A wind-solar-storage station group reactive power collaborative optimization regulation method

Abstract

The application discloses a kind of wind and light storage station group reactive power collaborative optimization regulation methods, it is related to electric power system new energy regulation technical field, including acquisition station group each node voltage amplitude, injection reactive power and new energy power generation unit active power output etc. Real-time operating condition data, calculate public connection point reactive power shortage and divide overvoltage suppression, under-voltage support, steady-state optimization operation mode, through improved collaborative optimization algorithm, in combination with operating mode, voltage amplitude and active power output generate each controllable reactive source's collaborative regulation instruction sequence and issue execution.This method can match station group different operating conditions to carry out reactive power regulation, improve node voltage abnormal fluctuation, improve the cooperativity of each controllable reactive source regulation action, so that the reactive power distribution of station group is more balanced, and the reactive power shortage of public connection point is accurately compensated.

Description

Technical Field

[0001] This invention belongs to the field of power system new energy control technology, specifically a reactive power coordinated optimization control method for wind, solar and energy storage power station clusters. Background Technology

[0002] Conventional reactive power control technologies for wind, solar, and energy storage power plant clusters often rely on fixed control logic for reactive power regulation. During the control process, only basic operating parameters of the power plant cluster are collected, without specific calculations for reactive power deficits at common coupling points. Operating modes are often based on a single steady-state control mode, without differentiating different control conditions based on the actual voltage status of the power plant cluster. Reactive power control command generation often uses conventional optimization algorithms, which are calculated based on only a single operating parameter. The generated control commands are only formulated for a single controllable reactive power source, failing to form a control command system adapted to the entire power plant cluster.

[0003] In practical applications, this type of control method cannot match the different operating conditions of power plant clusters, including overvoltage, undervoltage, and steady-state operation. There are discrepancies between reactive power regulation and the actual operating needs of the power plant cluster, node voltage amplitudes are prone to abnormal fluctuations, and the adaptability between active power output and reactive power regulation of new energy generation units is poor. The regulation actions of each controllable reactive power source are independent, making it difficult to maintain a balanced reactive power distribution across the entire power plant cluster, and reactive power deficits at the point of common coupling cannot be accurately compensated.

[0004] During the operation of wind, solar and energy storage power station clusters, there is a lack of technical means to classify overvoltage suppression, undervoltage support and steady-state optimization operation modes based on the reactive power deficit at the common coupling point. It is impossible to combine the operation mode, node voltage amplitude and active power output of new energy sources to generate a coordinated control command sequence that is compatible with each controllable reactive power source. As a result, the coordinated regulation effect of reactive power in the power station cluster is difficult to achieve the expected results. Summary of the Invention

[0005] This invention aims to solve at least one of the technical problems existing in the prior art; Therefore, this invention proposes a method for coordinated optimization and control of reactive power in a wind-solar-storage storage station cluster, comprising: Real-time operating condition data of wind, solar and energy storage stations are collected. The real-time operating condition data includes the voltage amplitude, injected reactive power and active power output of each node in the station group. The reactive power deficit of the common connection point of the power station group is calculated based on the real-time operating condition data, and the current operating mode is determined based on the reactive power deficit. The operating modes include overvoltage suppression mode, undervoltage support mode and steady-state optimization mode. An improved collaborative optimization algorithm is invoked to generate a sequence of collaborative control instructions for each controllable reactive power source within the power station group, based on the current operating mode, the voltage amplitude, and the active power output. The coordinated control command sequence is sent to the corresponding controllable reactive power source execution unit to adjust the reactive power distribution of the power station group.

[0006] Further, the reactive power deficit at the common connection point of the power station group is calculated based on the real-time operating condition data, and the current operating mode is determined based on the reactive power deficit, including: Based on the reactive power and voltage control targets issued by the power grid dispatch center, obtain the voltage target value and allowable fluctuation range of the common connection point of the power station group; The difference between the measured voltage value at the corresponding common coupling point in the voltage amplitude and the target voltage value is calculated as the voltage deviation. By combining the equivalent impedance between the power station group and the main grid, the voltage deviation is converted into the reactive power compensation required to maintain voltage stability. The reactive power compensation is the reactive power deficit. The reactive power deficit is compared with preset overvoltage deficit thresholds and undervoltage deficit thresholds: If the reactive power deficit is greater than the overvoltage deficit threshold, then the current operating mode is determined to be the overvoltage suppression mode. If the reactive power deficit is less than the undervoltage deficit threshold, then the current operating mode is determined to be the undervoltage support mode. If the reactive power deficit is between the undervoltage deficit threshold and the overvoltage deficit threshold, then the current operating mode is determined to be the steady-state optimization mode.

[0007] Furthermore, the invocation of the improved collaborative optimization algorithm, based on the current operating mode, the voltage amplitude, and the active power output, generates a sequence of collaborative control instructions for each controllable reactive power source within the power station cluster, including: Read the status information of all controllable reactive power sources in the power station group. The status information includes the current generating capacity of the static var generator, the four-quadrant operating limit of the energy storage converter, and the reactive power regulation capability of the rotor-side converter of the doubly fed wind turbine. Based on the current operating mode, different objective functions and constraints are set for the collaborative optimization algorithm; Under the overvoltage suppression mode, the objective function is set to minimize the total active power loss of the power station group, and the constraints include that the voltage of each node does not exceed the upper limit and the upper limit of the reactive power absorption capacity of the controllable reactive power source. In the undervoltage support mode, the objective function is set to minimize the sum of squares of the total voltage deviation of the power station group, and the constraints include that the voltage of each node is not lower than the lower limit and the upper limit of the reactive power generation capacity of the controllable reactive power source. Under the steady-state optimization mode, the objective function is set to minimize the overall operating cost of the reactive power regulation equipment in the power station group, and the constraints include that the voltage of each node is within the acceptable range and the supply and demand of reactive power are balanced. Based on the set objective function and constraints, the improved collaborative optimization algorithm is used to solve the problem and calculate the set of reactive power output setpoints of each controllable reactive power source that optimizes the objective function under the constraints. Based on the set of reactive power output settings and the current actual output of each controllable reactive power source, a sequence of coordinated control instructions is generated, which includes equipment identification, target output value and adjustment rate.

[0008] Furthermore, the working principle of the improved collaborative optimization algorithm includes: The improved collaborative optimization algorithm is based on the multi-agent consensus theory, which models each controllable reactive power source in the power station cluster as an agent, and defines a local objective function and adjacency communication relationship for each agent. The local objective function of each agent consists of the adjustment cost function of its corresponding controllable reactive power source in the current operating mode and the output consistency function with neighboring agents; In each iteration of the calculation, each agent collects its own state information and the state information of neighboring agents obtained through the adjacency communication relationship. The state information includes the reactive power output setpoint. Each agent independently updates its own reactive power output setting value based on the collected state information using a predefined iterative update rule. The iterative update rule ensures that the reactive power output setting values ​​of all agents evolve toward a Pareto optimal solution that satisfies global constraints. Through multiple iterations, the reactive power output setpoints of all intelligent agents will reach a consistent and stable state. The set of output values ​​in this stable state is the set of reactive power output setpoints for each controllable reactive power source.

