A power distribution network voltage control method and system

Abstract

The application discloses a power distribution network voltage control method and system, and belongs to the technical field of power distribution network voltage control. The application obtains the topology operation data of the power distribution network, constructs a voltage model and a voltage sensitivity matrix set, divides the power distribution network into several sub-areas based on the voltage model and the voltage sensitivity matrix set, and decomposes the power distribution network into the sub-areas. Then, a global optimization model and a sub-area collaborative optimization model are constructed in a preset hour time scale control period and a preset minute time scale control period, respectively, so that the slow-acting equipment and the inverter are hierarchically coordinated in different time scales. Finally, voltage control is performed based on the slow-acting equipment control plan and the inverter control data in a preset second time scale control period, so that the inverter can respond to fast voltage fluctuations in a second time scale, and the slow-acting equipment adjusts the voltage baseline in a hour time scale, thereby improving the accuracy and stability of the power distribution network voltage control.

Description

Technical Field

[0001] This invention belongs to the field of distribution network voltage control technology, and particularly relates to a distribution network voltage control method and system. Background Technology

[0002] As a crucial link connecting users in the power system, the voltage control of the distribution network directly impacts power supply reliability and the safe operation of user equipment. With the large-scale integration of distributed photovoltaic (PV) systems, the power flow pattern of the distribution network has evolved from traditional unidirectional power supply to a coexistence of bidirectional power flow and localized reverse power flow. For example, during peak PV output and off-peak load periods, voltage exceeding limits at the feeder and transformer substation ends becomes increasingly prominent; in scenarios with rapid changes in cloud cover and localized shading, voltage exhibits short-term rapid fluctuations. Currently, the resources available for voltage regulation in the distribution network are becoming increasingly abundant, including inverters with reactive power regulation capabilities and a large number of slow-acting devices. Slow-acting devices refer to voltage regulation equipment in the distribution network, such as on-load tap-changing transformers, capacitor banks, and line voltage regulators, which cannot be frequently operated, have slow operating speeds, and long response times. Therefore, how to control different control resources has become the key to voltage control in the distribution network.

[0003] Current distribution network voltage control schemes are mainly divided into two categories. One category uses the local voltage reactive power droop control curve of the inverter, where a single inverter independently provides reactive power support or absorption based on the local voltage deviation. Although this scheme has a fast response speed, the large number of control resources in the distribution network leads to a lack of coordination between the various inverters, which can easily cause problems such as reactive power circulation, control conflicts, and voltage oscillations, thus affecting the accuracy and stability of distribution network voltage control. The other category involves centralized optimization by the master station or control center, which uses power flow calculations to provide control targets for different control resources. However, this method controls slow-moving equipment and inverters on the same time scale, resulting in frequent actions of slow-moving equipment or mutually canceling adjustment behaviors with the inverters, affecting the stable and accurate control of the distribution network voltage. Summary of the Invention

[0004] The present invention aims to provide a distribution network voltage control method and system to solve the above-mentioned technical problems and improve the accuracy and stability of distribution network voltage control.

[0005] To address the aforementioned technical problems, embodiments of the present invention provide a power distribution network voltage control method, comprising: Obtain the topology operation data of the distribution network, and construct the voltage model and voltage sensitivity matrix set of the distribution network based on the topology operation data; Based on the voltage model and voltage sensitivity matrix set, the distribution network is divided into several regions to obtain the distribution network. Based on the voltage model and voltage sensitivity matrix set, a global optimization model of the distribution network is constructed in a preset hourly time scale control cycle. Based on the global optimization model, the slow-motion equipment control plan of the distribution network and the zonal adjustment data of each zone are obtained. Based on the voltage sensitivity matrix set and the zone adjustment data, a zone collaborative optimization model is constructed for each zone within a preset minute time scale control cycle; based on the zone collaborative optimization model, the inverter control data for each zone is determined; Based on the slow-motion equipment control plan and inverter control data, the voltage of the distribution network is controlled within a preset second-level time scale control cycle.

[0006] Understandably, compared to existing technologies, this invention provides a precise mathematical foundation for subsequent voltage control by acquiring topology operation data of the distribution network and constructing a voltage model and voltage sensitivity matrix set. Based on this, the distribution network is divided into several zones according to the voltage model and voltage sensitivity matrix set, enabling independent coordinated control within each zone and avoiding the wide-area disturbances and computational burden caused by global unified control. Subsequently, a global optimization model and a zone-based collaborative optimization model are constructed at preset hourly and minute-based control cycles, respectively, enabling slow-moving devices and inverters to operate under different time scales. Layered coordination on a time scale avoids the frequent operation and control conflicts of slow-moving devices caused by scheduling slow-moving devices and fast inverters on the same time scale. Finally, voltage control is performed based on the control plan of slow-moving devices and inverter control data within a preset second-level time scale control cycle. This allows the inverter to respond promptly to rapid voltage fluctuations on a second-level time scale, while the slow-moving devices adjust the voltage baseline on an hourly time scale. By partitioning and coordinating the distribution network in the spatial dimension and decoupling it in the time dimension, mutual interference and control conflicts between different control resources are avoided, thereby improving the accuracy and stability of voltage control in the distribution network.

[0007] As a preferred embodiment, the step of acquiring the topology operation data of the distribution network and constructing the voltage model and voltage sensitivity matrix set of the distribution network based on the topology operation data includes: Obtain the topology operation data of the distribution network, the topology operation data including: source node voltage values ​​of the distribution network; Using the source node voltage value as a reference boundary, a vector of squared changes in node voltage of the distribution network is constructed according to a preset linearized power flow recursion algorithm; Construct the node net active power injection change vector, node net reactive power injection change vector, slow-action equipment control quantity change vector, and inverter reactive power control quantity change vector of the distribution network. Based on the node net active power injection change vector and the node voltage square change vector, a first sensitivity matrix is ​​constructed between the node voltage square change and the node net active power injection change of the distribution network. Based on the node net reactive power injection change vector and the node voltage square change vector, a second sensitivity matrix is ​​constructed between the node voltage square change and the node net reactive power injection change of the distribution network. Based on the control quantity change vector of the slow-action device and the square change vector of the node voltage, a third sensitivity matrix is ​​constructed between the square change of the node voltage and the control state change of the slow-action device in the distribution network. The voltage sensitivity matrix set of the distribution network is formed based on the first sensitivity matrix, the second sensitivity matrix, and the third sensitivity matrix; The first voltage model of the distribution network is constructed based on the voltage sensitivity matrix set, the node net active power injection change vector, the node net reactive power injection change vector, and the slow-acting equipment control quantity change vector. Based on the topology operation data, construct the node inverter mapping matrix between inverters and nodes in the distribution network; The first voltage model is transformed based on the node inverter mapping matrix and the inverter reactive power control change vector to obtain the second voltage model of the distribution network. The voltage model of the distribution network is formed based on the first voltage model and the second voltage model.

[0008] In the above scheme, the source node voltage value is used as the reference boundary. A node voltage squared change vector is constructed based on a preset linearized power flow recursion algorithm. Simultaneously, based on topology operation data, vectors for node net active power injection change, node net reactive power injection change, slow-acting device control quantity change, and inverter reactive power control quantity change are constructed respectively. Based on these, three sensitivity matrices are constructed for node voltage squared change, node net active power injection change, and node net reactive power injection change, as well as slow-acting device control state change, forming a complete set of voltage sensitivity matrices. This allows for accurate characterization of the effects of different control resources and disturbance factors on each node of the distribution network. The degree and direction of voltage influence are determined. Based on this, a first voltage model is constructed using a set of voltage sensitivity matrices, node net active power injection change vectors, node net reactive power injection change vectors, and slow-acting device control quantity change vectors. The first voltage model is then transformed using the node inverter mapping matrix and inverter reactive power control quantity change vectors to obtain a second voltage model. This ensures that the final voltage model can describe both the influence of slow-acting devices and the influence of inverters on voltage, thus providing an accurate basis for subsequent regional division and multi-time-scale voltage control, thereby improving the accuracy and stability of distribution network voltage control.

[0009] As a preferred embodiment, the distribution network is divided into several zones based on the voltage model and voltage sensitivity matrix set, including: Within each preset hourly timescale control cycle, the second sensitivity matrix is ​​transformed to obtain a fourth sensitivity matrix between the node voltage amplitude change and the node net reactive power injection change of the distribution network. Based on the node inverter mapping matrix and the second sensitivity matrix in the second voltage model, the influence intensity of the unit reactive power output change of any inverter in the distribution network on the voltage amplitude change of any node is determined, and the node inverter voltage coupling matrix is ​​constructed based on the influence intensity. The node inverter voltage coupling matrix is ​​normalized to obtain the normalized node inverter voltage coupling matrix. The similarity of the common influence between any two nodes on the same inverter is calculated based on the normalized node inverter voltage coupling matrix, and the real-time node coupling adjacency matrix of the distribution network is constructed based on the common influence similarity. Obtain the historical node coupling adjacency matrix and historical topology switch state data of the previous hourly timescale control cycle; Based on the real-time node coupling adjacency matrix, the historical node coupling adjacency matrix, and the historical topology switch status data, the distribution network is divided into several regions to obtain the distribution network.

[0010] In the above scheme, a fourth sensitivity matrix is ​​obtained by transforming the second sensitivity matrix within each preset hourly timescale control cycle. Then, based on the node inverter mapping matrix and the second sensitivity matrix, the influence intensity of the unit reactive power output change of any inverter on the voltage amplitude change of any node is determined, and a node inverter voltage coupling matrix is ​​constructed. This allows for precise quantification of the voltage regulation capability distribution of each inverter across all nodes in the network. After normalizing the node inverter voltage coupling matrix, the similarity of the common influence between any two nodes on the same inverter is calculated based on the normalized node inverter voltage coupling matrix, and a real-time node coupling adjacency matrix is ​​constructed. This accurately reflects the influence of the same inverter on the voltage regulation capability distribution between nodes. The voltage control coupling strength is influenced by the device cluster. Finally, historical node coupling adjacency matrices and historical topology switch status data are obtained. Based on the real-time node coupling adjacency matrices, historical node coupling adjacency matrices, and historical topology switch status data, the distribution network is divided into regions. This ensures that the partitioning results can reflect the electrical coupling characteristics under the current operating state and can also use historical information to compare and verify state changes. By dividing the region based on voltage control sensitivity, it is ensured that the electrical coupling between nodes within the same region is strong and the cross-region coupling is weak. This avoids inter-regional control conflicts caused by unreasonable partitioning, improves the accuracy and stability of regional coordinated control, and thus improves the accuracy and stability of distribution network voltage control.

[0011] As a preferred embodiment, the distribution network is divided into several zones based on the real-time node coupling adjacency matrix, the historical node coupling adjacency matrix, and the historical topology switch data, including: Based on the topology operation data, the real-time topology switch status data of the distribution network is determined, and based on the real-time topology switch status data and historical topology switch data, the degree of topology change of the distribution network is determined. The coupling variation degree of the distribution network is determined based on the historical node coupling adjacency matrix and the real-time node coupling adjacency matrix. When the coupling change degree is less than the first preset threshold and the topology change degree does not meet the preset topology change condition, several partitions of the previous hourly time scale control cycle are taken as several partitions of the distribution network in the current hourly time scale control cycle. When the coupling change degree is greater than or equal to the first preset threshold or the topology change degree meets the preset topology change condition, the normalized graph Laplace matrix of the distribution network is constructed based on the real-time node coupling adjacency matrix. The nodes are clustered based on a preset clustering algorithm and the normalized graph Laplace matrix to obtain several partitions of the power distribution network. Based on the topology operation data, several topology boundary nodes for each partition are determined; The cross-region control sensitivity between any node in each of the partitions and any inverter in other partitions is calculated based on the normalized node inverter voltage coupling matrix. Based on the cross-region control sensitivity, several control boundary nodes of each partition are determined. The topological boundary nodes and control boundary nodes of each partition are used as the boundary nodes of each partition to complete the regional division of the power distribution network.

[0012] In the above scheme, the degree of topology change is determined by comparing real-time topology switch status data and historical topology switch status data, and the degree of coupling change is determined by comparing historical node coupling adjacency matrices and real-time node coupling adjacency matrices. This achieves quantitative monitoring of changes in electrical coupling relationships and topology connections in the distribution network. Based on this, when the degree of coupling change is less than a first preset threshold or the degree of topology change is less than a second preset threshold, the partitioning results of the previous hourly timescale control cycle are used. This avoids repeatedly performing computationally intensive partitioning operations such as spectral clustering when the distribution network operating status has not changed significantly, reducing the computational burden on the control system. When the degree of coupling change is greater than or equal to the first preset threshold or the degree of topology change meets the preset topology... When conditions change, a normalized graph Laplace matrix is ​​constructed and nodes are clustered using a pre-defined clustering algorithm to re-divide the partitions. This allows the partitioning results to adapt promptly to changes in the distribution network topology and electrical coupling relationships. Furthermore, by determining the topological boundary nodes and control boundary nodes of each partition as boundary nodes, the partition boundaries reflect both the physical topology connections and the impact of cross-regional control sensitivity. This avoids regulation conflicts caused by strong cross-regional control influences, ensuring strong electrical coupling between nodes within the same partition and weak cross-regional coupling. This avoids inter-partition control conflicts caused by unreasonable partitioning, improves the accuracy and stability of partitioned coordinated control, and consequently improves the accuracy and stability of distribution network voltage control.

[0013] As a preferred embodiment, the global optimization model of the distribution network under a preset hourly time scale control cycle is constructed based on the voltage model and the voltage sensitivity matrix set. Based on the global optimization model, the slow-action equipment control plan of the distribution network and the zonal adjustment data of each zone are obtained, including: Within the hourly timescale control period, the net injected power prediction, network loss estimate, and distribution network node voltage target vector of the distribution network are obtained. Based on the net injected power prediction value, a node net injected power prediction change vector is constructed, and based on the topology operation data, the distribution network is constructed to obtain the slow-moving equipment state vector. Based on the slow-motion device state vector, the distribution network node voltage target vector, and the network loss estimate, a global optimization model of the distribution network is constructed for the hourly time scale control period. The first voltage model is transformed based on the predicted change vector of the node net injected power to obtain the voltage equality constraint of the global optimization model; The global optimization model is solved based on the voltage equality constraint, the preset voltage safety constraint, and the slow-action device action constraint to obtain the slow-action device control plan and the node hourly time scale voltage vector of the distribution network. The node voltage baseline vector of the distribution network is determined based on the node hourly time-scale voltage vector, and the partition voltage baseline vector of each partition is determined based on the node voltage baseline vector. The boundary node voltage offset budget for each partition is determined based on the partition voltage baseline vector and the preset voltage limit value of each partition. Obtain the reactive power output data of each inverter in each of the partitions, and determine the reactive power regulation budget of each partition based on the reactive power output data and the node inverter voltage coupling matrix; Obtain the active power output data of each inverter in each partition, and determine the active power reduction budget for each partition based on the active power output data; The partition voltage baseline vector, boundary node voltage offset budget, reactive power adjustment budget, and active power reduction budget of each partition are used as the partition adjustment data for each partition.

