Virtual power plant distributed resource cluster abnormity monitoring system, method and equipment

Abstract

The invention relates to the technical field of virtual power plant operation monitoring, and discloses a virtual power plant distributed resource cluster anomaly monitoring system, method and equipment, and the system comprises a resource node data collection module, an anomaly trend score calculation module, an association graph model construction module, a single node influence calculation module and a comprehensive anomaly early warning output module. In the prior art, a single-point alarm mode depends on a fixed threshold value or a rule base, and especially under the conditions that distributed resources in a virtual power plant are complex in type and space coupling and controller sharing exist between nodes, accurate identification and dynamic early warning of trend anomalies, cooperative faults and potential propagation paths are difficult to realize. According to the method and the device, the multi-dimensional parameter modeling and the graph neural network anomaly propagation analysis are fused, so that the anomaly in the distributed resource cluster is identified and dynamically evaluated, and the early warning accuracy and the intelligent operation and maintenance efficiency of the virtual power plant are improved.

Description

Technical Field

[0001] The present invention belongs to the technical field of virtual power plant operation monitoring, and in particular relates to a virtual power plant distributed resource cluster abnormality monitoring system, method and equipment. Background Art

[0002] Currently, virtual power plants (VPPs), a key technology for distributed energy aggregation and flexible scheduling, have widely integrated a variety of distributed resources, including photovoltaics, energy storage, electric vehicles, and adjustable loads. Given the widespread distribution, complex structure, and ever-changing state of these resources, real-time monitoring of their operating status and identification of anomalies are crucial for ensuring the stable operation of VPPs. However, existing monitoring technologies still primarily rely on single-point data collection and static rule-based judgment. For example, systems typically determine whether a node is abnormal by setting fixed thresholds, such as triggering an alarm when the temperature exceeds a certain value, or by comparing historical device logs to identify abnormal behavior. These approaches have significant limitations: First, they cannot identify trending anomalies, such as "sub-healthy" conditions where device parameters remain within their limits but continuously fluctuate; second, they lack the ability to model inter-node relationships, making it difficult to identify coordinated failures and anomaly propagation paths; and third, the system's response to anomalies in a large number of heterogeneous nodes is often delayed, making dynamic updates and coordinated alerts impossible. Especially as virtual power plants continue to expand in size and the number of nodes increases dramatically, traditional methods cannot fully meet the needs of real-time processing and intelligent identification of high-frequency, multi-source, and strongly correlated data. They suffer from problems such as response delays, high false alarm rates, and missed risks, which seriously affect scheduling safety and system stability. Therefore, there is an urgent need for a new distributed resource anomaly monitoring method that can achieve fusion identification of single-point and group anomalies, trend warning, and propagation path analysis despite strong resource heterogeneity, spatial-structural coupling between nodes, and dynamic evolution of operating states. This approach can improve the anomaly perception, intelligent warning, and scheduling decision-making efficiency of virtual power plants. Summary of the Invention

[0003] In response to the above-mentioned technical deficiencies, the purpose of the present invention is to propose a virtual power plant distributed resource cluster abnormality monitoring system, method and equipment, aiming to solve the existing technology of relying on fixed thresholds or rule bases to achieve single-point alarms, especially in virtual power plants where the distributed resource types are complex, there is spatial coupling and controller sharing between nodes, and it is difficult to achieve accurate identification and dynamic warning of trend anomalies, collaborative failures and potential propagation paths. Technical problems.

[0004] In order to solve the above technical problems, the present invention adopts the following technical solutions: The present invention provides a virtual power plant distributed resource cluster abnormality monitoring system, The virtual power plant distributed resource cluster abnormality monitoring system includes: The resource node data acquisition module is used to collect the real-time operating parameters of various distributed resource nodes in the virtual power plant, and standardize the real-time operating parameters to obtain multi-dimensional node operation feature vectors; The abnormal trend score calculation module is used to calculate the node abnormal trend score based on the multi-dimensional node operation feature vector using an abnormal scoring method based on the trend change rate, time series covariance analysis and local abnormal factor fusion; The association graph model building module is used to obtain the physical structure information of the node and form a spatial structure behavior association graph model of the distributed resource cluster based on the node abnormal trend score and the node physical structure information; A single-node impact calculation module is used to calculate the propagation impact score of a single node based on the constructed spatial structure behavior association graph model combined with the graph neural network; The comprehensive anomaly warning output module is used to integrate the node anomaly trend score and the propagation impact score, calculate the node comprehensive risk score, divide the warning level according to the comprehensive risk score, and output the warning result.

