A distributed regional power distribution autonomous regulation system with complementary coordination of resources

Abstract

The application relates to the technical field of power distribution autonomous regulation and control, in particular to a regional power distribution autonomous regulation and control system based on complementary coordination of distributed resources. The system comprises: an initial partition module that clusters nodes to generate initial regional division; a dynamic re-partition module that locates affected areas and reallocates boundary nodes; an autonomous regulation and control module that updates control parameters according to re-partition information, separates parameters according to fast and slow time scales, and sets regional capacity labels in combination with supply-demand balance and callable load evaluation results; and a complementary coordination module that determines coordinated scheduling areas according to the capacity labels, drives reactive power flow through fine-tuning voltage, and solves the problem of insufficient reactive power resources in the areas. The application solves the problem that the fixed partition boundary of the existing power distribution network cannot adapt to changes in operating conditions, and improves the adaptability and control robustness of the partition strategy to complex and variable operating conditions of the power distribution network.

Description

Technical Field

[0001] This application relates to the field of autonomous power distribution control technology, specifically to a regional autonomous power distribution control system with distributed resource complementarity and collaboration. Background Technology

[0002] With the integration of numerous distributed resources, such as photovoltaic, wind power, energy storage, and charging piles, different types of energy within these distributed resources exhibit varying probabilistic fluctuation characteristics and power operation characteristics. This has led to the distribution network gradually displaying features such as active, bidirectional, and highly uncertain characteristics. Therefore, dividing a large-scale regional distribution network into several smaller regions and implementing collaborative autonomy among these smaller regions has gradually become the mainstream choice.

[0003] When conducting autonomous regulation of regional power distribution, it is necessary to take into account both the instability of distributed resources and the actual operating conditions of the power distribution network. However, existing methods mostly adopt offline zoning strategies, which generally suffer from fixed boundaries of autonomous units and cannot be dynamically adjusted according to changes in operating conditions such as photovoltaic output, load levels, and network topology. In addition, the few methods that involve dynamic zoning often rely on the setting of empirical parameters of the model, which reduces the adaptability and control robustness of the zoning strategy to the complex and ever-changing operating conditions of the power distribution network, making it difficult to meet the actual needs of dynamic coordinated regulation of the power distribution network. Summary of the Invention

[0004] To address the aforementioned technical problems, the purpose of this application is to provide a regional power distribution autonomous control system with distributed resource complementarity and collaboration, the specific technical solution of which is as follows:

[0005] This application proposes a distributed resource complementary and collaborative regional power distribution autonomous control system, the system comprising:

[0006] The initial partitioning module is used in the offline initialization phase to construct the electrical distance matrix and voltage-reactive power sensitivity matrix based on the fixed line parameters and topology of the distribution network, so as to cluster the nodes in the distribution network topology and obtain the initial region partitioning results.

[0007] The dynamic rezoning module is used to determine whether structural changes or statistically significant mode drifts that require adjustment of the zone boundaries have occurred during the operation of the distribution network through dual-channel trigger detection. Based on the judgment results, the affected areas and nodes are identified, and the affected boundary nodes are redistributed.

[0008] The autonomous control module is used to receive specific information about the rezoning after dynamic rezoning is completed, update control parameters, separate control parameters according to fast and slow time scales, and set regional capacity labels based on the regional load supply and demand balance relationship and the available load assessment results.

[0009] The complementary coordination module is used to determine the coordinated scheduling area of ​​the reactive power resource shortage area based on the regional capacity label, and to drive reactive power flow by fine-tuning the voltage in the two areas.

[0010] Preferably, the process of constructing the electrical distance matrix and the voltage-reactive power sensitivity matrix is ​​as follows:

[0011] Assume the electrical distance matrix is A square matrix of order, in which, This represents the number of all nodes in the distribution network topology. The rows and columns of the electrical distance matrix correspond to the node numbers in the distribution network topology, and the elements in the electrical distance matrix are the electrical distances between nodes.

[0012] Based on the impact of reactive power injection changes at any node on the voltages of other nodes, a voltage-reactive power sensitivity matrix is ​​constructed. An dimensional square matrix, where the elements are the voltage-reactive power sensitivity values ​​between nodes.

[0013] Preferably, in the dynamic repartitioning module, the dual-channel trigger detection includes channel A1 and channel A2:

[0014] Channel A1 is used to monitor topology changes, including explicit and implicit topology changes;

[0015] Channel A2 is used to monitor slow mode drift, and the monitoring object is the boundary node after the initial region division.

[0016] Preferably, the explicit topology change is determined by monitoring the SOE change signals of all operable switches in the distribution network; the implicit topology change is identified by the equivalent impedance of the branch where the node is located in the distribution network. When the rate of change of the equivalent impedance of the branch is greater than a preset threshold and the correlation coefficient of the voltage waveform at both ends of the branch drops sharply, the implicit topology change is confirmed to have occurred.

[0017] Preferably, the process of slow mode drift monitoring in channel A2 is as follows:

[0018] For the boundary nodes after the initial region division, the deviation statistics of the boundary nodes are calculated using the voltage data within the preset observation window to construct the judgment index of the boundary nodes.

[0019] The EWMA control chart is used to monitor the judgment indicators of the boundary node. When the EWMA value exceeds the preset control limit and meets the preset duration requirement, it is determined that the voltage operation mode of the boundary node has slowly drifted.

[0020] Preferably, the redistribution of affected boundary nodes includes:

[0021] Electrical connectivity constraint checks and regional resource constraint checks are performed on the boundary nodes after the initial region division. Boundary nodes that pass the aforementioned two constraint checks are recorded as affected boundary nodes.

[0022] Calculate the degree of coupling affiliation of the affected boundary node to each of its adjacent regions, and take the region corresponding to the maximum degree of coupling affiliation as the optimal candidate region for the affected boundary node;

[0023] Node migration is performed when the benefit of migrating an affected boundary node to its optimal candidate region meets a preset criterion; otherwise, node migration is not performed.

[0024] Preferably, the calculation process for the degree of coupling affiliation is as follows:

[0025] Affected boundary nodes its adjacent areas Degree of coupling between The expression is: In the formula, Indicates the affected boundary nodes to its adjacent areas The normalized result of the quartiles of the electrical distances between all nodes within the node; Indicates the affected boundary nodes The normalized result of the mean voltage-reactive power sensitivity between all nodes in its adjacent region; , These represent the preset first weight and the preset second weight, respectively. .

