The central control platform regularly collects historical data, optimizes the vulnerability index calculation model, and updates the weighting coefficients.

[0138] 18. The system continuously monitors the status of the distribution network and repeats the above steps to achieve dynamic voltage recovery and flicker suppression.

[0139] The above description is merely a preferred embodiment of the present invention and is not intended to limit the present invention in any way. Although the present invention has been disclosed above with reference to preferred embodiments, it is not intended to limit the present invention. Any person skilled in the art can make some modifications or alterations to the above-disclosed technical content to create equivalent embodiments without departing from the scope of the present invention. Any simple modifications, equivalent changes and alterations made to the above embodiments based on the technical essence of the present invention without departing from the scope of the present invention shall still fall within the scope of the present invention.

Claims

1. A dynamic voltage recovery and flicker suppression method based on photo-storage synergy, characterized in that, Includes the following steps: Step 1: Real-time acquisition of the dynamic topology and node impedance spectrum of the distribution network; Step 2: Calculate the vulnerability index of each node based on the dynamic topology and node impedance spectrum; Step 3: Generate a risk heat map of the distribution network based on the vulnerability index; Step 4: Based on the risk heat map, perform proactive power scheduling on the energy storage system connected to the distribution network to suppress voltage disturbances; Among them, the dynamic topology structure collects switch status signals in real time by deploying intelligent monitoring units at key nodes of the distribution network and updates the topology connection relationship using graph theory algorithms. The node impedance spectrum is obtained by injecting multi-frequency test signals into the distribution network and measuring the voltage and current response to calculate the impedance amplitude and impedance angle at each frequency. The vulnerability index is calculated by integrating impedance ratio, electrical distance, load sensitivity and voltage deviation through mathematical formulas. The risk heat map maps the vulnerability index into color values ​​for real-time visualization through a graphical interface. Proactive power dispatch identifies high-risk nodes based on the risk heat map and sends power control commands to nearby energy storage systems.

2. The dynamic voltage recovery and flicker suppression method based on photo-storage synergy according to claim 1, characterized in that, The real-time acquisition of the dynamic topology of the distribution network in step one specifically includes: Intelligent monitoring units are deployed at substation outlets, main branch boxes, and grid connection points of distributed photovoltaic power generation systems in the distribution network. These units are configured to collect and upload the opening and closing status of all controllable switching devices at key nodes in real time. These controllable switching devices include circuit breakers and load switches. A central control platform receives status signals from all intelligent monitoring units. Based on the adjacency matrix method in graph theory, the central control platform abstracts the distribution network as a set of nodes and edges. It dynamically adjusts the element values ​​in the adjacency matrix according to the switch status, thereby reconstructing the distribution network topology in real time. The update frequency of the dynamic topology is synchronized with switch status change events; that is, when any switch status change is detected, a topology update calculation is immediately triggered to maintain consistency between the topology and the actual power grid. The intelligent monitoring units include remote terminal units or feeder terminal units. Status signals are transmitted to the central control platform via a communication network. The topology update calculation includes verifying node connectivity and updating the electrical distance matrix.

3. The dynamic voltage recovery and flicker suppression method based on photo-storage synergy according to claim 1, characterized in that, The real-time acquisition of nodal impedance spectra in step one specifically includes: The inverters of distributed power sources connected in the distribution network or a dedicated signal injection device inject a set of predefined multi-frequency sinusoidal test signals into the power grid. The amplitude of the test signals is lower than a set threshold to ensure grid safety. Voltage and current waveforms of multiple nodes in the distribution network are acquired simultaneously. The voltage and current components at the test signal frequencies are extracted using signal processing algorithms. For each node and each test frequency, the impedance value is calculated according to Ohm's law. The impedance amplitude is the ratio of the voltage component amplitude to the current component amplitude, and the impedance angle is the difference between the voltage component phase and the current component phase. The impedance amplitudes and impedance angles at all frequencies are combined to form the impedance spectrum of the node. The frequency range of the multi-frequency test signals covers the typical voltage disturbance frequency band caused by load fluctuations and changes in the output of distributed power sources. The signal processing algorithms include Fast Fourier Transform or Wavelet Analysis. The impedance spectrum is stored in a database for subsequent analysis.