[0005] Preferably, in the resource node data acquisition module, the real-time operating parameters include active power, energy storage battery temperature, three-phase voltage fluctuation rate and communication packet loss rate; active power is used to indicate the output power of the device at time t, the energy storage battery temperature is used to indicate the temperature of the current battery unit of the node, the three-phase voltage fluctuation rate is used to indicate the standard deviation of the voltage change of the node, and the communication packet loss rate is used to indicate the communication data packet loss rate between the node and the dispatching center.

[0006] Preferably, in the abnormal trend score calculation module, the step of calculating the node abnormal trend score based on the multi-dimensional node operation feature vector using an abnormal scoring method based on the trend change rate, time series covariance analysis and local abnormal factor fusion specifically includes: Based on the multi-dimensional node operation feature vector, the historical observation values of each node at the current moment and the previous moment are extracted to construct a multi-dimensional time series including power series, temperature series, voltage fluctuation series and communication packet loss series; Based on the multidimensional time series, the normalized trend change rate is used to calculate the current node fluctuation trend. According to the current node fluctuation trend, the instantaneous mutation score is calculated. Based on the multidimensional time series, the covariance matrix is constructed to calculate the deviation of the collaborative relationship within the current time window. Based on the multidimensional time series, the density detection method is used to calculate the behavioral consistency score of the current node and its neighboring nodes. The instantaneous mutation score, collaborative relationship deviation and behavioral consistency score are weighted and fused to form the final node abnormal trend score.

[0007] Preferably, in the association graph model construction module, the step of forming a spatial structure behavior association graph model of a distributed resource cluster based on the node abnormality trend score and the node physical structure information specifically includes: Obtain the physical coordinates between nodes from the node physical structure information, and calculate the geographical distance based on the physical coordinates between nodes; Determine whether two nodes are controlled by the same controller based on their geographical distance and form a structural coupling tag value; The behavioral synergy strength correlation coefficient is calculated based on the node abnormal trend score, and a spatial structural behavioral association graph model of distributed resource clusters is formed according to the behavioral synergy strength correlation coefficient, structural coupling mark value and geographical distance.

[0008] Preferably, in the association graph model construction module, the calculation formula of the synergy strength correlation coefficient is: in, is the correlation coefficient of the synergy strength between node i and node j, is the structural coupling mark value between node i and node j, Score the node abnormal trend between node i and node j, 、 and is the weight adjustment coefficient, is the geographical distance between node i and node j.

[0009] Preferably, in the single node impact calculation module, the step of calculating the propagation impact score of a single node based on the constructed spatial structure behavior association graph model combined with the graph neural network specifically includes: Based on the constructed spatial structure behavior association graph model, a target node n is selected. With the target node n as the center, its neighbor node set and influence channel are extracted. The influence channel includes geographical distance, structural coupling mark value and node abnormal trend score. Update the current state of each node in the neighbor node set according to the influence channel, and use the graph attention mechanism to calculate the influence weight of any node j in the neighbor node set on the target node n; The abnormal information of the neighbor node set is aggregated through the influence weight, and the target node n is updated to obtain the propagation influence score.

[0010] Preferably, in the comprehensive abnormality warning output module, the steps of fusing the node abnormality trend score and the propagation impact score, calculating the node comprehensive risk score, and dividing the warning level according to the comprehensive risk score and outputting the warning result specifically include: Integrate node anomaly trend score and propagation impact score to calculate node comprehensive risk score , ,in, The node abnormality trend score of the target node n at time t is used to reflect whether the node's own operating status has fluctuated in the near future; Score the propagation influence of the target node n to reflect the potential influence of the node in the spatial structure graph; It is the fusion weight coefficient, which is used to indicate the proportion of attention paid to trends and spreads, and is adjusted based on historical operation and maintenance data or experience; Based on the comprehensive risk score of the node Classify warning levels, generate warning results and push them to the front end. The warning levels include: when When , the node status is determined to be normal; when When the node status is judged as moderate warning, manual inspection or configuration of preventive tasks are recommended; when When the node status is judged as a high-risk abnormality, the automatic isolation logic is linked or the emergency operation and maintenance process is started.

