Distributed photovoltaic optimization coordination control system and method

Abstract

The invention discloses a distributed photovoltaic optimization coordination control system and method, and relates to the technical field of distributed photovoltaic power generation, and the control system comprises a multi-dimensional data collection module, a self-adaptive cluster division module, a robust dominant node election module, a three-dimensional voltage state evaluation module, and a hierarchical coordination control execution module. According to the distributed photovoltaic optimization coordination control system and method, real-time operation data, meteorological prediction data and power grid load prediction data of each node are continuously monitored, comprehensive information support is provided for a control strategy, and when it is detected that voltage abnormity is caused by sudden change of illumination intensity in a certain area, the real-time operation data of each node, meteorological prediction data and power grid load prediction data are monitored. And immediately finishing cluster structure adjustment in an extremely short time by adopting a fuzzy clustering algorithm with constraint conditions based on the electrical distance and the voltage fluctuation trend characteristic value, and ensuring that the node voltage change characteristics in the same cluster are highly consistent.

Description

TECHNICAL FIELD

[0001] The present application relates to the technical field of distributed photovoltaic power generation, and particularly relates to a distributed photovoltaic optimization coordination control system and method. BACKGROUND

[0002] Under the trend of global energy transformation, distributed photovoltaic power generation is widely used in urban buildings, rural farmland and other scenarios due to its advantages of local consumption and flexible deployment. However, with the rapid growth of distributed photovoltaic installed capacity, a large number of photovoltaic devices are connected to the power distribution network, which brings many challenges to the stable operation of the power grid.

[0003] The output of distributed photovoltaic power generation has significant intermittency and randomness. In a day, in the morning, as the sun rises, the power generation of the photovoltaic system gradually increases; at noon, the power reaches a peak when the sunlight is sufficient; in the evening, the power rapidly decreases as the sun sets; if sudden weather changes occur, such as sudden clouds and sudden rain, the power generation will fluctuate sharply, or even be greatly reduced instantaneously; at the same time, the power distribution network itself has a complex structure, with extensive line distribution and numerous nodes, and the power load in each region is dynamically changing at all times, further exacerbating the complexity of the power grid operation state.

[0004] At present, many distributed photovoltaic projects still use traditional control methods; some projects only adjust the reactive power according to the voltage deviation, and blindly increase the reactive power output when the voltage of a certain node is monitored to be lower than the set value; this single control strategy does not fully consider the voltage state of adjacent nodes and the overall carrying capacity of the power grid, and not only cannot solve the voltage problem, but also may cause new voltage imbalance, or even lead to voltage collapse in the local power grid; moreover, due to the lack of prediction of future power changes and load fluctuations, there is obvious hysteresis in the voltage regulation process, and it is difficult to respond effectively at the first time when the voltage is abnormal.

[0005] Some projects use a static photovoltaic cluster division method, which divides the clusters according to the power grid topology structure and photovoltaic distribution at the initial stage of project construction, and then does not adjust them; however, in the actual operation process, with the continuous access of new distributed photovoltaic devices, the upgrading of power grid lines and the change of power load characteristics, the original cluster division cannot adapt to the new operation scenario; for example, the newly connected photovoltaic devices may significantly change the power characteristics of a cluster, causing the control strategy originally suitable for the cluster to fail, resulting in a significant reduction in voltage regulation effect, and the reactive power resources cannot be reasonably allocated and efficiently utilized, causing a lot of waste, and therefore, we propose a distributed photovoltaic optimization coordination control system and method. SUMMARY

[0006] (I) Technical problems solved

[0007] In view of the deficiencies of the prior art, the application provides a distributed photovoltaic optimization coordination control system and method, which realizes accurate control of the distributed photovoltaic through multi-source data fusion, dynamic cluster division and hierarchical collaborative control strategy, effectively solves the problems of voltage fluctuation and over-limit, and improves the stability of power grid operation and the power generation efficiency of the distributed photovoltaic.

[0008] (II) Technical solutions

[0009] To achieve the above object, the application is implemented by the following technical solutions:

[0010] A distributed photovoltaic optimization coordination control system, the control system comprising:

[0011] A multi-dimensional data acquisition module: used for acquiring real-time voltage data, power flow data, meteorological prediction data of future period light intensity and temperature of each node of the distributed photovoltaic system, and load prediction data of each region of the power grid in the future period; at the same time, based on historical data, the electrical distance value for measuring the closeness of electrical connection between nodes, the voltage sensitivity value reflecting the influence degree of the voltage of the node on the change of reactive power, the voltage deviation value indicating the degree of deviation of the actual voltage of the node from the standard voltage, and the voltage fluctuation trend value embodying the voltage change trend are calculated;

[0012] An adaptive cluster division module: based on the electrical distance value and the voltage fluctuation trend value, a fuzzy clustering algorithm with constraint conditions is adopted to divide the distributed photovoltaic nodes into multiple control clusters; the constraint condition is that the voltage change amount difference of any two nodes in the same cluster does not exceed a pre-set voltage change amount threshold, so as to ensure the consistency of the voltage characteristics of the nodes in the cluster;

[0013] A robust leading node election module: by comprehensively evaluating the voltage sensitivity value and the operation reliability of the node, a leading control node is selected from each cluster; the operation reliability evaluation includes analysis of the historical voltage stability, communication link stability and protection device action reliability of the node;

[0014] A three-dimensional voltage state evaluation module: based on the voltage deviation value, the voltage change rate value and the prediction error value, an evaluation model is constructed to divide the state of each cluster into three categories of emergency over-limit, early warning over-limit and normal operation, and determine the corresponding control priority;

[0015] A hierarchical collaborative control execution module:

[0016] For the emergency over-limit cluster, active power and reactive power joint optimization control is executed, the maximum power tracking mode, active power reduction ratio and reactive power output capacity of the photovoltaic inverter are dynamically adjusted through the voltage change trend in the future period;

[0017] For the early warning out-of-limit cluster, a flexible coupling control relationship is established with the adjacent normal cluster by dynamically adjusting the reactive power droop control coefficient.

[0018] For the normal cluster, the reactive power reserve capacity is pre-allocated based on the prediction data.

[0019] The adaptive cluster division module and the hierarchical collaborative control execution module form a closed-loop control through real-time data interaction: when the system operating state changes, the adaptive cluster division module recalculates and adjusts the cluster structure, and the hierarchical collaborative control execution module optimizes the collaborative control strategy between clusters based on the updated cluster division result, to realize rapid and stable adjustment of the voltage of the distributed photovoltaic system.

[0020] Preferably, the fuzzy clustering algorithm of the adaptive cluster division module is executed as follows:

[0021] Initialize the cluster center and the node membership matrix;

[0022] Calculate the weighted distance of each node to each cluster center, wherein the weight is dynamically determined according to the electrical distance value and the voltage fluctuation trend value of the node;

[0023] Update the cluster center and the membership matrix through iteration until the preset convergence condition is met;

[0024] Verify the voltage correlation of the nodes in each cluster, and if the difference between the voltage changes of the nodes in the same cluster exceeds the pre-set voltage change threshold, the cluster is re-divided.