4. The dynamic voltage recovery and flicker suppression method based on photo-storage synergy according to claim 1, characterized in that, The calculation of the vulnerability index of each node in step two specifically includes: collecting the following real-time parameters for each node in the distribution network: High-frequency impedance, low-frequency impedance, electrical distance from the disturbance source, local load sensitivity, and current voltage deviation are considered. High-frequency impedance is obtained through impedance spectrum analysis of the average impedance amplitude in the high-frequency range. Low-frequency impedance is obtained through impedance spectrum analysis of the impedance amplitude at the power frequency. Electrical distance is calculated as the sum of the impedance amplitudes of all branches along the path from the node to the nearest known disturbance source. Local load sensitivity is defined as the ratio of the change in active power at the node to the change in voltage, derived through real-time measurement or load characteristic curves. Current voltage deviation is calculated as the absolute value of the difference between the real-time voltage measurement at the node and the rated voltage. The vulnerability index is calculated using the following mathematical formula: Among them, VI i Z is the vulnerability index of node i. high,i Z is the impedance magnitude of node i in the high-frequency range. low,i D is the impedance magnitude of node i at the power frequency. i S is the electrical distance between node i and the disturbance source. i It is the local load sensitivity of node i, ΔV i Let w1, w2, w3, and w4 be the current voltage deviation of node i, and w1, w2, w3, and w4 be weighting coefficients that satisfy w1 + w2 + w3 + w4 = 1. The high-frequency range is defined as 100Hz to 1000Hz, and the power frequency is defined as 50Hz. The weighting coefficients are obtained through training with historical data.

5. The dynamic voltage recovery and flicker suppression method based on photo-storage synergy according to claim 4, characterized in that, The process of determining the weighting coefficients w1, w2, w3, and w4 includes: Time-series data is extracted from the historical operation database of the power distribution network. Each data record includes the impedance ratio, electrical distance, load sensitivity, voltage deviation, and whether a voltage disturbance event was recorded at that moment. The data is divided into training and test sets. A classification model is trained using the training set data. The input features are impedance ratio, the reciprocal of electrical distance, load sensitivity, and voltage deviation. The output is the probability of voltage disturbance occurrence. The classification model used is a random forest or a neural network. During training, the model parameters are adjusted through backpropagation or decision tree generation algorithms. After training, the model's performance on the test set is evaluated, and the average importance score of each feature is calculated. The feature importance scores are normalized so that the sum of the four scores is 1. These normalized scores are set as weight coefficients. The weight coefficients are embedded in the control system for real-time calculation of the vulnerability index. The classification model is trained monthly to ensure adaptability.

6. The dynamic voltage recovery and flicker suppression method based on photo-storage synergy according to claim 1, characterized in that, The generation of the risk heat map of the distribution network in step three specifically includes: The central control platform maintains a fusion database of a power distribution network geographic information system model and an electrical topology model. It uses a graphics rendering engine to read the current topology from the fusion database and displays a simplified single-line diagram of the power distribution network on a large monitoring screen. Nodes are represented by circles, and branches by lines. For each node, the platform reads the latest calculated vulnerability index from its real-time data buffer and designs a color mapping function. This function normalizes the vulnerability index to the zero-to-one interval and then obtains the corresponding color in the color space through linear interpolation, where zero corresponds to green and one corresponds to red. The color of each node's circle is set as the output value of the color mapping function. The graphics rendering engine redraws the entire single-line diagram at a rate of at least one frame per second to ensure the real-time nature of the risk heatmap. The system provides interactive functionality, allowing users to click on nodes to view detailed vulnerability index values ​​and component decompositions. The color mapping is implemented using the HSL color model, and the node size scales with the vulnerability index value.

