Power state evaluation method and system based on power grid model

Abstract

The application discloses a power state evaluation method and system based on a power grid model, and relates to the technical field of power state evaluation.The method comprises the following steps: performing power equipment modeling, configuring a standard grid size based on the monitoring accuracy of the power state; performing equipment structure feature and function feature analysis, and performing adaptive grid distribution according to the importance rating order; performing abnormality analysis by taking the grid as the minimum analysis unit, and constructing a power state abnormality identifier; establishing a first shrinkage constraint and a second shrinkage constraint; performing equipment feature adaptive grid granularity reconstruction; performing power state abnormality identifier verification, and generating a state evaluation result.The method solves the technical problems of the prior art, such as fixed power state evaluation grid division, difficulty in adapting to the importance difference of equipment, and lack of dynamic optimization after abnormality detection, which leads to insufficient power state evaluation accuracy, real-time performance and reliability, and achieves the technical effect of improving the power state evaluation accuracy, real-time performance and reliability.

Description

TECHNICAL FIELD

[0001] The present application relates to the technical field of power state evaluation, and particularly relates to a power state evaluation method and system based on a power grid model. BACKGROUND

[0002] State monitoring and evaluation of a power system is a key to ensuring safe and stable operation of a power grid. Traditional power state evaluation mainly relies on data collection and analysis of fixed monitoring points. However, due to the wide distribution and complex structure of power equipment, and the large differences in functions and importance of different equipment, the existing state evaluation method is difficult to achieve fine and dynamic state evaluation. In addition, with the rapid development of Internet of Things and intelligent sensing technology, the amount of data generated by the power system increases exponentially, making it difficult to efficiently process massive data and accurately identify abnormal states. The existing power grid model usually adopts a fixed-size grid division method, which is difficult to adapt to the importance and functional differences of different equipment, resulting in uneven resource allocation or insufficient local monitoring accuracy. Moreover, the region clustering and grid optimization ability after abnormal detection is insufficient, affecting the accuracy and reliability of the final evaluation results.

[0003] Therefore, in the related art, there is a technical problem that the power state evaluation grid division is fixed, difficult to adapt to the importance difference of equipment, and lacks dynamic optimization after abnormal detection, resulting in insufficient accuracy, real-time performance and reliability of power state evaluation. SUMMARY

[0004] The power state evaluation method and system based on the power grid model provided by the present application solve the technical problem of insufficient accuracy, real-time performance and reliability of power state evaluation due to the fixed power state evaluation grid division, the difficulty to adapt to the importance difference of equipment, and the lack of dynamic optimization after abnormal detection in the prior art, and achieve the technical effect of improving the accuracy, real-time performance and reliability of power state evaluation.

[0005] The application provides a power state evaluation method based on a power grid model, which comprises: performing power equipment modeling of a monitoring area, configuring a standard grid size based on monitoring accuracy of a power state; performing device structure feature and function feature analysis based on the power equipment modeling result, performing importance rating using the analysis result, performing adaptive grid distribution under the standard grid size according to the importance rating order, and establishing an initial grid distribution result; performing abnormal analysis using the initial grid distribution result as a minimum analysis unit, constructing a power state abnormal identifier, and performing regional clustering on the power state abnormal identifier, establishing a first shrinkage constraint according to the regional clustering result, and establishing a second shrinkage constraint of a grid according to the power state abnormal identifier; inputting the first shrinkage constraint, the second shrinkage constraint, and the initial grid distribution result into the power grid model, performing device feature adaptive grid granularity reconstruction, and establishing a reconstructed grid distribution result; and performing power state abnormal identifier verification using the reconstructed grid distribution result, and generating a state evaluation result.

[0006] In a possible implementation, the inputting the first shrinkage constraint, the second shrinkage constraint, and the initial grid distribution result into the power grid model, performing device feature adaptive grid granularity reconstruction, and establishing a reconstructed grid distribution result comprises: after the initial grid distribution result is imported into the power grid model, performing shrinkage grid positioning using the second shrinkage constraint, and establishing a shrinkage grid positioning result; activating a shrinkage analysis layer of the power grid model, performing shrinkage constraint fusion after receiving the first shrinkage constraint and the second shrinkage constraint, and establishing a joint shrinkage constraint; calling a device feature according to the shrinkage grid positioning result, performing grid granularity reconstruction of the shrinkage grid positioning result under device feature adaptation based on the joint shrinkage constraint, and establishing a reconstructed grid distribution result.