[0014] In the above scheme, by obtaining the predicted net injected power of the distribution network within the hourly timescale control cycle and constructing the predicted change vector of the node net injected power, the global optimization model can predict the power change trend over a future period on an hourly timescale, thereby planning the control schedule of slow-acting equipment in advance and avoiding frequent operation of slow-acting equipment due to lack of foresight. Based on this, a global optimization model is constructed using the state vector of the slow-acting equipment, the target voltage vector of the distribution network nodes, and the estimated network loss. The voltage equality constraint obtained by transforming the node net injected power prediction change vector onto the first voltage model allows the global optimization model to simultaneously consider the comprehensive optimization objectives of minimizing voltage deviation, minimizing the operation cost of slow-acting equipment, and minimizing network loss. Solving this global optimization model... After modeling, a control plan for slow-moving devices is obtained. This plan enables slow-moving devices to operate optimally on an hourly timescale, avoiding frequent triggering by second-level voltage fluctuations. Simultaneously, based on the node hourly timescale voltage vector, node voltage baseline vectors and zone voltage baseline vectors are determined. Based on these, the boundary node voltage offset budget, reactive power regulation budget, and active power reduction budget for each zone are determined. These zone regulation data provide clear spatial boundaries and resource limits for subsequent minute-level zone collaborative optimization. Through global optimization of the hourly timescale control cycle, control separation between slow-moving devices and inverters is achieved in the time dimension, thereby eliminating mutual interference between fast and slow resources on the time scale and improving the stability and accuracy of distribution network voltage control.

[0015] As a preferred embodiment, the step of constructing a zone collaborative optimization model for each zone within a preset minute timescale control cycle based on the voltage sensitivity matrix set and zone adjustment data; and determining the inverter control data for each zone based on the zone collaborative optimization model, includes: Within a minute-scale control cycle, a first partition sensitivity matrix is ​​determined based on the second sensitivity matrix to determine the node voltage amplitude and inverter reactive power regulation of each partition; a second partition sensitivity matrix is ​​determined based on the first sensitivity matrix to determine the node voltage amplitude and inverter active power reduction of each partition. Obtain the node active net injection disturbance data and node reactive net injection disturbance data of the distribution network, and construct the partition voltage disturbance vector of each partition based on the node active net injection disturbance data and node reactive net injection disturbance data. Construct the inverter reactive power adjustment vector and inverter active power reduction vector for each of the partitions. Based on the partition voltage baseline vector, partition voltage disturbance vector, first partition sensitivity matrix, second partition sensitivity matrix, inverter reactive power adjustment vector and inverter active power reduction vector, construct the partition node voltage prediction vector for each of the partitions. Based on the partition node voltage prediction vector, boundary node voltage offset budget, reactive power regulation budget, and active power reduction budget of each partition, a partition collaborative optimization model is constructed for each partition, and the inverter control data of each partition is determined based on the partition collaborative optimization model.

[0016] In the above scheme, by determining the first-zone sensitivity matrix of node voltage amplitude and inverter reactive power regulation in each zone and the second-zone sensitivity matrix of node voltage amplitude and inverter active power reduction in each zone during a minute-scale control cycle, the influence of inverter reactive power regulation and active power reduction on the node voltage within each zone can be accurately characterized. Simultaneously, by acquiring the net active power injection disturbance data and net reactive power injection disturbance data of the distribution network nodes, and combining them with the fourth and fifth sensitivity matrices to construct the zone voltage disturbance vector, the subsequent zone node voltage prediction model can fully consider... The impact of real-time disturbances such as load fluctuations and photovoltaic output deviations on voltage has improved the accuracy of voltage prediction. Based on this, a zone node voltage prediction vector is constructed using zone voltage baseline vector, zone voltage disturbance vector, first zone sensitivity matrix, second zone sensitivity matrix, inverter reactive power adjustment vector, and inverter active power reduction vector. This enables each zone to independently predict the voltage change trend within the zone on a minute-level time scale, thereby significantly improving the control response speed and accuracy. At the same time, it avoids the communication burden and computational delay problems caused by global centralized optimization, thus improving the efficiency and accuracy of distribution network voltage control.

[0017] As a preferred embodiment, the step of constructing a partitioned collaborative optimization model for each partition based on the partition node voltage prediction vector, boundary node voltage offset budget, reactive power regulation budget, and active power reduction budget for each partition, and determining the inverter control data for each partition based on the partitioned collaborative optimization model, includes: Obtain the historical inverter reactive power regulation vector of the previous minute-scale control cycle; Calculate the first difference vector between the inverter reactive power regulation vector and the historical inverter reactive power regulation vector, and calculate the second difference vector between the partition node voltage prediction vector and the partition voltage baseline vector; With the goal of minimizing the sum of the first difference vector, the second difference vector, and the inverter active power reduction vector, a partition collaborative optimization objective function is constructed for each of the partitions. Based on the boundary node voltage offset budget, reactive power regulation budget, and active power reduction budget, construct a set of partition collaborative optimization constraints for each partition; Based on the set of partitioned collaborative optimization constraints and the objective function of partitioned collaborative optimization, a partitioned collaborative optimization model is formed. The partitioned collaborative optimization model is solved to obtain the optimal reactive power adjustment vector and the optimal active power reduction vector of the inverter in each partition. Based on the inverter's optimal reactive power adjustment vector and the inverter's optimal active power reduction vector, the target active power output and target reactive power output of each inverter in each of the partitions are determined. Based on the target reactive power output and target active power output, determine the set of voltage reactive power curve parameters and voltage active power reduction curve parameters for each inverter in each of the partitions. Based on the target active power output, target reactive power output, voltage-reactive curve parameter set, and voltage-active power reduction curve parameter, and combined with the pre-acquired reactive power regulation rate of change limit and active power reduction rate of change limit for each inverter, the inverter control data for each of the aforementioned zones is determined.

[0018] In the above scheme, the historical inverter reactive power regulation vector of the previous minute-scale control cycle is obtained, and the first difference vector between the current inverter reactive power regulation vector and the historical inverter reactive power regulation vector, and the second difference vector between the partition node voltage prediction vector and the partition voltage baseline vector are calculated respectively. Based on this, a partition collaborative optimization objective function is constructed with the goal of minimizing the sum of the first difference vector, the second difference vector, and the inverter active power reduction vector. This objective function simultaneously includes three objective components: minimizing voltage deviation, smoothing reactive power regulation, and minimizing active power reduction. A partition collaborative optimization constraint set is constructed through the boundary node voltage offset budget, reactive power regulation budget, and active power reduction budget, so that the optimization result can effectively reduce the voltage deviation within the partition and ensure a smooth transition of reactive power regulation. To reduce equipment wear and tear, and to strictly control active power reduction to minimize solar power curtailment losses, the optimal reactive power regulation vector and optimal active power reduction vector of the inverter are obtained by solving the zoned collaborative optimization model. Based on this, the target active power output and target reactive power output of each inverter are determined, and the voltage-reactive power curve parameter set and voltage-active power reduction curve parameters are further determined. Combined with the pre-acquired reactive power regulation change rate limit and active power reduction change rate limit, the inverter control data is determined, enabling inverters in each zone to coordinate and allocate reactive power regulation tasks in an optimal manner on a minute-level time scale. This avoids reactive power contention and regulation conflicts between inverters in the zone. At the same time, the hard constraint of the active power reduction budget ensures that the amount of solar power curtailment is within a controllable range. Thus, while ensuring voltage compliance, control smoothness is also taken into account, thereby improving the stability and accuracy of distribution network voltage control.

[0019] As a preferred embodiment, the voltage control of the distribution network based on the slow-motion equipment control plan and inverter control data within a preset second-scale control cycle includes: Within the control cycle on the second-level time scale, the real-time voltage measurement values ​​of each inverter and the current control status of each slow-motion device are acquired; The initial reactive power output value of each inverter is determined based on the real-time voltage measurement value and the voltage reactive power curve parameter set. Obtain the current active power output value and rated capacity of each inverter, and determine the current reactive power output limit of each inverter based on the current active power output value and rated capacity; The historical reactive power output value of each inverter in the previous second-level time scale control cycle is obtained. Based on the current reactive power output limit, the historical reactive power output value and the reactive power regulation change rate limit, the initial reactive power output value is corrected to obtain the actual reactive power output value of each inverter. The initial active power output of each inverter is calculated based on the real-time voltage measurement value and the voltage active power reduction curve parameters. Obtain the historical active power output value of each inverter in the previous second-level time scale control cycle, and determine the actual active power output value of each inverter based on the active power reduction change rate limit and the historical active power output value. The average voltage of the feeder key nodes and the target voltage of the slow-action equipment of the distribution network are obtained. When the average voltage of the feeder key nodes and the target voltage of the slow-action equipment meet the preset control conditions of the slow-action equipment, the slow-action equipment of the distribution network is controlled based on the control plan. Based on the actual active power output and actual reactive power output of each inverter, the output of each inverter is controlled to complete the voltage control of the distribution network.

[0020] In the above scheme, by acquiring the real-time voltage measurement values ​​of each inverter within a control cycle on a second-level time scale, and combining this with the voltage-reactive power curve parameter set to determine the initial reactive power output value, the inverter can autonomously adjust reactive power based on the local real-time voltage. Simultaneously, by acquiring the current active power output value and rated capacity of the inverter, the current reactive power output limit is determined. Based on the current reactive power output limit, historical reactive power output values, and reactive power adjustment rate of change limit, the initial reactive power output value is corrected using ramp constraints to obtain the actual reactive power output value. This physically limits the rate of change of reactive power output, avoiding voltage oscillations caused by sudden reactive power changes. Next, the initial active power output value is calculated based on the real-time voltage measurement values ​​and voltage-active power reduction curve parameters, and based on the active power reduction rate of change limit and... Historical active power output values ​​determine actual active power output values, making the active power reduction process smooth and controllable. Regarding slow-moving device control, by acquiring the average voltage of key feeder nodes and comparing it with the target voltage of slow-moving devices, control is only applied to slow-moving devices based on the control plan when preset control conditions are met. This ensures that slow-moving devices do not respond to second-level voltage fluctuations, enabling inverters to safely and smoothly complete rapid voltage regulation on a second-level timescale. Slow-moving devices only operate under slow-moving device control conditions, achieving spatial zonal coordination and temporal layered decoupling of the distribution network. This avoids mutual interference and control conflicts between different control resources, thereby improving the accuracy and stability of distribution network voltage control.

[0021] As a preferred embodiment, after voltage control of the distribution network, the following is also included: Within the preset evaluation window, obtain the voltage measurement value after control of each node in the distribution network, the available active power after control of each inverter, and the actual number of actions of each slow-action device. The voltage over-limit index of the distribution network is calculated based on the voltage measurement value after control. The slow-moving equipment index of the power distribution network is calculated based on the actual number of actions. Based on the controlled active power output value and the actual active power output value, the curtailment index of the distribution network can be determined. The boundary exceedance index of the distribution network is determined based on the boundary node voltage offset budget, the zone voltage baseline vector, and the voltage measurement after control. The inverter response deviation index of the distribution network is determined based on the actual reactive power output, actual active power output, target reactive power output, and target active power output of each inverter. A comprehensive evaluation index is generated based on the voltage over-limit index, slow-motion equipment index, curtailment index, boundary over-limit index, and inverter response deviation index. The distribution network voltage control evaluation results are generated based on the comprehensive evaluation indicators, and the distribution network voltage control is adjusted for the next time based on the voltage over-limit indicators, slow-action equipment indicators, curtailment indicators, boundary over-limit indicators, and inverter response deviation indicators.

[0022] In the above scheme, by acquiring the measured voltage values ​​of each node in the distribution network after control, the available active power of each inverter after control, and the actual number of actions of each slow-acting device within a preset evaluation window, a comprehensive measured data foundation is provided for closed-loop evaluation. Based on this, voltage over-limit index, slow-acting device index, curtailment index, boundary over-limit index, and inverter response deviation index are calculated respectively. These five indicators comprehensively reflect the distribution network voltage control effect from different dimensions. Then, a comprehensive evaluation index is generated based on the above five indicators, which can quantitatively evaluate the overall effect of the current voltage control and avoid the one-sidedness of single-indicator evaluation. Furthermore, it can be adjusted accordingly in the next distribution network voltage control, so that the distribution network voltage control forms a complete control closed loop of execution, evaluation, feedback, and adjustment. This allows the distribution network voltage control to be continuously optimized according to the actual operating effect, thereby continuously improving the accuracy and stability of distribution network voltage control in long-term operation.

[0023] Accordingly, embodiments of the present invention provide a distribution network voltage control system, including: a distribution network voltage modeling module, a distribution network partitioning module, an hourly time scale control module, a minute time scale control module, and a second-level time scale control module; The distribution network voltage modeling module is used to acquire the topology operation data of the distribution network, and to construct the voltage model and voltage sensitivity matrix set of the distribution network based on the topology operation data; The distribution network partitioning module is used to divide the distribution network into regions based on the voltage model and the voltage sensitivity matrix set, thereby obtaining several partitions of the distribution network. The hourly time-scale control module is used to construct a global optimization model of the distribution network in a preset hourly time-scale control cycle based on the voltage model and the voltage sensitivity matrix set, and to obtain the slow-motion equipment control plan of the distribution network and the zonal adjustment data of each zone based on the global optimization model. The minute-scale control module is used to construct a zone collaborative optimization model for each zone within a preset minute-scale control cycle based on the voltage sensitivity matrix set and zone adjustment data; and to determine the inverter control data for each zone based on the zone collaborative optimization model. The second-level timescale control module is used to perform voltage control on the distribution network within a preset second-level timescale control cycle based on the slow-motion equipment control plan and inverter control data.