[0025] Preferably, the process of the hierarchical collaborative control execution module establishing a flexible coupling control relationship between the early warning out-of-limit cluster and the adjacent normal cluster includes:

[0026] Real-time monitoring of the voltage interaction between clusters, calculation and update of the voltage sensitivity matrix between clusters, which is used to measure the influence degree of reactive power change of one cluster on the voltage of another cluster;

[0027] When the voltage of a certain early warning out-of-limit cluster reaches the pre-set early warning voltage threshold, the collaborative voltage regulation mechanism of the adjacent normal cluster is triggered;

[0028] According to the electrical connection characteristics (including line length, line impedance and other parameters) and the voltage sensitivity value between clusters, the responsibility weight of each adjacent cluster in the collaborative voltage regulation process is dynamically allocated according to the pre-set weight allocation rule, forming an adaptive voltage support network.

[0029] Preferably, the cooperative working mechanism of the adaptive cluster division module and the hierarchical collaborative control execution module is:

[0030] When the power grid topology changes (such as line switching, transformer tap adjustment) or the photovoltaic system operating state changes (such as sudden change of light intensity, large fluctuation of load), the adaptive cluster division module completes the redivision of the cluster structure within the first preset time;

[0031] The hierarchical collaborative control execution module adjusts the control parameters of each cluster within the second preset time based on the updated cluster division result, and updates the flexible coupling control relationship;

[0032] Through the above-mentioned collaborative mechanism, the response speed and control accuracy of the voltage regulation of the distributed photovoltaic system are improved.

[0033] Preferably, when the hierarchical collaborative control execution module performs joint active power and reactive power optimization control on the emergency out-of-limit cluster, it specifically includes:

[0034] A multi-objective optimization model containing voltage deviation minimization, reactive power reserve maximization, and active power reduction minimization is established;

[0035] According to the current operating state of the system, the priority weights of each optimization objective are dynamically adjusted;

[0036] A predictive control algorithm is used to solve the multi-objective optimization model to generate an optimal control sequence for multiple future control periods to regulate the active output and reactive compensation capacity of the photovoltaic inverter.

[0037] Preferably, the specific method for the robust master node election module to evaluate node operating reliability includes:

[0038] The frequency, amplitude, and duration of voltage fluctuations of the node in the past period of time are counted to evaluate the voltage stability of the node;

[0039] The number of communication interruptions, average data transmission delay time, and data packet loss rate of the node communication link in the past period of time are calculated to evaluate the stability of the node communication link;

[0040] The number of actions and correct action rate of the node protection device in the past period of time are counted to evaluate the action reliability of the node protection device;

[0041] The above three evaluation results are quantified and comprehensively calculated to obtain the operating reliability score of the node, and the node with the highest score is selected as the master control node.

[0042] Preferably, the data acquisition module is also used to send a signal to the hierarchical collaborative control execution module at night or when the reactive power deficiency of the out-of-limit cluster reaches a pre-set reactive power deficiency threshold, so that the photovoltaic inverter of the out-of-limit cluster switches to a special operating mode for fast reactive power compensation to maintain the voltage stability of the photovoltaic microgrid.

[0043] Preferably, the specific determination criteria of the three-dimensional voltage state evaluation module for cluster state division are as follows:

[0044] Emergency out-of-limit state: the voltage deviation degree value is greater than the first preset voltage deviation threshold value, and the voltage change rate value is greater than the first preset voltage change rate threshold value;

[0045] Early warning out-of-limit state: the voltage deviation degree value is greater than the second preset voltage deviation threshold value and not greater than the first preset voltage deviation threshold value, or the voltage change rate value is greater than the second preset voltage change rate threshold value and not greater than the first preset voltage change rate threshold value;

[0046] Normal operation state: the voltage deviation degree value is not greater than the second preset voltage deviation threshold value, and the voltage change rate value is not greater than the second preset voltage change rate threshold value.

[0047] Preferably, the system further comprises a fault diagnosis and recovery module for real-time monitoring of the operation state of each node of the system; when voltage abnormal fluctuation of a node is detected, the fault type and location are determined by comparing the electrical parameter difference between the node and adjacent nodes, and combining a historical fault case library; after the fault is eliminated, the leading node is automatically re-assigned and the control strategy is reconfigured according to the latest results of the adaptive cluster division module.

[0048] A distributed photovoltaic optimization and coordination control method, the control method comprising the following steps:

[0049] Collecting real-time voltage data, power flow data, weather forecast data and grid load forecast data of each node of the distributed photovoltaic system, calculating electrical distance values, voltage sensitivity values, voltage deviation degree values and voltage fluctuation trend values between nodes;

[0050] Based on the electrical distance values and voltage fluctuation trend values, using a fuzzy clustering algorithm with constraints, the distributed photovoltaic nodes are divided into multiple control clusters;

[0051] By comprehensively evaluating the voltage sensitivity values and operation reliability of the nodes, the leading control nodes are determined from each cluster;

[0052] Based on the voltage deviation degree values, voltage change rate values and prediction error values, the states of each cluster are divided into three categories: emergency out-of-limit, early warning out-of-limit and normal operation, and the corresponding control priorities are determined;

[0053] The corresponding collaborative control strategies are started for the divided categories, as follows:

[0054] For the emergency out-of-limit cluster, active power and reactive power joint optimization control is performed;

[0055] For the early warning over-limit cluster, a flexible coupling control relationship is established with the adjacent normal cluster;

[0056] For the normal cluster, reactive power reserve capacity is pre-allocated based on prediction data;

[0057] Among them, the control cluster division and the coordinated control strategy work together: when the system operating state changes, the adaptive cluster division step recalculates and adjusts the cluster structure, and the hierarchical coordinated control step optimizes the coordinated control strategy between clusters based on the updated cluster division result, to realize the rapid and stable adjustment of the voltage of the distributed photovoltaic system.

[0058] (Three) beneficial effects

[0059] 1. By continuously monitoring real-time operating data, meteorological prediction data and power grid load prediction data of each node, comprehensive information support is provided for the control strategy, so that when voltage abnormalities are detected in a certain area due to sudden changes in light intensity, based on electrical distance and voltage fluctuation characteristic value, a fuzzy clustering algorithm with constraints is used to complete cluster structure adjustment in a very short time, ensuring that the voltage change characteristics of nodes in the same cluster are highly consistent; At the same time, it is quickly determined that the cluster is in an emergency over-limit state, and the hierarchical coordinated control execution module immediately starts active-reactive joint optimization control, dynamically adjusts the maximum power tracking mode, active power reduction ratio and reactive power output capacity of the photovoltaic inverter by predicting the voltage change trend in the future period; Compared with traditional methods, this series of linkage response mechanism greatly shortens the voltage over-limit elimination time, realizes rapid and accurate regulation and control of the grid voltage, and effectively guarantees the stability and reliability of power supply.