[0011] The present invention also provides a method for monitoring abnormalities of a distributed resource cluster of a virtual power plant, comprising: Step S10: collecting real-time operating parameters of various distributed resource nodes in the virtual power plant, and standardizing the real-time operating parameters to obtain a multi-dimensional node operating feature vector; Real-time operating parameters include active power, energy storage battery temperature, three-phase voltage fluctuation rate, and communication packet loss rate. Active power represents the output power of the device at time t, energy storage battery temperature represents the current battery cell temperature of the node, three-phase voltage fluctuation rate represents the standard deviation of the node's voltage change, and communication packet loss rate represents the communication data packet loss rate between the node and the dispatch center. Step S20: Based on the multi-dimensional node operation feature vector, an abnormality scoring method based on the trend change rate, time series covariance analysis and local abnormality factor fusion is used to calculate the node abnormality trend score; The steps for calculating the node abnormal trend score based on the multi-dimensional node operation feature vector using an abnormal scoring method based on the trend change rate, time series covariance analysis and local abnormal factor fusion include: Based on the multi-dimensional node operation feature vector, the historical observation values of each node at the current moment and the previous moment are extracted to construct a multi-dimensional time series including power series, temperature series, voltage fluctuation series and communication packet loss series; Based on the multidimensional time series, the normalized trend change rate is used to calculate the current node fluctuation trend. According to the current node fluctuation trend, the instantaneous mutation score is calculated. Based on the multidimensional time series, the covariance matrix is constructed to calculate the deviation of the collaborative relationship within the current time window. Based on the multidimensional time series, the density detection method is used to calculate the behavioral consistency score of the current node and its neighboring nodes. The instantaneous mutation score, collaborative relationship deviation, and behavioral consistency score are weighted and integrated to form the final node abnormal trend score; Step S30: Obtain node physical structure information, and form a spatial structure behavior association graph model of the distributed resource cluster based on the node abnormality trend score and the node physical structure information; Step S40: Calculate the propagation impact score of a single node based on the constructed spatial structure behavior association graph model combined with the graph neural network; Step S50: Integrate the node abnormal trend score and the propagation impact score to calculate the node comprehensive risk score, divide the warning level according to the comprehensive risk score, and output the warning result.

[0012] The present invention also provides a virtual power plant distributed resource cluster abnormality monitoring device, including: a memory, a processor, and a virtual power plant distributed resource cluster abnormality monitoring program stored on the memory and capable of running on the processor. When the virtual power plant distributed resource cluster abnormality monitoring program is executed by the processor, a virtual power plant distributed resource cluster abnormality monitoring method is implemented.

[0013] The beneficial effect of the present invention is that compared with the existing technology that relies on fixed thresholds or rule bases to achieve single-point alarms, it is difficult to achieve accurate identification and dynamic early warning of trend anomalies, collaborative failures and potential propagation paths, especially under the conditions of complex distributed resource types, spatial coupling and controller sharing between nodes in virtual power plants. Because this application integrates multi-dimensional parameter modeling and graph neural network anomaly propagation analysis to construct a set of spatial structure and behavior integrated anomaly monitoring mechanism, it realizes the identification and dynamic evaluation of anomalies in distributed resource clusters, and improves the accuracy of early warning and intelligent operation and maintenance efficiency of virtual power plants. BRIEF DESCRIPTION OF THE DRAWINGS

[0014] In order to more clearly illustrate the embodiments of the present invention or the technical solutions in the prior art, the following briefly introduces the drawings required for use in the embodiments or the description of the prior art. Obviously, the drawings described below are only some embodiments of the present invention. For ordinary technicians in this field, other drawings can be obtained based on these drawings without paying any creative work.

[0015] Figure 1This is a system diagram of a first embodiment of a virtual power plant distributed resource cluster abnormality monitoring system of the present invention.

[0016] Figure 2 This is a schematic diagram of equipment for a virtual power plant distributed resource cluster abnormality monitoring system according to the present invention. DETAILED DESCRIPTION

[0017] The following will clearly and completely describe the technical solutions in the embodiments of the present invention in conjunction with the accompanying drawings. Obviously, the described embodiments are only part of the embodiments of the present invention, not all of the embodiments. Based on the embodiments of the present invention, all other embodiments obtained by ordinary technicians in this field without making creative efforts are within the scope of protection of the present invention.

[0018] Example 1: Figure 1 1 is a flow chart of the first embodiment of the abnormality monitoring system for distributed resource clusters of virtual power plants according to the present invention, which proposes the first embodiment of the abnormality monitoring system for distributed resource clusters of virtual power plants according to the present invention.

[0019] In the first embodiment, as Figure 1 As shown, the virtual power plant distributed resource cluster abnormality monitoring system includes: The resource node data acquisition module is used to collect the real-time operating parameters of various distributed resource nodes in the virtual power plant, and standardize the real-time operating parameters to obtain multi-dimensional node operation feature vectors; It should be noted that in the resource node data acquisition module, the real-time operating parameters include active power, energy storage battery temperature, three-phase voltage fluctuation rate and communication packet loss rate; active power is used to indicate the output power of the device at time t, energy storage battery temperature is used to indicate the temperature of the current battery unit of the node, three-phase voltage fluctuation rate is used to indicate the standard deviation of the voltage change of the node, and communication packet loss rate is used to indicate the communication data packet loss rate between the node and the dispatching center.