[0009] Furthermore, the iterative update rules for each agent are as follows: The difference between the current reactive power output setpoint of the agent and the average reactive power output setpoint of all neighboring agents in the previous iteration is calculated as the consistency error. Based on the objective function weights corresponding to the current operating mode, calculate the reactive power adjustment amount that only considers the gradient direction of the local objective function. The consistency error and the reactive power adjustment are weighted and fused to obtain the total adjustment for the current round; The total adjustment is applied to the reactive power setting value of the current agent to complete this iteration update; After the update is completed, the new reactive power output setting is broadcast to all neighboring agents through the adjacency communication relationship.

[0010] Furthermore, under the steady-state optimization mode, the method for calculating the comprehensive operating cost of the reactive power regulation equipment of the power station group includes: Obtain the unit adjustment cost parameters of various controllable reactive power sources, including the switching loss cost coefficient of static var generators, the cycle life loss cost coefficient of energy storage converters, and the additional copper loss cost coefficient of doubly-fed wind turbine generators. Calculate the absolute value of the reactive power change of each controllable reactive power source from its current actual output to its reactive power setpoint. Multiply the absolute value of the reactive power change of each controllable reactive power source by its corresponding unit adjustment cost parameter to obtain the individual adjustment cost of the controllable reactive power source. The sum of the individual adjustment costs of all controllable reactive power sources within the power station group is the comprehensive operating cost.

[0011] Furthermore, in the overvoltage suppression mode or the undervoltage support mode, the node voltage in the constraint condition does not exceed the upper limit or fall below the lower limit, and its specific implementation includes: Construct the node admittance matrix of the power station group, and based on the active power output and the current reactive power output setpoints, use the forward-backward substitution method to perform power flow calculations and predict the voltage values ​​of each node. The predicted voltage values ​​of each node are compared with the preset upper or lower voltage limits. If the predicted voltage value of a node exceeds the limit, the degree of the exceeding of the limit of the corresponding node voltage is converted into a penalty term for the adjustment requirement of the nearby controllable reactive power source, and the penalty term is added to the objective function of the collaborative optimization algorithm. In the next round of iterative solution, the collaborative optimization algorithm will tend to generate a set of reactive power output setpoints that can eliminate the node voltage overruns corresponding to the penalty term.

[0012] Furthermore, the coordinated control command sequence is sent to the corresponding controllable reactive power source execution unit to adjust the reactive power distribution of the power station group, including: The coordinated control instruction sequence is encoded according to a preset communication protocol to form a standardized downlink control message; The downlink control message is broadcast and sent through the data bus of the monitoring system of the station group; The local controllers of each controllable reactive power source within the power station group receive and parse the downlink control message, and extract the instruction content that matches their own equipment identifier; The local controller of each controllable reactive power source generates a specific pulse width modulation signal or switching signal according to the target output value and adjustment rate in the instruction content, drives the power device to operate, and realizes the adjustment of reactive power output. After the instruction is executed, the local controller of each controllable reactive power source feeds back the actual execution result to the monitoring system, forming a closed-loop control.

[0013] Further, based on the objective function weights corresponding to the current operating mode, the reactive power adjustment considering only the gradient direction of the local objective function is calculated. The steps include: Obtain the objective function weight coefficients preset for each agent, corresponding to the current operating mode. The objective function weight coefficients define the relative importance of the adjustment cost term and the global optimization objective in the local objective function. Within each iteration calculation cycle, the agent calculates the gradient vector of its local objective function with respect to its own reactive power output setpoint, based on its own local objective function. The calculated gradient vector is multiplied by a negative learning rate step size to determine an initial adjustment direction and magnitude aimed at locally optimizing its own objective function; Multiply the initial adjustment direction and magnitude by the objective function weight coefficient to obtain the weighted gradient adjustment amount. This adjustment amount is the reactive power adjustment amount that only considers the gradient direction of its own local objective function.

[0014] Furthermore, the construction of the node admittance matrix of the power station group, and the prediction of the voltage value of each node by using the forward-backward substitution method based on the active power output and the current reactive power output setpoints, includes: The node admittance matrix is ​​calculated based on the topology of the power station group, line parameters, and transformer parameters. The active power output is used as the known quantity of the active power injected into each node. The set of reactive power output settings obtained by the improved collaborative optimization algorithm in the current iteration is used as the known quantity of the injected reactive power of each controllable reactive power source corresponding node. The reactive power of uncontrollable nodes or load nodes is set as a given value or an estimated value. Set the root node as the common connection point of the power station group, and set its voltage amplitude and phase angle as the reference values; Starting from the terminal node directly connected to the root node, calculate the current of each branch forward based on the node admittance matrix and the voltage of the adjacent parent node. The voltage of each node is updated by back-substituting from the last node to the root node based on the calculated branch current and line impedance. Repeat the forward calculation and the back-up update steps until the voltage difference between two iterations of each node is less than the preset convergence threshold. Use the voltage value of each node obtained from the last iteration as the predicted voltage value of each node.

[0015] Compared with the prior art, the beneficial effects of the present invention are: Based on the reactive power deficit at the common coupling point (CCP) of the wind, solar, and energy storage power plant cluster, overvoltage suppression, undervoltage support, and steady-state optimization modes are determined. This allows for the determination of control directions tailored to different voltage operating states within the cluster. The reactive power regulation's operating condition matching aligns with the real-time operating status of the cluster, controlling abnormal fluctuations in node voltage amplitude. The compensation direction for reactive power deficit at the CCP remains consistent with the actual needs of the cluster, and reactive power regulation actions under different operating conditions can accurately correspond to abnormal voltage states. The adaptability of active power output and reactive power regulation between new energy generation units is improved, voltage deviation during cluster operation is mitigated, and the targeted reactive power regulation is tailored to real-time operating condition changes within the cluster. The reactive power regulation logic under different control modes can independently adapt to corresponding operating condition requirements.