[0024] Understandably, compared to existing technologies, this system acquires the topology operation data of the distribution network and constructs a voltage model and voltage sensitivity matrix set, providing a precise mathematical foundation for subsequent voltage control. Based on this, the distribution network is divided into several zones according to the voltage model and voltage sensitivity matrix set, enabling independent coordinated control within each zone and avoiding the wide-area disturbances and computational burden caused by global unified control. Subsequently, a global optimization model and a zone-based collaborative optimization model are constructed at preset hourly and minute-based control cycles, respectively, enabling slow-moving devices and inverters to operate under different time scales. Layered coordination on a time scale avoids the frequent operation and control conflicts of slow-moving devices caused by scheduling slow-moving devices and fast inverters on the same time scale. Finally, voltage control is performed based on the control plan of slow-moving devices and inverter control data within a preset second-level time scale control cycle. This allows the inverter to respond promptly to rapid voltage fluctuations on a second-level time scale, while the slow-moving devices adjust the voltage baseline on an hourly time scale. By partitioning and coordinating the distribution network in the spatial dimension and decoupling it in the time dimension, mutual interference and control conflicts between different control resources are avoided, thereby improving the accuracy and stability of voltage control in the distribution network. Attached Figure Description

[0025] Figure 1 A flowchart illustrating the steps of a power distribution network voltage control method provided in an embodiment of the present invention; Figure 2 This is a schematic diagram of a power distribution network voltage control system provided in an embodiment of the present invention. Detailed Implementation

[0026] The technical solutions of the embodiments of the present invention will be clearly and completely described below with reference to the accompanying drawings. Obviously, the described embodiments are only some embodiments of the present invention, and not all embodiments. Based on the embodiments of the present invention, all other embodiments obtained by those skilled in the art without creative effort are within the scope of protection of the present invention.

[0027] Example 1 Please refer to Figure 1 , Figure 1 The flowchart of a power distribution network voltage control method provided in this embodiment of the invention includes steps S101 to S105.

[0028] Step S101: Obtain the topology operation data of the distribution network, and construct the voltage model and voltage sensitivity matrix set of the distribution network based on the topology operation data.

[0029] Step S102: Based on the voltage model and voltage sensitivity matrix set, the distribution network is divided into regions to obtain several partitions of the distribution network.

[0030] Step S103: Based on the voltage model and voltage sensitivity matrix set, construct a global optimization model of the distribution network in a preset hourly time scale control cycle, and obtain the slow-motion equipment control plan of the distribution network and the zonal adjustment data of each zone based on the global optimization model.

[0031] Step S104: Construct a zone collaborative optimization model for each zone within a preset minute time scale control cycle based on the voltage sensitivity matrix set and zone adjustment data; determine the inverter control data for each zone based on the zone collaborative optimization model.

[0032] Step S105: Based on the slow-motion equipment control plan and inverter control data, perform voltage control on the distribution network within a preset second-level time scale control cycle.

[0033] This embodiment acquires the topology operation data of the distribution network and constructs a voltage model and a voltage sensitivity matrix set, providing a precise mathematical foundation for subsequent voltage control. Based on this, the distribution network is divided into several zones according to the voltage model and voltage sensitivity matrix set, enabling independent coordinated control within each zone and avoiding the wide-area disturbances and computational burden caused by global unified control. Subsequently, a global optimization model and a zone-based collaborative optimization model are constructed at preset hourly and minute-scale control periods, respectively, realizing the coordinated control of slow-moving devices and inverters at different time scales. Layered coordination avoids the frequent operation and control conflicts of slow-moving devices caused by scheduling slow-moving devices and fast inverters on the same time scale. Finally, voltage control is performed based on the control plan of slow-moving devices and inverter control data on a preset second-level time scale control cycle. This allows the inverter to respond promptly to rapid voltage fluctuations on a second-level time scale, while the slow-moving devices adjust the voltage baseline on an hour-level time scale. Through the spatial partitioning coordination and time-level layered decoupling of the distribution network, mutual interference and control conflicts between different control resources are avoided, thereby improving the accuracy and stability of distribution network voltage control.

[0034] In this embodiment, acquiring the topology operation data of the distribution network and constructing the voltage model and voltage sensitivity matrix set of the distribution network based on the topology operation data includes: Obtain the topology operation data of the distribution network, the topology operation data including: source node voltage values ​​of the distribution network; Using the source node voltage value as a reference boundary, a vector of squared changes in node voltage of the distribution network is constructed according to a preset linearized power flow recursion algorithm; Construct the node net active power injection change vector, node net reactive power injection change vector, slow-action equipment control quantity change vector, and inverter reactive power control quantity change vector of the distribution network. Based on the node net active power injection change vector and the node voltage square change vector, a first sensitivity matrix is ​​constructed between the node voltage square change and the node net active power injection change of the distribution network. Based on the node net reactive power injection change vector and the node voltage square change vector, a second sensitivity matrix is ​​constructed between the node voltage square change and the node net reactive power injection change of the distribution network. Based on the control quantity change vector of the slow-action device and the square change vector of the node voltage, a third sensitivity matrix is ​​constructed between the square change of the node voltage and the control state change of the slow-action device in the distribution network. The voltage sensitivity matrix set of the distribution network is formed based on the first sensitivity matrix, the second sensitivity matrix, and the third sensitivity matrix; The first voltage model of the distribution network is constructed based on the voltage sensitivity matrix set, the node net active power injection change vector, the node net reactive power injection change vector, and the slow-acting equipment control quantity change vector. Based on the topology operation data, construct the node inverter mapping matrix between inverters and nodes in the distribution network; The first voltage model is transformed based on the node inverter mapping matrix and the inverter reactive power control change vector to obtain the second voltage model of the distribution network. The voltage model of the distribution network is formed based on the first voltage model and the second voltage model.

[0035] In one optional embodiment, the topology operation data of the distribution network includes a set of nodes, a set of branches, and source node voltage values, as well as the resistance and reactance of each branch in the branch set; wherein, the set of nodes is defined as... ,in, This indicates the number of nodes. Node 0 represents the source node, which specifically refers to the feeder head node, the low-voltage side bus node of the substation, or the low-voltage side bus node of the transformer in the distribution area. Its physical meaning is the upstream power supply node that supplies power to the feeder or distribution area to be controlled. The remaining nodes, excluding node 0, constitute the set of non-source nodes. The number of non-source nodes is denoted as Define the set of branches as follows: ,in and These represent the nodes at both ends of the branch. This indicates the reference direction of the branch; therefore, the resistance and reactance of any branch are expressed as follows: and .

[0036] Furthermore, the topology operation data also includes the inverter set. and slow motion equipment as ;in Indicates the first One inverter, This refers to the number of inverters; Indicates the first A slow-motion device, The quantity of slow-motion equipment includes voltage regulation equipment such as on-load tap changer transformers, line voltage regulators, and capacitor banks. (This refers to the number of slow-motion devices.) Define its control state as When slow motion devices When it is an on-load tap-changing transformer, Indicates the tap position; when the slow-motion device... When it is a capacitor bank, Indicates the switching gear or switching status; when the slow-motion device... When used as a line voltage regulator, This indicates the voltage adjustment level or voltage compensation setting. Data sources include the current tap position, switching status, and adjustable range transmitted from the power distribution master station, voltage regulator controller, capacitor switching controller, or field terminal.

[0037] Therefore, in this embodiment, the source node voltage value is defined as node 0 during the control cycle. The per-unit voltage at the source is The specific sources include: when a voltage measuring device is installed at the feeder end, it can be directly acquired from SCADA (Supervisory Control and Data Acquisition) systems or distribution automation terminals commonly used in power grids; when the source-end voltage is controlled by an on-load tap-changing transformer, it is calculated from the substation bus voltage and transformer tap position; the square of the source-end node voltage is expressed as... Therefore, this embodiment uses the source node voltage value as the reference boundary and employs a linearized power flow recursion algorithm to construct the node voltage squared change vector of the distribution network; specifically, it defines the branch... Let its active power flow and reactive power flow be respectively and Define the square of the node voltage as , For nodes The voltage amplitude per unit value; next, this embodiment uses the LinDistFlow recursive formula (Linearized Distribution Flow) that ignores second-order loss terms to establish a baseline relationship between the square of the node voltage and the branch power flow, that is... ;in, , , and They are nodes and nodes The voltage amplitude per unit value. When When, in the above formula That is, the square of the source-end node voltage. .

[0038] The result of power flow calculation using the LinDistFlow recursive formula is used as the linearization reference point. The linearization reference point includes the nodes under the linearization reference point. voltage squared reference value Therefore, the vector representing the squared change in node voltage is: ; .

[0039] The topology operation data of the distribution network also includes: the load active power, load reactive power, distributed resource active power injection, and distributed resource reactive power injection for each node; the linearization reference point also includes: the nodes under the linearization reference point. Net active power injection baseline value and net reactive power injection reference value For nodes Its net active power injection and net reactive power injection are respectively and The specific calculation method is as follows: (1); (2); in, and Representing nodes respectively During the control cycle The active and reactive power of the load are derived from load measurement data from SCADA or distribution automation terminals. and Representing nodes respectively During the control cycle The distributed resource active power injection and distributed resource reactive power injection are derived from data transmitted from distributed power source grid connection monitoring or inverters. In particular, net active power injection and net reactive power injection are defined as "positive for injection into the distribution network and negative for absorption from the distribution network".

[0040] Therefore, the node net active power injection change vector can be constructed as shown in formula (3), and the node net reactive power injection change vector can be constructed as shown in formula (4), which satisfy the following conditions: , .

[0041] (3); (4); The control states of the slow-motion device have been described above; therefore, the control state vector of the slow-motion device is defined as follows: ,set up Show the first The change in the control quantity of each slow-motion device relative to the current control state; therefore, the change vector of the control quantity of the slow-motion device is: When the slow-motion equipment is an on-load tap-changing transformer, This indicates the change in tap position; when the slow-motion device is a capacitor bank. This indicates the switching position or the change in switching capacity; when the slow-acting equipment is a line voltage regulator... This indicates the amount of change in the voltage compensation setting.

[0042] The topology operation data of the distribution network also includes: the reactive power output value of each inverter. The linearization reference point also includes: the inverter at the linearization reference point. reactive power output reference value Therefore, setting Indicates inverter During the control cycle reactive power output value Inverter relative to linearized reference point reactive power output reference value The change in reactive power output, i.e. Positive values ​​indicate that the inverter injects reactive power into the distribution network, while negative values ​​indicate that the inverter absorbs reactive power from the distribution network. Therefore, the reactive power control variable change vector of the inverter is constructed as follows: .

[0043] After obtaining the node voltage squared change vector, node net active power injection change vector, node net reactive power injection change vector, slow-action device control quantity change vector, and inverter reactive power control quantity change vector, this embodiment calculates the first sensitivity matrix, second sensitivity matrix, and third sensitivity matrix using the LinDistFlow first-order linearized sensitivity algorithm, the AC power flow Jacobian matrix inversion algorithm, or the disturbance trial calculation algorithm. Specifically, the first sensitivity matrix is ​​used to represent the relationship between the node voltage squared change and the node net active power injection change, and its third sensitivity matrix is ​​used to represent the relationship between the node voltage squared change and the node net active power injection change. Each element represents a node. The change in net active power injection per unit affects the node The linear approximation of the effect of the square of the voltage is expressed in the following form: The second sensitivity matrix is ​​used to represent the relationship between the squared change in node voltage and the change in net reactive power injection at the node. Each element represents a node. The change in unit net reactive power injection on the node The linear approximation of the effect of the square of the voltage is expressed in the following form: The third sensitivity matrix is ​​used to represent the relationship between the squared change of the node voltage and the change in the control state of the slow-moving device. The element represents the first element. The change in the unit control state of a slow-motion device affects the node. The linear approximation of the effect of the square of the voltage is expressed in the following form: Therefore, the first sensitivity matrix, the second sensitivity matrix, and the third sensitivity matrix form the voltage sensitivity matrix set of the distribution network.

[0044] Simultaneously, considering the influence of the source node voltage value, this embodiment also constructs a source-end sensitivity vector for the effect of the source node voltage value change on the square of the voltage of each non-source node, which is expressed as: Therefore, the first voltage model is expressed as shown in formula (5). (5); Specifically, if the source node voltage remains constant during the control cycle, then Therefore, the first voltage model can be simplified to: ; Based on this first voltage model, the topology operation data also includes the connection relationship between inverters and nodes; therefore, this embodiment constructs a node-inverter mapping matrix to represent the connection relationship between inverters and nodes; the node-inverter mapping matrix is ​​expressed as: ; Indicates inverter With nodes The connection relationship; if the inverter Access Node ,but If the inverter Unconnected nodes ,but .

[0045] The change in node reactive power injection caused by inverter reactive power control is expressed as: Substituting this into the first voltage model shown in formula (5), we can transform it to obtain the second voltage model, as shown in formula (6). (6); Therefore, the voltage model of the distribution network is shown in formulas (5) and (6).

[0046] In the above embodiments, using the source node voltage value as the reference boundary, a node voltage squared change vector is constructed according to a preset linearized power flow recursion algorithm. Simultaneously, based on topology operation data, vectors for node net active power injection change, node net reactive power injection change, slow-acting device control quantity change, and inverter reactive power control quantity change are constructed respectively. Based on these, three sensitivity matrices are constructed for node voltage squared change, node net active power injection change, and node net reactive power injection change, as well as slow-acting device control state change, forming a complete set of voltage sensitivity matrices. This allows for accurate characterization of the effects of different control resources and disturbance factors on each node of the distribution network. The degree and direction of voltage influence are determined. Based on this, a first voltage model is constructed using a set of voltage sensitivity matrices, node net active power injection change vectors, node net reactive power injection change vectors, and slow-acting device control quantity change vectors. The first voltage model is then transformed using the node inverter mapping matrix and inverter reactive power control quantity change vectors to obtain a second voltage model. This ensures that the final voltage model can describe both the influence of slow-acting devices and the influence of inverters on voltage, thus providing an accurate basis for subsequent regional division and multi-time-scale voltage control, thereby improving the accuracy and stability of distribution network voltage control.

[0047] In this embodiment, the distribution network is divided into several zones based on the voltage model and voltage sensitivity matrix set, including: Within each preset hourly timescale control cycle, the second sensitivity matrix is ​​transformed to obtain a fourth sensitivity matrix between the node voltage amplitude change and the node net reactive power injection change of the distribution network. Based on the node inverter mapping matrix and the second sensitivity matrix in the second voltage model, the influence intensity of the unit reactive power output change of any inverter in the distribution network on the voltage amplitude change of any node is determined, and the node inverter voltage coupling matrix is ​​constructed based on the influence intensity. The node inverter voltage coupling matrix is ​​normalized to obtain the normalized node inverter voltage coupling matrix. The similarity of the common influence between any two nodes on the same inverter is calculated based on the normalized node inverter voltage coupling matrix, and the real-time node coupling adjacency matrix of the distribution network is constructed based on the common influence similarity. Obtain the historical node coupling adjacency matrix and historical topology switch state data of the previous hourly timescale control cycle; Based on the real-time node coupling adjacency matrix, the historical node coupling adjacency matrix, and the historical topology switch status data, the distribution network is divided into several regions to obtain the distribution network.

