Reactive voltage control method and device for photovoltaic power station cluster

Abstract

The application relates to the technical field of voltage control, and discloses a reactive power voltage control method and device for a photovoltaic power station cluster; the reactive power voltage control method is applied to a cluster coordination controller and specifically comprises the following steps: S11: real-time operation data of the photovoltaic power station cluster is acquired, the photovoltaic power station cluster is dynamically partitioned, and leading nodes of each control sub-region are determined, so that the entire cluster is divided into a plurality of sub-regions with the maximum modularity index of the cluster as the target; S12: for each control sub-region, a double-layer optimization model with the minimum voltage deviation of the leading node of the sub-region and the equalization of the reactive power output of each photovoltaic substation as the target is constructed. The application adds a setting value checking and redistribution link based on sensitivity priority after double-layer optimization solving, solves the problem that the adjustment capacity of each reactive power source is not fully utilized or is overloaded, and makes the utilization of each reactive power resource in the region more reasonable and sufficient.

Description

Technical Field

[0001] This invention relates to the technical field of voltage control, and more specifically, to a method and apparatus for reactive power voltage control of photovoltaic power plant clusters. Background Technology

[0002] With the continuous and rapid growth of photovoltaic installed capacity in my country and the comprehensive implementation of county-wide photovoltaic pilot projects, the penetration rate of distributed photovoltaic grid connection is increasing. The clustered development and grid connection of photovoltaic power stations has become an important trend in the development of new energy power systems. Due to the significant volatility and intermittency of photovoltaic power generation, the reactive power and voltage control of the power system after large-scale photovoltaic grid connection faces severe challenges. At present, there are relatively rich research results on reactive power and voltage control of photovoltaic power stations. Existing technologies are mainly divided into two categories: centralized control and decentralized control. The centralized control method collects the operating data of each photovoltaic substation to a unified control center, and the control center uniformly solves the global optimization problem and then issues control commands, which has the advantage of good global optimality. In addition, the control strategy based on cluster partitioning has also received widespread attention. Its core idea is to divide the large-scale photovoltaic system into several control sub-regions according to the electrical coupling relationship between photovoltaic nodes, and implement voltage control in each sub-region.

[0003] However, existing reactive power and voltage control technologies for photovoltaic power plant clusters have the following technical problems: (1) Existing cluster partitioning methods based on modularity index mainly consider the electrical coupling strength between nodes, neglecting the balance of reactive power regulation capacity of each photovoltaic substation. This can easily lead to significant differences in reactive power regulation capacity among the control sub-regions after partitioning. Some sub-regions cannot effectively respond to voltage regulation requirements due to insufficient regulation capacity, while other sub-regions waste resources due to excessive regulation capacity. (2) Although centralized control has global optimality, as the scale of photovoltaic power plant clusters continues to expand, centralized control needs to collect massive amounts of operational data, which leads to a sharp increase in communication burden and an exponential increase in computational complexity. Moreover, once the central controller fails, the entire cluster will lose control capability. While the existing distributed control has a lighter communication burden, it lacks effective coordination between sub-regions and is difficult to achieve cluster-level global optimization.

[0004] Based on the above problems, there is an urgent need for a reactive power and voltage control method and device for photovoltaic power plant clusters that can take into account both global optimization and efficient distributed solution. Summary of the Invention

[0005] The purpose of this invention is to provide a reactive power and voltage control method and device for photovoltaic power plant clusters. By introducing a reactive power regulation capacity balance weight to correct the electrical distance and then performing optimal modular partitioning, the imbalance phenomenon of insufficient regulation capacity failing to effectively respond to regulation needs and some sub-regions having excess capacity but being idle is avoided.

[0006] This invention is implemented as follows: a reactive power and voltage control method for photovoltaic power plant clusters, applied to a cluster coordination controller, specifically includes the following steps: S11: Obtain real-time operation data of the photovoltaic power station cluster, dynamically partition the photovoltaic power station cluster, and determine the dominant node of each control sub-region. With the goal of maximizing the modularity index of the cluster, the entire cluster is divided into several sub-regions. S12: For each control sub-region, a two-layer optimization model is constructed with the objectives of minimizing the voltage deviation of the dominant node in the sub-region and balancing the reactive power output of each photovoltaic substation. A distributed solution algorithm is used to solve the two-layer optimization model. S13: Based on the solution results, generate voltage regulation commands for each photovoltaic substation and reactive power setpoints for each reactive power source. Verify the setpoints according to the voltage-reactive power sensitivity matrix. If the setpoint of a reactive power source exceeds its reactive power regulation capability boundary, then all reactive power sources that exceed the boundary in that sub-region will be processed uniformly. After uniform processing, the voltage regulation commands and reactive power setpoints will be sent to each photovoltaic substation. S14: After receiving the voltage regulation command and reactive power setpoint, each photovoltaic substation uses the model predictive control algorithm to perform tracking control. At the same time, the model predictive control algorithm coordinates according to the actual response rate differences of each reactive power source. After each photovoltaic substation completes local control, it feeds back the actual voltage and reactive power output data to the cluster coordinating controller in real time. S15: After receiving real-time data from each photovoltaic substation, the cluster coordination controller calculates the deviation between the actual voltage at the cluster grid connection point and the voltage target value issued by the grid dispatch. If the absolute value of the deviation does not exceed the preset dead zone threshold, the current voltage is deemed qualified. If the deviation exceeds the dead zone threshold, a secondary adjustment at the cluster level is triggered.