[0016] An improved collaborative optimization algorithm is invoked to generate a sequence of controllable reactive power source coordinated control commands by combining the operating mode, voltage amplitude of each node, and active power output of the new energy power generation unit. The controllable reactive power source adjustment commands are mutually adapted, and there are no conflicts in reactive power adjustment actions, resulting in a more stable reactive power distribution within the power plant cluster. After each controllable reactive power source executes the control commands, the overall reactive power distribution of the power plant cluster remains stable, the reactive power deviation at the point of common coupling remains within a small range, and the node voltage amplitude can be kept within a reasonable operating range. The coordinated control command sequence can simultaneously adapt to the adjustment needs of multiple controllable reactive power sources, and the execution process of reactive power control is more in line with the overall operating logic of the power plant cluster, continuously improving the coordination of reactive power adjustment. Attached Figure Description

[0017] Figure 1 This is a state diagram of a reactive power coordinated optimization and control method for a wind-solar-storage storage station group as described in this invention. Figure 2 This is a flowchart for determining the current operating mode based on reactive power deficit comparison; Figure 3 A flowchart illustrating the work done on the improved collaborative optimization algorithm. Detailed Implementation

[0018] The technical solution of the present invention will be clearly and completely described below with reference to the embodiments. 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.

[0019] This invention provides a method for coordinated optimization and control of reactive power in a wind-solar-storage storage station cluster, the specific method including: See Figure 1The system collects real-time operating data of the wind, solar, and energy storage power storage cluster, including voltage amplitude, injected reactive power, and active power output of the renewable energy generation units at each node within the cluster. Based on this data, the reactive power deficit at the common connection point (CCP) of the cluster is calculated, and the current operating mode is determined. These modes include overvoltage suppression, undervoltage support, and steady-state optimization. An improved collaborative optimization algorithm is then applied. This algorithm, based on the determined operating mode, node voltage amplitude, and active power output, performs collaborative optimization calculations to generate a coordinated control command sequence for all controllable reactive power sources within the cluster. This sequence is then distributed to the corresponding controllable reactive power source execution units, which drive equipment actions to optimize the overall reactive power distribution of the cluster.

[0020] In one embodiment of the present invention, see [reference] Figure 2 Based on the reactive power and voltage control targets issued by the power grid dispatch center, the target voltage value and its allowable fluctuation range at the point of common coupling (PCC) of the power substation group are first obtained. Then, the difference between the real-time measured PCC voltage value and the aforementioned target voltage value is calculated; this difference is the voltage deviation. Combining the equivalent impedance between the power substation group and the main grid, the calculated voltage deviation is converted into the reactive power compensation required to maintain PCC voltage stability; this reactive power compensation is defined as the reactive power deficit. The calculated reactive power deficit is compared with preset overvoltage deficit thresholds and undervoltage deficit thresholds. If the reactive power deficit is greater than the overvoltage deficit threshold, the current operating mode is determined to be overvoltage suppression mode. If the reactive power deficit is less than the undervoltage deficit threshold, the current operating mode is determined to be undervoltage support mode. If the reactive power deficit is between the undervoltage deficit threshold and the overvoltage deficit threshold, the current operating mode is determined to be steady-state optimization mode.

[0021] In specific implementation, the method for calculating the reactive power deficit at the common connection point of the power station group and determining the current operating mode based on the reactive power deficit is as follows: Based on the reactive voltage control target issued by the power grid dispatch center, the voltage target value of the common connection point of the power station group is obtained. and allowable fluctuation range From the collected real-time operating condition data, read the measured voltage value of the corresponding common coupling point. Calculate the measured voltage value With voltage target value The difference is used to obtain the voltage deviation. ,Right now By combining the equivalent impedance between the power station cluster and the main grid, the voltage deviation is... This is converted into the reactive power compensation required to maintain voltage stability; this reactive power compensation is the reactive power deficit. .

[0022] In some embodiments, voltage deviation Converted to reactive power deficit This is achieved through the following relationship: in: Indicates the rated voltage of the point of common coupling. This represents the equivalent reactance between the power station group and the main grid. (Equivalent reactance) The equivalent reactance is obtained through offline calculation or online identification. Offline calculation is based on the electrical wiring diagram and equipment parameters of the substation group, calculating the equivalent impedance from the point of common coupling (PCC) towards the main grid. Online identification is based on historical operating data, estimating the equivalent reactance by analyzing the relationship between PCC voltage changes and reactive power injection changes. .

[0023] Optionally, after obtaining the voltage deviation, the reactive power deficit can be... Compared with the preset overvoltage deficit threshold Undervoltage threshold Compare. Overvoltage deficit threshold. It is a positive number, the undervoltage threshold. It is a negative number. If there is a reactive power deficit... Greater than the overvoltage deficit threshold If so, the current operating mode is determined to be overvoltage suppression mode. If there is a reactive power deficit... Less than the undervoltage deficit threshold If so, the current operating mode is determined to be undervoltage support mode. If there is a reactive power deficit... Given the undervoltage threshold With overvoltage deficiency threshold If the current operating mode is between these values, then the current operating mode is determined to be the steady-state optimization mode.

[0024] Understandable, allowable fluctuation range Defined voltage target value A suitable range nearby. When the measured voltage value... Falling within the allowable fluctuation range Within this time, voltage deviation The absolute value is small, and the reactive power deficit obtained after conversion is small. The absolute value is also relatively small, usually falling within the undervoltage deficit threshold. With overvoltage deficiency threshold Between these values, a steady-state optimization mode is triggered. When the measured voltage value... Exceeding the upper limit of the allowable fluctuation range When this occurs, it indicates a risk of overvoltage, and the voltage deviation at this time... If the value is positive and relatively large, the reactive power deficit after conversion is significant. A positive value that is large may exceed the overvoltage deficit threshold. This triggers the overvoltage suppression mode. When the measured voltage value... Below the lower limit of the allowable fluctuation range When this occurs, it indicates a risk of undervoltage, and the voltage deviation at this time... If the value is negative and the absolute value is large, the reactive power deficit after conversion is significant. If the value is negative and the absolute value is large, it may be below the undervoltage deficit threshold. This triggers the undervoltage support mode.

[0025] In practical implementation, the overvoltage deficit threshold and undervoltage threshold The setting needs to consider the requirements for safe operation of the power grid and the regulation capability of the equipment. Overvoltage deficit threshold. The setting corresponds to the critical point at which the system needs to significantly absorb reactive power to suppress voltage rise, the undervoltage deficit threshold. The settings correspond to the critical point at which the system needs to significantly generate reactive power to support voltage. The specific values ​​of these two thresholds are set through static security analysis of the power grid or based on operational experience, and can be adjusted according to dispatch instructions. The entire judgment process is automatic and continuous. The three steps of collecting real-time operating condition data, calculating reactive power deficit, and determining the operating mode are performed cyclically according to a preset control cycle. The length of the control cycle is determined based on the dynamic characteristics of the power grid and the performance of the communication system. The result of the operating mode determination directly determines the objective function and constraints adopted by the subsequent collaborative optimization algorithm. Overvoltage suppression mode, undervoltage support mode, and steady-state optimization mode correspond to different control priority objectives and safety boundaries, ensuring that the control strategy matches the current power grid state.