[0048] In the above embodiments, a fourth sensitivity matrix is ​​obtained by transforming the second sensitivity matrix within each preset hourly timescale control cycle. Then, based on the node inverter mapping matrix and the second sensitivity matrix, the influence intensity of the unit reactive power output change of any inverter on the voltage amplitude change of any node is determined, and a node inverter voltage coupling matrix is ​​constructed. This allows for precise quantification of the voltage regulation capability distribution of each inverter across all nodes in the network. After normalizing the node inverter voltage coupling matrix, the similarity of the common influence between any two nodes on the same inverter is calculated based on the normalized node inverter voltage coupling matrix, and a real-time node coupling adjacency matrix is ​​constructed. This accurately reflects the influence between nodes on the same inverter. The voltage control coupling strength caused by the influence of transformer groups is analyzed. Finally, historical node coupling adjacency matrices and historical topology switch status data are obtained. Based on the real-time node coupling adjacency matrices, historical node coupling adjacency matrices, and historical topology switch status data, the distribution network is divided into regions. This ensures that the partitioning results can reflect the electrical coupling characteristics under the current operating state and can also use historical information to compare and verify state changes. By dividing the region based on voltage control sensitivity, strong electrical coupling between nodes within the same region and weak cross-region coupling are ensured. This avoids inter-regional control conflicts caused by unreasonable partitioning, improves the accuracy and stability of regional coordinated control, and thus improves the accuracy and stability of distribution network voltage control.

[0049] In this embodiment, the distribution network is divided into several partitions based on the real-time node coupling adjacency matrix, the historical node coupling adjacency matrix, and historical topology switch data, including: Based on the topology operation data, the real-time topology switch status data of the distribution network is determined, and based on the real-time topology switch status data and historical topology switch data, the degree of topology change of the distribution network is determined. The coupling variation degree of the distribution network is determined based on the historical node coupling adjacency matrix and the real-time node coupling adjacency matrix. When the coupling change degree is less than the first preset threshold and the topology change degree does not meet the preset topology change condition, several partitions of the previous hourly time scale control cycle are taken as several partitions of the distribution network in the current hourly time scale control cycle. When the coupling change degree is greater than or equal to the first preset threshold or the topology change degree meets the preset topology change condition, the normalized graph Laplace matrix of the distribution network is constructed based on the real-time node coupling adjacency matrix. The nodes are clustered based on a preset clustering algorithm and the normalized graph Laplace matrix to obtain several partitions of the power distribution network. Based on the topology operation data, several topology boundary nodes for each partition are determined; The cross-region control sensitivity between any node in each of the partitions and any inverter in other partitions is calculated based on the normalized node inverter voltage coupling matrix. Based on the cross-region control sensitivity, several control boundary nodes of each partition are determined. The topological boundary nodes and control boundary nodes of each partition are used as the boundary nodes of each partition to complete the regional division of the power distribution network.

[0050] In the above embodiments, the degree of topology change is determined by comparing real-time topology switch status data and historical topology switch status data, and the degree of coupling change is determined by comparing historical node coupling adjacency matrices and real-time node coupling adjacency matrices. This achieves quantitative monitoring of changes in electrical coupling relationships and topology connections in the distribution network. Based on this, when the degree of coupling change is less than a first preset threshold or the degree of topology change is less than a second preset threshold, the partitioning results of the previous hourly timescale control cycle are used. This avoids repeatedly performing computationally intensive partitioning operations such as spectral clustering when the distribution network operating status has not changed significantly, reducing the computational burden on the control system. When the degree of coupling change is greater than or equal to the first preset threshold or the degree of topology change meets the preset topology... When conditions change, a normalized graph Laplace matrix is ​​constructed and nodes are clustered using a pre-defined clustering algorithm to re-divide the partitions. This allows the partitioning results to adapt promptly to changes in the distribution network topology and electrical coupling relationships. Furthermore, by determining the topological boundary nodes and control boundary nodes of each partition as boundary nodes, the partition boundaries reflect both the physical topology connections and the impact of cross-regional control sensitivity. This avoids regulation conflicts caused by strong cross-regional control influences, ensuring strong electrical coupling between nodes within the same partition and weak cross-regional coupling. This avoids inter-partition control conflicts caused by unreasonable partitioning, improves the accuracy and stability of partitioned coordinated control, and consequently improves the accuracy and stability of distribution network voltage control.

[0051] In an optional embodiment, this embodiment describes the specific implementation of step S102; the hourly time scale control period is defined as: , This represents an optimized cross-section within a control period on an hourly timescale. To optimize the number of optimized sections within the window; Within each preset hourly timescale control cycle, the voltage square sensitivity is converted into voltage amplitude sensitivity. Let... For linearized reference points, the nodes Given the voltage amplitude reference value, the fourth sensitivity matrix is ​​expressed as: ; This indicates that a diagonal matrix is ​​formed using the elements within the brackets as diagonal elements. Next, based on the node inverter mapping matrix and the second sensitivity matrix in the second voltage model, the influence intensity of the unit reactive power output change of any inverter in the distribution network on the voltage amplitude change of any node is determined, specifically as follows: Therefore, the node inverter voltage coupling matrix can be expressed as: ,Right now The The elements are denoted as The larger this value, the better the inverter. For nodes The stronger the voltage regulation effect; To avoid coupling comparison deviations caused by different inverter capacities or different node dimensions, the node inverter voltage coupling matrix is ​​adjusted. After normalization, the normalized node inverter voltage coupling matrix is ​​obtained: in: ; To prevent extremely small positive numbers with a denominator of zero, After obtaining the normalized node inverter voltage coupling matrix, the similarity of the common influence between any two nodes on the same inverter is calculated, and the real-time node coupling adjacency matrix of the distribution network is constructed based on the similarity of the common influence. Specifically, an initial node coupling adjacency matrix is ​​first defined. Its satisfaction , Represents a node and nodes The degree of similarity between them due to the common influence of the same inverter; Larger values ​​indicate a larger node and nodes All nodes are strongly influenced by the same inverter; subsequently, the initial node is coupled with an adjacency matrix. Cosine normalization is performed to obtain the real-time node coupling adjacency matrix. Its satisfaction ; Real-time node coupling adjacency matrix For a symmetric matrix, its first... element Represents a node With nodes The normalized coupling strength in voltage control typically ranges from [value range missing]. . The larger the value, the more suitable the two nodes are to be assigned to the same control partition; The smaller the value, the weaker the control coupling between the two nodes.

[0052] To adapt to changes in topology and operating status, this embodiment obtains the historical node coupling adjacency matrix of the previous hourly timescale control cycle. and historical topology switch status data Then calculate the degree of topological change and the degree of coupling change; The degree of topological change is expressed as: ; The degree of topological change; This refers to a set of topology switches that can change the energized topology of feeders or the connectivity of nodes, including feeder outlet circuit breakers, sectionalizing switches, tie switches, and ring main unit switches that can change the power supply path, but excluding capacitor switching switches that only change the reactive power compensation capacity. This is real-time topology switch status data; it is set to 1 when closed and 0 when open. This is historical topology switch status data; The degree of coupling change is expressed as: ;in, The coupling change degree represents the degree of change of the real-time node coupling adjacency matrix of the hourly timescale control period relative to the historical node coupling adjacency matrix of the previous hourly timescale control period. This represents the Frobenius norm.

[0053] Set the first preset threshold as The preset topology change conditions are: And the duration is not less than 30 seconds; when the coupling change degree is less than the first preset threshold and the topology change degree does not meet the preset topology change condition, several partitions of the previous hourly time scale control cycle are taken as several partitions of the distribution network in the current hourly time scale control cycle. When the coupling change degree is greater than or equal to the first preset threshold, it indicates that the voltage control coupling relationship has changed significantly, requiring re-partitioning and boundary node updates; or when the topology change degree meets the preset topology change condition, it indicates that a topology switch that changes the power supply path or node connectivity has undergone a confirmed change from closed to open or from open to closed, requiring re-partitioning and boundary node updates; at this time, based on the real-time node coupling adjacency matrix... Construct degree matrix , (7); Then through the degree matrix Construct the normalized graph Laplacian matrix: (8); in, The normalized graph Laplacian matrix; It is the identity matrix; It is the square root inverse of the degree matrix; Then, the normalized spectral clustering algorithm was used to... Perform eigenvalue decomposition and take the first few features. The eigenvectors corresponding to the smallest eigenvalues ​​form the embedding matrix, and the row vectors of the embedding matrix are then processed. -means clustering yields several partitions of the power distribution network: ,in, Indicates the first One partition, Indicates the number of partitions.

[0054] It should be noted that the zoning of the distribution network should meet the following requirements. and , That is, each non-source node is assigned to only one partition; if a partition does not contain a controllable inverter or does not meet the physical connectivity requirements after clustering, the partition is merged with the adjacent partition with the strongest node coupling, or assigned to the adjacent partition with the closest physical connection, so as to ensure that subsequent partition control is actually executable.

[0055] Specifically, the number of partitions It can be pre-set by maintenance personnel based on the feeder size and communication architecture, or it can be determined based on the eigenvalue gaps of the normalized graph Laplace matrix. If determined based on the eigenvalue gaps of the normalized graph Laplace matrix, it is expressed as follows: ; for The The feature values ​​are arranged in ascending order; Maximum number of partitions allowed Based on this, several topology boundary nodes for each partition are determined using the aforementioned topology operation data. For each partition... First, determine the topology boundary nodes based on the topology operation data: (9); in, For partitioning The set of topological boundary nodes; Indicates a branch During the control cycle It is in a closed or electrically connected state; that is, if the nodes within the partition... A node that is directly connected to other partitions via a closed branch will then... It was identified as a topological boundary node.

[0056] Inverter The grid connection node is ,like Then the inverter Belongs to the partition and define partitions The internal inverter set is as follows: For nodes Define its partition Cross-regional control sensitivity for: (10); in, Represents a node With partitions Sensitivity to cross-regional control between internal inverters.

[0057] Therefore, the node is calculated. Maximum cross-region control sensitivity: ; when At that time, the node It is identified as a control boundary node; therefore, a control boundary node is represented as follows: ;in, For partitioning The set of control boundary nodes.

[0058] Finally, the topology boundary nodes and control boundary nodes of each partition are used as the boundary nodes of each partition: ;in, For partitioning The set of boundary nodes.

[0059] In this embodiment, the construction of a global optimization model of the distribution network within a preset hourly timescale control cycle based on the voltage model and voltage sensitivity matrix set, and the obtaining of the slow-motion equipment control plan and the zonal adjustment data of each zone based on the global optimization model, includes: Within the hourly timescale control period, the net injected power prediction, network loss estimate, and distribution network node voltage target vector of the distribution network are obtained. Based on the net injected power prediction value, a node net injected power prediction change vector is constructed, and based on the topology operation data, the distribution network is constructed to obtain the slow-moving equipment state vector. Based on the slow-motion device state vector, the distribution network node voltage target vector, and the network loss estimate, a global optimization model of the distribution network is constructed for the hourly time scale control period. The first voltage model is transformed based on the predicted change vector of the node net injected power to obtain the voltage equality constraint of the global optimization model; The global optimization model is solved based on the voltage equality constraint, the preset voltage safety constraint, and the slow-action device action constraint to obtain the slow-action device control plan and the node hourly time scale voltage vector of the distribution network. The node voltage baseline vector of the distribution network is determined based on the node hourly time-scale voltage vector, and the partition voltage baseline vector of each partition is determined based on the node voltage baseline vector. The boundary node voltage offset budget for each partition is determined based on the partition voltage baseline vector and the preset voltage limit value of each partition. Obtain the reactive power output data of each inverter in each of the partitions, and determine the reactive power regulation budget of each partition based on the reactive power output data and the node inverter voltage coupling matrix; Obtain the active power output data of each inverter in each partition, and determine the active power reduction budget for each partition based on the active power output data; The partition voltage baseline vector, boundary node voltage offset budget, reactive power adjustment budget, and active power reduction budget of each partition are used as the partition adjustment data for each partition.

[0060] In one optional embodiment, this embodiment describes the specific implementation process of step S103. Obtain the predicted net injected power, estimated network loss, and target voltage vector of the distribution network nodes.

[0061] Control cycle on an hourly time scale In the process of optimizing the cross-section This involves obtaining the predicted net injected power, estimated network losses, and target voltage vectors for the distribution network nodes. The predicted net injected power includes the predicted net injected active power. Net injected reactive power prediction value The estimated network loss is expressed as: The predicted net injected power and estimated network loss are conventional techniques used in the power grid field, and will not be specifically described in this embodiment. Distribution network node voltage target vector. Typically, 1.00 pu can be used, or the value can be set within the range of 0.98 pu to 1.02 pu depending on the operating strategy.

[0062] Therefore, the predicted change vector of net injected power at nodes includes the predicted change vector of net injected active power at nodes. And the predicted change vector of net reactive power injected at nodes Its satisfaction ; ; The net active power injection reference vector is formed by the net active power injection reference values ​​of the nodes under the linearized reference point. The reference vector for the net reactive power injection at the node under the linearized reference point; The slow-motion device state vector has already been defined above as follows: Therefore, the global optimization model is expressed as shown in formula (11). (11); in, The objective function of the global optimization model; This represents the hourly time-scale voltage vector of the node to be solved; This is a node voltage deviation weight matrix, used to penalize each node voltage deviation from the target value with different degrees of severity. Its specific value can be set by technicians. The motion cost weight for slow-motion devices is used to suppress frequent actions of slow-motion devices. The control state vector of the slow-motion device for the previous hourly timescale control cycle; The L1 norm of the control state change of the slow-motion device is used to penalize the amplitude of the device's movement. For network loss weight; The cycle length is controlled on an hourly time scale.

[0063] The first voltage model is transformed based on the predicted change vector of the node net injected power, and the voltage equality constraint is expressed as shown in formula (12). (12); in, Indicates optimization of cross-section The node voltage square vector is determined by both the slow-motion equipment control plan and the net injected power prediction. This represents the square vector of the nodal voltages at the linearized reference point in the optimized profile. The value of ; , , and The first sensitivity matrix, second sensitivity matrix, third sensitivity matrix, and source-end sensitivity vector are respectively located in the optimization profile. The value of ; This refers to the change in control state of slow-motion equipment. For the source node voltage value in the optimized section The change in quantity.

[0064] Voltage safety constraints are set as follows: (13); and satisfy ; in, To optimize the cross-section Next node The voltage amplitude; To optimize the cross-section Next node The square of the voltage; and These are the lower and upper limits of the node voltage, respectively, which are derived from the operating procedures. To provide a voltage safety margin, in this embodiment, we can take... ; The motion constraints for the slow-motion device are set as shown in formulas (14) to (16). (14); (15); (16); This represents the feasible control set for slow-motion devices; Indicates the first A slow-motion device is used to optimize the cross-section. The number of times the action is performed; Indicates the maximum number of allowed actions; Indicates the distance to the first The time it takes for a slow-motion device to perform one action; This indicates the minimum dwell time of the slow-motion device; Solving the global optimization model yields the control plan for slow-moving equipment in the distribution network. and node hour-scale voltage vector ; Transform the node hour-scale voltage vector Defined as the node voltage baseline vector of the distribution network Node voltage baseline vector Indicates pressing on a slow-motion device When the load and photovoltaics are operating at their predicted average values ​​on an hourly timescale, the reference voltage level that each node in the distribution network should reach. Based on the partition set obtained above Construct a node selection matrix for each partition. , used from node voltage baseline vector Extracting from partitions partition voltage baseline vector , which is represented as ; partition voltage baseline vector It contains the voltage baseline values ​​of each node within the partition; For boundary nodes Its voltage upper margin and lower margin are respectively: and ;in, Boundary nodes Voltage margin; Boundary nodes Voltage margin; Boundary nodes The voltage baseline value.