[0020] It is understandable that through standardization processing, such as normalization, the scale inconsistency caused by differences in device types can be effectively eliminated, and the model's adaptability to different types of resource nodes can be enhanced, thereby improving the generalization effect and robustness of the anomaly detection algorithm.

[0021] It should be understood that this module has good interface compatibility and protocol expansion capabilities, can connect to the data output interface of mainstream smart meters, inverters, RTU modules or edge controllers, and has industrial feasibility for online deployment and low-latency sampling.

[0022] For example, in an actual virtual power plant deployment, the resource node data acquisition module collects power, voltage, and temperature data every 30 seconds, and performs preliminary standardized processing at the acquisition end to achieve decoupled operation of edge computing and the main center, providing a stable data foundation for subsequent trend analysis and anomaly judgment.

[0023] The abnormal trend score calculation module is used to calculate the node abnormal trend score based on the multi-dimensional node operation feature vector using an abnormal scoring method based on the trend change rate, time series covariance analysis and local abnormal factor fusion; It should be noted that in the abnormal trend score calculation module, the steps for calculating the node abnormal trend score based on the multi-dimensional node operation feature vector using an abnormal scoring method based on the trend change rate, time series covariance analysis and local abnormal factor fusion are as follows: Based on the multi-dimensional node operation feature vector, the historical observation values of each node at the current moment and the previous moment are extracted to construct a multi-dimensional time series including power series, temperature series, voltage fluctuation series and communication packet loss series; Based on the multidimensional time series, the normalized trend change rate is used to calculate the current node fluctuation trend. According to the current node fluctuation trend, the instantaneous mutation score is calculated. Based on the multidimensional time series, the covariance matrix is constructed to calculate the deviation of the collaborative relationship within the current time window. Based on the multidimensional time series, the density detection method is used to calculate the behavioral consistency score of the current node and its neighboring nodes. Among them, the degree of trend mutation is calculated by normalized first-order difference, and the formula is: in, is the time series value of the current node t obtained from the current node fluctuation trend, is the time series value of the node at time t-1, is a non-zero constant; The calculation formula for the deviation from the synergy relationship is: in, is the Frobenius norm function, Used to represent the covariance matrix, is the preset benchmark covariance matrix; The behavioral consistency score is calculated as follows: The instantaneous mutation score, collaborative relationship deviation and behavioral consistency score are weighted and fused to form the final node abnormal trend score.

[0024] It is understandable that the trend scoring module does not judge anomalies based on a single indicator alone, but integrates the abnormal characteristics of multiple time dimensions, structural dimensions and group behavior dimensions to achieve a more comprehensive trend perception; this scoring is not only applicable to "extreme value exceeding the limit" scenarios, but is also more suitable for identifying "boundary wandering" or "gradual deterioration" abnormal states.

[0025] For example, in a virtual power plant test case, the power of a certain energy storage node dropped by 18% and the temperature continued to rise in the past 10 minutes, but did not cross the hard threshold. The traditional monitoring system judged it to be "normal". However, after the fusion score calculation of this module, the abnormal trend score reached 0.68, successfully judging the node as "sub-healthy" and triggering a moderate warning in advance, greatly improving the timeliness and accuracy of abnormality identification.

[0026] The association graph model building module is used to obtain the physical structure information of the node and form a spatial structure behavior association graph model of the distributed resource cluster based on the node abnormal trend score and the node physical structure information; It should be noted that in the association graph model construction module, the steps of forming a spatial structure behavior association graph model of a distributed resource cluster based on the node anomaly trend score and the node physical structure information specifically include: obtaining the physical coordinates between nodes from the node physical structure information, and calculating the geographical distance based on the physical coordinates between the nodes; judging whether the two nodes are controlled by the same controller based on the geographical distance, and forming a structural coupling mark value; calculating the behavioral synergy strength correlation coefficient based on the node anomaly trend score, and forming a spatial structure behavior association graph model of a distributed resource cluster based on the behavioral synergy strength correlation coefficient, the structural coupling mark value and the geographical distance.

[0027] In the association graph model construction module, the calculation formula for the synergy strength correlation coefficient is: in, is the correlation coefficient of the synergy strength between node i and node j, is the structural coupling mark value between node i and node j, Score the node abnormal trend between node i and node j, 、 and is the weight adjustment coefficient, is the geographical distance between node i and node j.