[0026] Preferably, the regional capacity label includes a net load satisfaction coefficient obtained based on the regional load supply and demand balance relationship, and a load dispatchability coefficient obtained based on the callable load assessment results.

[0027] Preferably, the step of determining the coordinated scheduling area for areas with insufficient reactive power resources based on the regional capability label includes:

[0028] Based on the cross-regional coordination efficiency between the region with insufficient reactive power resources and its neighboring regions, and combined with the regional capacity labels of its neighboring regions, the comprehensive selection label value of each neighboring region of the region with insufficient reactive power resources is calculated.

[0029] Among adjacent regions that meet the preset necessary constraints, the adjacent region with the largest comprehensive selection label value is selected as the coordination and scheduling region for cross-regional coordination.

[0030] The preferred process for acquiring areas with insufficient reactive power resources is as follows:

[0031] Obtain the reactive power deficit in each region, and record the regions where the reactive power deficit exceeds the preset reactive power deficit threshold as regions with insufficient reactive power resources.

[0032] This application has the following beneficial effects:

[0033] This application constructs an electrical distance matrix and a voltage-reactive power sensitivity matrix, and combines a spectral clustering algorithm to divide network nodes, thereby obtaining initial region division results. This process uses an optimization algorithm to automatically optimize clustering parameters, effectively overcoming the drawbacks of traditional offline partitioning that relies on manual experience. It ensures that the partitioning results meet the constraints of topological connectivity and controllability, while possessing optimal intra-class compactness and inter-class separation, providing a reliable benchmark framework for subsequent dynamic repartitioning.

[0034] Furthermore, this application constructs a "dual-channel trigger detection-local node redistribution" mechanism during distribution network operation. Channel A1 monitors second-level topology changes, while channel A2 captures slow-changing mode drift, accurately identifying the timing when partition boundaries need to be adjusted. When repartitioning is triggered, matrix incremental update technology is used to quickly correct electrical parameters, and constraint verification and coupling affiliation calculation are performed only on the affected boundary nodes. Local migration is performed under the premise of ensuring topology connectivity and resource sufficiency, effectively solving the problem that traditional fixed partitioning cannot adapt to changes in operating conditions, avoiding the oscillation risk caused by network-wide re-clustering, and significantly improving the dynamic adaptability of the autonomous region of the distribution network to distributed resource fluctuations and network structure changes.

[0035] Furthermore, after dynamic rezoning is completed, this application receives zoning change information and updates control parameters separately according to fast and slow time scales. By calculating the net load satisfaction coefficient and load dispatchability coefficient, a regional capacity label is constructed, realizing a quantitative assessment of the regional supply and demand balance and response capability. At the same time, the node with the highest self-sensitivity in the region is selected as the voltage reference point, and the reactive power source output allocation weight is calculated by combining the sensitivity sub-matrix and the real-time available reactive power margin, ensuring the rationality of control command allocation and providing accurate regional control models and parameter support for subsequent coordinated control.

[0036] Ultimately, this application identifies regions with insufficient reactive power resources based on regional capacity labels. By calculating the inter-regional coordination efficiency index and comprehensively selecting label values, and under the premise of satisfying resource self-sufficiency and complementarity constraints, it selects the coordinated dispatching regions with the strongest electrical coupling and abundant resources. Then, it adopts a bidirectional voltage fine-tuning strategy, calculates the amplitude correction based on voltage-reactive power sensitivity, and constructs a voltage gradient between the supporting and supported regions to drive the inter-regional flow of reactive power. This effectively solves the problem of local reactive power shortage, realizes resource complementarity and voltage coordinated control between regions, and improves the adaptability and control robustness of the zoning strategy to the complex and variable operating conditions of the distribution network. Attached Figure Description

[0037] Figure 1 A block diagram of a distributed resource complementary and collaborative regional power distribution autonomous control system provided in one embodiment of this application;

[0038] Figure 2 This is a schematic diagram of the reactive power resource shortage area screening process provided in one embodiment of this application. Detailed Implementation

[0039] The following description, in conjunction with the accompanying drawings, details a specific scheme for a distributed resource complementary and collaborative regional power distribution autonomous control system provided in this application.

[0040] Please see Figure 1 The diagram illustrates a block diagram of a distributed resource complementary and collaborative regional power distribution autonomous control system according to an embodiment of this application. The system includes an initial partitioning module 101, a dynamic repartitioning module 102, an autonomous control module 103, and a complementary coordination module 104.

[0041] Application scenario description:

[0042] This embodiment applies to a regional distribution network with a high proportion of photovoltaic and wind power integration, specifically an IEEE 33-node distribution network, comprising 20 nodes connected to distributed resources. Of these, 15 nodes are connected to distributed photovoltaic power, 5 nodes to distributed wind power, and 3 nodes are equipped with energy storage. This embodiment achieves the goal of dynamically triggering rezoning and coordinating control between regions by constructing a closed-loop method of "dynamic zoning-coordinated control." The implementation process is as follows:

[0043] The initial partitioning module 101 is used to construct an electrical distance matrix and a voltage-reactive power sensitivity matrix based on the fixed line parameters and topology of the distribution network during the offline initialization phase, so as to cluster the nodes in the distribution network topology and obtain the initial region partitioning results.

[0044] In the offline initialization phase, modeling is first performed based on the fundamental physical model of the distribution network. The topology of the distribution network consists of buses, feeders, sectionalizing switches, and tie switches, representing a connected graph where nodes are linked by lines. Line parameters include line resistance, reactance, and ground susceptance, which determine the electrical characteristics of the lines. Based on the aforementioned fixed line parameters and topology, the electrical distance matrix and voltage-reactive power sensitivity matrix between any two nodes are calculated. Clustering is performed using electrical distance as the primary feature and sensitivity as the secondary feature to obtain the initial region division results, which serve as the benchmark for subsequent dynamic partitioning.

[0045] 1. Construction of the electrical distance matrix:

[0046] The electrical distance matrix is ​​constructed based on the magnitude of the equivalent impedance between nodes. The smaller the magnitude of the equivalent impedance between two nodes, the closer the electrical distance and the tighter the electrical coupling. The specific construction process of the electrical distance matrix includes steps i-iii:

[0047] i. Calculate the node admittance matrix between nodes based on the node number and branch parameters (including resistance, reactance, and ground admittance, etc.). .