[0065] Therefore, partitioning The offset budget above the boundary voltage and the offset budget below the boundary voltage are respectively and ;in, For partitioning Offset budget on boundary voltage; For partitioning Offset budget under boundary voltage; This is the boundary margin reduction factor, used to reserve a safety margin for measurement errors, prediction errors, and disturbances in adjacent zones. In this embodiment, it is taken as... ; partition The boundary voltage upper offset budget and the boundary voltage lower offset budget constitute the boundary node voltage offset budget of the partition.

[0066] The partitioning has been described above. The internal inverter assembly is Therefore, inverter In optimizing the cross section The available margins for reactive power injection and reactive power absorption are respectively and ;in, For inverter Available margin for reactive power injection; For inverter The available margin for reactive power absorption; and These are the upper and lower limits of reactive power output allowed by the inverter, respectively, derived from the equipment nameplate or grid connection protocol; The planned reactive power for hourly time-scale control cycles originates from inverter operation measurements or scheduling plans.

[0067] This yields the partitioned physical reactive power budget, which is expressed as: and ;in, For partitioning Physical reactive power injection budget; For partitioning The physical reactive power absorption budget; to avoid boundary node voltage exceeding limits due to reactive power regulation within the region, the boundary constraint reactive power injection budget under boundary constraints is calculated through the node inverter voltage coupling matrix. and boundary constraint reactive power absorption budget Specifically, it is expressed as and Finally, the reactive power adjustment budget is expressed as the reactive power injection budget. and reactive power absorption budget Its satisfaction and ; The effective budget reductions are: ;in, For partitioning The successful budget cuts; For inverter In optimizing the cross section The active power output data comes from photovoltaic power forecasts or available power transmitted from the equipment side; This is the minimum active power output allowed by the inverter, derived from equipment parameters or grid connection protocol.

[0068] Ultimately, the data was adjusted by partition. It is represented as follows: .

[0069] In the above embodiments, by obtaining the predicted net injected power of the distribution network within the hourly timescale control cycle and constructing the predicted change vector of the node net injected power, the global optimization model can predict the power change trend over a future period on an hourly timescale, thereby planning the control schedule of slow-acting equipment in advance and avoiding frequent operation of slow-acting equipment due to lack of foresight. Based on this, a global optimization model is constructed using the state vector of the slow-acting equipment, the target voltage vector of the distribution network nodes, and the estimated network loss. The voltage equality constraint obtained by transforming the node net injected power prediction change vector into the first voltage model allows the global optimization model to simultaneously consider the comprehensive optimization objectives of minimizing voltage deviation, minimizing the operation cost of slow-acting equipment, and minimizing network loss. Solving this global optimization model... After modeling, a control plan for slow-moving devices is obtained. This plan enables slow-moving devices to operate optimally on an hourly timescale, avoiding frequent triggering by second-level voltage fluctuations. Simultaneously, based on the node hourly timescale voltage vector, node voltage baseline vectors and zone voltage baseline vectors are determined. Based on these, the boundary node voltage offset budget, reactive power regulation budget, and active power reduction budget for each zone are determined. These zone regulation data provide clear spatial boundaries and resource limits for subsequent minute-level zone collaborative optimization. Through global optimization of the hourly timescale control cycle, control separation between slow-moving devices and inverters is achieved in the time dimension, thereby eliminating mutual interference between fast and slow resources on the time scale and improving the stability and accuracy of distribution network voltage control.

[0070] In this embodiment, the step of constructing a zone collaborative optimization model for each zone within a preset minute timescale control cycle based on the voltage sensitivity matrix set and zone adjustment data, and determining the inverter control data for each zone based on the zone collaborative optimization model, includes: Within a minute-scale control cycle, a first partition sensitivity matrix is ​​determined based on the second sensitivity matrix to determine the node voltage amplitude and inverter reactive power regulation of each partition; a second partition sensitivity matrix is ​​determined based on the first sensitivity matrix to determine the node voltage amplitude and inverter active power reduction of each partition. Obtain the node active net injection disturbance data and node reactive net injection disturbance data of the distribution network, and construct the partition voltage disturbance vector of each partition based on the node active net injection disturbance data and node reactive net injection disturbance data. Construct the inverter reactive power adjustment vector and inverter active power reduction vector for each of the partitions. Based on the partition voltage baseline vector, partition voltage disturbance vector, first partition sensitivity matrix, second partition sensitivity matrix, inverter reactive power adjustment vector and inverter active power reduction vector, construct the partition node voltage prediction vector for each of the partitions. Based on the partition node voltage prediction vector, boundary node voltage offset budget, reactive power regulation budget, and active power reduction budget of each partition, a partition collaborative optimization model is constructed for each partition, and the inverter control data of each partition is determined based on the partition collaborative optimization model.

[0071] In the above embodiments, by determining the first partition sensitivity matrix of the node voltage amplitude and inverter reactive power regulation amount, and the second partition sensitivity matrix of the node voltage amplitude and inverter active power reduction amount, respectively, within a minute-scale control cycle, the influence degree and direction of inverter reactive power regulation and active power reduction on the node voltage within each partition can be accurately characterized. Simultaneously, by acquiring the net active power injection disturbance data and net reactive power injection disturbance data of the distribution network nodes, and combining them with the fourth and fifth sensitivity matrices to construct the partition voltage disturbance vector, the subsequent partition node voltage prediction model can fully consider... The system considers the impact of real-time disturbances such as load fluctuations and photovoltaic output deviations on voltage, thus improving the accuracy of voltage prediction. Based on this, a zone node voltage prediction vector is constructed using zone voltage baseline vector, zone voltage disturbance vector, first zone sensitivity matrix, second zone sensitivity matrix, inverter reactive power adjustment vector, and inverter active power reduction vector. This enables each zone to independently predict the voltage change trend within its zone on a minute-level time scale, thereby significantly improving the control response speed and accuracy. At the same time, it avoids the communication burden and computational delay problems caused by global centralized optimization, thus improving the efficiency and accuracy of distribution network voltage control.

[0072] In this embodiment, the step of constructing a partitioned collaborative optimization model for each partition based on the partition node voltage prediction vector, boundary node voltage offset budget, reactive power regulation budget, and active power reduction budget for each partition, and determining the inverter control data for each partition based on the partitioned collaborative optimization model, includes: Obtain the historical inverter reactive power regulation vector for the previous minute-scale control cycle; Calculate the first difference vector between the inverter reactive power regulation vector and the historical inverter reactive power regulation vector, and calculate the second difference vector between the partition node voltage prediction vector and the partition voltage baseline vector; With the goal of minimizing the sum of the first difference vector, the second difference vector, and the inverter active power reduction vector, a partition collaborative optimization objective function is constructed for each of the partitions. Based on the boundary node voltage offset budget, reactive power regulation budget, and active power reduction budget, construct a set of partition collaborative optimization constraints for each partition; Based on the set of partitioned collaborative optimization constraints and the objective function of partitioned collaborative optimization, a partitioned collaborative optimization model is formed. The partitioned collaborative optimization model is solved to obtain the optimal reactive power adjustment vector and the optimal active power reduction vector of the inverter in each partition. Based on the inverter's optimal reactive power adjustment vector and the inverter's optimal active power reduction vector, the target active power output and target reactive power output of each inverter in each of the partitions are determined. Based on the target reactive power output and target active power output, determine the set of voltage reactive power curve parameters and voltage active power reduction curve parameters for each inverter in each of the partitions. Based on the target active power output, target reactive power output, voltage-reactive curve parameter set, and voltage-active power reduction curve parameter, and combined with the pre-acquired reactive power regulation rate of change limit and active power reduction rate of change limit for each inverter, the inverter control data for each of the aforementioned zones is determined.

[0073] In the above embodiments, the historical inverter reactive power regulation vector of the previous minute-scale control cycle is obtained, and the first difference vector between the current inverter reactive power regulation vector and the historical inverter reactive power regulation vector, and the second difference vector between the partition node voltage prediction vector and the partition voltage baseline vector are calculated respectively. Based on this, a partition collaborative optimization objective function is constructed with the goal of minimizing the sum of the first difference vector, the second difference vector, and the inverter active power reduction vector. This objective function simultaneously includes three objective components: minimizing voltage deviation, smoothing reactive power regulation, and minimizing active power reduction. A partition collaborative optimization constraint set is constructed through the boundary node voltage offset budget, reactive power regulation budget, and active power reduction budget, so that the optimization result can effectively reduce the voltage deviation within the partition and ensure a smooth transition of reactive power regulation. To reduce equipment wear and tear, and to strictly control the amount of active power reduction to reduce curtailment losses, the optimal reactive power regulation vector and optimal active power reduction vector of the inverter are obtained after solving the regional collaborative optimization model. Based on this, the target active power output and target reactive power output of each inverter are determined, and the voltage-reactive power curve parameter set and voltage-active power reduction curve parameters are further determined. Combined with the pre-acquired reactive power regulation change rate limit and active power reduction change rate limit, the inverter control data is determined, so that the inverters in each region can coordinate and allocate reactive power regulation tasks in an optimal manner on a minute-level time scale, avoiding reactive power competition and regulation conflicts between inverters in the region. At the same time, the hard constraint of the active power reduction budget ensures that the amount of curtailment is within a controllable range, thereby ensuring control smoothness while guaranteeing voltage compliance, and thus improving the stability and accuracy of distribution network voltage control.

[0074] In one optional embodiment, this embodiment describes the specific implementation of step S104. Let the minute-level moments within the minute-scale control period be... ,and Belongs to the optimized cross section The time window; The sensitivity matrix for the first partition is: ;in, For partitioning The first partition sensitivity matrix, its... Each element represents the inverter within the partition. The unit reactive power regulation for the first zone The influence of the voltage amplitude at each node; The fourth sensitivity matrix at minute-level time. The value of ; Select a matrix for the partitioned inverters, used to select from the entire network of inverters. Extracting from partitions The inverter.

[0075] The sensitivity matrix for the second partition is: ;in, For partitioning The second partition sensitivity matrix, its first Each element represents the inverter within the partition. The unit active power reduction for the first partition The impact of node voltage amplitude; the negative sign indicates that active power reduction is equivalent to a reduction in active power injection at the node, that is, active power reduction will reduce the voltage rise caused by active power injection; , The first sensitivity matrix at minute-level time. The value of ; For minute-level moments Benchmark node The voltage amplitude; Obtain the net active power injection disturbance data of the distribution network nodes. and node reactive net injection disturbance data ,in, This indicates the net active power injection disturbance at the node caused by load forecasting errors, actual photovoltaic output deviations, and uncontrolled distributed resource fluctuations, which originates from the short-time forecast residual update algorithm or Kalman filter estimation. This represents the net reactive power injection disturbance at the node caused by load reactive power fluctuations, uncontrolled reactive power device operation, or prediction errors; therefore, the zone voltage disturbance vector... for: (17); This indicates the change in source voltage. The source-end sensitivity vector at minute-level time intervals The value of ; This represents the measurement correction term formed by the deviation between the real-time measured voltage and the model-predicted voltage.

[0076] The reactive power regulation vector of the inverter in each zone is: ;in, For partitioning The inverter reactive power regulation vector; Indicates inverter In minutes The reactive power regulation relative to the linear reference point; a positive value indicates that the inverter injects reactive power into the distribution network, and a negative value indicates that the inverter absorbs reactive power from the distribution network.

[0077] The inverter active power reduction vector is: ,in, For partitioning Internal inverter active power reduction vector; Indicates inverter In minutes The active power reduction, taking a non-negative value, when This indicates that the inverter does not reduce active power. Therefore, the node voltage prediction vector is expressed as: ; Obtain the historical inverter reactive power regulation vector for the previous minute-scale control cycle. Calculate the first difference vector between the inverter reactive power adjustment vector and the historical inverter reactive power adjustment vector. Calculate the second difference vector between the partition node voltage prediction vector and the partition voltage baseline vector. With the goal of minimizing the sum of the first difference vector, the second difference vector, and the inverter active power reduction vector, a partition collaborative optimization objective function is constructed for each partition, which is expressed as shown in formula (18). (18); in, This is a zoned voltage weighting matrix, typically a diagonal matrix, used to increase the voltage control weight of end nodes, critical load nodes, historical over-limit nodes, and boundary nodes. Its specific values ​​are set by technical personnel. For the partitioned minute-level voltage reference vector, it is usually taken as ; The reactive power regulation smoothing weight is used to suppress frequent changes in the inverter's reactive power command. Represents the L2 norm; The cost weighting for active light reduction is used to limit unnecessary light wastage.

[0078] Based on the boundary node voltage offset budget, reactive power regulation budget, and active power reduction budget, a partition collaborative optimization constraint set is constructed for each of the partitions. The partition collaborative optimization constraint set includes: Intra-region linearized voltage model constraints: ; Voltage safety constraints are: ;in, for A 1-dimensional vector; and These are the upper and lower limits of voltage operation; The boundary voltage constraint is: ;in, Select a matrix for the boundary nodes to extract the voltage of the boundary nodes from the voltage vector of the partition nodes; For partitioning The number of boundary nodes; and This is a budget for the voltage offset at the boundary nodes; this constraint is used to limit the excessive voltage disturbances that zonal control may cause to adjacent zonals or the feeder backbone. The reactive power budget constraint for each zone is: and ;in, ; This means that the sum of the reactive power injection of all inverters within the zone must not exceed the reactive power injection budget. ; This means that the sum of the reactive power absorption of all inverters within the zone must not exceed the reactive power absorption budget. ; The effective budget reduction constraint is: ; The inverter capacity circle constraint is: ;in, For inverter In minutes The reactive power output; This is the reactive power regulation quantity; The rated apparent capacity of the inverter is derived from the equipment nameplate. The active power output of the inverter plan comes from the scheduling plan or the results of upper-level optimization. This is the active power reduction; this constraint is used to ensure that the inverter's reactive and active power outputs together meet the apparent capacity limit. The upper and lower limits of reactive power for the inverter are: ;in, and These are the upper and lower limits, respectively, for the inverter to generate and absorb reactive power. The inverter reactive power ramping constraint is: ;in, For inverter The reactive power value in the previous minute-level control cycle; The upper limit of the reactive power change rate of the inverter is derived from the inverter equipment parameters; The control cycle length is in the minute range; The effective reduction of climbing constraints is ;in, This is the amount of active power reduction in the previous minute-level control cycle; The upper limit of the active power reduction rate of change is derived from the inverter equipment parameters; Based on the aforementioned set of partitioned collaborative optimization constraints and the objective function of partitioned collaborative optimization, a partitioned collaborative optimization model is formed. When retaining the capacity circle constraint, this model can be transformed into a second-order cone programming problem, solved using the second-order cone interior-point method, the primal-dual interior-point method, or the alternating direction multiplier method. If the capacity circle constraint is linearized, it can be transformed into a quadratic programming problem, solved using the quadratic programming interior-point method or the active set algorithm. The solution yields the optimal reactive power regulation vector for the inverter in each partition. and the inverter's optimal active power reduction vector ; Based on the optimal reactive power regulation vector of the inverter and the inverter's optimal active power reduction vector Determine the target active power output and target reactive power output of each inverter within each of the aforementioned zones, i.e. and ;in, For inverter The goal is to achieve nothing but wasted effort; For inverter The optimal reactive power regulation; For inverter The goal is to contribute one's efforts; Contribute to the plan; For inverter The optimal active power reduction; Based on the target reactive power output and target active power output, determine the set of voltage reactive power curve parameters and the set of voltage active power reduction curve parameters for each inverter in each of the aforementioned zones; specifically, based on the target active power output... Calculate the current available reactive power limit of the inverter: and ;in, For inverter In minutes Maximum reactive power injection capability; For inverter The maximum reactive power absorption capacity; then, assume the inverter's grid connection point is at the optimized predicted voltage. , its origin Middle and grid connection nodes The corresponding element is obtained.