[0028] It should be noted that the structure in the spatial structure behavior association graph model includes: a node set consisting of physical resource nodes and an edge set reflecting the connection relationship between nodes. Among them, the node set is used to represent each distributed resource unit (such as photovoltaic, energy storage, load nodes, etc.), and each node contains its physical location, equipment attributes, and an abnormal trend score calculated based on time series; the edge set is used to characterize the spatial connection relationship and behavioral similarity between nodes. The weight of the edge is determined by the physical topological relationship and the trend score similarity, forming a spatial association graph with both physical structure and behavioral characteristics, which is used to reveal the abnormal transmission path and co-evolution pattern between resource nodes.

[0029] It can be understood that this module significantly enhances the graph model's ability to depict the collaborative relationships between actual devices by integrating the three factors of "spatial proximity", "structural commonality" and "behavioral correlation", thereby avoiding the limitations of traditional graph models that are only based on physical connections or topological structures and lack behavioral characterization.

[0030] It should be understood that this graph model is not only suitable for static structure modeling, but also for graph structure updates during dynamic system evolution, such as node online / offline, communication structure changes, and other scenarios. It has good flexibility and scalability.

[0031] For example, in a scenario where photovoltaic energy storage operates collaboratively in a certain region, multiple energy storage nodes are physically close to each other and share an inverter controller. Furthermore, they exhibit a synchronized power decline trend during high-temperature weather, with a synergistic strength correlation coefficient of 0.86. These nodes are automatically mapped as strongly connected paths and identified as high-risk abnormal channels in subsequent propagation models, effectively avoiding missed detections caused by isolated judgments.

[0032] A single-node impact calculation module is used to calculate the propagation impact score of a single node based on the constructed spatial structure behavior association graph model combined with the graph neural network; It should be noted that in the single-node impact calculation module, the steps for calculating the propagation impact score of a single node based on the constructed spatial structure behavior association graph model combined with the graph neural network include: Based on the constructed spatial structure behavior association graph model, a target node n is selected. With the target node n as the center, its neighbor node set and influence channel are extracted. The influence channel includes geographical distance, structural coupling mark value and node abnormal trend score. Update the current state of each node in the neighbor node set according to the influence channel, and use the graph attention mechanism to calculate the influence weight of any node j in the neighbor node set on the target node n; The abnormal information of the neighbor node set is aggregated through the influence weight, and the target node n is updated to obtain the propagation influence score.

[0033] The current state is updated by comprehensively constructing a state representation for each neighboring node based on its current anomaly trend score, connected structural attribute information (such as device type and capacity level), and spatial distance from the target node. Dynamic adjustments are made based on trend data from the current time window, ensuring that the neighboring node state accurately reflects its immediate influence on the target node. Aggregating anomaly information from a set of neighboring nodes using influence weights is done by weightedly merging the state information of neighboring nodes based on the influence weights of each neighboring node, as determined by the graph attention mechanism, to obtain a comprehensive anomaly information representation of the neighboring node set at the current moment. This anomaly information is then combined with the historical state of the target node to update the target node's propagation influence score, thereby reflecting the potential anomaly transmission risk it faces within the network structure.

[0034] It is understandable that by introducing the attention mechanism of graph neural networks, the module no longer relies on the traditional "equal-weighted average" or "static edge weight model", but instead uses the behavioral characteristics and structural relationships between nodes to adaptively learn the strength of influence, significantly improving the accuracy of identifying abnormal propagation paths.

[0035] It should be understood that the propagation impact score not only reflects whether a node is currently abnormal, but also reflects whether it has the characteristics of a "propagation source" or a "passively affected person" in the network structure. It can be used to identify the essential difference between "isolated anomalies" and "linked anomalies" and effectively assist the power grid dispatching system in identifying high-risk infectious nodes.

[0036] For example, during the operation of a multi-node energy storage system, the power of a node decreased and the anomaly score increased, but its neighboring nodes remained stable, with a propagation impact score of 0.21. The system identified it as an "isolated anomaly"; while another node had an anomaly score of 0.33, and its neighboring nodes had multiple synchronous score increases, with a propagation impact score of 0.78. It was judged as a "potential propagation source node". The system automatically set the priority investigation level and generated control linkage recommendations.

[0037] The comprehensive anomaly warning output module is used to integrate the node anomaly trend score and the propagation impact score, calculate the node comprehensive risk score, divide the warning level according to the comprehensive risk score, and output the warning result.