[0060] 2. The reliable and voltage-sensitive leading control nodes are selected from each cluster to lay the foundation for accurate control; when a cluster is determined to be in a pre-warning over-limit state, a flexible coupling control relationship is established with the adjacent normal cluster by dynamically adjusting the reactive droop control coefficient; then, according to the electrical connection characteristics and voltage sensitivity values between clusters, the responsibility weight of each adjacent cluster in the coordinated voltage regulation process is dynamically allocated according to the pre-set weight allocation rule, forming an adaptive voltage support network; In this process, active power reduction of photovoltaic equipment in the cluster is not required, but the reactive power resources of adjacent clusters are fully utilized to realize voltage regulation. Through actual application verification, this method can significantly improve the power generation efficiency of distributed photovoltaic, effectively reduce the loss of power generation, and improve energy utilization efficiency.

[0061] 3. A closed-loop control mechanism is formed through real-time data interaction between the adaptive cluster division module and the hierarchical cooperative control execution module; when a change in the power grid topology is detected, the cluster structure can be re-divided within a first preset time, and then the hierarchical cooperative control execution module rapidly adjusts the control parameters of each cluster within a second preset time based on the updated cluster division result, and updates the flexible coupling control relationship; this rapid response and dynamic adjustment capability enables the system to always remain stable under complex and variable operating conditions, greatly enhances the adaptability of the power grid to high proportion of distributed photovoltaic access, effectively reduces the risk of power grid operation, and provides a solid technical guarantee for the stable development of smart grids. BRIEF DESCRIPTION OF DRAWINGS

[0062] The above description is only a summary of the technical solutions of the present application. In order to more clearly understand the technical means of the present application, and to implement the content of the specification, the following will be described in detail with the preferred embodiments of the present application and with the aid of the accompanying drawings.

[0063] Figure 1 The overall structure flow chart of the embodiment of the present application. DETAILED DESCRIPTION

[0064] The embodiment of the present application provides a distributed photovoltaic optimization and coordination control system and method, which realizes precise control of distributed photovoltaic through multi-source data fusion, dynamic cluster division and hierarchical cooperative control strategy, effectively solves the problems of voltage fluctuation and over-limit, and improves the stability of power grid operation and the power generation efficiency of distributed photovoltaic.

[0065] Embodiment 1: As shown in the following table, the technical solution in the embodiment of the present application is to effectively solve the problems of voltage fluctuation and over-limit, and the overall idea is as follows: Figure 1

[0066] In view of the problems in the prior art, the present application provides a distributed photovoltaic optimization and coordination control system, and the specific content of the system is as follows:

[0067] 1. Multi-dimensional data acquisition and processing:

[0068] Data acquisition, as the foundation of system operation, undertakes the key task of acquiring and processing multiple types of data; the data collected covers real-time operation data, future environment data and future power data, and accurate acquisition and effective processing of these data provide indispensable basis for subsequent control strategy formulation. The specific data collected is as follows:

[0069] ​Real-time operation data is obtained through intelligent monitoring terminals deployed at distributed photovoltaic nodes and key positions of the power grid. Using devices with detection functions such as voltage transformers, current transformers, and power measurement devices, the node voltage amplitude, phase, active power, reactive power, and power flow data can be collected in real time at a preset sampling period. These data can immediately reflect the current operating state of the photovoltaic system and are an important basis for determining whether the photovoltaic equipment and photovoltaic system can operate normally, providing a real-time information basis for subsequent fault diagnosis and operation optimization.

[0070] The collected data also includes future environmental data and future power data. The future environmental data mainly includes meteorological information such as light intensity and temperature, and its acquisition relies on the Numerical Weather Prediction (NWP) system of the meteorological department to predict the future changes of meteorological elements. A dedicated data communication interface can be used to connect with the meteorological department database, and meteorological prediction data within 24 hours with a time resolution of 15 minutes can be downloaded regularly according to pre-set rules and protocols. After obtaining the original data, format conversion and analysis are needed, and then data cleaning algorithms are used to remove outliers and missing values. For missing values, linear interpolation or weighted average method based on adjacent period data is used for supplementation. For outliers, statistical analysis and machine learning algorithms are used for identification and correction. Finally, the processed data is stored in the system database. These data can help the system to predict the power generation potential and possible problems of the photovoltaic system in advance, such as predicting the power generation power according to the light intensity and correcting the power generation efficiency of the photovoltaic module according to the temperature change, so that the system can adjust the power generation plan in advance and improve the energy utilization efficiency.

[0071] The future power data, i.e. the load prediction data of each region of the power grid, is mainly obtained by relying on the load prediction method of the power system and the power dispatching information system to estimate the future power load. Through data interaction with the Energy Management System (EMS) of the power dispatching center, the power load prediction data of each region within 24 hours with a time resolution of 30 minutes is obtained. After obtaining, data preprocessing is also carried out, consistency check is carried out to ensure the consistency of data in time and space, then noise and fluctuations are eliminated through data smoothing algorithm, and then related data is associated and integrated. Through these data, the system can understand the trend of future power grid load, so as to better coordinate the power generation of distributed photovoltaic system and the power demand of power grid, realize the balance of supply and demand of power, avoid voltage fluctuation and energy waste caused by mismatch between power generation and power consumption, and the application is not limited to the above-mentioned acquisition method, other systems or methods that can obtain corresponding data can also be used.

[0072] After obtaining these basic data, the basic data need to be processed, and the electrical distance D between node i and node j is calculated by the admittance matrix method ij , and the calculation formula is as follows:

[0073]

[0074] In the formula, Y ij is the mutual admittance between node i and node j, reflecting the electrical connection strength; ΔV k is the voltage change of node k under unit power injection; V k is the rated voltage of node k; n is the total number of system nodes, α ij is the impedance temperature correction coefficient of line ij between node i and node j, which is related to real-time environmental temperature T, and is calculated through the function α ij =1+k t (T-T0), kt is the temperature coefficient of line material, and T0 is the standard temperature; β ij is the transformer ratio error correction coefficient of line ij, which is determined by the difference between the actual transformer ratio and the preset rated ratio; the obtained value is used to measure the electrical connection tightness between nodes, and provides a key basis for adaptive cluster division; through the calculation of electrical distance, the electrical coupling relationship of each node in the power grid can be clearly presented, so that nodes with close electrical connection can be classified into the same cluster during cluster division, thereby improving the pertinence and effectiveness of subsequent control strategies.