[0079] Calculate the droop slope of the initial voltage reactive power curve: ;in, The droop slope of the initial voltage reactive power curve; Unproductive efforts towards a goal; Predict the voltage at the grid connection point; This is the baseline voltage of the inverter's grid connection node; The dead zone width for voltage reactive power curve control is derived from the control strategy setting. To prevent small positive numbers with a denominator of zero; By adjusting the initial voltage-reactive power curve droop slope to within the inverter's allowable range, the voltage-reactive power curve droop slope is obtained, i.e. ;in, The downward slope of the voltage reactive power curve; The projection operator indicates that the variable is restricted to an interval. Inside; and These are the minimum and maximum voltage reactive power curve droop slopes that the inverter can be configured with, respectively. Next, in order for the inverter to output voltage near the current grid connection point voltage... Determine the local voltage reference value based on the target reactive power direction: If (Inverters need to absorb reactive power to suppress overvoltage), then define ;like (The inverter needs to inject reactive power to support low voltage), then define ;in, This is the local voltage reference value for the inverter; It is a positive number close to zero, used to determine whether the target reactive power is zero; To prevent small positive numbers with a denominator of zero, the four inflection points of the voltage-reactive power curve are calculated based on this: , , ,and ;in, , , , These are the four inflection point voltages of the voltage-reactive power curve; Specifically, this allows for the determination of the inverter's local voltage reactive power curve control rate. This is the grid connection point voltage measured locally by the inverter; ; This results in a set of voltage-reactive power curve parameters: ; When there is an active power reduction, i.e. The parameters for generating the voltage active power reduction curve are as follows: ; First, generate the active power reduction initiation voltage for the voltage active power reduction curve: ;in, To reserve margin for active power reduction triggers, it is advisable to... to Secondly, the reduction slope of the voltage active power reduction curve is generated. Specifically, it is as follows: ;in, and These are the minimum and maximum allowable cutoff slopes for the inverter, respectively; simultaneously, the minimum allowable active power output of the inverter is obtained. For inverters that allow for full active power reduction, the following can be adopted: For equipment that needs to maintain a minimum output, the following can be adopted: ; For the minimum output ratio, for example ; The rated active power of the inverter; The final inverter control data for each of the aforementioned partitions is represented as follows: ; and These are the limits for the rate of change of reactive power regulation and the rate of change of active power reduction, respectively.

[0080] In this embodiment, the voltage control of the distribution network based on the slow-motion device control plan and inverter control data within a preset second-level time scale control cycle includes: Within the control cycle on the second-level time scale, the real-time voltage measurement values ​​of each inverter and the current control status of each slow-motion device are acquired; The initial reactive power output value of each inverter is determined based on the real-time voltage measurement value and the voltage reactive power curve parameter set. Obtain the current active power output value and rated capacity of each inverter, and determine the current reactive power output limit of each inverter based on the current active power output value and rated capacity; The historical reactive power output value of each inverter in the previous second-level time scale control cycle is obtained. Based on the current reactive power output limit, the historical reactive power output value and the reactive power regulation change rate limit, the initial reactive power output value is corrected to obtain the actual reactive power output value of each inverter. The initial active power output of each inverter is calculated based on the real-time voltage measurement value and the voltage active power reduction curve parameters. Obtain the historical active power output value of each inverter in the previous second-level time scale control cycle, and determine the actual active power output value of each inverter based on the active power reduction change rate limit and the historical active power output value. The average voltage of the feeder key nodes and the target voltage of the slow-action equipment of the distribution network are obtained. When the average voltage of the feeder key nodes and the target voltage of the slow-action equipment meet the preset control conditions of the slow-action equipment, the slow-action equipment of the distribution network is controlled based on the control plan. Based on the actual active power output and actual reactive power output of each inverter, the output of each inverter is controlled to complete the voltage control of the distribution network.

[0081] In one optional embodiment, this embodiment describes the specific implementation of step S105. The second-level control time within the second-level timescale control period. Obtain the inverter Real-time voltage measurement value The data source is the inverter's local voltage sampling module or the power quality monitoring device at the grid connection point; Next, the inverter determines its initial reactive power output value based on the real-time voltage measurement and the voltage-reactive power curve parameter set. Specifically, it is as follows: ; Considering the inverter capacity circle constraint, calculate the current reactive power output limit, including the current reactive power output upper limit. and the current lower limit of reactive power output Specifically: and ; Then the initial reactive power output value Within the capacity circle, the first reactive power output value is obtained, which is expressed as: ;in, This is the first reactive power output value; Indicates will Limited to the range Inside; To avoid sudden changes in reactive power output, a reactive power ramping constraint is then added to obtain the actual reactive power output value. Its specific representation is shown in the following formula: ; in, This represents the historical reactive power output value of the previous control cycle on a second-level time scale. The control cycle length is in the second range; This is the limit for the rate of change of reactive power regulation.

[0082] When reactive power regulation has reached the inverter's reactive power capacity limit, or when there is still an overvoltage risk within the zone under the current reactive power regulation conditions, active power reduction is required. The specific process is as follows: First, define the first voltage inflection point of the voltage active power reduction curve parameters. Second voltage inflection point Among them, the first voltage inflection point Indicates inverter The voltage threshold at which active power reduction begins; the second voltage inflection point. Indicates inverter The voltage termination threshold that reduces active output to the minimum allowable value; both are per-unit values ​​and satisfy... Specifically, Set as ; ; The width of the reduction interval of the voltage active power reduction curve can be taken as... ; Next, set For inverter At any moment Available active power represents the maximum active power that the inverter can output without active power reduction under current constraints of illumination, DC-side input, equipment capacity, and reactive power output; for example, for a photovoltaic inverter, The data sources include the maximum power output, DC-side voltage and current measurements, irradiance and component temperature estimates, or short-time predictions for different zones, all transmitted from the inverter's MPPT module. It can be determined using the following formula: ;in, The maximum active power that the inverter's MPPT module can generate is determined based on the current DC-side operating conditions. The rated active power of the inverter; The rated apparent capacity of the inverter; This represents the actual reactive power output of the inverter. Under the above definition, the initial active power output of the inverter Represented as: ; To avoid abrupt changes in active power output, this embodiment further superimposes an active power ramp constraint to obtain the actual active power output value. : ; Actual active power output Simultaneously satisfy ,in, This represents the historical active power output of the inverter during the previous second-level control cycle. This represents the upper limit of available active power. To achieve decoupling across multiple time scales, slow-motion equipment does not directly respond to second-level voltage fluctuations, but instead employs a dwell and hysteresis triggering mechanism. Specifically, the set of key feeder nodes is defined as follows: Therefore, the average voltage at the critical node of the feeder is expressed as: ; This refers to the number of critical nodes in the feeder. For nodes In seconds The voltage value; the target voltage of slow-motion equipment is defined as... Its specific value is set according to actual needs. The voltage hysteresis bandwidth for triggering actions of slow-motion devices; The minimum dwell time for slow-motion equipment; therefore, the control conditions for slow-motion equipment are set as follows: And the duration of this deviation exceeds Slow motion devices are only allowed to move if the number of slow motion actions does not exceed the allowed limit. The slow-action equipment performs a tap adjustment, switching, or compensation setting adjustment according to the control plan. If the voltage deviation is only a short-term disturbance, or has not yet met the dwell time condition, the slow-action equipment remains in its original state, and the inverter equipment layer continues to perform second-level rapid compensation. This avoids frequent operation of the slow-action equipment due to short-term photovoltaic fluctuations or instantaneous load changes.

[0083] In the above embodiments, by acquiring the real-time voltage measurement values ​​of each inverter within a second-scale control cycle and combining them with the voltage-reactive power curve parameter set to determine the initial reactive power output value, the inverter can autonomously adjust reactive power based on the local real-time voltage. Simultaneously, by acquiring the current active power output value and rated capacity of the inverter, the current reactive power output limit is determined. Based on the current reactive power output limit, historical reactive power output values, and reactive power adjustment rate of change limit, the initial reactive power output value is corrected using ramp constraints to obtain the actual reactive power output value. This physically limits the rate of change of reactive power output, avoiding voltage oscillations caused by sudden reactive power changes. Next, the initial active power output value is calculated based on the real-time voltage measurement values ​​and voltage-active power reduction curve parameters, and based on the active power reduction rate of change limit... The actual active power output is determined by comparing the historical active power output value with the actual active power output value, making the active power reduction process smooth and controllable. In terms of slow device control, by obtaining the average voltage of key nodes in the feeder and comparing it with the target voltage of the slow-acting device, the slow-acting device is controlled only when the preset slow-acting device control conditions are met. This ensures that the slow-acting device will not respond to second-level voltage fluctuations, enabling the inverter to safely and smoothly complete rapid voltage regulation on a second-level time scale. The slow-acting device only operates under the slow-acting device control conditions, realizing the spatial zonal coordination and temporal layered decoupling of the distribution network. This avoids mutual interference and control conflicts between different control resources, thereby improving the accuracy and stability of the distribution network voltage control.

[0084] In this embodiment, after voltage control of the distribution network, the method further includes: Within the preset evaluation window, obtain the voltage measurement value after control of each node in the distribution network, the available active power after control of each inverter, and the actual number of actions of each slow-action device. The voltage over-limit index of the distribution network is calculated based on the voltage measurement value after control. The slow-moving equipment index of the power distribution network is calculated based on the actual number of actions. Based on the controlled active power output value and the actual active power output value, the curtailment index of the distribution network can be determined. The boundary exceedance index of the distribution network is determined based on the boundary node voltage offset budget, the zone voltage baseline vector, and the voltage measurement after control. The inverter response deviation index of the distribution network is determined based on the actual reactive power output, actual active power output, target reactive power output, and target active power output of each inverter. A comprehensive evaluation index is generated based on the voltage over-limit index, slow-motion equipment index, curtailment index, boundary over-limit index, and inverter response deviation index. The distribution network voltage control evaluation results are generated based on the comprehensive evaluation indicators, and the distribution network voltage control is adjusted for the next time based on the voltage over-limit indicators, slow-action equipment indicators, curtailment indicators, boundary over-limit indicators, and inverter response deviation indicators.

[0085] In an optional embodiment, the evaluation window is set to ; To evaluate the number of sampling points per second within the evaluation window; to obtain the post-control voltage measurement values ​​of each node in the distribution network. The available active power after controlling each inverter and the actual number of movements of each slow-motion device ; Therefore, the voltage over-limit indicator is expressed as: ; The larger the value, the more severe the voltage exceedance within the evaluation window; The slow-motion device specifications are expressed as follows: ; For slow motion equipment The action cost coefficient can be set according to the equipment type, mechanical life, action maintenance cost, or operating procedure; The larger the value, the more frequently the slow-motion device operates; The curtailment index is expressed as: ; Indicates inverter At any moment The usable active power output value after control; This indicates the active energy reduction caused by voltage control within the evaluation window; Boundary violation indicators are expressed as follows: ; in, Represents boundary nodes The measured voltage value after control; Used to measure whether partition control causes voltage disturbances to boundary nodes that exceed the budget; The inverter response deviation index is expressed as: ; The active power deviation weight can be set by technical personnel; the inverter response deviation index is used to identify whether the inverter has response hysteresis, limit saturation, communication abnormality or command non-execution. A comprehensive evaluation index is obtained by weighting and summing the voltage over-limit index, slow-action equipment index, curtailment index, boundary over-limit index, and inverter response deviation index. It is represented as: ; in, , , , and These are the weighting coefficients for voltage over-limit indicators, slow-motion equipment indicators, curtailment indicators, boundary over-limit indicators, and inverter response deviation indicators, respectively. The specific values ​​can be set by technical personnel. Comprehensive evaluation indicators This serves as the evaluation result for the distribution network voltage control in this area; simultaneously, the next distribution network voltage control is adjusted based on the comprehensive evaluation indicators; specifically, the safety threshold for voltage over-limit indicators is obtained. Voltage over-limit threshold Slow-motion device indicator thresholds Waste Light Index Threshold Boundary limit exceedance threshold Inverter response deviation index threshold ; like (If the voltage exceeds the limit severely), then for the node that exceeds the voltage limit, increase its partition voltage weight matrix in the partitioned collaborative optimization objective function shown in formula (18). The weights in; if (If slow-motion devices operate too frequently), then for slow-motion devices that operate too frequently, increase their slow-motion device action cost weight in the global optimization model shown in the final formula (11); and extend their slow device dwell time, or increase the hysteresis bandwidth; if And the voltage over-limit index is relatively small ( If ), then increase the active power reduction cost weight in the partitioned collaborative optimization objective function shown in formula (18); if (If the boundary disturbance is too large), then reduce the range of the boundary voltage budget; if If the inverter response deviation is large, the inverter's response reliability coefficient will be reduced, thus decreasing the workload of the inverter in the next distribution network voltage control. In particular, the above adjustments are merely illustrative examples provided in this embodiment of the invention; those skilled in the art can make corresponding adjustments according to actual needs.