[0038] It should be noted that the steps in the comprehensive anomaly warning output module, which integrate the node anomaly trend score and the propagation impact score, calculate the node comprehensive risk score, divide the warning level according to the comprehensive risk score, and output the warning result, specifically include: Integrate node anomaly trend score and propagation impact score to calculate node comprehensive risk score , ,in, The node abnormality trend score of the target node n at time t is used to reflect whether the node's own operating status has fluctuated in the near future; Score the propagation influence of the target node n to reflect the potential influence of the node in the spatial structure graph; It is the fusion weight coefficient, which is used to indicate the proportion of attention paid to trends and spreads, and is adjusted based on historical operation and maintenance data or experience; Based on the comprehensive risk score of the node Classify warning levels, generate warning results and push them to the front end. The warning levels include: when When , the node status is determined to be normal; when When the node status is judged as moderate warning, manual inspection or configuration of preventive tasks are recommended; when When the node status is judged as a high-risk abnormality, the automatic isolation logic is linked or the emergency operation and maintenance process is started.

[0039] It is understandable that this module integrates risk information from two aspects: single-point trend anomalies and network structure influence. It can not only accurately identify potential failures of the equipment itself, but also determine whether it is the "source of abnormal propagation" in the system, thereby significantly improving the comprehensiveness and accuracy of the early warning.

[0040] It should be understood that comprehensive risk scoring can not only be used to predict node status, but also for system-level abnormal situation analysis, such as: counting the proportion of high-risk nodes, predicting the potential impact range, assisting in generating scheduling priority maps, etc., which helps to achieve active safety control and intelligent operation and maintenance upgrades of virtual power plants.

[0041] For example, in a city-level virtual power plant monitoring platform, the abnormal trend score of a certain energy storage node was 0.41, and the propagation impact score was 0.52. After weight fusion, the risk score was 0.488. The system automatically marked it as a "moderate warning" and added it to the "preventive maintenance queue task". Inspectors were arranged in advance to handle the situation on site, and finally the battery was replaced before a fault occurred, avoiding the risk of system shutdown.

[0042] As mentioned above, all calculation parameters used in this patent for processes such as propagation impact scoring, node evaluation, and optimized scheduling are dimensionless. For example, original indicators with physical dimensions, such as power, electricity price, anomaly score, and distance, are converted into dimensionless forms with a [0,1] interval or zero mean and unit variance through normalization, standard deviation scaling, or range mapping. This method can effectively eliminate the dimensional differences between parameters in each dimension, preventing certain features from dominating the model calculation process due to their large numerical magnitude, thereby improving the rationality of parameter fusion and the robustness of optimization calculations.

[0043] Embodiment 2: In addition, the present invention provides a method for monitoring abnormalities in a distributed resource cluster of a virtual power plant, which adopts a system for monitoring abnormalities in a distributed resource cluster of a virtual power plant in the above embodiment, and can solve the technical problem of monitoring abnormalities in a distributed resource cluster of a virtual power plant. Compared with the prior art, the beneficial effects of the method for monitoring abnormalities in a distributed resource cluster of a virtual power plant provided by the present invention are the same as the beneficial effects of the system for monitoring abnormalities in a distributed resource cluster of a virtual power plant provided by the above embodiment, and the other technical features of the method for monitoring abnormalities in a distributed resource cluster of a virtual power plant are the same as the features disclosed in the above embodiment method, and are not described in detail here.

[0044] Example 3: The present invention provides a virtual power plant distributed resource cluster abnormality monitoring device, please refer to Figure 2A device for monitoring abnormalities in a distributed resource cluster of a virtual power plant includes: at least one processor; and a memory communicatively connected to the at least one processor; wherein the memory stores instructions executable by the at least one processor, and the instructions are executed by the at least one processor to enable the at least one processor to perform the method for monitoring abnormalities in a distributed resource cluster of a virtual power plant described in the first embodiment. The device for monitoring abnormalities in a distributed resource cluster of a virtual power plant in this embodiment of the present invention may include, but is not limited to, mobile terminals such as mobile phones, laptops, digital broadcast receivers, PDAs (Personal Digital Assistants), PADs (Portable Application Descriptions), PMPs (Portable Media Players), in-vehicle terminals (e.g., in-vehicle navigation terminals), and fixed terminals such as digital TVs and desktop computers. The device for monitoring abnormalities in a distributed resource cluster of a virtual power plant is merely an example and should not limit the functionality or scope of use of the embodiments of the present invention. The device for monitoring abnormalities in a distributed resource cluster of a virtual power plant may include a processing device 1001 that can perform various appropriate actions and processes based on a program stored in a read-only memory 1002 or a program loaded from a storage device 1003 into a random access memory 1004. The random access memory 1004 also stores various programs and data required for the operation of a virtual power plant distributed resource cluster abnormality monitoring device. The processing device 1001, read-only memory 1002, and random access memory 1004 are interconnected via a bus 1005. An I / O interface 1006 is also connected to the bus. Typically, the following systems may be connected to the I / O interface 1006: an input device 1007 including, for example, a touchscreen, touchpad, keyboard, mouse, image sensor, microphone, accelerometer, gyroscope, etc.; an output device 1008 including, for example, a liquid crystal display (LCD), speaker, vibrator, etc.; storage device 1003; and a communication device 1009. The communication device 1009 allows the virtual power plant distributed resource cluster abnormality monitoring device to communicate wirelessly or wired with other devices to exchange data. While the figure illustrates a virtual power plant distributed resource cluster abnormality monitoring device with various systems, it should be understood that implementation or presence of all illustrated systems is not required. More or fewer systems may alternatively be implemented or present.