[0007] In a possible implementation, the performing grid granularity reconstruction of the shrinkage grid positioning result under device feature adaptation based on the joint shrinkage constraint and establishing a reconstructed grid distribution result comprises: establishing a feature space of a grid in each shrinkage grid positioning result according to the analysis result; after the joint shrinkage constraint is mapped to the feature space, performing grid granularity reconstruction under the joint shrinkage constraint, establishing a benchmark reconstruction result; performing an adaptation decision of the benchmark reconstruction result and the feature according to the feature space, and establishing a compensation reconstruction result using the adaptation decision; and establishing a reconstructed grid distribution result according to the compensation reconstruction result.

[0008] In a possible implementation, the adaptive grid distribution under the standard grid size constraint in the order of the importance rating establishes an initial grid distribution result, including: dividing the importance rating into levels to establish N level spaces, the N level spaces being importance level spaces, N being an integer greater than 1; sequentially distributing grids based on the level space order under the standard grid size constraint, and saving the distributed grids as frozen grids; performing traversal segmentation of each level space order under the constraint of the frozen grids to establish the initial grid distribution result.

[0009] In a possible implementation, the power state anomaly identification verification using the reconstructed grid distribution result generates a state evaluation result, including: performing perception data labeling of each reconstructed grid on the reconstructed grid distribution result; establishing a multi-dimensional state vector set of the reconstructed grid based on the perception data labeling result to generate a reconstructed grid state feature cluster; performing multi-scale cross-validation of the power state anomaly identification using the reconstructed grid state feature cluster, and generating a state evaluation result based on the multi-scale cross-validation result.

[0010] In a possible implementation, the perception data labeling of each reconstructed grid on the reconstructed grid distribution result includes: performing data richness identification on the perception data labeling result to establish a first perception defect; performing multi-dimensional data verification on the perception data labeling result to establish a second perception defect; performing additional data acquisition according to the first perception defect and the second perception defect, and using the additional data acquisition result to compensate the perception data labeling result.

[0011] In a possible implementation, after the regional clustering of the power state anomaly identification, a first shrinkage constraint is established according to the regional clustering result, including: establishing a comprehensive judgment distance of the anomaly, the comprehensive judgment distance including a first distance and a second distance, the first distance being a position Euclidean distance, and the second distance being a similarity distance of the anomaly type; performing clustering identification of the power state anomaly identification according to the comprehensive judgment distance to establish the regional clustering result.

[0012] In a possible implementation, the first shrinkage constraint is established according to the regional clustering result, including: extracting a boundary identification of a clustering cluster according to the regional clustering result to define a region contour; calculating a tightness factor of each clustering cluster using the region contour, the tightness factor representing a concentration degree of the abnormal region; and establishing the first shrinkage constraint using the tightness factor and the regional clustering result.

[0013] In a possible implementation, the generation of the state evaluation result further includes: performing pre-warning signal matching based on the state evaluation result to construct a pre-warning signal set; and performing visual identification of the pre-warning signal set in a monitoring region.

[0014] The application also provides a power state evaluation system based on a power grid modeling model, which is used to implement any of the power state evaluation methods based on the power grid modeling model. The system comprises: a power equipment modeling unit, configured to perform power equipment modeling of a monitoring area, and configure a standard grid size based on monitoring accuracy of a power state; a grid distribution unit, configured to perform device structure feature and function feature analysis based on the power equipment modeling result, perform adaptive grid distribution under the constraint of the standard grid size according to the importance rating order after importance rating using the analysis result, and establish an initial grid distribution result; an anomaly analysis unit, configured to perform anomaly analysis taking a grid as a minimum analysis unit after associating the initial grid distribution result with an original perception data set, and construct a power state anomaly identifier; a shrinkage constraint establishment unit, configured to perform regional clustering on the power state anomaly identifier, establish a first shrinkage constraint according to the regional clustering result, and establish a second shrinkage constraint of a grid according to the power state anomaly identifier; a grid granularity reconstruction unit, configured to input the first shrinkage constraint, the second shrinkage constraint, and the initial grid distribution result into the power grid modeling model, perform device feature adaptive grid granularity reconstruction, and establish a reconstructed grid distribution result; and a state evaluation result generation unit, configured to perform power state anomaly identifier verification using the reconstructed grid distribution result, and generate a state evaluation result.