[0075] The voltage sensitivity S ij of node i to node j reactive power injection is calculated as: wherein, Based on the principle of power flow calculation, the Taylor expansion and linearization of power flow equation are carried out, and the Newton-Raphson method is used for iterative solution, to obtain the sensitivity value, which is used to guide the reactive power regulation strategy and help the system to accurately adjust the reactive power output to stabilize the voltage; accurate calculation of voltage sensitivity can help the system to quickly locate the nodes with greater influence on voltage, and these nodes can be operated preferentially during reactive power regulation, greatly improving the efficiency and accuracy of voltage regulation.

[0076] The voltage deviation degree δ i of node i is defined as: wherein V i is the real-time voltage of node, V ref is the standard voltage, and σ iis the voltage fluctuation safety margin coefficient of node i, with a value range of [0.8, 1.2]. It is dynamically adjusted according to factors such as the importance of the node and the characteristics of the surrounding load. This formula is used to quantify the degree to which the voltage deviates from the normal range and provide data support for voltage status assessment. By calculating the voltage deviation, the system can intuitively determine whether the voltage of each node is normal, promptly detect problems such as voltage exceeding the limit, and provide a decision-making basis for subsequent hierarchical coordinated control.

[0077] ARIMA (autoregressive integrated moving average model) is used to process the historical voltage data V(t). The ARIMA model stabilizes the time series data by performing a differential operation, and then combines the autoregressive (AR) and moving average (MA) processes to build a prediction model. Its general form can be expressed as ARIMA(p,d,q), where p is the autoregressive order, reflecting the linear relationship between the current series value and the past p series values; d is the difference order, which is used to convert the non-stationary time series into a stationary series; q is the moving average order, which reflects the linear relationship between the current series value and the past q error terms.

[0078] For the historical voltage data V(t), the stationary sequence Δ is obtained after d-order difference d V(t), its ARIMA model expression is as follows:

[0079]

[0080] Where, Δ d V(t) is the sequence value of the historical voltage data V(t) after d-order difference; φ i is the autoregressive coefficient, Δ d V(ti) is the sequence value of the past i-th moment, θ j is the sliding average coefficient, ε(tj) is the jth error term in the past; ε(t) is a white noise sequence with a mean of 0 and a variance of σ 2 The white noise sequence represents random fluctuations that cannot be explained by the model; it is the information not captured by the autoregressive part and the sliding average part of the model, reflecting the uncertainty and randomness in the data.

[0081] Among them, in time series analysis, many actual data (such as historical voltage data) are often non-stationary, and directly modeling them will lead to inaccurate model results; through differential operation, non-stationary series can be converted into stationary series to meet the modeling requirements of ARIMA model; for example, if d = 1, then ΔV(t) = V(t)-V(t-1), that is, the original series is differentiated by the first order; if d = 2, then another difference is performed on the basis of the first order difference. In distributed photovoltaic systems, voltage data will be affected by various factors such as changes in light intensity and load fluctuations and show non-stationary characteristics; the stationary series Δ obtained by differential operationd V(t), which can eliminate the non-stationary factors such as trend and seasonality in the data, laying a foundation for the subsequent accurate establishment of the ARIMA model.

[0082] p reflects the linear relationship between the current sequence value Δ d V(t) and the past p sequence values Δ d V(t-1), Δ d V(t-2), …, Δ d V(t-p); that is, the current voltage difference sequence value can be represented by the linear combination of the voltage difference sequence values at the past p time points.

[0083] Meaning: Autoregressive part, where φ i is the autoregressive coefficient, representing the influence weight of the sequence value Δ d V(t-i) at the past i time point on the current sequence value Δ d V(t); these coefficients can be determined by the model parameter estimation method, and different φ i values reflect the difference in the influence degree of data at different time points on the current data; in the voltage data processing of distributed photovoltaic systems, the autoregressive part can capture the temporal dependence of the voltage sequence; for example, if p = 2, the current voltage difference sequence value is related to the voltage difference sequence values at the past two time points, and through appropriate φ1 and φ2, this dependence can be described to predict the future voltage trend.

[0084] q reflects the linear relationship between the current sequence value and the past error terms; it reflects the influence of past random fluctuations on the current sequence value.

[0085] Meaning: Moving average part, where θ j is the moving average coefficient, used to measure the influence degree of the past j error term ε(t-j) on the current sequence value Δ d V(t); these coefficients are also determined by the model parameter estimation method; in the actual operation of photovoltaic systems, voltage data will be affected by various random factors (such as sudden small load changes, slight fluctuations in environmental factors, etc.), resulting in random errors; the moving average part can model and predict these random fluctuations by considering the past error terms, making the model better fit the actual voltage data and improving the prediction accuracy.

[0086] The system estimates model parameters by least square method, predicts future voltage variation trend, and provides prospective information for control strategy; voltage fluctuation trend prediction enables the system to perceive voltage variation in advance, and takes corresponding control measures before voltage appears abnormal fluctuation, thereby enhancing system stability and reliability. The above calculation methods are all feasible schemes, and are not limited to the above calculation methods.

[0087] 2. Adaptive cluster division

[0088] Based on the electrical distance and voltage fluctuation trend values obtained by processing the data acquisition module, a fuzzy clustering algorithm with constraint conditions is used to divide the distributed photovoltaic nodes into multiple control clusters.

[0089] Voltage variation threshold setting: the voltage variation difference of nodes in the same cluster needs to meet ΔV max -ΔV min ≤0.05V ref This constraint condition; wherein, ΔV max and ΔV min are the maximum and minimum values of the voltage variation of nodes in the cluster, V ref is the standard voltage; the threshold value can be determined according to a large amount of historical operation data statistical analysis and research on the influence of power grid stability; in actual power grid operation, when the voltage variation difference of nodes in the cluster exceeds the threshold value, it means that the voltage response characteristics of nodes are too different, and if a unified control strategy is used, it may lead to poor control effect of some nodes, and even cause new voltage problems; by setting this threshold value, it can ensure that the voltage characteristics of nodes in the divided cluster have high consistency, and lay a foundation for subsequent accurate control.

[0090] Specific execution process of fuzzy clustering algorithm: a fuzzy C-means clustering algorithm (FCM) with constraints is used, and the objective function J is: Wherein, c is the number of clusters; n is the number of nodes; u ij is the membership degree of node j to cluster i (0≤u ij ≤1); d ij is the distance from node j to the center of cluster i; is the mth power operation of membership degree u ij ; m is the fuzzy weight (usually m=2), and λ is the penalty factor for balancing clustering compactness and constraint conditions; g ij is the constraint condition function, g ij =0 when node j meets the voltage variation difference constraint with cluster i, otherwise g ij is a penalty value related to the difference size, and the constraint condition is

[0091] g ijAs a quantitative indicator to measure whether node j meets the voltage variation difference constraint in cluster i, its calculation is closely related to the voltage variation difference between nodes. The specific calculation formula is:

[0092]

[0093] In the formula, ΔV max,j and ΔV min,j are the maximum and minimum values of the voltage variation of node j in cluster i and other nodes; V ref is the standard voltage; κ is the penalty coefficient, which is a constant determined according to the requirements of power grid operation stability and actual test data, and generally takes a value in the range of 1-10. The formula shows that when the node meets the voltage variation difference constraint, g ij takes a value of 0, which has no additional effect on the objective function; when the constraint is not met, g ij is calculated according to the size of the threshold value, and the more it exceeds, the greater the penalty value.