[0048] Based on the network topology of the distribution network, let the number of network nodes be... Node admittance matrix for A complex matrix of order 1, whose elements Determined by the admittance of the node branches, the specific construction method is as follows:

[0049] For non-diagonal elements If node With nodes If there are branch roads directly connecting them, then It equals the negative value of the branch admittance, that is If there is no direct branch connection between the two nodes, then ;

[0050] For diagonal elements Its value is equal to that of the node. The sum of the admittances of all connected branches, plus the admittance of the branch to ground at node i (i.e., half of the line-to-ground capacitance and the admittance of the parallel reactor connected to the node).

[0051] In summary, the nodal admittance matrix Represented as:

[0052] ii. Calculate the impedance matrix between nodes : ;

[0053] The diagonal elements in the matrix represent the nodal self-impedance. , The elements represent the self-impedances of nodes 1 and n, respectively; the off-diagonal elements represent the mutual impedances between nodes. , Both represent the mutual impedance between node 1 and node n, and , where n is the number of nodes.

[0054] iii. Electrical distance matrix between nodes Construction:

[0055] The equivalent impedance between two nodes is the sum of the self-impedances of the two nodes minus twice the mutual impedance; equivalent impedance The electrical distance between nodes is the magnitude of the equivalent impedance. The electrical distance between all nodes is constructed symmetric electrical distance matrix :

[0056] 2. Construction of the voltage-reactive power sensitivity matrix:

[0057] Based on the impact of changes in reactive power injection at a node on the voltage of other nodes, a voltage-reactive power sensitivity matrix is ​​constructed. The elements in the matrix represent voltage-reactive power sensitivity, which expresses the response of the voltage amplitude of each node in the distribution network topology to a small change in reactive power injection at a certain node. The construction process includes steps i-iii:

[0058] i. Calculate the active and reactive power injection equations of the nodes using the admittance matrix between nodes and the voltage magnitude and phase angle of the nodes: ; ;

[0059] In the formula, These represent the active power and reactive power injected at node i, respectively. These represent the voltage magnitudes at nodes i and j, respectively. The phase difference between nodes i and j; These are the real and imaginary parts of the element in the i-th row and j-th column of the admittance matrix Y, respectively.

[0060] ii. Perform total differential processing on the active and reactive power injection equations P and Q of the nodes to obtain four Jacobian submatrices H, N, J, and L: ;

[0061] In the formula, the elements of the four Jacobian submatrices are defined as follows:

[0062] This is the vector of changes in active power at the nodes. ;

[0063] This is the vector of reactive power changes at the nodes. ;

[0064] This is the vector of changes in the node voltage phase angle. ;

[0065] This is a vector representing the change in the magnitude of the node voltage. ;

[0066] Matrix: Composed of the partial derivatives of active power with respect to voltage phase angle, elements This reflects the impact of phase angle changes on active power;

[0067] Matrix: Composed of the partial derivatives of active power with respect to voltage amplitude, elements This reflects the impact of voltage amplitude changes on active power;

[0068] Matrix: Composed of the partial derivatives of reactive power with respect to voltage phase angle, elements This reflects the impact of phase angle changes on reactive power;

[0069] Matrix: Composed of the partial derivatives of reactive power with respect to voltage amplitude, elements This reflects the impact of voltage amplitude changes on reactive power.

[0070] iii. Extract the sensitivity relationship between voltage and reactive power, and construct a voltage-reactive power sensitivity matrix. Invert the Jacobian matrix equation to convert the state variables ( Move the control variable () to the left side of the equation. Move to the right and use the block matrix inversion formula to obtain the sensitivity matrix: ;

[0071] From this, the change in voltage amplitude can be obtained. Relationship with power change: ;

[0072] In the formula, This is the voltage phase angle-active power sensitivity matrix, which reflects the impact of active power changes on the voltage phase angle; This is the voltage phase angle-reactive power sensitivity matrix, reflecting the impact of reactive power changes on the voltage phase angle; This is the voltage-active power sensitivity matrix, which reflects the impact of active power changes on voltage amplitude;

[0073] This is the voltage-reactive power sensitivity matrix, which reflects the impact of reactive power changes on voltage amplitude.

[0074] When only reactive power voltage control is considered, and the impact of active power fluctuations is ignored (i.e., assuming...), When ), the above formula simplifies to At this time, the matrix This is the voltage-reactive power sensitivity matrix, whose elements... Represents a node When the reactive power injection changes by a unit, the node The degree of response generated by the voltage amplitude, the diagonal elements in the matrix This represents the impact of reactive power changes at a node on its own voltage. An element value greater than 0 indicates that when reactive power injection is added to a node, the voltage of the corresponding node increases. The larger the element value, the tighter the electrical coupling between the two nodes and the stronger the control capability. An element value close to 0 indicates that the electrical distance between the two nodes is large and the impact is weak.

[0075] 3. Clustering and initial partitioning results:

[0076] Based on electrical distance matrix Voltage-Reactive Power Sensitivity Matrix Cluster the n nodes to obtain the initial partitioning results.

[0077] First, based on the electrical distance matrix The element values ​​corresponding to two nodes and the matrix The ratio of the largest element value in the middle determines the normalized metric distance between the two nodes. .

[0078] Secondly, based on matrix The reactive coupling distance between the nodes is constructed using the corresponding elements, denoted as . The smaller the distance, the more likely the two nodes should be assigned to the same autonomous region. With nodes reactive coupling distance between , where max() represents the maximum value function.

[0079] After that, and Perform linear weighting ( The weight is set to 0.7. (The weight is set to 0.3), and the comprehensive distance is obtained. A similarity matrix is ​​constructed using comprehensive distance. ,in exp() represents an exponential function with base to the natural constant. The width parameter (i.e., bandwidth) of the Gaussian kernel function is used to control the rate at which similarity decays. It is a variable to be optimized and is automatically determined through an optimization algorithm.

[0080] Considering the non-convex structure of distribution network node connections, this embodiment utilizes a spectral clustering algorithm based on the similarity matrix. and Gaussian kernel bandwidth The constructed normalized Laplace matrix is ​​used for node clustering.