[0007] Furthermore, in S11, real-time operating data of the photovoltaic power plant cluster is acquired, and the photovoltaic power plant cluster is dynamically partitioned, including: The real-time operating data includes the grid connection point voltage of each photovoltaic substation, the output active and reactive power of each photovoltaic power generation unit, the current reactive power output of the dynamic reactive power compensation device, and the target value of the cluster grid connection point voltage issued by the grid dispatch. The acquired data is filtered and denoised, and bad data is identified. After removing abnormal data, a valid running dataset is formed. The data is then preprocessed by aligning and normalizing according to the timestamp to obtain valid running data. Based on effective operating data, the reactive voltage sensitivity and electrical distance between each photovoltaic substation are calculated, and the ratio of the current adjustable reactive capacity of each substation to the maximum adjustable capacity is introduced as the balance weight to make weighted correction to the electrical distance.

[0008] Furthermore, with the goal of maximizing the modularity of the cluster, the entire cluster is divided into several sub-regions, including: The improved modularity index comprehensively considers the reactive power-voltage sensitivity, electrical distance, and reactive power regulation capacity balance among the photovoltaic substations, and divides the photovoltaic power station cluster into several control sub-regions with strong electrical coupling and similar reactive power regulation characteristics. Using the corrected electrical distance to construct an adjacency matrix, with the goal of maximizing the cluster modularity, the entire photovoltaic power station cluster is dynamically divided into several control sub-regions. The number of photovoltaic substations contained in each control sub-region is limited to 2 to 5. At the same time, within each control sub-region, the substations are sorted according to their sensitivity to changes in the region's voltage, and the substation with the highest sensitivity is selected as the dominant node of that region.

[0009] Furthermore, in S12, a two-level optimization model is constructed with the objectives of minimizing the voltage deviation of the dominant node in the sub-region and balancing the reactive power output of each photovoltaic substation, including: For each control sub-region, the upper part of the two-layer optimization model aims to minimize the global voltage deviation of the entire control sub-region. It takes the voltage setpoint of each photovoltaic substation as the decision variable and simultaneously considers the safety upper and lower limits of voltage amplitude, the limit of the change amplitude of setpoint in adjacent cycles, and the power factor constraint of the entire sub-region, so as to optimize the voltage setpoint trajectory of each substation from a global perspective. The lower part of the two-layer optimization model aims to minimize the reactive power output increment of all reactive power sources in each photovoltaic substation. It takes the reactive power adjustment of each photovoltaic power generation unit inverter and dynamic reactive power compensation device as decision variables, and considers the upper and lower limits of the output of each reactive power source, the ramp rate limit, and the active and reactive power capacity coupling limit of the inverter to achieve the minimization of the voltage deviation of the dominant node and the equalization of reactive power output of each substation.

[0010] Furthermore, a distributed solution algorithm is used to solve the two-level optimization model, including: The two-level optimization model is transformed into local subproblems for each control subregion. Each local subproblem contains decision variables within the subregion and boundary coupling variables related to adjacent subregions. Each control subregion solves its own local subproblem. During the solution process, the subregions only exchange voltage magnitude and phase angle information of the boundary nodes. In solving their respective local subproblems, the penalty coefficient is dynamically adjusted according to the ratio of the original residual to the dual residual in two adjacent iterations. Each sub-region independently judges whether the local original residual and dual residual have reached the convergence condition. The iteration is terminated after all sub-regions have converged, and the optimal voltage setpoint and reactive power output increment of each sub-region are output.