[0094] In the fuzzy clustering algorithm iteration process, g ij and the penalty factor λ jointly act on the objective function J; when node j does not meet the voltage variation difference constraint, u ij ·g ij term will increase the value of the objective function J; the optimization goal of the algorithm is to minimize J, so in order to reduce the value of J, the algorithm will automatically adjust the membership degree u ij so that the nodes that do not meet the constraint are more inclined to be divided into other clusters that meet the constraint; the introduction of g ij significantly improves the rationality and practicality of the clustering results; by imposing a penalty on nodes that do not meet the constraint, it effectively avoids dividing nodes with large differences in voltage response characteristics into the same cluster, making the final formed cluster more in line with actual control requirements.

[0095] The algorithm execution steps are as follows:

[0096] Initialization phase: First, determine the number of clusters c according to experience or trial method, for example, for a small power grid area containing 50 distributed photovoltaic nodes, c can be initially set to 5; randomly generate the membership degree matrix U of each node to different clusters = [u ij ], ensuring that At the same time, according to the characteristic data such as electrical distance and voltage fluctuation trend of nodes, randomly select c nodes as the initial cluster center Z = [z i ], and set the initial penalty factor λ0.

[0097] Iteration calculation phase: In each iteration, first update the cluster center Z according to the current membership degree matrix U; for each cluster i, its new cluster center z i is calculated as where x j is the feature vector of node j (contains data such as electrical distance, voltage fluctuation trend, etc.); Then, the membership matrix U is recalculated according to the new cluster center Z; For node j and cluster i, its membership u ij The update formula is where d ij , d kj are the distances from node j to cluster i and from node j to cluster k center respectively; In the calculation process, it is checked in real time whether the difference ΔV max -ΔV min ≤0.05V ref This constraint is met; If not, adjust the cluster center or membership appropriately, for example, for the cluster that exceeds the threshold, select a node close to the mean of the voltage variation within the cluster as the new cluster center, and calculate the membership again until the constraint is met.

[0098] Convergence judgment stage: repeatedly iterate the calculation process, calculate the difference ΔJ of the objective function J of this iteration and the last iteration, when ΔJ is less than the pre-set convergence threshold (such as 10 -5 ), it is considered that the algorithm converges, and the iteration is stopped; At this time, the membership matrix U obtained is the final clustering attribution of the node, and the adaptive cluster division of the distributed photovoltaic node is completed.

[0099] Each cluster construction process: after the fuzzy clustering algorithm iteration convergence is completed, according to the final membership matrix U, each node is divided into the corresponding cluster; For the cluster i with the maximum membership u ij , node j belongs to this cluster; In this way, the nodes in the entire distributed photovoltaic network are divided into multiple clusters with similar electrical characteristics and voltage response trends, providing a basis for subsequent targeted collaborative control strategy.

[0100] 3. Robust leader node election:

[0101] The robust leader node election module undertakes the important task of selecting key control nodes from each cluster; This module evaluates the voltage sensitivity value and operation reliability of each node to ensure that the selected dominant control nodes can play a core role in voltage regulation and have stable and reliable operation ability.

[0102] In the evaluation index calculation, the voltage sensitivity value is calculated by the formula The formula is calculated based on the principle of power flow calculation, linearized by Taylor expansion and solved by Newton-Raphson method iteration, and reflects the sensitivity of the voltage of node i to the change of reactive power injection of node j; the higher the value, the more significant the voltage change of the node under the influence of reactive power regulation, and the more key role it plays in cluster voltage regulation.

[0103] The operation reliability evaluation is carried out from three dimensions: in the voltage stability evaluation, the frequency, amplitude and duration of voltage fluctuation of the system within the past 24 hours are traced back, and the voltage fluctuation complexity index ξi is introduced, the calculation formula is:

[0104]

[0105] In the formula, ΔV i (t) is the voltage change of node i at time t, is the average voltage change, N is the number of sampling points, and the index is used to measure the complexity of voltage fluctuation, which, combined with the original frequency, amplitude and duration index, more comprehensively evaluates the voltage stability; for example, the voltage stability score of a node with frequent voltage fluctuations and amplitude exceeding the standard voltage by 5% for more than 15 minutes will be affected; the communication link stability evaluation counts the number of communication interruptions, average delay time and data packet loss rate in the past 7 days, and if there are multiple communication interruptions, high delay and high packet loss rate, the reliability score of the node in communication will be reduced; the protection device action reliability evaluation counts the number of protection device actions and the correct action rate in the past 1 month, and the node with frequent misoperation and low correct action rate will have a lower score in this dimension; the system assigns weights to each dimension (voltage stability 0.4, communication link stability 0.3, protection device action reliability 0.3), and the weighted sum gives the node operation reliability score.

[0106] In the specific election process, the module first arranges the nodes in the cluster in descending order according to the voltage sensitivity value, and selects the top 30% of the nodes as the preliminary candidate nodes, ensuring that the candidate nodes have strong voltage regulation sensitivity; then, according to the operation reliability score, the candidate nodes are further screened, and the nodes with high scores are preferentially selected as the leading control nodes; if the voltage sensitivity is similar, the operation reliability score is used as the final decision basis to ensure the stable operation ability of the leading nodes.

[0107] 4. Three-dimensional voltage state evaluation:

[0108] The three-dimensional voltage state evaluation module constructs a comprehensive evaluation model based on the voltage deviation value, voltage change rate value and prediction error value, and realizes accurate division and control priority generation of each cluster state.

[0109] Evaluation index calculation:

[0110] Voltage deviation: calculated by formula where V i is the real-time voltage of node i, V ref is the standard voltage of the corresponding node; this index intuitively quantifies the degree to which the node voltage deviates from the standard value, and is the basic data for evaluating the voltage state.

[0111] Voltage rate of change: calculated by formula where V i (t) and V i (t-Δt) are the voltage values of node i at time t and t-Δt, respectively, Δt is the time interval (set to 5 minutes by the system), μ i is the load mutation coefficient, based on the real-time changes of the loads connected to the node, with a value range of [0.9, 1.1]; v i is the photovoltaic output fluctuation coefficient, according to the real-time output fluctuation of the photovoltaic equipment connected to the node, with a value range of [0.8, 1.2]; the voltage rate of change reflects the fluctuation amplitude of the voltage per unit time, and is used to measure the degree of voltage change.