[0015] The power state evaluation method and system based on the power grid modeling model can perform power equipment modeling, configure a standard grid size based on monitoring accuracy of a power state, perform device structure feature and function feature analysis, perform adaptive grid distribution according to the importance rating order, perform anomaly analysis taking a grid as a minimum analysis unit, construct a power state anomaly identifier, establish a first shrinkage constraint and a second shrinkage constraint, perform device feature adaptive grid granularity reconstruction, perform power state anomaly identifier verification, and generate a state evaluation result. The technical problems of fixed grid division in power state evaluation, difficulty in adapting to device importance differences, and lack of dynamic optimization after anomaly detection, which result in insufficient accuracy, real-time performance, and reliability of power state evaluation, are solved, and the technical effects of improving the accuracy, real-time performance, and reliability of power state evaluation are achieved. BRIEF DESCRIPTION OF DRAWINGS

[0016] In order to more clearly illustrate the technical solutions of the embodiments of the present disclosure, the drawings of the embodiments of the present disclosure will be briefly introduced below. In the present application, a flowchart is used to illustrate the operations performed by the system according to the embodiments of the present application. It should be understood that the preceding or following operations are not necessarily performed in sequence. On the contrary, various steps can be processed in reverse order or simultaneously as needed. Meanwhile, other operations can be added to these processes, or a step or several steps can be removed from these processes.

[0017] Figure 1A power state evaluation method based on the power grid model is provided in the embodiments of the present application.

[0018] Figure 2 A power state evaluation system structure diagram based on the power grid model is provided in the embodiments of the present application.

[0019] Label explanation: power equipment modeling unit 10, grid distribution unit 20, abnormality analysis unit 30, contraction constraint establishment unit 40, grid granularity reconstruction unit 50, state evaluation result generation unit 60. DETAILED DESCRIPTION

[0020] The above description is only a summary of the technical solutions of the present application. In order to make the technical means of the present application more clear, the embodiments of the present application can be implemented according to the content of the description, and in order to make the above and other purposes, characteristics and advantages of the present application more obvious and easy to understand, the following specific embodiments of the present application are described.

[0021] In order to make the purposes, technical solutions and advantages of the present application more clear, the following will combine the drawings to make a further detailed description of the present application. The described embodiments should not be regarded as limiting the present application. All other embodiments obtained by those skilled in the art without making creative labor are within the scope of protection of the present application.

[0022] In the following description, "some embodiments" are related to a subset of all possible embodiments, but it can be understood that "some embodiments" can be the same subset or different subset of all possible embodiments, and can be combined with each other without conflict. The term "first\second" is only to distinguish similar objects, and does not represent the specific order of the objects. The terms "include" and "have" and any variations, are intended to cover non-exclusive inclusion, for example, a process, method, system, product or server including a series of steps or units does not have to be limited to those steps or units clearly listed, but can include other steps or modules not clearly listed or inherent to these processes, methods, products or devices. Unless otherwise defined, all technical and scientific terms used herein have the same meaning as understood by those skilled in the art of the technology to which the present application belongs. The terms used herein are only for the purpose of describing the embodiments of the present application.

[0023] The embodiments of the present application provide a power state evaluation method based on a power grid model, as shown in Figure 1 The method comprises the following steps:

[0024] Step S100, the power equipment modeling of the monitoring area is performed, and the standard grid size is configured based on the monitoring accuracy of the power state.