[0081] Let the preset value of the number of clusters be 1. Randomly set different options within the selectable range. Combining to maximize the evaluation indicators of zoning rationality Optimize the objective function.

[0082] Regarding parameters Definition: This represents the total number of regions (i.e., the number of clusters) into which the distribution network is divided, and is a positive integer. For the IEEE 33-bus system, it is set to... The selectable range is .

[0083] Regarding parameters Definition: Gaussian kernel bandwidth is a parameter used in spectral clustering algorithms to adjust the distribution characteristics of the similarity matrix and is an evaluation index for the rationality of partitioning. The formula for quantitatively evaluating cluster quality is as follows: ;

[0084] The symbols in the formula are defined as follows:

[0085] Total number of regions (i.e., number of clusters);

[0086] : No. The average of the comprehensive distances between all nodes within the cluster is used to characterize the _th cluster. Intra-class density of each region;

[0087] : No. The cluster centers of each region and the rest The sum of the comprehensive distances between the cluster centers of the regions is used to characterize the . Inter-class separation degree between each region and other regions;

[0088] The value of the partition rationality evaluation is as follows: the larger the value, the greater the distance between classes and the smaller the distance within classes, and the better the clustering effect.

[0089] The optimization process must satisfy the following constraints:

[0090] Topological connectivity constraints: Partitions must satisfy topological connectivity to ensure that each partition can form a connected subgraph; otherwise, they cannot operate independently.

[0091] Controllability constraint: Each zone must include a reactive power compensation device to ensure that each zone is controllable and can participate in the coordinated control of the system.

[0092] After the optimization process is completed, the optimal Gaussian kernel bandwidth will be obtained. Number of clusters Substitute the data into the spectral clustering model to output the initial partitioning results.

[0093] Finally, the boundary nodes in the clustering results are determined based on the heterogeneity of the neighbors: for each node, if there are at least two nodes in different clusters among its neighbors, then the node is marked as a boundary node.

[0094] The dynamic rezoning module 102 is used to determine whether there is a structural change or statistically significant mode drift that requires adjustment of the partition boundary during the operation of the distribution network by means of dual-channel trigger detection. Based on the judgment result, the affected areas and nodes are determined, and the affected boundary nodes are redistributed.

[0095] During the operation of the distribution network, existing autonomous regional methods often remain unchanged after the initial partitioning, neglecting the problem that changes in operating conditions such as sudden drops in distributed photovoltaic power may cause the complementary relationship between nodes in the same area to fail, or that energy storage resources in adjacent areas cannot be accessed across fixed boundaries. If a single node is merged based solely on the distance change calculated by measuring the real-time injected power and voltage amplitude of the node, the node migration will change the cluster center of the two areas, thereby changing the coupling reference benchmark of all remaining boundary nodes, which can easily trigger cascading migrations or even oscillations.

[0096] To address this, this embodiment constructs a dynamic rezoning mechanism of "dual-channel trigger detection - local node reallocation". During distribution network operation, SCADA telemetry data and switch quantity change signals are received. A dual-channel trigger detection method is used to determine whether structural changes requiring adjustment of zone boundaries or statistically significant mode drift have occurred. Based on the determination results, the affected areas and nodes are identified. Reallocation is performed only on the affected boundary nodes, and a final dynamic allocation scheme is determined through a decision fusion strategy. The allocation change information is then transmitted to the autonomous module. The specific process is as follows:

[0097] 1. Channel A1: Monitoring of topology changes.

[0098] Channel A1 monitors topology changes on a second-by-second basis, including explicit and implicit topology changes, to identify critical electrical coupling events that trigger repartitioning.

[0099] 1.1 Explicit topology change monitoring:

[0100] Explicit topology changes are detected by monitoring the SOE (Sequence of Events) change signals of all operable switches in the distribution network. A change signal indicating a switch state change is detected when the change stabilizes for a set time. (In this embodiment, it is set to 2 seconds) After that, it is confirmed that an explicit topological change has occurred.

[0101] 1.2 Implicit Topological Change Monitoring:

[0102] Implicit topology changes are identified by the equivalent impedance of the branch where the node is located in the distribution network.

[0103] First, the equivalent impedance of the branch is estimated based on the electrical parameters of the nodes at both ends of the branch. In this embodiment, the PMU synchronous measurement method is used to obtain the voltage difference between the two nodes at both ends of the branch by utilizing the synchronous phasor measurement unit (PMU) installed at the node. and branch current Calculate the equivalent impedance of the branch according to Ohm's law. : ;

[0104] In the formula, The voltage difference between the two nodes of the branch is calculated by subtracting the voltage of the end node from the initial node voltage.

[0105] Secondly, maintain an online equivalent impedance monitor for each branch. Definition for The equivalent impedance at time t. The equivalent impedance at the start of the distribution network operation; when the absolute rate of change of the branch equivalent impedance... Greater than the threshold And the duration of this state exceeds the time threshold. Preliminary assessment indicates that the branch may be experiencing a change in condition: ;

[0106] In this embodiment, the threshold Set to 25%, time threshold Set to 5 seconds.

[0107] Simultaneously, an auxiliary criterion is set: because the voltage waveforms at both ends of the branch are highly synchronized during normal operation, the Pearson correlation coefficient is usually greater than 0.95. Definition for The Pearson correlation coefficient between the voltage waveforms at the two nodes at the two ends of the branch at time point 1 is used to calculate its absolute rate of change compared to the previous time point. .like Greater than the threshold (Threshold in this embodiment) If 15% is taken, it is determined that the correlation coefficient drops sharply. When both the impedance mutation and the auxiliary criterion are valid, the implicit topological change is confirmed to prevent misjudgment caused by measurement anomalies.

[0108] Note: This embodiment does not directly use node voltages for implicit topology determination because node voltages have a low signal-to-noise ratio to topology changes and are affected by the overall network power flow, lacking attribution capability. Branch equivalent impedance estimation transforms the global coupling problem into a local identification problem, and combined with auxiliary conditions, reliable implicit topology identification can be achieved even without SOE signals.

[0109] 2. Incremental matrix update:

[0110] Once detected Topology changes are always present, immediately triggering the electrical distance matrix. Voltage-Reactive Power Sensitivity Matrix Incremental updates.