[0112] Prediction error value: calculated by formula where V i,pred is the voltage prediction value of node i, V i,real is the actual measured value, and ψ i is the confidence interval correction factor, according to the historical prediction accuracy of the voltage prediction model and the complexity of the current operating environment, with a value range of [0.8, 1.2]; this index reflects the accuracy of voltage prediction, helping to determine whether the current voltage change is as expected.

[0113] Evaluation model construction:

[0114] The analytic hierarchy process (AHP) is used to determine the weights of each index, and a comprehensive evaluation function Ec is constructed:

[0115] where, are the average values of the voltage deviation, voltage rate of change, and prediction error of each node in cluster c, respectively; ω1, ω2, and ω3 are the weights of the average values of the voltage deviation, voltage rate of change, and prediction error, respectively, and are preferably ω1=0.5, ω2=0.3, and ω3=0.2; this function integrates the three-dimensional indexes and comprehensively reflects the voltage state of the cluster.

[0116] Cluster state division:

[0117] According to the comprehensive evaluation value E c , the cluster state is divided as follows:

[0118] When E cWhen the voltage exceeds 8, it is determined to be an emergency over-limit, indicating that the voltage within the cluster deviates significantly from the standard and fluctuates dramatically, potentially posing a threat to the safe and stable operation of the power grid. At this time, the system quickly captures the urgency of the voltage anomaly, providing a basis for the subsequent activation of high-intensity control strategies.

[0119] When 3 <E c When the value is ≤8, it is judged as a warning limit violation, indicating that the cluster voltage is showing an abnormal trend and measures must be taken in advance to prevent the problem from worsening. This judgment can trigger a response mechanism at the budding stage of voltage problems and prevent them from happening.

[0120] When E c When ≤3, it is judged to be normal operation, indicating that the cluster voltage is in a stable and controllable state.

[0121] Control priority generation:

[0122] Based on the cluster status classification results, corresponding control priorities are generated; the emergency over-limit cluster has the highest priority, and a rapid voltage regulation and power balancing strategy must be immediately initiated; the warning over-limit cluster has the second highest priority, and flexible coordinated voltage regulation measures are adopted; the normal operation cluster has the lowest priority, and a preventive control strategy of pre-allocating reactive power reserve capacity is implemented; this priority system ensures the rational allocation of system resources, gives priority to handling the most urgent voltage problems, and improves overall control efficiency and system stability.

[0123] 5. Hierarchical collaborative control strategy:

[0124] The hierarchical collaborative control execution module executes different control strategies according to the cluster status determined by the three-dimensional voltage status assessment module to achieve collaborative optimization operation of each cluster.

[0125] Emergency over-limit cluster coordinated control strategy: When a cluster is judged to be in an emergency over-limit state, it means that its voltage deviates significantly from the standard and fluctuates dramatically, posing a threat to the safe and stable operation of the power grid. In this case, the system constructs a multi-objective optimization model:

[0126] minF=kl1×∑ i∈E |V i -V ref |+kl2×Σ i∈E (Q reserue,i -Q i )+kl3×∑ i∈E P cut,i ;

[0127] Where E is the set of emergency over-limit cluster nodes, V i is the real-time voltage of node i, V refis the standard voltage of the corresponding node (here refers to the standard voltage corresponding to node i); kl1, kl2, kl3 are the weight coefficients of the voltage deviation target, reactive power reserve target, and active power reduction target respectively (dynamically adjusted according to the system state, kl1+kl2+kl3=1); Q reserve,i is the reactive reserve capacity of node i; Q i is the reactive power output of node i; P cut,i is the active power reduction of node i; the model predictive control (MPC) algorithm is used, with a control cycle of 15 minutes, to roll-optimize the control sequence for the next four cycles.

[0128] During the execution process, each node receives control instructions through the dominant control node; the dominant node coordinates the actions of the reactive compensation devices and active power regulation equipment of each node based on the results calculated by the optimization model; for example, it prioritizes nodes with sufficient reactive power reserves for reactive compensation to quickly increase the voltage; if the voltage still cannot return to normal, the active power of some nodes is reduced according to the set priority to maintain power balance and voltage stability in the power grid; at the same time, each node feeds back the execution effect to the dominant node in real time, forming a closed-loop control to ensure the effectiveness of the control strategy; actual application shows that this strategy has shortened the voltage recovery time of the emergency over-limit cluster from 12 seconds using the traditional method to 4.5 seconds, greatly improving the power grid's ability to respond to sudden voltage crises.

[0129] Collaborative control strategy of early warning over-limit clusters: For early warning over-limit clusters, the system establishes an inter-cluster voltage sensitivity matrix S = [S ij ] c×c , when the voltage sensitivity S ij When >0.1, flexible coupling control is started: Where ΔQ i is the reactive regulation of cluster i; Sλ is the regulation coefficient; is the target voltage change of cluster j; ΔV j is the actual voltage change of cluster j.

[0130] During the specific implementation process, when a warning over-limit cluster detects an abnormal voltage trend, its dominant control node first sends the voltage status information to the dominant node of the adjacent cluster with higher voltage sensitivity; the adjacent cluster calculates a reasonable reactive power regulation amount based on the received information and its own operating status, and adjusts the reactive power output of the nodes within the cluster to achieve voltage support for the target cluster.

[0131] Normal operation cluster cooperative control strategy: for clusters in normal operation state, the system pre-allocates reactive power reserve capacity based on prediction data; the dominant control node of each node calculates the voltage fluctuation that the cluster may face according to the future illumination intensity, temperature and power grid load prediction data for a period of time (such as 1 hour in the future); then, according to certain rules (such as allocation according to node capacity and voltage sensitivity), the reactive power reserve capacity is allocated to each node in the cluster.

[0132] During operation, each node monitors its own voltage and reactive power in real time; when a slight disturbance of the power grid or photovoltaic system causes voltage fluctuation, the node preferentially uses its reserved reactive power reserve for adjustment to suppress voltage fluctuation within the minimum range; at the same time, the dominant control node continuously monitors the overall voltage state of the cluster, and if it finds that the local node adjustment capacity is insufficient, it promptly coordinates other nodes for cooperative adjustment to ensure the voltage stability of the entire cluster and provide stable power support for the power grid; statistics show that after adopting this strategy, the voltage fluctuation range of the normal operation cluster when facing small disturbances is controlled within ±0.01 pu, and the power supply quality of the power grid is significantly improved.

[0133] 6. Fault diagnosis and recovery:

[0134] Fault diagnosis and recovery is a key component to ensure the reliable operation of distributed photovoltaic systems, which builds a complete fault defense system through three core functions of real-time monitoring, intelligent diagnosis and rapid recovery.