[0086] In the above embodiments, by acquiring the measured voltage values ​​of each node in the distribution network after control, the available active power of each inverter after control, and the actual number of actions of each slow-acting device within a preset evaluation window, a comprehensive measured data foundation is provided for closed-loop evaluation. Based on this, voltage over-limit index, slow-acting device index, curtailment index, boundary over-limit index, and inverter response deviation index are calculated respectively. These five indicators comprehensively reflect the distribution network voltage control effect from different dimensions. Then, a comprehensive evaluation index is generated based on the above five indicators, which can quantitatively evaluate the overall effect of the current voltage control and avoid the one-sidedness of single-indicator evaluation. Furthermore, it can be adjusted accordingly in the next distribution network voltage control, so that the distribution network voltage control forms a complete control closed loop of execution, evaluation, feedback, and adjustment. This allows the distribution network voltage control to be continuously optimized according to the actual operating effect, thereby continuously improving the accuracy and stability of distribution network voltage control in long-term operation.

[0087] Example 2 Please refer to Figure 2 , Figure 2 A schematic diagram of a distribution network voltage control system provided in an embodiment of the present invention includes: a distribution network voltage modeling module 201, a distribution network partitioning module 202, an hourly time scale control module 203, a minute time scale control module 204, and a second-level time scale control module 205. The distribution network voltage modeling module 201 is used to acquire the topology operation data of the distribution network and construct the voltage model and voltage sensitivity matrix set of the distribution network based on the topology operation data. The power distribution network partitioning module 202 is used to divide the power distribution network into regions based on the voltage model and the voltage sensitivity matrix set, thereby obtaining several partitions of the power distribution network. The hourly time-scale control module 203 is used to construct a global optimization model of the distribution network in a preset hourly time-scale control cycle based on the voltage model and the voltage sensitivity matrix set, and to obtain the slow-motion equipment control plan of the distribution network and the zonal adjustment data of each zone based on the global optimization model. The minute-scale control module 204 is used to construct a zone collaborative optimization model for each zone within a preset minute-scale control cycle based on the voltage sensitivity matrix set and zone adjustment data; and to determine the inverter control data for each zone based on the zone collaborative optimization model. The second-level time scale control module 205 is used to perform voltage control on the power distribution network within a preset second-level time scale control period based on the slow-motion equipment control plan and inverter control data.

[0088] In this embodiment, the distribution network voltage modeling module 201 includes: a distribution network voltage modeling unit; The distribution network voltage modeling unit is used to acquire the topology operation data of the distribution network, which includes: the source node voltage values ​​of the distribution network; Using the source node voltage value as a reference boundary, a vector of squared changes in node voltage of the distribution network is constructed according to a preset linearized power flow recursion algorithm; Construct the node net active power injection change vector, node net reactive power injection change vector, slow-action equipment control quantity change vector, and inverter reactive power control quantity change vector of the distribution network. Based on the node net active power injection change vector and the node voltage square change vector, a first sensitivity matrix is ​​constructed between the node voltage square change and the node net active power injection change of the distribution network. Based on the node net reactive power injection change vector and the node voltage square change vector, a second sensitivity matrix is ​​constructed between the node voltage square change and the node net reactive power injection change of the distribution network. Based on the control quantity change vector of the slow-action device and the square change vector of the node voltage, a third sensitivity matrix is ​​constructed between the square change of the node voltage and the control state change of the slow-action device in the distribution network. The voltage sensitivity matrix set of the distribution network is formed based on the first sensitivity matrix, the second sensitivity matrix, and the third sensitivity matrix; The first voltage model of the distribution network is constructed based on the voltage sensitivity matrix set, the node net active power injection change vector, the node net reactive power injection change vector, and the slow-acting equipment control quantity change vector. Based on the topology operation data, construct the node inverter mapping matrix between inverters and nodes in the distribution network; The first voltage model is transformed based on the node inverter mapping matrix and the inverter reactive power control change vector to obtain the second voltage model of the distribution network. The voltage model of the distribution network is formed based on the first voltage model and the second voltage model.

[0089] In this embodiment, the power distribution network partitioning module 202 includes: a power distribution network partitioning unit; The distribution network partitioning unit is used to transform the second sensitivity matrix within each preset hourly time scale control cycle to obtain a fourth sensitivity matrix between the node voltage amplitude change and the node net reactive power injection change of the distribution network. Based on the node inverter mapping matrix and the second sensitivity matrix in the second voltage model, the influence intensity of the unit reactive power output change of any inverter in the distribution network on the voltage amplitude change of any node is determined, and the node inverter voltage coupling matrix is ​​constructed based on the influence intensity. The node inverter voltage coupling matrix is ​​normalized to obtain the normalized node inverter voltage coupling matrix. The similarity of the common influence between any two nodes on the same inverter is calculated based on the normalized node inverter voltage coupling matrix, and the real-time node coupling adjacency matrix of the distribution network is constructed based on the common influence similarity. Obtain the historical node coupling adjacency matrix and historical topology switch state data of the previous hourly timescale control cycle; Based on the real-time node coupling adjacency matrix, the historical node coupling adjacency matrix, and the historical topology switch status data, the distribution network is divided into several regions to obtain the distribution network.

[0090] In this embodiment, the power distribution network zoning unit includes: a power distribution network zoning subunit; The distribution network partition subunit is used to determine the real-time topology switch status data of the distribution network based on the topology operation data, and to determine the topology change degree of the distribution network based on the real-time topology switch status data and historical topology switch data. The coupling variation degree of the distribution network is determined based on the historical node coupling adjacency matrix and the real-time node coupling adjacency matrix. When the coupling change degree is less than the first preset threshold and the topology change degree does not meet the preset topology change condition, several partitions of the previous hourly time scale control cycle are taken as several partitions of the distribution network in the current hourly time scale control cycle. When the coupling change degree is greater than or equal to the first preset threshold or the topology change degree meets the preset topology change condition, the normalized graph Laplace matrix of the distribution network is constructed based on the real-time node coupling adjacency matrix. The nodes are clustered based on a preset clustering algorithm and the normalized graph Laplace matrix to obtain several partitions of the power distribution network. Based on the topology operation data, several topology boundary nodes for each partition are determined; The cross-region control sensitivity between any node in each of the partitions and any inverter in other partitions is calculated based on the normalized node inverter voltage coupling matrix. Based on the cross-region control sensitivity, several control boundary nodes of each partition are determined. The topological boundary nodes and control boundary nodes of each partition are used as the boundary nodes of each partition to complete the regional division of the power distribution network.

[0091] In this embodiment, the hourly timescale control module 203 includes: an hourly timescale control unit; The hourly time-scale control unit is used to acquire the predicted net injected power, estimated network loss, and target voltage vector of the distribution network nodes within the hourly time-scale control period. Based on the net injected power prediction value, a node net injected power prediction change vector is constructed, and based on the topology operation data, the distribution network is constructed to obtain the slow-moving equipment state vector. Based on the slow-motion device state vector, the distribution network node voltage target vector, and the network loss estimate, a global optimization model of the distribution network is constructed for the hourly time scale control period. The first voltage model is transformed based on the predicted change vector of the node net injected power to obtain the voltage equality constraint of the global optimization model; The global optimization model is solved based on the voltage equality constraint, the preset voltage safety constraint, and the slow-action device action constraint to obtain the slow-action device control plan and the node hourly time scale voltage vector of the distribution network. The node voltage baseline vector of the distribution network is determined based on the node hourly time-scale voltage vector, and the partition voltage baseline vector of each partition is determined based on the node voltage baseline vector. The boundary node voltage offset budget for each partition is determined based on the partition voltage baseline vector and the preset voltage limit value of each partition. Obtain the reactive power output data of each inverter in each of the partitions, and determine the reactive power regulation budget of each partition based on the reactive power output data and the node inverter voltage coupling matrix; Obtain the active power output data of each inverter in each partition, and determine the active power reduction budget for each partition based on the active power output data; The partition voltage baseline vector, boundary node voltage offset budget, reactive power adjustment budget, and active power reduction budget of each partition are used as the partition adjustment data for each partition.

[0092] In this embodiment, the minute timescale control module 204 includes: a minute timescale control unit; The minute-scale control unit is used to determine, within a minute-scale control cycle, a first partition sensitivity matrix based on the second sensitivity matrix, the node voltage amplitude of each partition and the inverter reactive power regulation amount; and a second partition sensitivity matrix based on the first sensitivity matrix, the node voltage amplitude of each partition and the inverter active power reduction amount. Obtain the node active net injection disturbance data and node reactive net injection disturbance data of the distribution network, and construct the partition voltage disturbance vector of each partition based on the node active net injection disturbance data and node reactive net injection disturbance data. Construct the inverter reactive power adjustment vector and inverter active power reduction vector for each of the partitions. Based on the partition voltage baseline vector, partition voltage disturbance vector, first partition sensitivity matrix, second partition sensitivity matrix, inverter reactive power adjustment vector and inverter active power reduction vector, construct the partition node voltage prediction vector for each of the partitions. Based on the partition node voltage prediction vector, boundary node voltage offset budget, reactive power regulation budget, and active power reduction budget of each partition, a partition collaborative optimization model is constructed for each partition, and the inverter control data of each partition is determined based on the partition collaborative optimization model.

[0093] In this embodiment, the minute timescale control unit includes: a minute timescale control subunit; The minute-timescale control subunit is used to obtain the historical inverter reactive power adjustment vector of the previous minute-timescale control cycle. Calculate the first difference vector between the inverter reactive power regulation vector and the historical inverter reactive power regulation vector, and calculate the second difference vector between the partition node voltage prediction vector and the partition voltage baseline vector; With the goal of minimizing the sum of the first difference vector, the second difference vector, and the inverter active power reduction vector, a partition collaborative optimization objective function is constructed for each of the partitions. Based on the boundary node voltage offset budget, reactive power regulation budget, and active power reduction budget, construct a set of partition collaborative optimization constraints for each partition; Based on the set of partitioned collaborative optimization constraints and the objective function of partitioned collaborative optimization, a partitioned collaborative optimization model is formed. The partitioned collaborative optimization model is solved to obtain the optimal reactive power adjustment vector and the optimal active power reduction vector of the inverter in each partition. Based on the inverter's optimal reactive power adjustment vector and the inverter's optimal active power reduction vector, the target active power output and target reactive power output of each inverter in each of the partitions are determined. Based on the target reactive power output and target active power output, determine the set of voltage reactive power curve parameters and voltage active power reduction curve parameters for each inverter in each of the partitions. Based on the target active power output, target reactive power output, voltage-reactive curve parameter set, and voltage-active power reduction curve parameter, and combined with the pre-acquired reactive power regulation rate of change limit and active power reduction rate of change limit for each inverter, the inverter control data for each of the aforementioned zones is determined.

[0094] In this embodiment, the second-level time scale control module 205 includes: a second-level time scale control unit; The second-level time scale control unit is used to acquire the real-time voltage measurement value of each inverter and the current control status of each slow-motion device within the second-level time scale control cycle. The initial reactive power output value of each inverter is determined based on the real-time voltage measurement value and the voltage reactive power curve parameter set. Obtain the current active power output value and rated capacity of each inverter, and determine the current reactive power output limit of each inverter based on the current active power output value and rated capacity; The historical reactive power output value of each inverter in the previous second-level time scale control cycle is obtained. Based on the current reactive power output limit, the historical reactive power output value and the reactive power regulation change rate limit, the initial reactive power output value is corrected to obtain the actual reactive power output value of each inverter. The initial active power output of each inverter is calculated based on the real-time voltage measurement value and the voltage active power reduction curve parameters. Obtain the historical active power output value of each inverter in the previous second-level time scale control cycle, and determine the actual active power output value of each inverter based on the active power reduction change rate limit and the historical active power output value. The average voltage of the feeder key nodes and the target voltage of the slow-action equipment of the distribution network are obtained. When the average voltage of the feeder key nodes and the target voltage of the slow-action equipment meet the preset control conditions of the slow-action equipment, the slow-action equipment of the distribution network is controlled based on the control plan. Based on the actual active power output and actual reactive power output of each inverter, the output of each inverter is controlled to complete the voltage control of the distribution network.

[0095] In this embodiment, the second-level time scale control module 205 further includes: a voltage control evaluation unit; The voltage control evaluation unit is used to obtain the voltage measurement value after control of each node in the distribution network, the available active power after control of each inverter, and the actual number of actions of each slow-acting device within a preset evaluation window. The voltage over-limit index of the distribution network is calculated based on the voltage measurement value after control. The slow-moving equipment index of the power distribution network is calculated based on the actual number of actions. Based on the controlled active power output value and the actual active power output value, the curtailment index of the distribution network can be determined. The boundary exceedance index of the distribution network is determined based on the boundary node voltage offset budget, the zone voltage baseline vector, and the voltage measurement after control. The inverter response deviation index of the distribution network is determined based on the actual reactive power output, actual active power output, target reactive power output, and target active power output of each inverter. A comprehensive evaluation index is generated based on the voltage over-limit index, slow-motion equipment index, curtailment index, boundary over-limit index, and inverter response deviation index. The distribution network voltage control evaluation results are generated based on the comprehensive evaluation indicators, and the distribution network voltage control is adjusted for the next time based on the voltage over-limit indicators, slow-action equipment indicators, curtailment indicators, boundary over-limit indicators, and inverter response deviation indicators.

[0096] This embodiment acquires the topology operation data of the distribution network and constructs a voltage model and a voltage sensitivity matrix set, providing a precise mathematical foundation for subsequent voltage control. Based on this, the distribution network is divided into several zones according to the voltage model and voltage sensitivity matrix set, enabling independent coordinated control within each zone and avoiding the wide-area disturbances and computational burden caused by global unified control. Subsequently, a global optimization model and a zone-based collaborative optimization model are constructed at preset hourly and minute-scale control periods, respectively, realizing the coordinated control of slow-moving devices and inverters at different time scales. Layered coordination avoids the frequent operation and control conflicts of slow-moving devices caused by scheduling slow-moving devices and fast inverters on the same time scale. Finally, voltage control is performed based on the control plan of slow-moving devices and inverter control data on a preset second-level time scale control cycle. This allows the inverter to respond promptly to rapid voltage fluctuations on a second-level time scale, while the slow-moving devices adjust the voltage baseline on an hour-level time scale. Through the spatial partitioning coordination and time-level layered decoupling of the distribution network, mutual interference and control conflicts between different control resources are avoided, thereby improving the accuracy and stability of distribution network voltage control.

[0097] In summary, this invention provides a precise mathematical foundation for subsequent voltage control by acquiring the topology operation data of the distribution network and constructing a voltage model and a voltage sensitivity matrix set. Based on this, the distribution network is divided into several regions according to the voltage model and voltage sensitivity matrix set, enabling independent coordinated control within each region and avoiding the wide-area disturbances and computational burden caused by global unified control. Subsequently, a global optimization model and a regional collaborative optimization model are constructed at preset hourly and minute-scale control periods, respectively, enabling slow-moving devices and inverters to operate at different time scales. The hierarchical coordination avoids the frequent operation and control conflicts of slow-moving devices caused by scheduling slow-moving devices and fast inverters on the same time scale. Finally, voltage control is performed based on the control plan of slow-moving devices and the control data of inverters within a preset second-level time scale control cycle. This allows inverters to respond promptly to rapid voltage fluctuations on a second-level time scale, while slow-moving devices adjust the voltage baseline on an hourly time scale. Through the spatial partitioning coordination and time-level hierarchical decoupling of the distribution network, mutual interference and control conflicts between different control resources are avoided, thereby improving the accuracy and stability of distribution network voltage control.