[0026] In one embodiment of the present invention, when the improved collaborative optimization algorithm is invoked to generate the instruction sequence, the real-time status information of all controllable reactive power sources within the power plant cluster is first read. This status information includes the current generating capacity of the static var generator (SGF), the four-quadrant operating limits of the energy storage converter, and the reactive power regulation capability of the rotor-side converter of the doubly-fed induction generator (DFIG). Next, based on the determined current operating mode, different objective functions and constraints are set for the collaborative optimization algorithm. In overvoltage suppression mode, the objective function is set to minimize the total active power loss of the power plant cluster, and the constraints include ensuring that the voltage of each node does not exceed the upper limit and the upper limit of the reactive power absorption capacity of the controllable reactive power sources. In undervoltage support mode, the objective function is set to minimize the sum of squares of the total voltage deviation of the power plant cluster, and the constraints include ensuring that the voltage of each node does not fall below the lower limit and the upper limit of the reactive power generation capacity of the controllable reactive power sources. In steady-state optimization mode, the objective function is set to minimize the comprehensive operating cost of the reactive power regulation equipment in the power plant cluster, and the constraints include ensuring that the voltage of each node is within the acceptable range and that the supply and demand of reactive power are balanced. The comprehensive operating cost of the reactive power regulation equipment in the power station group is calculated using the following method: The unit regulation cost parameters of various controllable reactive power sources are obtained, including the switching loss cost coefficient of the static var generator, the cycle life depreciation cost coefficient of the energy storage converter, and the additional copper loss cost coefficient of the doubly-fed induction generator (DFIG). The absolute value of the reactive power change of each controllable reactive power source from its current actual output to its reactive power output setpoint is calculated. The absolute value of the reactive power change of each controllable reactive power source is multiplied by its corresponding unit regulation cost parameter to obtain the individual regulation cost of that controllable reactive power source. The sum of the individual regulation costs of all controllable reactive power sources in the power station group is the comprehensive operating cost. Based on the set objective function and constraints, the improved collaborative optimization algorithm is used to solve the problem, calculating the set of reactive power output setpoints for each controllable reactive power source that optimizes the objective function while satisfying the constraints. Finally, based on the set of reactive power output setpoints and the current actual output of each controllable reactive power source, a collaborative control command sequence containing equipment identification, target output value, and regulation rate is generated.

[0027] In practical implementation, the specific methods for generating the coordinated control command sequence for each controllable reactive power source within the power station cluster and calculating the comprehensive action cost under steady-state optimization mode are as follows. When invoking the improved coordinated optimization algorithm, the real-time status information of all controllable reactive power sources within the power station cluster is first read. The status information includes the current generating capacity of the static var generator, the four-quadrant operating limits of the energy storage converter, and the reactive power regulation capability of the rotor-side converter of the doubly-fed induction generator (DFIG). This status information is acquired through the real-time data acquisition unit of the power station monitoring system and transmitted to the optimization calculation module in the form of data frames.

[0028] In some embodiments, differentiated objective functions and constraints are set for the improved collaborative optimization algorithm based on the current operating mode determined in real time. In overvoltage suppression mode, the objective function is set to minimize the total active power loss of the power station group. Constraints include ensuring that the voltage of each node does not exceed the upper limit and the upper limit of the reactive power absorption capacity of controllable reactive power sources. The upper limit of reactive power absorption capacity is determined by the maximum capacitive reactive power output capacity of equipment such as static var generators and energy storage converters. In undervoltage support mode, the objective function is set to minimize the sum of squares of the total voltage deviation of the power station group. Constraints include ensuring that the voltage of each node does not fall below the lower limit and the upper limit of the reactive power generation capacity of controllable reactive power sources. The upper limit of reactive power generation capacity is determined by the maximum inductive reactive power output capacity of equipment such as static var generators and doubly-fed induction generator rotor-side converters. In steady-state optimization mode, the objective function is set to minimize the comprehensive operating cost of the reactive power regulation equipment in the power station group. Constraints include ensuring that the voltage of each node is within the acceptable range and that reactive power supply and demand are balanced. The acceptable voltage range is specified by the power grid operation regulations.

[0029] Optionally, in steady-state optimization mode, a comprehensive operating cost calculation method for the reactive power regulation equipment of the power station group is activated. Pre-set unit regulation cost parameters for various controllable reactive power sources are obtained, including the switching loss cost coefficient of the static var generator (SVM), the cycle life depreciation cost coefficient of the energy storage converter, and the additional copper loss cost coefficient of the doubly-fed induction generator (DFIG) wind turbine. The switching loss cost coefficient is related to the switching frequency and on-state loss of the power devices in the SVM; the cycle life depreciation cost coefficient is related to the charge / discharge depth and cycle number model of the battery connected to the energy storage converter; and the additional copper loss cost coefficient is related to the additional losses caused by the rotor winding current of the DFIG wind turbine. For each controllable reactive power source, the absolute value of the reactive power change from the current actual reactive power output to the reactive power output setpoint calculated by the improved collaborative optimization algorithm is calculated.

[0030] It can be understood that the overall operating cost is a quantitative reflection of the economic efficiency of regulating all controllable reactive power sources within the power station cluster. The individual regulation cost of each controllable reactive power source is obtained by multiplying the absolute value of its reactive power change by the unit regulation cost parameter corresponding to that source. The sum of the individual regulation costs of all controllable reactive power sources is the objective function of the steady-state optimization model: the overall operating cost of the reactive power regulation equipment in the power station cluster. The calculation can be expressed as: in: This represents the total number of controllable reactive power sources within the power station group. Indicates the first The unit adjustment cost parameter of a controllable reactive power source Indicates the first The absolute value of the reactive power change of a controllable reactive power source. This formula allows the optimization algorithm to prioritize the use of reactive power sources with low adjustment costs, under the premise of satisfying voltage and power balance constraints, thereby achieving a more economical redistribution of reactive power.

[0031] In practical implementation, an improved collaborative optimization algorithm is used to solve the problem based on the set objective function and constraints. The improved collaborative optimization algorithm performs iterative calculations under the set objective function and constraints, ultimately outputting a set of reactive power output setpoints for each controllable reactive power source that optimizes the objective function value while satisfying all constraints. Based on the calculated set of reactive power output setpoints, and compared with the current actual output of each controllable reactive power source obtained in real time from the monitoring system, a final collaborative control command sequence is generated. The collaborative control command sequence is a structured data list. Each command in the list contains three key fields: device identifier, used to uniquely identify a controllable reactive power source execution unit within the power station group; target output value, i.e., the reactive power setpoint that the device needs to achieve; and adjustment rate, used to guide the device controller to smoothly adjust the output at a specified rate of change to avoid impacting the power grid.