[0011] Furthermore, in S13, the setpoint is verified according to the voltage-reactive power sensitivity matrix. If the setpoint of a certain reactive power source exceeds its reactive power regulation capability boundary, then all reactive power sources that exceed the boundary in that sub-region are uniformly processed, including: If the set value of a certain reactive power source exceeds its reactive power regulation capacity boundary, the total excess of all reactive power sources that exceed the boundary in that sub-region is counted, and the excess is reduced proportionally according to the ratio of the remaining adjustable capacity of each reactive power source to the total remaining adjustable capacity, so that the excess part is fairly distributed to each reactive power source with remaining regulation capacity. If, after proportional reduction, there are still individual reactive power source settings that exceed the limit, the setting value of the reactive power source will be forcibly set to its boundary value, and the excess portion will be redistributed to other reactive power sources in the same sub-region that still have remaining regulation capacity in order of the regulation sensitivity of each reactive power source to the dominant node voltage from high to low, until the setting value of all reactive power sources meets their own regulation capacity constraints. If there is still an unallocated excess after all reactive power sources have reached their adjustment limits, this excess portion will be reported to the cluster coordination controller, which will then initiate cross-subregion reactive power support based on global information.

[0012] Furthermore, in S14, after each photovoltaic substation receives the voltage regulation command and reactive power setpoint, it executes tracking control using a model predictive control algorithm, including: Each photovoltaic substation is based on a pre-set prediction time domain and control time domain, with the goal of minimizing the deviation between the voltage prediction trajectory and the set value over multiple future control cycles. At the same time, it takes into account the minimization of reactive power output changes of each reactive power source and the stability of control input changes, and constructs a comprehensive performance evaluation index. In the comprehensive performance evaluation index, the priority relationship between voltage tracking accuracy, reactive power regulation cost and smoothness of action is adjusted by different weighting coefficients, and the actual response rate constraint of each reactive power source is added to match the difference in regulation capability of different types of reactive power sources. Within each control cycle, the model predictive control algorithm updates the initial state of the predictive model based on the measured state information at the current moment. Under the conditions of satisfying the upper and lower limits of the output of each reactive power source, the ramp rate, and the inverter capacity coupling, it solves for the optimal control sequence in the control time domain that minimizes the performance evaluation index. When the next control cycle arrives, the algorithm re-predicts and optimizes based on the latest measured state and executes the optimal control sequence of the previous control cycle.

[0013] Furthermore, in S15, if the deviation exceeds the dead zone threshold, a cluster-level secondary adjustment is triggered, including: When the deviation between the actual voltage at the cluster grid connection point and the voltage target value issued by the grid dispatch exceeds the preset dead zone threshold, the adjustment sensitivity of each control sub-region to the voltage at the cluster grid connection point is first calculated. Then, combined with the actual reactive power output change of each sub-region in the current cycle, the contribution responsibility of each sub-region to the current voltage deviation is quantified to obtain the deviation contribution of each sub-region. Then, the adjustment priority is determined according to the deviation contribution of each sub-region. The sub-region with the larger deviation contribution gets the higher adjustment priority. At the same time, the total reactive power compensation required for the entire cluster is calculated. Based on the current available reactive power regulation margin of each sub-region, the total compensation amount is allocated to each sub-region in a differentiated manner, so that the sub-region with a larger deviation contribution and a larger current regulation margin undertakes more compensation tasks. The allocated compensation amount is then sent to the corresponding control sub-region, triggering each sub-region to re-execute the optimization and control process until the cluster voltage deviation converges to within the dead zone threshold.

[0014] Furthermore, each sub-region is triggered to re-execute the optimization and control process until the cluster voltage deviation converges to within the dead zone threshold, including: After receiving the reactive power compensation amount from the cluster coordinating controller, each control sub-region adds the compensation amount to the total reactive power demand of its region and restarts its respective two-layer optimization solution and model predictive control tracking process. After each sub-region completes a new round of control, it feeds back the actual voltage and reactive power output data to the cluster coordination controller, recalculates the voltage deviation at the cluster grid connection point and makes a convergence judgment. If the deviation still exceeds the dead zone threshold, it continues to iterate and adjust. If the voltage deviation at the grid connection point of the cluster still fails to converge to within the dead zone threshold after a preset number of adjustments, it is determined that the current disturbance level has exceeded the limit of the cluster's own reactive power regulation capability, i.e., it is determined to be a cluster-level voltage runaway state. The cluster coordination controller immediately sends an alarm signal to the power grid dispatch center containing information on the deviation, the number of adjustments, and the current reactive power margin, and requests the intervention of external dispatch instructions. The dispatch center then decides on subsequent measures based on the overall network situation.