[0135] Real-time monitoring mechanism: this module uses the millisecond-level electrical parameters (voltage, current, power, etc.) provided by the data acquisition module to continuously monitor each node of the system; by establishing a node operating state feature library and dynamically setting monitoring thresholds, for example, for voltage parameters, in addition to setting a regular ±5% deviation threshold, a dynamic threshold range is also calculated based on historical operating data of the node, when the voltage fluctuation of a certain node exceeds the threshold, the fault early warning process is triggered immediately.

[0136] Fault location and diagnosis algorithm:

[0137] Comparison of electrical parameter differences: when an abnormal voltage node is found, the module first retrieves the electrical parameters of its adjacent nodes at the same time, calculates the difference of voltage deviation, phase difference, power fluctuation and other indicators, for example, when the voltage of a certain node drops by 15% and the voltage of the adjacent node only drops by 3%, it is preliminarily determined that there is a local fault in the node.

[0138] Historical case library matching: after fault feature extraction, the system automatically retrieves the historical fault case library (containing 500+ actual fault cases) and calculates the similarity using a fuzzy matching algorithm; if the similarity is more than 80%, the corresponding fault type handling plan is directly called.

[0139] Fault Type Discrimination Tree: For complex faults, the system uses a multi-layer discrimination tree algorithm for diagnosis; the first layer classifies according to the voltage fluctuation direction (rise / fall), the second layer combines power trend judgment, and the third layer refines the fault type through harmonic content analysis; for example, when the voltage of a certain node rises and is accompanied by a sharp increase in harmonic content, it is determined to be a "harmonic resonance fault" through discrimination tree analysis.

[0140] Fault Recovery Process:

[0141] Fault Isolation: After diagnosing the fault, the system immediately sends instructions to the hierarchical collaborative control execution module to quickly disconnect the fault node from the power grid, preventing the spread of the fault.

[0142] Dominant Node Reconstruction: After troubleshooting, the module re-evaluates the voltage sensitivity and operational reliability indicators of each node in the cluster based on the latest results of the adaptive cluster division module; using an improved analytic hierarchy process (AHP), it considers 6 dimensions and 18 indicators to re-elect the dominant node.

[0143] Control Strategy Adaptive Adjustment: Based on the new cluster structure and dominant node configuration, the system automatically reconstructs the collaborative control strategy; for example, for the cluster after fault recovery, it reallocates reactive power reserve capacity and adjusts voltage regulation priority to ensure the system quickly recovers and operates stably.

[0144] Collaboration Mechanism with Other Modules:

[0145] Data Acquisition: Real-time acquisition of electrical parameters, while feeding back special data requirements during faults, such as increasing the sampling frequency of fault nodes.

[0146] Adaptive Cluster Division: After fault recovery, the control network is reconstructed based on the latest cluster division results to ensure the rationality of node affiliation.

[0147] Hierarchical Collaborative Control Execution: Linkage to execute fault isolation, device restart, etc., to achieve closed-loop control of fault handling.

[0148] Technical Effects and Advantages: The application of this module significantly improves the system's fault handling capability.

[0149] III. Analysis of Technical Collaboration and Combination Effects:

[0150] Each technical module does not operate in isolation, but through data sharing and functional complementation, forming a strong synergistic effect; the data acquisition module provides basic data for other modules, the adaptive cluster division module optimizes the control unit based on electrical distance and voltage fluctuation trend, the robust dominant node election module ensures efficient transmission of instructions, the three-dimensional voltage state evaluation module provides decision-making basis for hierarchical collaborative control, and the fault diagnosis and recovery module ensures reliable operation of the system.

[0151] Through multi-module collaborative innovation, the key problem of distributed photovoltaic access to the power grid has been effectively solved; the system has achieved technological breakthroughs in data acquisition and processing, cluster division, node election, status assessment, collaborative control and fault handling, and has been verified to have significantly improved voltage stability, energy utilization efficiency and system reliability.

[0152] Example 2: Based on Example 1, this embodiment of the present application provides a distributed photovoltaic optimization and coordination control method, the overall idea is as follows:

[0153] Collect real-time voltage data, power flow data, weather forecast data, and grid load forecast data for each node in the distributed photovoltaic system, and calculate the electrical distance value, voltage sensitivity value, voltage deviation value, and voltage fluctuation trend value between nodes;

[0154] Based on the electrical distance value and the voltage fluctuation trend value, a fuzzy clustering algorithm with constraints is used to divide the distributed photovoltaic nodes into multiple control clusters;

[0155] By comprehensively evaluating the voltage sensitivity value and operational reliability of the nodes, the dominant control node is determined from each cluster;

[0156] Based on the voltage deviation value, voltage change rate value and prediction error value, each cluster status is divided into three categories: emergency over-limit, warning over-limit and normal operation, and the corresponding control priority is determined;

[0157] Enable the corresponding collaborative control strategy for the divided categories, as follows;

[0158] For emergency over-limit clusters, perform joint optimization control of active power and reactive power;

[0159] For the warning over-limit cluster, a flexible coupling control relationship is established with the adjacent normal cluster;

[0160] For normal clusters, reactive reserve capacity is pre-allocated based on forecast data;

[0161] Among them, the control cluster division and collaborative control strategy work together: when the system operating state changes, the adaptive cluster division step recalculates and adjusts the cluster structure, and the hierarchical collaborative control step optimizes the collaborative control strategy between clusters based on the updated cluster division results to achieve rapid and stable regulation of the distributed photovoltaic system voltage.

[0162] It should be pointed out finally that the above embodiments are merely examples for clearly illustrating the present application and are not intended to limit the embodiments. Based on the above description, other different forms of changes or variations can be made by those skilled in the art. It is unnecessary and impossible to enumerate all the embodiments. The changes or variations derived therefrom are still within the protection scope of the present application.

Claims

1. A distributed photovoltaic optimization and coordination control system, characterized in that: The control system includes: The multi-dimensional data acquisition module collects real-time operating data, weather forecast data, and grid load forecast data for each node in the distributed photovoltaic system. It also calculates the electrical distance between nodes, voltage sensitivity, voltage deviation, and voltage fluctuation trend based on historical data. Adaptive cluster division module, which divides distributed photovoltaic nodes into multiple control clusters based on electrical distance values ​​and voltage fluctuation trend values ​​using a fuzzy clustering algorithm; The robust dominant node election module selects the dominant control node from each cluster by comprehensively evaluating the voltage sensitivity value and operational reliability of each node; The three-dimensional voltage status assessment module builds an assessment model based on voltage deviation values, voltage change rate values, and prediction error values. It divides each cluster status into three categories: emergency over-limit, warning over-limit, and normal operation, and generates corresponding control priorities. The hierarchical collaborative control execution module executes the corresponding collaborative control strategy according to the type of cluster division, and recalculates and adjusts the cluster structure when the system operation status changes. At the same time, based on the updated cluster division results, it adjusts the collaborative control strategy between clusters.