[0032] In one embodiment of the present invention, see [reference] Figure 3The improved collaborative optimization algorithm, based on multi-agent consensus theory, models each controllable reactive power source within the power station cluster as an agent, defining a local objective function and adjacency communication relationship for each agent. Each agent's local objective function consists of its corresponding controllable reactive power source's adjustment cost function under the current operating mode and its output consensus function with neighboring agents. In each iteration, each agent collects its own state information and that of neighboring agents obtained through the adjacency communication relationship. This collected state information includes the reactive power output setpoint. Based on the collected state information, each agent independently updates its own reactive power output setpoint using a predefined iterative update rule. This iterative update rule ensures that the reactive power output setpoints of all agents evolve towards a Pareto optimal solution that satisfies global constraints. The specific steps of the iterative update rule are: calculating the difference between the current reactive power output setpoint of the agent and the average reactive power output setpoint of all neighboring agents in the previous iteration; this difference serves as the consensus error. For each agent, pre-defined objective function weight coefficients corresponding to the current operating mode are obtained. These weight coefficients define the relative importance of the adjustment cost term and the global optimization objective in the local objective function. Within each iteration cycle, the agent calculates the gradient vector of its local objective function with respect to its own reactive power output setting. The calculated gradient vector is multiplied by a negative learning rate step to determine an initial adjustment direction and magnitude aimed at locally optimizing its own objective function. This initial adjustment direction and magnitude are multiplied by the objective function weight coefficients to obtain a weighted gradient adjustment amount, which is the reactive power output adjustment amount considering only the gradient direction of its own local objective function. The consistency error and the reactive power output adjustment amount are weighted and fused to obtain the total adjustment amount for the current round. This total adjustment amount is applied to the current agent's reactive power output setting to complete the iteration update. After the update, the new reactive power output setting is broadcast to all neighboring agents through the adjacency communication relationship. Through multiple iterations, the reactive power output setpoints of all intelligent agents will reach a consistent and stable state. The set of output values ​​in this stable state is the set of reactive power output setpoints for each controllable reactive power source.

[0033] In practical implementation, the improved cooperative optimization algorithm is constructed and applied based on the multi-agent consensus theory. This algorithm models each controllable reactive power source within the power plant cluster as an agent with independent computational and communication capabilities. A local objective function and adjacency communication relationship are defined for each agent. The local objective function is composed of the adjustment cost function of the corresponding controllable reactive power source under the current operating mode and the output consistency function with neighboring agents. The adjacency communication relationship defines the network topology that allows information exchange between agents, typically based on the electrical connection structure or communication network structure of the power plant cluster.

[0034] In some embodiments, at the start of each iteration, each agent obtains its own state information through its internal state record and collects the state information of all neighboring agents defined by adjacency communication relationships through a communication link. The collected state information includes the reactive power output setpoints of the agent itself and its neighboring agents. Based on all the collected state information, each agent independently updates its own reactive power output setpoint using a predefined iterative update rule. The design of the iterative update rule ensures that the reactive power output setpoints of all agents can collaboratively evolve towards a Pareto optimal solution that satisfies global constraints during the iteration process.

[0035] Optionally, the iterative update rules for each agent are executed according to specific steps. The difference between the current agent's own reactive power output setpoint and the average reactive power output setpoint of all neighboring agents in the previous iteration is calculated. This difference serves as the consistency error, which drives the output setpoints of the agent group to tend towards consistency. Pre-set objective function weight coefficients corresponding to the current operating mode are obtained for each agent. These weight coefficients define the relative importance of the adjustment cost term and the global optimization objective in the local objective function. The values ​​of these weight coefficients differ under different operating modes; for example, in steady-state optimization mode, more emphasis is placed on adjustment costs, while in emergency voltage mode, more emphasis is placed on voltage safety objectives.

[0036] It can be understood that within each iteration, the agent calculates the gradient vector of its local objective function with respect to its own reactive power output setpoint. This gradient vector is multiplied by a negative learning rate step to determine an initial adjustment direction and magnitude aimed at locally optimizing its own objective function. The initial adjustment direction and magnitude are then multiplied by the objective function weight coefficients to obtain a weighted gradient adjustment, which represents the reactive power output adjustment considering only the gradient direction of its own local objective function. The consistency error is then weighted and fused with the reactive power output adjustment to obtain the total adjustment for the current iteration. This total adjustment is applied to the agent's current reactive power output setpoint, completing the iteration update. After the update, the agent broadcasts the new reactive power output setpoint to all neighboring agents via adjacency communication.

[0037] In practical implementation, the agent's state update process can be described mathematically. (Agent) In the reactive power setpoint during the next iteration Calculated according to the following update rules: in: Represents intelligent agents In the The reactive power output setting value at the next iteration; Represents intelligent agents A collection of neighboring intelligent agents; Indicates the number of neighboring intelligent agents; Represents the gain coefficient of the consistency term; Representing neighboring intelligent agents In the The reactive power output setting value at the next iteration; Represents intelligent agents The objective function weight coefficients; Indicates the learning rate step size; Represents intelligent agents Local objective function exist The gradient at the point; the first part inside the parentheses on the right side of the formula corresponds to the consistency error term, and the second part corresponds to the reactive power adjustment considering the gradient of the local objective function.

[0038] In some embodiments, the iterative update process of multiple agents can be displayed round by round. The table below shows the changes in the reactive power setpoint of each agent in four consecutive iterations of a simplified system with three agents under an improved cooperative optimization algorithm, with the final setpoints converging. See Table 1: Table 1: Examples of Multi-Agent Iterative Update Process Optionally, through multiple iterations, the reactive power output setpoints of all agents will reach a consistent stability. The set of output values ​​in this stable state is the set of reactive power output setpoints for each controllable reactive power source that satisfies the requirements of the global optimization problem. The entire iterative process is distributed, with each agent relying only on local information and limited neighbor communication to update its own state. The improved cooperative optimization algorithm avoids the computational bottlenecks and single-point-of-failure risks that may exist in centralized optimization through this distributed cooperative mechanism.

[0039] In one embodiment of the present invention, under overvoltage suppression mode or undervoltage support mode, the voltage of each node in the constraint condition does not exceed the upper limit or fall below the lower limit. This is specifically implemented through the following steps: Constructing a node admittance matrix for the power station group, the node admittance matrix is ​​calculated based on the topology, line parameters, and transformer parameters of the power station group. Using the active power output as a known quantity of injected active power for each node, and using the set of reactive power output setpoints obtained by the improved collaborative optimization algorithm in the current iteration as a known quantity of injected reactive power for the corresponding node of each controllable reactive power source, the reactive power of uncontrollable nodes or load nodes is set as a given value or estimated value. Setting the root node as the common connection point of the power station group, and setting its voltage amplitude and phase angle as reference values. Starting from the end node directly connected to the root node, calculating the current of each branch based on the node admittance matrix and the voltage of the adjacent parent node. Back-substituting from the end node to the root node, updating the voltage of each node based on the calculated branch current and line impedance. Repeat the forward calculation and back-update steps until the voltage difference between two iterations of each node is less than a preset convergence threshold. Use the node voltage value obtained in the last iteration as the predicted node voltage value. Based on the active power output and the current reactive power output setpoints, perform power flow calculations using the forward-backdate method to predict the voltage value of each node. Compare the predicted node voltage values ​​with preset upper or lower voltage limits. If a node's predicted voltage value exceeds the limit, convert the degree of voltage exceedance into a penalty term for adjusting the demand of nearby controllable reactive power sources, and add this penalty term to the objective function of the collaborative optimization algorithm. In the next iteration, the collaborative optimization algorithm will tend to generate a set of reactive power output setpoints that can eliminate the node voltage exceedance corresponding to the penalty term.