[0111] 2.1 Matrix Update:

[0112] Single branch interruption or commissioning affects admittance matrix The effect is manifested in the increase or decrease of the admittance of the corresponding elements at the two ends of the branch, and its corrected rank does not exceed 2. Therefore, the Sherman-Morrison rank-2 correction formula is used to correct the impedance matrix. Incremental updates are performed to avoid full-dimensional matrix inversion. The update formula is as follows (assuming there is a branch between node i and node j, the branch between node i and node j is used). (Example: Opening / closing) ;

[0113] The definitions of the symbols and variables in the formula are as follows:

[0114] : The nodal impedance matrix before the update;

[0115] : The updated node impedance matrix;

[0116] Branches that have changed Its own impedance value;

[0117] : Admittance matrix correction, physically representing the change in the diagonal elements of the nodal admittance matrix after a branch is disconnected. For a branch... The switching operation is equal to the negative value of the branch admittance, i.e. ;

[0118] : are matrices respectively The Column vector, the first Column vectors;

[0119] : are matrices respectively The row vector, first row vector Row vectors;

[0120] Impedance matrix After the update is completed, the equivalent impedance is recalculated according to the method in the initial partitioning module 101 to obtain the updated electrical distance matrix. .

[0121] 2.2. Matrix Update:

[0122] Recalculate the Jacobian submatrix at the current tidal current section The sensitivity matrix was recalculated using the Schur complement formula. After the update, the system will automatically perform matrix property checks (such as symmetry and diagonal element positivity). If any abnormalities are found, the system will trigger a full matrix recalculation.

[0123] 3. Channel A2: Mode drift monitoring.

[0124] Channel A2 serves as a supplementary channel to capture slowly varying pattern drift without explicit topological events. After initial partitioning, boundary nodes are more susceptible to the combined effects of multi-source interference from adjacent partitions compared to non-boundary nodes. Therefore, boundary nodes within each region are the monitoring targets of channel A2.

[0125] 3.1. Calculation of Judgment Indicators:

[0126] For each boundary node, set the observation window. Taking boundary node g as an example, the judgment index of boundary node g at time t is calculated using the voltage data from SCADA telemetry within the observation window. : ;

[0127] To ensure the feasibility of the technical solution, the variables and calculation process in the formula are defined in detail as follows:

[0128] Observation window With sampling mechanism:

[0129] Observation window The time window is a sliding window with a length of 3 minutes; SCADA acquires the voltage of the boundary node g at time t. and the transformer cut-off voltage at time t .

[0130] Calculation of voltage deviation:

[0131] In the observation window Inside, calculation Voltage deviation at boundary node g at time t This voltage deviation is defined as the difference between the node voltage and the transformer switching voltage. The use of difference rather than ratio is to more intuitively reflect the fluctuation of line voltage drop.

[0132] Calculation of deviation statistics:

[0133] Collect observation window before time t The median of the voltage deviation of the boundary node g at all times within the time interval is used as the deviation statistic of the lower boundary node g at time t.

[0134] Historical baseline parameters and Definition:

[0135] Historical normal operating conditions time span: defined as the continuous time period during which the distribution network is in a typical daily operating mode. In this embodiment, the operating data of the past 30 days is selected as the historical sample library.

[0136] Benchmark calibration method: During the historical normal operating conditions of boundary node g, select a time period that is the same as the current observation window each day (e.g., 14:00-14:03 each day), and calculate the deviation statistic for that time period using the method described above. Make use of these 30 days average and standard deviation The deviation statistics of the lower boundary node g at time t are standardized, and the result is used as the judgment index of the lower boundary node g at time t.

[0137] 3.2. EWMA Control Chart Drift Determination:

[0138] Using Exponentially Weighted Moving Average (EWMA) control charts to monitor and judge indicators Tiny fluctuations, EWMA statistic of the boundary node g at time t The iterative formula is: ;

[0139] In the formula:

[0140] : respectively time, The EWMA statistic of the boundary node g at time t;

[0141] The preset weighting factor is set to 0.25 in this embodiment to maintain moderate smoothness and sensitivity to step drift.

[0142] Preset control limits Set as: ;

[0143] In the formula, To control the limit coefficient, this embodiment uses 3;

[0144] when Exceeding control limits When the range is reached, enter warning status;

[0145] Multi-node drift confirmation: If two or more boundary nodes in a partition are in a warning state and continuously exceed the limit in the same direction (simultaneously exceeding the upper or lower limit), it is determined that there is a slow drift phenomenon.

[0146] Single-node drift confirmation: If only one boundary node in the partition enters the warning, its one-way out-of-limit duration must exceed [a certain value]. (In this embodiment, it is set to 30 minutes) before it is determined to be drift, in order to prevent frequent triggering of partition changes due to single-point measurement abnormalities.

[0147] 4. Determination of the affected area and priority of the dual channels.

[0148] Affected areas: When channel A1 is triggered, the affected area is the partition to which the nodes at both ends of the branch where the topology change occurs belong; when channel A2 is triggered, the affected area is the partition to which the boundary node where the drift is confirmed belongs.

[0149] Affected Boundary Node Set: This set comprises all boundary nodes within the affected area (channel A1) or boundary nodes identified as drifting (channel A2). .

[0150] Execution priority: Channel A1 (topology change) has higher priority than channel A2. During the steady state, channel A2 is the only monitoring method. When channel A1 is triggered, channel A2 evaluation is suspended and topology events are handled first.

[0151] 5. Node migration and repartitioning decisions.

[0152] Based on the set of affected boundary nodes The specific steps for node migration are as follows:

[0153] 5.1. Constraint Check:

[0154] For the affected boundary nodes Then, electrical connectivity constraint checks and regional resource constraint checks are performed sequentially:

[0155] Electrical connectivity constraints: affected boundary nodes The nodes are topologically directly adjacent to the target migration region; and after the nodes migrate, the remaining nodes in the original region still satisfy topological connectivity, where the target region consists of the affected boundary nodes. The adjacent area to which the relocation is planned;

[0156] Regional resource constraints: Affected boundary nodes The current area has a remaining available reactive power capacity of no less than 30% of the area's reactive power deficit; and the affected boundary nodes... The amount of flexible load that can be called up shall not be less than 5% of the peak load demand of the target area.