[0015] Compared with the prior art, the reactive power and voltage control method and device for photovoltaic power plant clusters provided by the present invention have the following beneficial effects: 1. By introducing a weighted average of reactive power regulation capacity to correct electrical distances before performing modular optimal partitioning, the imbalance between insufficient regulation capacity leading to ineffective response to control demands and some sub-regions having excess capacity but being idle is avoided. At the same time, after the two-layer optimization solution, a setpoint verification and redistribution step based on sensitivity priority is added, which solves the problem of insufficient utilization or overload of the regulation capacity of each reactive power source, making the utilization of reactive power resources in the region more reasonable and full. 2. By exchanging only boundary node information and adaptively adjusting the penalty coefficient during the distributed solution process, and allowing different sub-regions to adopt differentiated convergence tolerances according to their own scale, the solution efficiency under large-scale clusters is improved. Model predictive control coordinates with the inherent differences in the response rates of inverters and dynamic compensation devices, so that fast and slow adjustment resources can be used to their fullest potential. In the cluster-level secondary adjustment, by quantifying the contribution responsibility of each sub-region to the voltage deviation and distributing the compensation amount differently according to the adjustment margin, adjustment conflicts and resource waste caused by allocation are avoided.

[0016] A reactive power and voltage control device for a photovoltaic power plant cluster includes a storage device and a processor. The storage device stores a computer program, and the processor runs the computer program to enable the reactive power and voltage control device to perform the aforementioned reactive power and voltage control method. Attached Figure Description

[0017] Figure 1 This is a flowchart illustrating the reactive power and voltage control method for photovoltaic power plant clusters proposed in this invention. Figure 2 This is a flowchart illustrating the process of verifying the setpoint according to the voltage-reactive power sensitivity matrix in the reactive power voltage control method for photovoltaic power plant clusters proposed in this invention. Figure 3 This is a schematic diagram of the reactive voltage control device for a photovoltaic power plant cluster proposed in this invention. Detailed Implementation

[0018] To make the objectives, technical solutions, and advantages of this invention clearer, the invention will be further described in detail below with reference to the accompanying drawings and embodiments. It should be understood that the specific embodiments described herein are merely illustrative and not intended to limit the invention.

[0019] The implementation of the present invention will be described in detail below with reference to specific embodiments.

[0020] In the accompanying drawings of this embodiment, the same or similar reference numerals correspond to the same or similar components. In the description of this invention, it should be understood that if terms such as "upper," "lower," "left," and "right" indicate the orientation or positional relationship based on the orientation or positional relationship shown in the drawings, they are only for the convenience of describing this invention and simplifying the description, and do not indicate or imply that the device or element referred to must have a specific orientation, or be constructed and operated in a specific orientation. Therefore, the terms used to describe positional relationships in the drawings are only for illustrative purposes and should not be construed as limiting this invention. For those skilled in the art, the specific meaning of the above terms can be understood according to the specific circumstances.

[0021] Reference Figure 1-2 As shown, the reactive power and voltage control method for a photovoltaic power plant cluster, applied to the cluster coordination controller, specifically includes the following steps; S11: Obtain real-time operation data of the photovoltaic power station cluster, dynamically partition the photovoltaic power station cluster, and determine the dominant node of each control sub-region. With the goal of maximizing the modularity index of the cluster, the entire cluster is divided into several sub-regions. In this embodiment, real-time operating data includes the grid connection point voltage of each photovoltaic substation, the active and reactive power of each power generation unit, the reactive power output of the dynamic compensation device, and the cluster voltage target value issued by the dispatching authority. The data is filtered and denoised, bad data is identified, and normalized preprocessing is performed. At the same time, the reactive power-voltage sensitivity between any two substations is calculated and the electrical distance is constructed. The ratio of the current adjustable reactive power capacity of each substation to its maximum capacity is introduced as the balance weight to adjust the electrical distance, so that substations with similar reactive power adjustment capabilities are more likely to be assigned to the same area. Then, the adjacency matrix is ​​constructed using the adjusted electrical distance, and greedy optimization partitioning is performed with the goal of maximizing modularity. Each area is limited to 2 to 5 substations, and the substation with the highest voltage sensitivity in the area is selected as the dominant node. S12: For each control sub-region, a two-layer optimization model is constructed with the objectives of minimizing the voltage deviation of the dominant node in the sub-region and balancing the reactive power output of each photovoltaic substation. A distributed solution algorithm is used to solve the two-layer optimization model. In this embodiment, the upper-level model aims to minimize the global voltage deviation of the entire sub-region, uses the voltage setpoint of each substation as the decision variable, and considers the upper and lower limits of voltage, the rate of change of setpoint, and power factor constraints. The objective function for minimizing the global voltage deviation at the upper level is:

[0022] The lower-level model aims to minimize the power output increment of each reactive source, with the reactive power regulation of the inverter and dynamic compensation device as decision variables, while also considering the upper and lower limits of output, ramp rate and inverter capacity coupling constraints. Objective function for minimizing the incremental reactive power output of the lower layer:

[0023] During the solution process, the model is transformed into a local subproblem for each sub-region. Each sub-region only exchanges the voltage magnitude and phase angle information of the boundary nodes to protect operational privacy. In the iteration, the penalty coefficient is adaptively adjusted according to the ratio of the original residual to the dual residual until all sub-regions converge, and the optimal voltage setpoint and reactive power increment are output. The distributed solution formula uses the ADMM augmented Lagrangian function:

[0024] Where N: the total number of photovoltaic substations within the control sub-region; V k : The voltage amplitude at the grid connection point of the kth substation; V k ref : Voltage setting value of the kth substation; N(k): The set of substations electrically adjacent to the k-th substation; V m : The voltage amplitude at the grid connection point of the m-th substation adjacent to this substation; ω1, ω2: Upper-level weighting coefficients, ω1+ω2=1, which respectively adjust the importance of voltage deviation tracking and adjacent station voltage smoothing; M: Total number of reactive power sources within the substation; ΔQ g : The power output increment of the g-th reactive power source; ΔQ g prev : The power output increment of the g-th reactive power source in the previous cycle; α g ,β g The lower-level weighting coefficients adjust the adjustment cost and output stability, respectively. x r : The set of local decision variables for the r-th sub-region; z: Global coupling variable; λ: Lagrange multiplier; A r : Coupling matrix, describing the relationship between local variables and global variables; T: Represents vector transpose This represents the inner product of the Lagrange multiplier vector and the coupled residual vector; ρ: Penalty coefficient, which is adaptively adjusted during iteration based on the ratio of the original residual to the dual residual; f r (x r ): Represents the local optimization objective of the r-th sub-region itself; g(z): Represents the objective function related to the global coupling variables; S13: Based on the solution results, generate voltage regulation commands for each photovoltaic substation and reactive power setpoints for each reactive power source. Verify the setpoints according to the voltage-reactive power sensitivity matrix. If the setpoint of a reactive power source exceeds its reactive power regulation capability boundary, then all reactive power sources that exceed the boundary in that sub-region will be processed uniformly. After uniform processing, the voltage regulation commands and reactive power setpoints will be sent to each photovoltaic substation. In this embodiment, the specific processing procedure is as follows: First, the total excess of all out-of-bounds reactive power sources is counted, and then reduced proportionally according to the ratio of the remaining adjustable capacity of each reactive power source to the total remaining capacity, so that the excess portion is fairly distributed; if there is still an out-of-bounds after reduction, the source is forcibly set as the boundary value, and the excess portion is redistributed to other sources in the same region that still have adjustment capabilities in order of the sensitivity of each source to the voltage of the dominant node from high to low, until all sources meet the constraints; if there is still an unallocated excess, the cluster coordination controller is reported to start cross-sub-region support. The formula for proportionally reducing the remaining adjustable capacity of each reactive power source to the total remaining capacity is as follows:

[0025] Where N is the total number of reactive power sources in the sub-region, Q i Q is the initial setting value for the i-th reactive power source. i ′ is the set value after proportional reduction; Q represents the remaining adjustable capacity of the i-th source. i max Q i min The upper and lower limits of adjustment for the i-th source, ΔQ total The total excess amount from all outbound sources is calculated using the following formula:

[0026] If the amplitude still exceeds the limit after reduction, then force limiting and adjust according to sensitivity. The excess is redistributed from high to low, and the unallocated portion is reported to the cluster coordination controller. Among them, V pcc The dominant node voltage amplitude in this control sub-region; Q i : The actual reactive power output of the i-th reactive power source in this sub-region; : Partial derivative sign; S i : The voltage-reactive power sensitivity of the i-th reactive power source to the dominant node voltage. The larger the value, the stronger the voltage regulation effect of the source, which is used as the basis for the priority of redistribution.