[0040] In practical implementation, under overvoltage suppression mode or undervoltage support mode, the specific process of ensuring that the voltage of each node does not exceed the limit under the constraints is as follows: Constructing the node admittance matrix of the power station group is the foundation for power flow analysis and voltage prediction. The node admittance matrix is ​​calculated based on the topology, line parameters, and transformer parameters of the power station group. The topology is determined by the real-time status of circuit breakers and disconnectors collected by the monitoring system. Line parameters include the resistance and reactance values ​​of overhead lines or cables, and transformer parameters include the rated turns ratio and short-circuit impedance percentage. Based on these power grid component parameters, the mutual admittance and self-admittance between each node in the network are calculated through circuit theory, ultimately forming the node admittance matrix describing the electrical connection relationship of the entire power station group network. The node admittance matrix is ​​a complex symmetric matrix.

[0041] In some embodiments, a forward-backward substitution method is used for power flow calculation to predict the voltage values ​​of each node. The collected active power output data of the new energy generation units is used as the known quantity of the injected active power of each grid-connected node. The set of reactive power output setpoints obtained by the improved collaborative optimization algorithm in the current iteration cycle is used as the known quantity of the injected reactive power of the corresponding nodes of each controllable reactive power source such as a static var generator (SVM) and energy storage converter. For nodes within the power plant group that do not have reactive power regulation capabilities or are purely load nodes, their injected reactive power is set to a typical value obtained from historical operating data statistics or an estimated value obtained through state estimation. The root node for power flow calculation is set as the common coupling point (CCP) of the power plant group. The voltage amplitude of the CCP is set to a reference value given by the dispatcher, and the voltage phase angle is set to zero degrees, which serves as the benchmark for the entire network power flow calculation.

[0042] Optionally, the forward-backward power flow calculation starts from the end node segment directly connected to the root node. Based on the constructed node admittance matrix and the voltage of adjacent parent nodes, the current of each branch is calculated forward, proceeding step by step from the root node to the end of the network. After completing the forward calculation, a backward substitution is performed from the end node of the network back to the root node. Based on the branch currents and corresponding line impedances obtained from the forward calculation steps, the voltage magnitude and phase angle of each node are updated. The forward calculation and backward substitution update steps are iteratively executed until the voltage difference obtained from two consecutive iterations for all nodes is less than a preset convergence threshold. The voltage values ​​of each node obtained from the last iteration are then used as the final predicted voltage values. The convergence threshold is set according to the required calculation accuracy, for example, 0.0001 (per unit).

[0043] It is understandable that after obtaining the predicted voltage values ​​for each node, these predicted voltage values ​​are compared with preset voltage safety upper or lower limits. These voltage safety upper and lower limits are set according to the power system safety operation regulations and substation grid connection technical specifications. If the comparison reveals that the predicted voltage value of a node exceeds the safety limit, the degree of voltage exceedance at the corresponding node is converted into a numerical penalty term, which is added to the objective function of the improved collaborative optimization algorithm. In the next iteration, the improved collaborative optimization algorithm, while minimizing the objective function, will tend to adjust the output setpoints of controllable reactive power sources to eliminate or reduce voltage exceedances that lead to an increase in the penalty term, thereby generating a set of reactive power output setpoints that satisfy voltage safety constraints. (Penalty term) One design form is: in: Represents a node The penalty value introduced by exceeding the voltage limit. The penalty coefficient is a sufficiently large positive number. Represents the nodes predicted by power flow calculation. voltage amplitude, This indicates the upper limit of voltage safety. This represents the lower limit of voltage safety. (Function) Ensure that the penalty term is positive only when the voltage exceeds the upper limit or falls below the lower limit.

[0044] In practice, the forward-backward power flow calculation and voltage limit check are nested in each major iteration of the improved collaborative optimization algorithm. The table below shows the comparison between the predicted node voltages and the safety limits after one inner-layer power flow calculation iteration. (See Table 2) Table 2: Node Voltage Prediction Results and Limit Exceedance Judgment Table In some embodiments, according to the examples in the table above, the predicted voltage value of node 5, 1.062 pu, exceeds the upper limit of the safe voltage range, 1.050 pu, while the predicted voltage value of node 8, 0.941 pu, is below the lower limit of the safe voltage range, 0.950 pu. The degree of voltage exceedance at these two nodes will be converted into a penalty term according to the aforementioned penalty term calculation formula. For example, for node 5, Penalty items and This will be added to the objective function of the improved co-optimization algorithm. In subsequent optimization iterations, when solving for the optimal set of reactive power output values, the improved co-optimization algorithm must consider the increased objective function value due to the presence of the penalty term. Its solution direction will drive the controllable reactive power sources that are electrically close to node 5 to increase the absorption of reactive power, and at the same time drive the controllable reactive power sources that are electrically close to node 8 to increase the output of reactive power. This will correct the predicted voltage values ​​of nodes 5 and 8 in the next round of power flow calculation, bringing them back to the safe and acceptable range.

[0045] In one embodiment of the present invention, the coordinated control command sequence is encoded according to a preset communication protocol to form a standardized downlink control message. This downlink control message is broadcast and sent through the data bus of the monitoring system of the power station group. The local controller of each controllable reactive power source in the power station group receives and parses the downlink control message, extracting the command content that matches its own equipment identifier. Based on the target output value and adjustment rate in the command content, the local controller of each controllable reactive power source generates a specific pulse width modulation signal or switching signal to drive the power devices to operate and realize the adjustment of reactive power output. After the command is executed, the local controller of each controllable reactive power source feeds back the actual execution result to the monitoring system, forming a closed-loop control.

[0046] In practical implementation, the method of issuing coordinated control command sequences to the corresponding controllable reactive power source execution units to adjust the reactive power distribution of the power station group follows a standardized command transmission and execution process. After the coordinated control command sequence is generated, it is first encoded according to a preset communication protocol. The communication protocol specifies the data frame format, field definitions, verification methods, and transmission rules. The encoding process assembles the coordinated control command sequence, which includes equipment identification, target output value, and adjustment rate information, according to the message structure of the selected protocol. It adds the frame header, address field, control field, information body, and frame check sequence required by the protocol, forming a standardized downlink control message that can be transmitted in the communication network.