[0157] 5.2. Calculation of Coupling Attribution Degree:

[0158] For affected boundary nodes that pass the constraint check ,calculate With each adjacent area Coupling degree : ;

[0159] In the formula:

[0160] , : These are the preset first weight and the preset second weight, respectively. In this embodiment, it is set , ;

[0161] Affected boundary nodes to its adjacent areas The normalized result of the quartiles of the electrical distances between all nodes is used to resist outlier interference;

[0162] Affected boundary nodes its adjacent areas Normalized result of the mean voltage-reactive power sensitivity among all nodes within the node;

[0163] When normalizing the quartiles and the mean voltage-reactive power sensitivity, the maximum value normalization method is used. The maximum value parameter for normalizing the quartiles is taken as the maximum value among the corresponding quartiles of all affected boundary nodes. The normalization process of the mean voltage-reactive power sensitivity is the same as that of the quartiles, and will not be described in detail here. The process of normalizing the data using the maximum value normalization method is a well-known technique, and will not be described in detail here either.

[0164] 5.3. Migration Criterion:

[0165] Define the affected boundary nodes Current region of origin The calculated optimal candidate region is (Right now (Maximum), the migration will be performed only if the following benefit conditions are met: ;

[0166] In the formula, As the lower limit of the migration benefit ratio, this embodiment takes 10%. If the coupling affiliation degree of the target area is higher than the sum of the affiliation degree of the original area and the lower limit of the benefit, then the migration action is performed; if the increase in coupling affiliation degree before and after the migration is less than 10%, then it is considered that the migration benefit is insufficient to offset the switching cost, and the status quo is maintained.

[0167] 5.4. Repartitioning Execution Constraints:

[0168] To maintain system stability, the following constraints are enforced:

[0169] Single migration limit: Maximum number of nodes that can be migrated in a single event. Select the top performers in descending order of their affiliation gain. Each node performs the migration.

[0170] Freeze-time mechanism: After the migration is implemented, the affected areas will enter a freeze-time period. (Empirical value is 15 minutes). During the freeze period, channel A1 still monitors topology events, but the evaluation scope excludes nodes within the freeze area and newly migrated nodes.

[0171] Cluster center update: After the migration is completed, the electrical distance centroid of the affected area is recalculated as the new cluster center, and the boundary nodes are updated.

[0172] The autonomous control module 103 is used to receive specific information about the rezoning after the dynamic rezoning is completed, update the control parameters, separate the control parameters according to the fast and slow time scales, and set regional capacity labels according to the regional load supply and demand balance relationship and the available load assessment results.

[0173] After dynamic repartitioning is completed, the autonomous module receives the specific information of the repartition, including the set of migrated boundary nodes and the updated region affiliation, which is used to update the control parameters. In this embodiment, the control parameters are separated according to the speed of the time scale:

[0174] Long-cycle update parameters: These are parameters related to the distributed resource complementarity and collaborative control between regions, and are set to be updated daily or weekly as needed; Short-cycle update parameters: These are control parameters related to the autonomous operation of the power grid within a zone, and are set to be updated every minute or every control time window as needed.

[0175] When coordinating and controlling each partition, it is first necessary to assess the load supply and demand balance and the amount of available load resources in each partition.

[0176] 1. Assessment of regional supply and demand capacity and response capability.

[0177] 1.1. Calculation of Net Load Satisfaction Factor:

[0178] By region For example, for each node within the region, obtain the historical time window. (In this embodiment, the value is taken as 30 days) The daily load curve and active power output curve of the node within this region are used for calculation. Net load satisfaction factor : ;

[0179] The parameters in the formula are defined as follows:

[0180] : respectively, regions In the historical time window The highest and lowest values ​​of the sum of the load power of all nodes within the area;

[0181] : respectively, regions In the historical time window The highest and lowest values ​​of net internal exchange power;

[0182] Net exchange power Defined as the difference between the net load power of the area and the active power output of distributed generation (DG) within the area, the calculation formula is: ;

[0183] in, For nodes within region R The load power, For nodes within region R The active power output of distributed power sources; This represents the set of all nodes within region R.

[0184] 1.2. Calculation of load dispatchability factor:

[0185] For the region For each node within the system, calculate its maximum responsive load and the load within historical time windows. The response achievement rate to coordinated control commands is used to characterize a node's ability to respond to system control commands; if a node has no response record within a time window, the average value of nodes of the same type within that region is taken. This calculation region is then used. Load dispatchability factor : ;

[0186] The parameters in the formula are defined as follows:

[0187] :area internal nodes Available load capacity (unit: kW);

[0188] :area internal nodes Response achievement rate, dimensionless parameter, range of values A higher value indicates a higher responsiveness to system commands.

[0189] The range of values ​​is The larger the value, the larger the region. The stronger the overall response capability of internal nodes to coordination and scheduling, the better the regional control of distributed resources across regions. The higher the response priority, the better.

[0190] 1.3. Region Label Definition:

[0191] Net load satisfaction factor and load dispatchability factor serve as regional labels, reflecting the region's coordination capability.

[0192] The smaller the net load satisfaction factor, the more difficult it is for reactive power sources in the corresponding area to meet the needs of the area, and the stronger the dependence on cross-regional coordination and control. Prioritizing service to this area can systematically reduce global imbalance.

[0193] The higher the load dispatchability coefficient, the more abundant the flexible load resources in the corresponding area, and the lower the cost and the higher the completion rate of executing demand response commands.

[0194] 2. Determination of zoned voltage reference points and power output allocation weights:

[0195] 2.1. Determine the voltage reference point:

[0196] This embodiment adopts a distributed autonomous architecture. The control master station does not directly control the equipment, but instead sends voltage correction values ​​as control commands to the regional controllers. The controllers in each region then control the reactive power equipment in their respective regions according to the control commands.

[0197] For each region, the reference point for calculating the voltage correction must first be determined, using the voltage-reactive power sensitivity matrix. The node corresponding to the maximum self-sensitivity in each region is selected as the voltage reference point.

[0198] Self-sensitivity: refers to the matrix diagonal elements in Its physical meaning is node. The degree to which changes in reactive power injection affect the amplitude of its own voltage.

[0199] Uniqueness verification: When two or more nodes have completely equal self-sensitivity values ​​and are all at their maximum values, a single node is selected according to a preset uniqueness rule (such as prioritizing the node with the largest rated capacity of the reactive power source corresponding to the node) to ensure that the voltage reference point interface output in this area has uniqueness.