[0027] S14: After receiving the voltage regulation command and reactive power setpoint, each photovoltaic substation uses the model predictive control algorithm to perform tracking control. At the same time, the model predictive control algorithm coordinates according to the actual response rate differences of each reactive power source. After each photovoltaic substation completes local control, it feeds back the actual voltage and reactive power output data to the cluster coordinating controller in real time. In this embodiment, during tracking control, a comprehensive performance index is constructed based on preset prediction and control time domains, taking into account the minimum voltage deviation, the minimum reactive power output change, and the smooth change of control input. The response rate constraints of each reactive power source are added to distinguish between fast and slow adjustment resources. In each control cycle, the algorithm updates the prediction model according to the current measured state and solves the optimal control sequence, but only executes the first instruction in the sequence. In the next cycle, the algorithm re-predicts and optimizes according to the new state to achieve rolling control. At the same time, the fast-responding dynamic compensation device is prioritized to deal with rapid fluctuations, while the inverter undertakes steady-state regulation. S15: After receiving real-time data from each photovoltaic substation, the cluster coordination controller calculates the deviation between the actual voltage at the cluster grid connection point and the voltage target value issued by the grid dispatch. If the absolute value of the deviation does not exceed the preset dead zone threshold, the current voltage is deemed qualified. If the deviation exceeds the dead zone threshold, a secondary adjustment at the cluster level is triggered. In this embodiment, during the secondary adjustment, the sensitivity of each sub-region to the voltage regulation of the cluster grid connection point is first calculated. Combined with the actual reactive power change in the current cycle of each sub-region, the deviation contribution of each sub-region is quantified. Then, the adjustment priority is determined according to the contribution, and the total reactive power compensation required by the cluster is calculated. Then, combined with the current available regulation margin of each sub-region, the compensation is allocated differentially, with sub-regions with larger contribution and larger margin bearing more compensation. After allocation, it is sent to each sub-region, and the optimization and control process is re-executed and iterated repeatedly until the deviation converges. If it still does not converge after a preset number of adjustments, it is determined to be a cluster-level voltage runaway. An alarm signal containing the deviation, number of adjustments, and margin information is immediately sent to the power grid dispatch center, and an external dispatch command is requested to intervene.

[0028] This technical solution corrects electrical distances by introducing a reactive power regulation capacity balance weight before performing modular optimal partitioning. This avoids the imbalance where some sub-regions have excess capacity but are idle due to insufficient regulation capacity. At the same time, after the two-layer optimization solution, a setpoint verification and redistribution step based on sensitivity priority is added to solve the problem of insufficient utilization or overload of the regulation capacity of each reactive power source, making the utilization of reactive power resources in the region more reasonable and full. Reference Figure 3As shown, the reactive power and voltage control device of the photovoltaic power station cluster includes a storage device and a processor. The storage device stores the computer program, and the processor runs the computer program to enable the reactive power and voltage control device to execute the aforementioned reactive power and voltage control method. By exchanging only boundary node information and adaptively adjusting the penalty coefficient during the distributed solution process, and allowing different sub-regions to adopt differentiated convergence tolerances according to their own scale, the solution efficiency under large-scale clusters is improved. Model predictive control coordinates with the inherent differences in the response rates of the inverter and the dynamic compensation device, so that fast and slow adjustment resources can be used to their fullest potential. In the cluster-level secondary adjustment, by quantifying the contribution responsibility of each sub-region to the voltage deviation and distributing the compensation amount differently according to the adjustment margin, adjustment conflicts and resource waste caused by allocation are avoided.

[0029] In this embodiment, the entire operation process can be automated by computer control. In each operation stage, sensors can be set up to provide signal feedback and ensure that the steps are performed sequentially. These are all conventional knowledge of current automation control, and will not be elaborated on in this embodiment.

[0030] The above description is only a preferred embodiment of the present invention and is not intended to limit the present invention. Any modifications, equivalent substitutions, and improvements made within the spirit and principles of the present invention should be included within the protection scope of the present invention.

Claims

1. A reactive power and voltage control method for photovoltaic power plant clusters, characterized in that, When applied to a cluster coordination controller, the specific steps include: S11: Obtain real-time operation data of the photovoltaic power station cluster, dynamically partition the photovoltaic power station cluster, and determine the dominant node of each control sub-region. With the goal of maximizing the modularity index of the cluster, the entire cluster is divided into several sub-regions. S12: For each control sub-region, a two-layer optimization model is constructed with the objectives of minimizing the voltage deviation of the dominant node in the sub-region and balancing the reactive power output of each photovoltaic substation. A distributed solution algorithm is used to solve the two-layer optimization model. S13: Based on the solution results, generate voltage regulation commands for each photovoltaic substation and reactive power setpoints for each reactive power source. Verify the setpoints according to the voltage-reactive power sensitivity matrix. If the setpoint of a reactive power source exceeds its reactive power regulation capability boundary, then all reactive power sources that exceed the boundary in that sub-region will be processed uniformly. After uniform processing, the voltage regulation commands and reactive power setpoints will be sent to each photovoltaic substation. S14: After receiving the voltage regulation command and reactive power setpoint, each photovoltaic substation uses the model predictive control algorithm to perform tracking control. At the same time, the model predictive control algorithm coordinates according to the actual response rate differences of each reactive power source. After each photovoltaic substation completes local control, it feeds back the actual voltage and reactive power output data to the cluster coordinating controller in real time. S15: After receiving real-time data from each photovoltaic substation, the cluster coordination controller calculates the deviation between the actual voltage at the cluster grid connection point and the voltage target value issued by the grid dispatch. If the absolute value of the deviation does not exceed the preset dead zone threshold, the current voltage is deemed qualified. If the deviation exceeds the dead zone threshold, a secondary adjustment at the cluster level is triggered.