[0047] In some embodiments, encoded downlink control messages are broadcast via the monitoring system data bus of the site cluster. The monitoring system data bus is the backbone of the communication network connecting various automated devices within the site to the monitoring master station, such as an industrial Ethernet network. The downlink control messages are encapsulated in Ethernet frames and sent via the data bus in a broadcast or multicast manner to all device nodes connected to the bus. The broadcast method ensures that all relevant local controllers of controllable reactive power sources receive the control messages in a timely manner, and each local controller determines whether to process the message based on the address information in the message.

[0048] Optionally, the local controllers of each controllable reactive power source within the power station cluster receive and parse downlink control messages. The local controller continuously monitors the data traffic on the monitoring system's data bus. When it detects a message that conforms to a preset protocol format and whose destination address matches its own logical address or broadcast address, the local controller receives the message. The local controller decodes the message according to the same communication protocol, stripping away encapsulation information such as frame headers and trailers, and extracting the instruction content contained in the message body. The local controller compares the decoded instruction content with its own device identifier. Only instruction content with a completely matching device identifier is adopted and prepared for execution by the local controller; instruction content with a mismatched device identifier is ignored.

[0049] It is understandable that the local controller of each controllable reactive power source generates specific drive signals based on the target output value and regulation rate in the instruction. For devices such as static var generators or energy storage converters that achieve reactive power regulation through power electronic converters, the local controller calculates the required reactive current reference value based on the target output value and generates a smoothly changing current command curve in conjunction with the regulation rate. The local controller uses pulse width modulation (PWM) technology to convert the current command into a PWM signal that controls the switching on and off of power devices such as insulated-gate bipolar transistors (IGBTs) or integrated gate commutated thyristors. For devices that can achieve reactive power regulation by switching capacitor banks or reactor banks, the local controller calculates the number of banks that need to be switched on or off based on the target output value and generates corresponding switching signals to drive circuit breakers or contactors. The PWM signal or switching signal directly drives the power devices or switching equipment to change the output voltage, current, or impedance of the equipment, thereby achieving precise regulation of reactive power output.

[0050] In practice, after the command is executed, a closed-loop control is formed. Each controllable reactive power source's local controller, after driving the power devices to complete their actions, measures the actual output voltage and current using local voltage and current sensors, and calculates the actual reactive power output value. The local controller assembles an uplink feedback message containing information such as the device identifier, command number, target output value, and actual reactive power output value of this command, according to a preset communication protocol. The uplink feedback message is uploaded to the monitoring system of the power station cluster via the monitoring system's data bus. The monitoring system receives and parses the uplink feedback messages from each controllable reactive power source execution unit, compares the actual execution results with the issued commands, confirms the command execution status, and thus forms a complete "command generation-issuance-execution-feedback" closed-loop control loop. The feedback information is recorded in the monitoring system's database for operators to monitor and provides actual system status information for subsequent control cycles.

[0051] The above embodiments are only used to illustrate the technical methods of the present invention and are not intended to limit it. Although the present invention has been described in detail with reference to preferred embodiments, those skilled in the art should understand that modifications or equivalent substitutions can be made to the technical methods of the present invention without departing from the spirit and scope of the technical methods of the present invention.

Claims

1. A method for reactive power coordinated optimization and control of a wind-solar-storage storage station cluster, characterized in that, The method includes: Real-time operating condition data of wind, solar and energy storage stations are collected. The real-time operating condition data includes the voltage amplitude, injected reactive power and active power output of each node in the station group. The reactive power deficit of the common connection point of the power station group is calculated based on the real-time operating condition data, and the current operating mode is determined based on the reactive power deficit. The operating modes include overvoltage suppression mode, undervoltage support mode and steady-state optimization mode. An improved collaborative optimization algorithm is invoked to generate a sequence of collaborative control instructions for each controllable reactive power source within the power station group, based on the current operating mode, the voltage amplitude, and the active power output. The coordinated control command sequence is sent to the corresponding controllable reactive power source execution unit to adjust the reactive power distribution of the power station group.

2. The reactive power coordinated optimization and control method for a wind-solar-storage storage station cluster according to claim 1, characterized in that, The reactive power deficit of the common connection point of the power station group is calculated based on the real-time operating condition data, and the current operating mode is determined based on the reactive power deficit, including: Based on the reactive power and voltage control targets issued by the power grid dispatch center, obtain the voltage target value and allowable fluctuation range of the common connection point of the power station group; The difference between the measured voltage value at the corresponding common coupling point in the voltage amplitude and the target voltage value is calculated as the voltage deviation. By combining the equivalent impedance between the power station group and the main grid, the voltage deviation is converted into the reactive power compensation required to maintain voltage stability. The reactive power compensation is the reactive power deficit. The reactive power deficit is compared with preset overvoltage deficit thresholds and undervoltage deficit thresholds: If the reactive power deficit is greater than the overvoltage deficit threshold, then the current operating mode is determined to be the overvoltage suppression mode. If the reactive power deficit is less than the undervoltage deficit threshold, then the current operating mode is determined to be the undervoltage support mode. If the reactive power deficit is between the undervoltage deficit threshold and the overvoltage deficit threshold, then the current operating mode is determined to be the steady-state optimization mode.

3. The reactive power coordinated optimization and control method for a wind-solar-storage storage station cluster according to claim 2, characterized in that, The improved collaborative optimization algorithm, based on the current operating mode, the voltage amplitude, and the active power output, generates a sequence of collaborative control instructions for each controllable reactive power source within the power station cluster, including: Read the status information of all controllable reactive power sources in the power station group. The status information includes the current generating capacity of the static var generator, the four-quadrant operating limit of the energy storage converter, and the reactive power regulation capability of the rotor-side converter of the doubly fed wind turbine. Based on the current operating mode, different objective functions and constraints are set for the collaborative optimization algorithm; Under the overvoltage suppression mode, the objective function is set to minimize the total active power loss of the power station group, and the constraints include that the voltage of each node does not exceed the upper limit and the upper limit of the reactive power absorption capacity of the controllable reactive power source. In the undervoltage support mode, the objective function is set to minimize the sum of squares of the total voltage deviation of the power station group, and the constraints include that the voltage of each node is not lower than the lower limit and the upper limit of the reactive power generation capacity of the controllable reactive power source. Under the steady-state optimization mode, the objective function is set to minimize the overall operating cost of the reactive power regulation equipment in the power station group, and the constraints include that the voltage of each node is within the acceptable range and the supply and demand of reactive power are balanced. Based on the set objective function and constraints, the improved collaborative optimization algorithm is used to solve the problem and calculate the set of reactive power output setpoints of each controllable reactive power source that optimizes the objective function under the constraints. Based on the set of reactive power output settings and the current actual output of each controllable reactive power source, a sequence of coordinated control instructions is generated, which includes equipment identification, target output value and adjustment rate.