[0200] 2.2. Constructing the sensitivity sub-matrix:

[0201] For each region, use the matrix And the set of nodes and the set of controllable reactive power sources within the region, to construct the sensitivity sub-matrix of reactive power sources within the region. .

[0202] Sensitivity submatrix of region R The dimension is ,in, For the region The number of nodes within, For the region The number of internally controllable reactive power sources.

[0203] matrix Each column in the table corresponds to a controllable reactive power source, and each row corresponds to a node. Column vectors represent the first The vector consisting of the reactive power output of each reactive source and the voltage-reactive sensitivity between all nodes.

[0204] 2.3. Calculate the power distribution weights for reactive power sources:

[0205] Based on the current reactive power output of the reactive power sources measured by SCADA and the maximum reactive power capacity at the current active power output level, determine the current available reactive power margin for each reactive power source, and combine it with the sensitivity sub-matrix. Calculate the output allocation weight of reactive power sources.

[0206] Define region Inner Available reactive power margin of each reactive power source for: ;

[0207] In the formula:

[0208] :area Inner The maximum reactive power capacity that a reactive power source is allowed to generate under the current active power output;

[0209] :area Inner The current measured reactive power output of the reactive power source;

[0210] area Inner Power distribution weight of each reactive power source The calculation formula is: ;

[0211] The parameters in the formula are defined as follows:

[0212] L2 norm;

[0213] :area The total number of internal power sources.

[0214] The complementary coordination module 104 is used to determine the coordinated scheduling area of ​​the reactive power resource shortage area based on the area capacity label, and to drive reactive power flow by fine-tuning the voltage in the two areas.

[0215] 1. Determination of the coordination and scheduling area.

[0216] 1.1 Identify areas with insufficient reactive power resources:

[0217] When reactive power resources are insufficient in a certain area, the adjacent area with the strongest electrical coupling is selected, and the voltage amplitude of the voltage reference point in both areas is adjusted in both directions to drive reactive power to flow from the area with sufficient resources to the area with insufficient resources.

[0218] Get area reactive power deficit The reactive power deficit threshold of region R Set as a region The region with a reactive power deficit exceeding 30% of the maximum reactive power deficit within the historical time window is designated as a region with insufficient reactive power resources.

[0219] Preferably, the schematic diagram of the reactive power resource shortage area screening process provided in this embodiment is as follows: Figure 2 As shown.

[0220] 1.2. Calculation of Cross-Regional Coordination Efficiency Index:

[0221] Based on the voltage-reactive power sensitivity matrix and electrical distance matrix Calculate regions with insufficient reactive power resources its adjacent areas Cross-regional coordination efficiency index This index is used to assess the abundance of adjustable reactive power resources in adjacent areas and the control effect on areas with insufficient reactive power resources. The higher the value, the higher the cross-regional coordination efficiency. The expression for the cross-regional coordination efficiency index is: ;

[0222] The parameters in the formula are defined as follows:

[0223] :matrix The elements in the text represent regions. Voltage reference node and region within Voltage-reactive sensitivity between internal voltage reference nodes;

[0224] :area The sum of available reactive power margins of all reactive power sources within the system;

[0225] :area Nodes and regions The mean of the electrical distance elements between nodes is used to characterize the degree of electrical isolation between two regions;

[0226] Power reference value, in this embodiment ;

[0227] Impedance reference value, in this embodiment .

[0228] 1.3. Comprehensive selection of label value calculation:

[0229] For each region with insufficient reactive power resources, calculate its cross-regional coordination efficiency index with all adjacent regions. First, determine the maximum value among the cross-regional coordination efficiency indices between the region T with insufficient reactive power resources and all its adjacent regions. Is it zero?

[0230] like This indicates that all adjacent areas have no available reactive power margin, terminates the current cross-regional collaborative control, and records and reports the resource shortage event.

[0231] like Obtain the area labels of all adjacent areas, and comprehensively calculate the adjacent areas of the reactive power resource-deficient area T. Comprehensive selection of tag values : ;

[0232] In the formula:

[0233] , : These are the adjacent regions of region T with insufficient reactive power resources. Net load satisfaction factor and load dispatchability factor;

[0234] : Weighted weights, satisfying and In this embodiment, empirical values ​​are used. .

[0235] 1.4. Determine the coordination and scheduling area:

[0236] Based on the comprehensive selection of label values ​​and necessary constraints, the coordinated scheduling area is determined.

[0237] Necessary constraints include:

[0238] Resource self-sufficiency constraint: The available reactive power margin in the coordinated scheduling area minus the reactive power resource shortage area awaiting support. After the shortfall, the remaining margin is sufficient to meet its own needs.

[0239] Complementarity constraint: The net load satisfaction factor of the coordinated scheduling area is greater than that of the area with insufficient reactive power resources.

[0240] For each region with insufficient reactive power resources, among the adjacent regions that meet the above constraints, the region with the largest comprehensive selection label value is selected as the coordination and scheduling region for cross-regional coordination.

[0241] 2. Determination of voltage amplitude correction amount.

[0242] After determining the coordinated scheduling area, calculate the voltage amplitude correction amount of the voltage reference point within the area.

[0243] In areas with insufficient reactive power resources and its coordination and scheduling area For example:

[0244] Correction direction: Coordinated scheduling area As the supporting party, it should increase the voltage amplitude of its reference point to "push out" reactive power; in areas with insufficient reactive power resources... As the recipient of aid, it should reduce the voltage amplitude of its reference point to "absorb" reactive power, thereby creating a voltage gradient between the two.

[0245] Correction calculation:

[0246] Coordinated scheduling area Voltage amplitude correction : ;

[0247] Areas with insufficient reactive power resources Voltage amplitude correction : ;

[0248] In the formula:

[0249] Minimum function, used to limit the adjustment range;

[0250] The upper limit of the voltage regulation step size during a single cross-regional coordinated control is set to prevent excessive voltage fluctuations. In this embodiment, the value is 0.5kV.