7. The dynamic voltage recovery and flicker suppression method based on photo-storage synergy according to claim 1, characterized in that, The proactive power scheduling of the energy storage system in step four specifically includes: From the risk heatmap, all nodes with vulnerability indices exceeding a preset threshold are extracted to form a high-risk node list. For each node in the list, Dijkstra's algorithm is used to calculate the shortest electrical path to all energy storage systems in the dynamic topology. The energy storage system with the shortest path is selected as the control target. The power imbalance of the node is evaluated, and the amount of power compensation that needs to be injected or absorbed is calculated. The power compensation amount is estimated using the following formula: ΔP=sign(V i -V nom )·|VI i |·K p Where sign is the sign function, V i It is the real-time voltage of the node, V nom It is the rated voltage, VI i It is a vulnerability index, K p It is the power gain coefficient; A control command is generated, which includes the identifier of the energy storage system, the power command value, and the duration of the action. The control command is sent to the local controller of the energy storage system via the IEC61850 or Modbus protocol. The energy storage system adjusts the output power of its converter according to the command to achieve forward power support. The power gain coefficient is adjusted according to the system stability.

8. The dynamic voltage recovery and flicker suppression method based on photo-storage synergy according to claim 1, characterized in that, The method also includes an active impedance shaping step: Nodes with consistently high vulnerability indices in the risk heatmap are identified, and it is checked whether these nodes are connected to distributed power sources with controllable inverters. If so, an impedance reshaping command is sent to the controller of the distributed power source. Impedance reshaping is achieved by adding a virtual impedance calculation module to the inverter's control system. The virtual impedance calculation module calculates in real time the virtual voltage drop that should be compensated in the output current based on the command parameters. This virtual voltage drop is proportional to the output current, and the proportionality coefficient is the complex impedance Z. v =R v +jωL v By setting R v A negative value achieves negative resistance characteristics within a specific frequency range, thereby offsetting the positive resistance of the line and reducing the overall impedance. The parameter R of the virtual impedance is... v and L v Dynamic adjustments are made based on the node impedance spectrum analysis results. To target the most sensitive disturbance frequencies, impedance shaping and proactive power scheduling work together to improve voltage stability. The virtual impedance element is inserted into the inverter's dual closed-loop control structure.

9. A dynamic voltage recovery and flicker suppression system based on photo-storage synergy for implementing the dynamic voltage recovery and flicker suppression method based on photo-storage synergy as described in any one of claims 1 to 8, characterized in that, The system includes: Multiple intelligent monitoring units are installed in substations, branch boxes, and distributed power access points of the distribution network. Each intelligent monitoring unit includes a status input module, an electrical quantity measurement module, and a communication module. The central control platform consists of a server cluster and runs topology management software, impedance analysis software, vulnerability calculation software, visualization software, and control decision software. At least one energy storage system includes a battery bank, a supercapacitor bank, and a bidirectional converter, which is connected to the distribution network through a circuit breaker. The communication network adopts wired Ethernet or wireless cellular network to provide data transmission between the intelligent monitoring units and the central control platform, and between the central control platform and the energy storage system. The intelligent monitoring units upload switch status and voltage and current data in real time. The central control platform processes the data and generates control commands, which are then sent to the energy storage system for execution via the communication network. The central control platform also includes a database to store historical data and real-time parameters.

10. A dynamic voltage recovery and flicker suppression system based on photo-storage synergy according to claim 9, characterized in that, The central control platform also includes an adaptive learning module for continuously optimizing the vulnerability index calculation model. The adaptive learning module initiates the optimization process at fixed time intervals. The optimization process includes collecting all historical data within the past time window. The data includes the impedance ratio, electrical distance, load sensitivity, voltage deviation, and whether voltage disturbances have occurred for each node. After cleaning and feature engineering the data, a new classifier is trained using the random forest algorithm. The input features are the impedance ratio, the reciprocal of the electrical distance, the load sensitivity, and the voltage deviation. The output is the disturbance probability. After training, the average Gini importance score of each feature in the random forest is calculated. The score is normalized to obtain a new weight coefficient. The difference between the new weight coefficient and the old weight coefficient is compared. If the difference exceeds a threshold, the vulnerability index calculation formula is updated with the new weight coefficient. At the same time, the old model is retained as a backup to back up when the new model fails to be validated. The adaptive learning module also generates an optimization report to record the history of weight coefficient changes. The time window is set to seven days and the threshold is set to 10%.

Images