4. The reactive power coordinated optimization and control method for a wind-solar-storage storage station cluster according to claim 3, characterized in that, The working principle of the improved collaborative optimization algorithm includes: The improved collaborative optimization algorithm is based on the multi-agent consensus theory, which models each controllable reactive power source in the power station cluster as an agent, and defines a local objective function and adjacency communication relationship for each agent. The local objective function of each agent consists of the adjustment cost function of its corresponding controllable reactive power source in the current operating mode and the output consistency function with neighboring agents; In each iteration of the calculation, each agent collects its own state information and the state information of neighboring agents obtained through the adjacency communication relationship. The state information includes the reactive power output setpoint. Each agent independently updates its own reactive power output setting value based on the collected state information using a predefined iterative update rule. The iterative update rule ensures that the reactive power output setting values ​​of all agents evolve toward a Pareto optimal solution that satisfies global constraints. Through multiple iterations, the reactive power output setpoints of all intelligent agents will reach a consistent and stable state. The set of output values ​​in this stable state is the set of reactive power output setpoints for each controllable reactive power source.

5. The reactive power coordinated optimization and control method for a wind-solar-storage storage station cluster according to claim 4, characterized in that, The specific iterative update rules for each agent are as follows: The difference between the current reactive power output setpoint of the agent and the average reactive power output setpoint of all neighboring agents in the previous iteration is calculated as the consistency error. Based on the objective function weights corresponding to the current operating mode, calculate the reactive power adjustment amount that only considers the gradient direction of the local objective function. The consistency error and the reactive power adjustment are weighted and fused to obtain the total adjustment for the current round; The total adjustment is applied to the reactive power setting value of the current agent to complete this iteration update; After the update is completed, the new reactive power output setting is broadcast to all neighboring agents through the adjacency communication relationship.

6. The reactive power coordinated optimization and control method for wind-solar-storage clusters according to claim 3, characterized in that, The method for calculating the comprehensive operating cost of the reactive power regulation equipment in the power station group under the said steady-state optimization mode includes: Obtain the unit adjustment cost parameters of various controllable reactive power sources, including the switching loss cost coefficient of static var generators, the cycle life loss cost coefficient of energy storage converters, and the additional copper loss cost coefficient of doubly-fed wind turbine generators. Calculate the absolute value of the reactive power change of each controllable reactive power source from its current actual output to its reactive power setpoint. Multiply the absolute value of the reactive power change of each controllable reactive power source by its corresponding unit adjustment cost parameter to obtain the individual adjustment cost of the controllable reactive power source. The sum of the individual adjustment costs of all controllable reactive power sources within the power station group is the comprehensive operating cost.

7. The reactive power coordinated optimization and control method for a wind-solar-storage storage station cluster according to claim 3, characterized in that, In the overvoltage suppression mode or the undervoltage support mode, the node voltages in the constraint conditions do not exceed the upper limit or fall below the lower limit, and their specific implementation includes: Construct the node admittance matrix of the power station group, and based on the active power output and the current reactive power output setpoints, use the forward-backward substitution method to perform power flow calculations and predict the voltage values ​​of each node. The predicted voltage values ​​of each node are compared with the preset upper or lower voltage limits. If the predicted voltage value of a node exceeds the limit, the degree of the exceeding of the limit of the corresponding node voltage is converted into a penalty term for the adjustment requirement of the nearby controllable reactive power source, and the penalty term is added to the objective function of the collaborative optimization algorithm. In the next round of iterative solution, the collaborative optimization algorithm will tend to generate a set of reactive power output setpoints that can eliminate the node voltage overruns corresponding to the penalty term.

8. The reactive power coordinated optimization and control method for a wind-solar-storage storage station cluster according to claim 1, characterized in that, The coordinated control command sequence is sent to the corresponding controllable reactive power source execution unit to adjust the reactive power distribution of the power station group, including: The coordinated control instruction sequence is encoded according to a preset communication protocol to form a standardized downlink control message; The downlink control message is broadcast and sent through the data bus of the monitoring system of the station group; The local controllers of each controllable reactive power source within the power station group receive and parse the downlink control message, and extract the instruction content that matches their own equipment identifier; The local controller of each controllable reactive power source generates a specific pulse width modulation signal or switching signal according to the target output value and adjustment rate in the instruction content, drives the power device to operate, and realizes the adjustment of reactive power output. After the instruction is executed, the local controller of each controllable reactive power source feeds back the actual execution result to the monitoring system, forming a closed-loop control.

9. The reactive power coordinated optimization and control method for a wind-solar-storage storage station cluster according to claim 5, characterized in that, Based on the objective function weights corresponding to the current operating mode, the reactive power adjustment amount is calculated, considering only the gradient direction of the local objective function. The steps include: Obtain the objective function weight coefficients preset for each agent, corresponding to the current operating mode. The objective function weight coefficients define the relative importance of the adjustment cost term and the global optimization objective in the local objective function. Within each iteration calculation cycle, the agent calculates the gradient vector of its local objective function with respect to its own reactive power output setpoint, based on its own local objective function. The calculated gradient vector is multiplied by a negative learning rate step size to determine an initial adjustment direction and magnitude aimed at locally optimizing its own objective function; Multiply the initial adjustment direction and magnitude by the objective function weight coefficient to obtain the weighted gradient adjustment amount. This adjustment amount is the reactive power adjustment amount that only considers the gradient direction of its own local objective function.

10. The reactive power coordinated optimization and control method for a wind-solar-storage storage station cluster according to claim 7, characterized in that, The node admittance matrix of the power station group is constructed, and based on the active power output and the current reactive power output setpoints, power flow calculation is performed using the forward-backward substitution method to predict the voltage values ​​of each node, including: The node admittance matrix is ​​calculated based on the topology of the power station group, line parameters, and transformer parameters. The active power output is used as the known quantity of the active power injected into each node. The set of reactive power output settings obtained by the improved collaborative optimization algorithm in the current iteration is used as the known quantity of the injected reactive power of each controllable reactive power source corresponding node. The reactive power of uncontrollable nodes or load nodes is set as a given value or an estimated value. Set the root node as the common connection point of the power station group, and set its voltage amplitude and phase angle as the reference values; Starting from the terminal node directly connected to the root node, calculate the current of each branch forward based on the node admittance matrix and the voltage of the adjacent parent node. The voltage of each node is updated by back-substituting from the last node to the root node based on the calculated branch current and line impedance. Repeat the forward calculation and the back-up update steps until the voltage difference between two iterations of each node is less than the preset convergence threshold. Use the voltage value of each node obtained from the last iteration as the predicted voltage value of each node.

Images