2. The reactive power and voltage control method for a photovoltaic power station cluster as described in claim 1, characterized in that, In S11, real-time operational data of the photovoltaic power plant cluster is acquired, and the cluster is dynamically partitioned, including: The real-time operating data includes the grid connection point voltage of each photovoltaic substation, the output active and reactive power of each photovoltaic power generation unit, the current reactive power output of the dynamic reactive power compensation device, and the target value of the cluster grid connection point voltage issued by the grid dispatch. The acquired data is filtered and denoised, and bad data is identified. After removing abnormal data, a valid running dataset is formed. The data is then preprocessed by aligning and normalizing according to the timestamp to obtain valid running data. Based on effective operating data, the reactive voltage sensitivity and electrical distance between each photovoltaic substation are calculated, and the ratio of the current adjustable reactive capacity of each substation to the maximum adjustable capacity is introduced as the balance weight to make weighted correction to the electrical distance.

3. The reactive power and voltage control method for a photovoltaic power station cluster as described in claim 2, characterized in that, With the goal of maximizing the modularity of the cluster, the entire cluster is divided into several sub-regions, including: The improved modularity index comprehensively considers the reactive power-voltage sensitivity, electrical distance, and reactive power regulation capacity balance among the photovoltaic substations, and divides the photovoltaic power station cluster into several control sub-regions with strong electrical coupling and similar reactive power regulation characteristics. Using the corrected electrical distance to construct an adjacency matrix, with the goal of maximizing the cluster modularity, the entire photovoltaic power station cluster is dynamically divided into several control sub-regions. The number of photovoltaic substations contained in each control sub-region is limited to 2 to 5. At the same time, within each control sub-region, the substations are sorted according to their sensitivity to changes in the region's voltage, and the substation with the highest sensitivity is selected as the dominant node of that region.

4. The reactive power and voltage control method for a photovoltaic power station cluster as described in claim 3, characterized in that, In S12, a two-level optimization model is constructed with the objectives of minimizing the voltage deviation of the dominant node in the sub-region and balancing the reactive power output of each photovoltaic substation, including: For each control sub-region, the upper part of the two-layer optimization model aims to minimize the global voltage deviation of the entire control sub-region. It takes the voltage setpoint of each photovoltaic substation as the decision variable and simultaneously considers the safety upper and lower limits of voltage amplitude, the limit of the change amplitude of setpoint in adjacent cycles, and the power factor constraint of the entire sub-region, so as to optimize the voltage setpoint trajectory of each substation from a global perspective. The lower part of the two-layer optimization model aims to minimize the reactive power output increment of all reactive power sources in each photovoltaic substation. It takes the reactive power adjustment of each photovoltaic power generation unit inverter and dynamic reactive power compensation device as decision variables, and considers the upper and lower limits of the output of each reactive power source, the ramp rate limit, and the active and reactive power capacity coupling limit of the inverter to achieve the minimization of the voltage deviation of the dominant node and the equalization of reactive power output of each substation.

5. The reactive power and voltage control method for a photovoltaic power station cluster as described in claim 4, characterized in that, A distributed solution algorithm is used to solve the two-level optimization model, including: The two-level optimization model is transformed into local subproblems for each control subregion. Each local subproblem contains decision variables within the subregion and boundary coupling variables related to adjacent subregions. Each control subregion solves its own local subproblem. During the solution process, the subregions only exchange voltage magnitude and phase angle information of the boundary nodes. In solving their respective local subproblems, the penalty coefficient is dynamically adjusted according to the ratio of the original residual to the dual residual in two adjacent iterations. Each sub-region independently judges whether the local original residual and dual residual have reached the convergence condition. The iteration is terminated after all sub-regions have converged, and the optimal voltage setpoint and reactive power output increment of each sub-region are output.