2. A distributed photovoltaic optimization and coordination control system according to claim 1, characterized in that: The fuzzy clustering algorithm of the adaptive clustering module has a constraint condition. The constraint condition is that the voltage change difference between any two nodes in the same cluster does not exceed the preset voltage change threshold. The specific execution process of the fuzzy clustering algorithm is as follows: Initialize cluster centers and node membership matrices; Calculate the weighted distance from each node to each cluster center, where the weight is dynamically determined based on the node's electrical distance value and voltage fluctuation trend value; The cluster centers and membership matrices are updated iteratively until the preset convergence conditions are met; Verify the voltage correlation of nodes in each cluster. If the voltage variation difference of nodes in the same cluster exceeds a preset voltage variation threshold, re-divide the cluster.

3. A distributed photovoltaic optimization and coordination control system according to claim 1, characterized in that: The process of the hierarchical collaborative control execution module establishing the flexible coupling control relationship between the warning limit-exceeding cluster and the adjacent normal cluster is as follows: Real-time monitoring of voltage interactions between clusters, calculation and update of the inter-cluster voltage sensitivity matrix, which measures the impact of reactive power changes in one cluster on the voltage of another cluster; When the voltage of the warning-exceeding cluster reaches the preset warning voltage threshold, the coordinated voltage regulation mechanism of the adjacent normal cluster is triggered; According to the electrical connection characteristics and voltage sensitivity values ​​between clusters, and in accordance with pre-set weight allocation rules, the responsibility weights of each adjacent cluster in the collaborative voltage regulation process are dynamically allocated to form an adaptive voltage support network.

4. A distributed photovoltaic optimization and coordination control system according to claim 1, characterized in that: The adaptive cluster division module and the hierarchical collaborative control execution module form a closed-loop control through real-time data interaction. The collaborative working mechanism between the two is as follows: When the grid topology changes or the operating status of the photovoltaic system changes, the cluster structure is re-divided within the first preset time; Based on the updated cluster division result, the control parameters of each cluster are adjusted within a second preset time, and the flexible coupling control relationship is updated.

5. A distributed photovoltaic optimization and coordination control system according to claim 1, characterized in that: The collaborative control strategy in the hierarchical collaborative control execution module includes the following: For emergency over-limit clusters, active power and reactive power joint optimization control is performed. By predicting the voltage change trend in the future period, the maximum power point tracking mode, active power reduction ratio and reactive power output capacity of the photovoltaic inverter are dynamically adjusted; For the warning over-limit cluster, a flexible coupling control relationship is established with the adjacent normal cluster by dynamically adjusting the reactive power droop control coefficient; For normal clusters, reactive reserve capacity is pre-allocated based on forecast data; The joint optimization control of active power and reactive power for the emergency over-limit cluster specifically includes: Establish a multi-objective optimization model that includes minimizing voltage deviation, maximizing reactive power reserve, and minimizing active power curtailment; Dynamically adjust the priority weight of each optimization goal according to the current operating status of the system; A predictive control algorithm is used to solve the multi-objective optimization model, generate an optimal control sequence for multiple future control cycles, and adjust the active output and reactive compensation capacity of the photovoltaic inverter.

6. A distributed photovoltaic optimization and coordination control system according to claim 1, characterized in that: The specific methods used by the robust leading node election module to evaluate node operation reliability include: Count the frequency, amplitude, and duration of voltage fluctuations at the node over a period of time to assess node voltage stability; Calculate the number of communication interruptions, average data transmission delay time, and data packet loss rate of the node communication link in the past period of time to evaluate the stability of the node communication link; Count the number of times the node protection device has been actuated and the correct action rate over the past period of time to evaluate the reliability of the node protection device; The above three evaluation results are quantified and comprehensively calculated to obtain the node's operation reliability score, and the node with the highest score is selected as the dominant control node.

7. A distributed photovoltaic optimization and coordination control system according to claim 1, characterized in that: The data acquisition module is also used to send a signal to the hierarchical collaborative control execution module during the night period, or when it detects that the reactive power shortage of the out-of-limit cluster reaches a preset reactive power shortage threshold, to switch the photovoltaic inverter of the out-of-limit cluster to an operating mode specifically for fast reactive power compensation, thereby controlling the voltage of the photovoltaic microgrid.

8. A distributed photovoltaic optimization and coordination control system according to claim 1, characterized in that: The specific criteria for cluster status classification in the three-dimensional voltage status assessment module are as follows: When the voltage deviation value is greater than the first preset voltage deviation threshold, and the voltage change rate value is greater than the first preset voltage change rate threshold, an emergency over-limit instruction is issued to divide the cluster into an emergency over-limit cluster; When the voltage deviation value is greater than the second preset voltage deviation threshold and not greater than the first preset voltage deviation threshold, or the voltage change rate value is greater than the second preset voltage change rate threshold and not greater than the first preset voltage change rate threshold, a warning limit-exceeding instruction is issued and the cluster is divided into a warning limit-exceeding cluster; On the contrary, when the voltage deviation value is not greater than the second preset voltage deviation threshold, and the voltage change rate value is not greater than the second preset voltage change rate threshold, a normal instruction is issued to classify the cluster into a normal cluster.

9. A distributed photovoltaic optimization and coordination control system according to claim 1, characterized in that: The control system also includes a fault diagnosis and recovery module for real-time monitoring of the operating status of each node in the system; When abnormal voltage fluctuations are detected at a node, the fault type and location are determined by comparing the electrical parameter differences between the node and adjacent nodes and combining them with the historical fault case library; After troubleshooting, the dominant nodes are automatically reallocated and the cooperative control strategy is reconstructed according to the latest results of the adaptive cluster partitioning module.

10. A distributed photovoltaic optimization and coordination control method, characterized in that: The method comprises the following steps: Collect real-time voltage data, power flow data, weather forecast data, and grid load forecast data for each node in the distributed photovoltaic system, and calculate the electrical distance value, voltage sensitivity value, voltage deviation value, and voltage fluctuation trend value between nodes; Based on the electrical distance values ​​and voltage fluctuation trend values, a fuzzy clustering algorithm with constraints is used to divide the distributed photovoltaic nodes into multiple control clusters. By comprehensively evaluating the voltage sensitivity value and operational reliability of the nodes, the dominant control node is determined from each cluster; Based on the voltage deviation value, voltage change rate value and prediction error value, each cluster status is divided into three categories: emergency over-limit, warning over-limit and normal operation, and the corresponding control priority is determined; For emergency over-limit clusters, perform joint optimization control of active power and reactive power; For the warning over-limit cluster, a flexible coupling control relationship is established with the adjacent normal cluster; For normal clusters, reactive reserve capacity is pre-allocated based on forecast data; Among them, when the system operation state changes, the adaptive cluster division step recalculates and adjusts the cluster structure, and the hierarchical collaborative control step adjusts the collaborative control strategy between clusters based on the updated cluster division results.

Images