[0251] The reactive power deficit in region T, where reactive power resources are insufficient;

[0252] :matrix The elements in the text represent the coordinated scheduling areas for regions with insufficient reactive power resources. Voltage reference node and reactive power resource shortage area Voltage-reactive sensitivity between internal voltage reference nodes;

[0253] , : are matrices respectively The elements in the text represent regions with insufficient reactive power resources. Internal voltage reference node's own voltage-reactive power sensitivity, coordinated scheduling area of ​​reactive power resource shortage region. Internal voltage reference node's own voltage-reactive sensitivity.

[0254] 3. Closed-loop control execution.

[0255] For regions with insufficient reactive power resources and regions requiring coordinated dispatch, each region independently implements autonomous control based on the aforementioned voltage amplitude correction. (The text then abruptly shifts to a seemingly unrelated topic: "Regions with insufficient reactive power resources...") For example, the voltage amplitude correction is superimposed on the voltage in the region with insufficient reactive power resources, serving as the input deviation of the PI controller. The PI controller outputs the total reactive power adjustment command, which is then distributed to each reactive power source device to complete the closed-loop control of voltage and reactive power.

Claims

1. A distributed resource-complementary and collaborative regional power distribution autonomous control system, characterized in that, The system includes: The initial partitioning module is used in the offline initialization phase to construct the electrical distance matrix and voltage-reactive power sensitivity matrix based on the fixed line parameters and topology of the distribution network, so as to cluster the nodes in the distribution network topology and obtain the initial region partitioning results. The dynamic rezoning module is used to determine whether structural changes or statistically significant mode drifts that require adjustment of the zone boundaries have occurred during the operation of the distribution network through dual-channel trigger detection. Based on the judgment results, the affected areas and nodes are identified, and the affected boundary nodes are redistributed. The autonomous control module is used to receive specific information about the rezoning after dynamic rezoning is completed, update control parameters, separate control parameters according to fast and slow time scales, and set regional capacity labels based on the regional load supply and demand balance relationship and the available load assessment results. The complementary coordination module is used to determine the coordinated scheduling area of ​​the reactive power resource shortage area based on the area capacity label, and drive reactive power flow by fine-tuning the voltage in the two areas. The process of constructing the electrical distance matrix and the voltage-reactive power sensitivity matrix is ​​as follows: Assume the electrical distance matrix is... A square matrix of order, in which, The electrical distance matrix represents the number of all nodes in the distribution network topology. The rows and columns of the electrical distance matrix correspond to the node numbers in the distribution network topology, and the elements in the electrical distance matrix represent the electrical distances between nodes. Based on the impact of reactive power injection changes at any node on the voltage of other nodes, a voltage-reactive power sensitivity matrix is ​​constructed; this matrix is ​​also... An dimensional square matrix, where the elements are the voltage-reactive power sensitivity values ​​between nodes; In the dynamic repartitioning module, the dual-channel trigger detection includes channel A1 and channel A2: channel A1 is used to monitor topology changes, including explicit topology changes and implicit topology changes; channel A2 is used to monitor slow mode drift, and the monitoring object is the boundary node after the initial region division. The process of reallocating affected boundary nodes includes: performing electrical connectivity constraint checks and regional resource constraint checks on the boundary nodes after the initial region division; recording the boundary nodes that pass the two constraint checks as affected boundary nodes; calculating the coupling affiliation degree of each affected boundary node to each of its adjacent regions; and selecting the region corresponding to the maximum coupling affiliation degree as the optimal candidate region for the affected boundary node; and performing node migration when the benefit of migrating the affected boundary node to its optimal candidate region meets a preset criterion, otherwise not performing node migration. The step of determining the coordination and scheduling area for areas with insufficient reactive power resources based on the regional capability labels includes: calculating the comprehensive selection label value of each adjacent area of ​​the area with insufficient reactive power resources based on the cross-regional coordination efficiency between the area with insufficient reactive power resources and its adjacent areas, and combining the regional capability labels of its adjacent areas; among the adjacent areas that meet the preset necessary constraints, selecting the adjacent area with the largest comprehensive selection label value as the coordination and scheduling area for cross-regional coordination. In the process of calculating the voltage amplitude correction of the voltage reference point within the region after determining the coordinated scheduling region, the coordinated scheduling region increases the voltage amplitude of its reference point; the region with insufficient reactive power resources decreases the voltage amplitude of its reference point, thereby forming a voltage gradient between the two.

2. The regional power distribution autonomous control system with distributed resource complementarity and coordination according to claim 1, characterized in that, The explicit topology change is determined by monitoring the SOE change signals of all operable switches in the distribution network; the implicit topology change is identified by the equivalent impedance of the branch where the node is located in the distribution network. When the rate of change of the equivalent impedance of the branch is greater than a preset threshold and the correlation coefficient of the voltage waveform at both ends of the branch drops sharply, the implicit topology change is confirmed to have occurred.

3. The regional power distribution autonomous control system with distributed resource complementarity and coordination according to claim 1, characterized in that, The process of slow mode drift monitored by channel A2 is as follows: For the boundary nodes after the initial region division, the deviation statistics of the boundary nodes are calculated using the voltage data within the preset observation window to construct the judgment index of the boundary nodes. The EWMA control chart is used to monitor the judgment indicators of the boundary node. When the EWMA value exceeds the preset control limit and meets the preset duration requirement, it is determined that the voltage operation mode of the boundary node has slowly drifted.

4. A regional power distribution autonomous control system with distributed resource complementarity and coordination according to claim 1, characterized in that, The calculation process for the degree of coupling affiliation is as follows: Affected boundary nodes its adjacent areas Degree of coupling between The expression is: In the formula, Indicates the affected boundary nodes to its adjacent areas The normalized result of the quartiles of the electrical distances between all nodes within the node; Indicates the affected boundary nodes The normalized result of the mean voltage-reactive power sensitivity between all nodes in its adjacent region; , These represent the preset first weight and the preset second weight, respectively. .

5. A distributed resource complementary and collaborative regional power distribution autonomous control system according to claim 1, characterized in that, The regional capacity label includes a net load satisfaction coefficient obtained based on the regional load supply and demand balance relationship, and a load dispatchability coefficient obtained based on the callable load assessment results.

6. A regional power distribution autonomous control system with distributed resource complementarity and coordination according to claim 1, characterized in that, The process of obtaining regions with insufficient reactive power resources is as follows: obtain the reactive power deficit in each region, and record the regions with reactive power deficits higher than the preset reactive power deficit threshold as regions with insufficient reactive power resources.

Images