[0045] It should be understood that the various parts disclosed in the present invention can be implemented using hardware, software, firmware, or a combination thereof. In the description of the above embodiments, specific features, structures, materials, or characteristics can be combined in any appropriate manner in any one or more embodiments or examples.

[0046] Obviously, those skilled in the art may make various changes and modifications to the present invention without departing from the spirit and scope of the present invention. Thus, if such changes and modifications fall within the scope of the claims and their equivalents, the present invention is intended to include such changes and modifications.

Claims

1. A virtual power plant distributed resource cluster abnormality monitoring system, characterized in that: The system includes: The resource node data acquisition module is used to collect the real-time operating parameters of various distributed resource nodes in the virtual power plant, and standardize the real-time operating parameters to obtain multi-dimensional node operation feature vectors; The abnormal trend score calculation module is used to calculate the node abnormal trend score based on the multi-dimensional node operation feature vector using an abnormal scoring method based on the trend change rate, time series covariance analysis and local abnormal factor fusion; The association graph model building module is used to obtain the physical structure information of the node and form a spatial structure behavior association graph model of the distributed resource cluster based on the node abnormal trend score and the node physical structure information; A single-node impact calculation module is used to calculate the propagation impact score of a single node based on the constructed spatial structure behavior association graph model combined with the graph neural network; The comprehensive anomaly warning output module is used to integrate the node anomaly trend score and the propagation impact score, calculate the node comprehensive risk score, divide the warning level according to the comprehensive risk score, and output the warning result.

2. A virtual power plant distributed resource cluster abnormality monitoring system according to claim 1, characterized in that: In the resource node data acquisition module, real-time operating parameters include active power, energy storage battery temperature, three-phase voltage fluctuation rate, and communication packet loss rate; active power is used to represent the output power of the device at time t, energy storage battery temperature is used to represent the temperature of the node's current battery unit, three-phase voltage fluctuation rate is used to represent the standard deviation of the node's voltage change, and communication packet loss rate is used to represent the communication data packet loss rate between the node and the dispatch center.

3. A virtual power plant distributed resource cluster abnormality monitoring system according to claim 2, characterized in that: In the abnormal trend score calculation module, the steps for calculating the node abnormal trend score are as follows: Based on the multi-dimensional node operation feature vector, the historical observation values of each node at the current moment and the previous moment are extracted to construct a multi-dimensional time series including power series, temperature series, voltage fluctuation series and communication packet loss series; Based on the multidimensional time series, the normalized trend change rate is used to calculate the current node fluctuation trend. According to the current node fluctuation trend, the instantaneous mutation score is calculated. Based on the multidimensional time series, the covariance matrix is constructed to calculate the deviation of the collaborative relationship within the current time window. Based on the multidimensional time series, the density detection method is used to calculate the behavioral consistency score of the current node and its neighboring nodes. The instantaneous mutation score, collaborative relationship deviation and behavioral consistency score are weighted and fused to form the final node abnormal trend score.

4. The virtual power plant distributed resource cluster abnormality monitoring system according to claim 1, characterized in that: In the association graph model construction module, the steps of forming a spatial structure behavior association graph model of a distributed resource cluster based on node anomaly trend scores and node physical structure information include: Obtain the physical coordinates between nodes from the node physical structure information, and calculate the geographical distance based on the physical coordinates between nodes; Determine whether two nodes are controlled by the same controller based on their geographical distance and form a structural coupling tag value; The behavioral synergy strength correlation coefficient is calculated based on the node abnormal trend score, and a spatial structural behavioral association graph model of distributed resource clusters is formed according to the behavioral synergy strength correlation coefficient, structural coupling mark value and geographical distance.

5. A virtual power plant distributed resource cluster abnormality monitoring system according to claim 4, characterized in that: In the association graph model construction module, the calculation formula for the synergy strength correlation coefficient is: in, is the correlation coefficient of the synergy strength between node i and node j, is the structural coupling mark value between node i and node j, Score the node abnormal trend between node i and node j, 、 and is the weight adjustment coefficient, is the geographical distance between node i and node j.