[0098] The specific embodiments described above further illustrate the purpose, technical solution, and beneficial effects of the present invention. It should be understood that the above descriptions are merely specific embodiments of the present invention and are not intended to limit the scope of protection of the present invention. In particular, it should be noted that any modifications, equivalent substitutions, improvements, etc., made within the spirit and principles of the present invention should be included within the scope of protection of the present invention for those skilled in the art.

Claims

1. A voltage control method for a power distribution network, characterized in that, include: Obtain the topology operation data of the distribution network, and construct the voltage model and voltage sensitivity matrix set of the distribution network based on the topology operation data; Based on the voltage model and voltage sensitivity matrix set, the distribution network is divided into several regions to obtain the distribution network. Based on the voltage model and voltage sensitivity matrix set, a global optimization model of the distribution network is constructed in a preset hourly time scale control cycle. Based on the global optimization model, the slow-motion equipment control plan of the distribution network and the zonal adjustment data of each zone are obtained. Based on the voltage sensitivity matrix set and the partition adjustment data, a partition collaborative optimization model is constructed for each partition within a preset minute time scale control period; Inverter control data for each partition is determined based on the partitioned collaborative optimization model. Based on the slow-motion equipment control plan and inverter control data, the voltage of the distribution network is controlled within a preset second-level time scale control cycle.

2. The distribution network voltage control method as described in claim 1, characterized in that, The process of acquiring the topology operation data of the distribution network and constructing the voltage model and voltage sensitivity matrix set of the distribution network based on the topology operation data includes: Obtain the topology operation data of the distribution network, the topology operation data including: source node voltage values ​​of the distribution network; Using the source node voltage value as a reference boundary, a node voltage squared change vector of the distribution network is constructed according to a preset linearized power flow recursion algorithm; Construct the node net active power injection change vector, node net reactive power injection change vector, slow-action equipment control quantity change vector, and inverter reactive power control quantity change vector of the distribution network. Based on the node net active power injection change vector and the node voltage square change vector, a first sensitivity matrix is ​​constructed between the node voltage square change and the node net active power injection change of the distribution network. Based on the node net reactive power injection change vector and the node voltage square change vector, a second sensitivity matrix is ​​constructed between the node voltage square change and the node net reactive power injection change of the distribution network. Based on the control quantity change vector of the slow-action device and the square change vector of the node voltage, a third sensitivity matrix is ​​constructed between the square change of the node voltage and the control state change of the slow-action device in the distribution network. The voltage sensitivity matrix set of the distribution network is formed based on the first sensitivity matrix, the second sensitivity matrix, and the third sensitivity matrix; The first voltage model of the distribution network is constructed based on the voltage sensitivity matrix set, the node net active power injection change vector, the node net reactive power injection change vector, and the slow-acting equipment control quantity change vector. Based on the topology operation data, construct the node inverter mapping matrix between inverters and nodes in the distribution network; The first voltage model is transformed based on the node inverter mapping matrix and the inverter reactive power control change vector to obtain the second voltage model of the distribution network. The voltage model of the distribution network is formed based on the first voltage model and the second voltage model.

3. The distribution network voltage control method as described in claim 2, characterized in that, The distribution network is divided into several zones based on the voltage model and voltage sensitivity matrix set, including: Within each preset hourly timescale control cycle, the second sensitivity matrix is ​​transformed to obtain a fourth sensitivity matrix between the node voltage amplitude change and the node net reactive power injection change of the distribution network. Based on the node inverter mapping matrix and the second sensitivity matrix in the second voltage model, the influence intensity of the unit reactive power output change of any inverter in the distribution network on the voltage amplitude change of any node is determined, and the node inverter voltage coupling matrix is ​​constructed based on the influence intensity. The node inverter voltage coupling matrix is ​​normalized to obtain the normalized node inverter voltage coupling matrix. The similarity of the common influence between any two nodes on the same inverter is calculated based on the normalized node inverter voltage coupling matrix, and the real-time node coupling adjacency matrix of the distribution network is constructed based on the common influence similarity. Obtain the historical node coupling adjacency matrix and historical topology switch state data of the previous hourly timescale control cycle; Based on the real-time node coupling adjacency matrix, the historical node coupling adjacency matrix, and the historical topology switch status data, the distribution network is divided into several regions to obtain the distribution network.

4. The distribution network voltage control method as described in claim 3, characterized in that, The distribution network is divided into several zones based on the real-time node coupling adjacency matrix, the historical node coupling adjacency matrix, and the historical topology switch data, including: Based on the topology operation data, the real-time topology switch status data of the distribution network is determined, and based on the real-time topology switch status data and historical topology switch data, the degree of topology change of the distribution network is determined. The coupling variation degree of the distribution network is determined based on the historical node coupling adjacency matrix and the real-time node coupling adjacency matrix. When the coupling change degree is less than the first preset threshold and the topology change degree does not meet the preset topology change condition, several partitions of the previous hourly time scale control cycle are taken as several partitions of the distribution network in the current hourly time scale control cycle. When the coupling change degree is greater than or equal to the first preset threshold or the topology change degree meets the preset topology change condition, the normalized graph Laplace matrix of the distribution network is constructed based on the real-time node coupling adjacency matrix. The nodes are clustered based on a preset clustering algorithm and the normalized graph Laplace matrix to obtain several partitions of the power distribution network. Based on the topology operation data, several topology boundary nodes for each partition are determined; The cross-region control sensitivity between any node in each of the partitions and any inverter in other partitions is calculated based on the normalized node inverter voltage coupling matrix. Based on the cross-region control sensitivity, several control boundary nodes of each partition are determined. The topological boundary nodes and control boundary nodes of each partition are used as the boundary nodes of each partition to complete the regional division of the power distribution network.

5. The distribution network voltage control method as described in claim 2, characterized in that, The global optimization model of the distribution network under a preset hourly time scale control cycle is constructed based on the voltage model and voltage sensitivity matrix set. Based on the global optimization model, the slow-action equipment control plan of the distribution network and the zonal adjustment data of each zone are obtained, including: Within the hourly timescale control period, the net injected power prediction, network loss estimate, and distribution network node voltage target vector of the distribution network are obtained. Based on the net injected power prediction value, a node net injected power prediction change vector is constructed, and based on the topology operation data, the distribution network is constructed to obtain the slow-moving equipment state vector. Based on the slow-motion device state vector, the distribution network node voltage target vector, and the network loss estimate, a global optimization model of the distribution network is constructed for the hourly timescale control period. The first voltage model is transformed based on the predicted change vector of the node net injected power to obtain the voltage equality constraint of the global optimization model; The global optimization model is solved based on the voltage equality constraint, the preset voltage safety constraint, and the slow-action device action constraint to obtain the slow-action device control plan and the node hourly time scale voltage vector of the distribution network. The node voltage baseline vector of the distribution network is determined based on the node hourly time-scale voltage vector, and the partition voltage baseline vector of each partition is determined based on the node voltage baseline vector. The boundary node voltage offset budget for each partition is determined based on the partition voltage baseline vector and the preset voltage limit value of each partition. Obtain the reactive power output data of each inverter in each of the partitions, and determine the reactive power regulation budget of each partition based on the reactive power output data and the node inverter voltage coupling matrix; Obtain the active power output data of each inverter in each partition, and determine the active power reduction budget for each partition based on the active power output data; The partition voltage baseline vector, boundary node voltage offset budget, reactive power adjustment budget, and active power reduction budget of each partition are used as the partition adjustment data for each partition.

6. The distribution network voltage control method as described in claim 5, characterized in that, The system constructs a partition collaborative optimization model for each partition within a preset minute timescale control period based on the voltage sensitivity matrix set and partition adjustment data. Based on the partitioned collaborative optimization model, the inverter control data for each partition is determined, including: Within a minute-scale control cycle, a first partition sensitivity matrix is ​​determined based on the second sensitivity matrix to determine the node voltage amplitude and inverter reactive power regulation of each partition; a second partition sensitivity matrix is ​​determined based on the first sensitivity matrix to determine the node voltage amplitude and inverter active power reduction of each partition. Obtain the node active net injection disturbance data and node reactive net injection disturbance data of the distribution network, and construct the partition voltage disturbance vector of each partition based on the node active net injection disturbance data and node reactive net injection disturbance data. Construct the inverter reactive power adjustment vector and inverter active power reduction vector for each of the partitions. Based on the partition voltage baseline vector, partition voltage disturbance vector, first partition sensitivity matrix, second partition sensitivity matrix, inverter reactive power adjustment vector and inverter active power reduction vector, construct the partition node voltage prediction vector for each of the partitions. Based on the partition node voltage prediction vector, boundary node voltage offset budget, reactive power regulation budget, and active power reduction budget of each partition, a partition collaborative optimization model is constructed for each partition, and the inverter control data of each partition is determined based on the partition collaborative optimization model.

7. The distribution network voltage control method as described in claim 6, characterized in that, The process involves constructing a partitioned collaborative optimization model for each partition based on the partition node voltage prediction vector, boundary node voltage offset budget, reactive power regulation budget, and active power reduction budget. Based on this partitioned collaborative optimization model, inverter control data for each partition is determined, including: Obtain the historical inverter reactive power regulation vector for the previous minute-scale control cycle; Calculate the first difference vector between the inverter reactive power regulation vector and the historical inverter reactive power regulation vector, and calculate the second difference vector between the partition node voltage prediction vector and the partition voltage baseline vector; With the goal of minimizing the sum of the first difference vector, the second difference vector, and the inverter active power reduction vector, a partition collaborative optimization objective function is constructed for each of the partitions. Based on the boundary node voltage offset budget, reactive power regulation budget, and active power reduction budget, construct a set of partition collaborative optimization constraints for each partition; Based on the set of partitioned collaborative optimization constraints and the objective function of partitioned collaborative optimization, a partitioned collaborative optimization model is formed. The partitioned collaborative optimization model is solved to obtain the optimal reactive power adjustment vector and the optimal active power reduction vector of the inverter in each partition. Based on the inverter's optimal reactive power adjustment vector and the inverter's optimal active power reduction vector, the target active power output and target reactive power output of each inverter in each of the partitions are determined. Based on the target reactive power output and target active power output, determine the set of voltage reactive power curve parameters and voltage active power reduction curve parameters for each inverter in each of the partitions. Based on the target active power output, target reactive power output, voltage-reactive curve parameter set, and voltage-active power reduction curve parameter, and combined with the pre-acquired reactive power regulation rate of change limit and active power reduction rate of change limit for each inverter, the inverter control data for each of the aforementioned zones is determined.

8. The distribution network voltage control method as described in claim 7, characterized in that, The voltage control of the distribution network based on the slow-motion equipment control plan and inverter control data within a preset second-level time scale control cycle includes: Within the control cycle on the second-level time scale, the real-time voltage measurement values ​​of each inverter and the current control status of each slow-motion device are acquired; The initial reactive power output value of each inverter is determined based on the real-time voltage measurement value and the voltage reactive power curve parameter set. Obtain the current active power output value and rated capacity of each inverter, and determine the current reactive power output limit of each inverter based on the current active power output value and rated capacity; The historical reactive power output value of each inverter in the previous second-level time scale control cycle is obtained. Based on the current reactive power output limit, the historical reactive power output value and the reactive power regulation change rate limit, the initial reactive power output value is corrected to obtain the actual reactive power output value of each inverter. The initial active power output of each inverter is calculated based on the real-time voltage measurement value and the voltage active power reduction curve parameters. Obtain the historical active power output value of each inverter in the previous second-level time scale control cycle, and determine the actual active power output value of each inverter based on the active power reduction change rate limit and the historical active power output value. The average voltage of the feeder key nodes and the target voltage of the slow-action equipment of the distribution network are obtained. When the average voltage of the feeder key nodes and the target voltage of the slow-action equipment meet the preset control conditions of the slow-action equipment, the slow-action equipment of the distribution network is controlled based on the control plan. Based on the actual active power output and actual reactive power output of each inverter, the output of each inverter is controlled to complete the voltage control of the distribution network.

9. A power distribution network voltage control method as described in claim 8, characterized in that, After voltage control of the distribution network, the following is also included: Within the preset evaluation window, obtain the voltage measurement value after control of each node in the distribution network, the available active power after control of each inverter, and the actual number of actions of each slow-action device. The voltage over-limit index of the distribution network is calculated based on the voltage measurement value after control. Calculate the slow-action equipment index of the power distribution network based on the actual number of actions; Based on the controlled active power output value and the actual active power output value, the curtailment index of the distribution network can be determined. The boundary exceedance index of the distribution network is determined based on the boundary node voltage offset budget, the zone voltage baseline vector, and the voltage measurement after control. The inverter response deviation index of the distribution network is determined based on the actual reactive power output, actual active power output, target reactive power output, and target active power output of each inverter. A comprehensive evaluation index is generated based on the voltage over-limit index, slow-motion equipment index, curtailment index, boundary over-limit index, and inverter response deviation index. The distribution network voltage control evaluation results are generated based on the comprehensive evaluation indicators, and the distribution network voltage control is adjusted for the next time based on the voltage over-limit indicators, slow-action equipment indicators, curtailment indicators, boundary over-limit indicators, and inverter response deviation indicators.

10. A power distribution network voltage control system, characterized in that, include: Distribution network voltage modeling module, distribution network partitioning module, hourly time scale control module, minute time scale control module, and second-level time scale control module; The distribution network voltage modeling module is used to acquire the topology operation data of the distribution network, and to construct the voltage model and voltage sensitivity matrix set of the distribution network based on the topology operation data; The distribution network partitioning module is used to divide the distribution network into regions based on the voltage model and the voltage sensitivity matrix set, thereby obtaining several partitions of the distribution network. The hourly time-scale control module is used to construct a global optimization model of the distribution network in a preset hourly time-scale control cycle based on the voltage model and the voltage sensitivity matrix set, and to obtain the slow-motion equipment control plan of the distribution network and the zonal adjustment data of each zone based on the global optimization model. The minute-scale control module is used to construct a partition collaborative optimization model for each partition within a preset minute-scale control period based on the voltage sensitivity matrix set and partition adjustment data. Inverter control data for each partition is determined based on the partitioned collaborative optimization model. The second-level timescale control module is used to perform voltage control on the distribution network within a preset second-level timescale control cycle based on the slow-motion equipment control plan and inverter control data.

Images