6. The reactive power and voltage control method for a photovoltaic power station cluster as described in claim 5, characterized in that, In S13, the setpoint is verified according to the voltage-reactive power sensitivity matrix. If the setpoint of a certain reactive power source exceeds its reactive power regulation capability boundary, all reactive power sources that exceed the boundary in that sub-region are uniformly processed, including: If the set value of a certain reactive power source exceeds its reactive power regulation capacity boundary, the total excess of all reactive power sources that exceed the boundary in that sub-region is counted, and the excess is reduced proportionally according to the ratio of the remaining adjustable capacity of each reactive power source to the total remaining adjustable capacity, so that the excess part is fairly distributed to each reactive power source with remaining regulation capacity. If, after proportional reduction, there are still individual reactive power source settings that exceed the limit, the setting value of the reactive power source will be forcibly set to its boundary value, and the excess portion will be redistributed to other reactive power sources in the same sub-region that still have remaining regulation capacity in order of the regulation sensitivity of each reactive power source to the dominant node voltage from high to low, until the setting value of all reactive power sources meets their own regulation capacity constraints. If there is still an unallocated excess after all reactive power sources have reached their adjustment limits, this excess portion will be reported to the cluster coordination controller, which will then initiate cross-subregion reactive power support based on global information.

7. The reactive power and voltage control method for a photovoltaic power station cluster as described in claim 6, characterized in that, In S14, after receiving the voltage regulation command and reactive power setpoint, each photovoltaic substation executes tracking control using a model predictive control algorithm, including: Each photovoltaic substation is based on a pre-set prediction time domain and control time domain, with the goal of minimizing the deviation between the voltage prediction trajectory and the set value over multiple future control cycles. At the same time, it takes into account the minimization of reactive power output changes of each reactive power source and the stability of control input changes, and constructs a comprehensive performance evaluation index. In the comprehensive performance evaluation index, the priority relationship between voltage tracking accuracy, reactive power regulation cost and smoothness of action is adjusted by different weighting coefficients, and the actual response rate constraint of each reactive power source is added to match the difference in regulation capability of different types of reactive power sources. Within each control cycle, the model predictive control algorithm updates the initial state of the predictive model based on the measured state information at the current moment. Under the conditions of satisfying the upper and lower limits of the output of each reactive power source, the ramp rate, and the inverter capacity coupling, it solves for the optimal control sequence in the control time domain that minimizes the performance evaluation index. When the next control cycle arrives, the algorithm re-predicts and optimizes based on the latest measured state and executes the optimal control sequence of the previous control cycle.

8. The reactive power and voltage control method for a photovoltaic power station cluster as described in claim 7, characterized in that, In S15, if the deviation exceeds the dead zone threshold, a cluster-level secondary adjustment is triggered, including: When the deviation between the actual voltage at the cluster grid connection point and the voltage target value issued by the grid dispatch exceeds the preset dead zone threshold, the adjustment sensitivity of each control sub-region to the voltage at the cluster grid connection point is first calculated. Then, combined with the actual reactive power output change of each sub-region in the current cycle, the contribution responsibility of each sub-region to the current voltage deviation is quantified to obtain the deviation contribution of each sub-region. Then, the adjustment priority is determined according to the deviation contribution of each sub-region. The sub-region with the larger deviation contribution gets the higher adjustment priority. At the same time, the total reactive power compensation required for the entire cluster is calculated. Based on the current available reactive power regulation margin of each sub-region, the total compensation amount is allocated to each sub-region in a differentiated manner, so that the sub-region with a larger deviation contribution and a larger current regulation margin undertakes more compensation tasks. The allocated compensation amount is then sent to the corresponding control sub-region, triggering each sub-region to re-execute the optimization and control process until the cluster voltage deviation converges to within the dead zone threshold.

9. The reactive power and voltage control method for a photovoltaic power station cluster as described in claim 8, characterized in that, Trigger each sub-region to re-execute the optimization and control process until the cluster voltage deviation converges to within the dead zone threshold, including: After receiving the reactive power compensation amount from the cluster coordinating controller, each control sub-region adds the compensation amount to the total reactive power demand of its region and restarts its respective two-layer optimization solution and model predictive control tracking process. After each sub-region completes a new round of control, it feeds back the actual voltage and reactive power output data to the cluster coordination controller, recalculates the voltage deviation at the cluster grid connection point and makes a convergence judgment. If the deviation still exceeds the dead zone threshold, it continues to iterate and adjust. If the voltage deviation at the grid connection point of the cluster still fails to converge to within the dead zone threshold after a preset number of adjustments, it is determined that the current disturbance level has exceeded the limit of the cluster's own reactive power regulation capability, i.e., it is determined to be a cluster-level voltage runaway state. The cluster coordination controller immediately sends an alarm signal to the power grid dispatch center containing information on the deviation, the number of adjustments, and the current reactive power margin, and requests the intervention of external dispatch instructions. The dispatch center then decides on subsequent measures based on the overall network situation.

10. A reactive power and voltage control device for a photovoltaic power station cluster, characterized in that, The device includes a storage device and a processor, the storage device being used to store a computer program, and the processor running the computer program to cause the reactive voltage control device to perform the reactive voltage control method according to any one of claims 1-9.

Images