6. A virtual power plant distributed resource cluster abnormality monitoring system according to claim 4, characterized in that: In the single-node impact calculation module, the steps for calculating the propagation impact score of a single node are based on the constructed spatial structure behavior association graph model combined with the graph neural network. Specifically, the steps include: Based on the constructed spatial structure behavior association graph model, a target node n is selected. With the target node n as the center, its neighbor node set and influence channel are extracted. The influence channel includes geographical distance, structural coupling mark value and node abnormal trend score. Update the current state of each node in the neighbor node set according to the influence channel, and use the graph attention mechanism to calculate the influence weight of any node j in the neighbor node set on the target node n; The abnormal information of the neighbor node set is aggregated through the influence weight, and the target node n is updated to obtain the propagation influence score.

7. The virtual power plant distributed resource cluster abnormality monitoring system according to claim 1, characterized in that: In the comprehensive anomaly warning output module, the steps of integrating the node anomaly trend score and the propagation impact score, calculating the node comprehensive risk score, dividing the warning level according to the comprehensive risk score, and outputting the warning result include: Integrate node anomaly trend score and propagation impact score to calculate node comprehensive risk score , ,in, The node abnormality trend score of the target node n at time t is used to reflect whether the node's own operating status has fluctuated in the near future; Score the propagation influence of the target node n to reflect the potential influence of the node in the spatial structure graph; It is the fusion weight coefficient, which is used to indicate the proportion of attention paid to trends and spreads, and is adjusted based on historical operation and maintenance data or experience; Based on the comprehensive risk score of the node Classify warning levels, generate warning results and push them to the front end. The warning levels include: when When , the node status is determined to be normal; when When the node status is judged as moderate warning, manual inspection or configuration of preventive tasks are recommended; when When the node status is judged as a high-risk abnormality, the automatic isolation logic is linked or the emergency operation and maintenance process is started.

8. A method for monitoring abnormalities of a distributed resource cluster of a virtual power plant, applied to a system for monitoring abnormalities of a distributed resource cluster of a virtual power plant according to any one of claims 1 to 7, characterized in that: Methods include: Step S10: collecting real-time operating parameters of various distributed resource nodes in the virtual power plant, and standardizing the real-time operating parameters to obtain a multi-dimensional node operating feature vector; Real-time operating parameters include active power, energy storage battery temperature, three-phase voltage fluctuation rate, and communication packet loss rate. Active power represents the output power of the device at time t, energy storage battery temperature represents the current battery cell temperature of the node, three-phase voltage fluctuation rate represents the standard deviation of the node's voltage change, and communication packet loss rate represents the communication data packet loss rate between the node and the dispatch center. Step S20: Based on the multi-dimensional node operation feature vector, an abnormality scoring method based on the trend change rate, time series covariance analysis and local abnormality factor fusion is used to calculate the node abnormality trend score; The steps for calculating the node abnormal trend score based on the multi-dimensional node operation feature vector using an abnormal scoring method based on the trend change rate, time series covariance analysis and local abnormal factor fusion include: Based on the multi-dimensional node operation feature vector, the historical observation values of each node at the current moment and the previous moment are extracted to construct a multi-dimensional time series including power series, temperature series, voltage fluctuation series and communication packet loss series; Based on the multidimensional time series, the normalized trend change rate is used to calculate the current node fluctuation trend. According to the current node fluctuation trend, the instantaneous mutation score is calculated. Based on the multidimensional time series, the covariance matrix is constructed to calculate the deviation of the collaborative relationship within the current time window. Based on the multidimensional time series, the density detection method is used to calculate the behavioral consistency score of the current node and its neighboring nodes. The instantaneous mutation score, collaborative relationship deviation, and behavioral consistency score are weighted and integrated to form the final node abnormal trend score; Step S30: Obtain node physical structure information, and form a spatial structure behavior association graph model of the distributed resource cluster based on the node abnormality trend score and the node physical structure information; Step S40: Calculate the propagation impact score of a single node based on the constructed spatial structure behavior association graph model combined with the graph neural network; Step S50: Integrate the node abnormal trend score and the propagation impact score to calculate the node comprehensive risk score, divide the warning level according to the comprehensive risk score, and output the warning result.

9. A virtual power plant distributed resource cluster abnormality monitoring device, applied to a virtual power plant distributed resource cluster abnormality monitoring system according to any one of claims 1 to 7, characterized in that: The virtual power plant distributed resource cluster abnormality monitoring device includes: a memory, a processor, and a virtual power plant distributed resource cluster abnormality monitoring program stored in the memory and capable of running on the processor.

Images