[0025] Preferably, the power equipment in the target monitoring area is digitally modeled, wherein the target area can be a substation, a distribution network, a transmission corridor, etc., and the power equipment can include but is not limited to transformers, circuit breakers, cables, poles, etc. Specifically, according to the geographical location, distribution density and topological connection relationship of the equipment recorded according to the power grid structure, the spatial modeling of the power equipment is completed, and the operating parameters, fault modes and health status indicators of the power equipment are defined and mapped on the power equipment spatial modeling. At the same time, different weights are given according to the criticality of the power equipment in the power grid, and finally the structured power equipment modeling result is obtained. Then, the standard grid size is configured based on the monitoring accuracy of the power state. Specifically, according to the operating state monitoring demand of the power equipment, the initial division scale of the grid, i.e. the standard grid size, is dynamically set, wherein important substations, high-voltage transmission lines, key equipment, etc. have high-precision demand areas, and smaller grids are used to improve data acquisition density and abnormal detection sensitivity; low-voltage distribution lines, non-critical equipment, etc. have ordinary monitoring areas, and larger grids are used to reduce the computational burden while meeting the basic monitoring needs. For example, a substation contains one core transformer and ten ordinary switch cabinets, then the surrounding area of the transformer is divided into a high-precision grid of 5m x 5m, and the switch cabinet area is divided into a standard grid of 10m x 10m. Limited by the density of sensor deployment, communication bandwidth, etc., the grid size and data acquisition feasibility need to be balanced to avoid resource waste or insufficient precision caused by fixed grid, so as to ensure the accuracy and efficiency of power state evaluation.

[0026] In step S200, based on the equipment modeling result, the structural and functional characteristics of the equipment are analyzed, the adaptive grid distribution under the constraint of the standard grid size is performed according to the importance rating order after the importance rating, and the initial grid distribution result is established.

[0027] Preferably, the structural and functional characteristics of the power equipment are combined to obtain the analysis result and dynamically divide the grid, so that the key area can obtain higher precision monitoring capability. Specifically, based on the equipment modeling result, the structural characteristics of the equipment are analyzed, i.e. the physical properties of the equipment are analyzed, including the types of equipment such as transformers, cables, circuit breakers, spatial distribution density, power grid topological structure connection relationship, installation environment, etc. For example, high-voltage transmission lines have complex structure and wide coverage, which need more detailed grid division; while low-voltage distribution cabinets are densely distributed but simple in structure, and the grid size can be appropriately relaxed. Based on the equipment modeling result, the functional characteristics are analyzed, i.e. the functional importance of the equipment in the power grid is evaluated, including whether it is a key node, whether it affects power supply reliability, historical fault frequency, etc. For example, a transformer that bears the core task of regional power supply has higher functional importance than an ordinary branch switch.

[0028] Preferably, the analysis result is used for importance rating, including evaluating the criticality of power equipment, such as backbone equipment and auxiliary equipment; the impact range of failure, such as whether the equipment failure will cause large-area power outage; the real-time operating state, such as whether it is in abnormal working condition such as overload and high temperature; then the power equipment is divided into different levels according to the importance, and each level is assigned a corresponding grid accuracy requirement, and the importance rating order is determined; then the adaptive grid distribution under the constraint of standard grid size is carried out according to the importance rating order, that is, using the four-tree or adaptive finite element grid algorithm, the grid density is dynamically adjusted based on the configured standard grid size, to ensure that the grid division conforms to the actual monitoring capability, including the sensor density and the calculation resource limit, for example, the area where the high importance power equipment such as main transformer substation is located, the minimum grid of 2m×2m is further refined based on the standard grid; the area where the low importance power equipment such as low-voltage meter box is located, the grid of 8m×8m is appropriately merged based on the standard grid; the area where the medium importance power equipment such as power distribution cabinet is located, the gradual grid distribution is adopted to avoid the data fault caused by the sudden change of granularity. Finally, the non-uniform grid distribution covering the monitoring area is generated as the initial grid distribution result, in which the grid density in the key equipment area is high and the grid density in the secondary area is low, thereby ensuring the use efficiency of optimization calculation and sensing resources and improving the sensitivity of power state anomaly detection.

[0029] Further, step S200 further includes step S210 of grading the importance rating to establish N level spaces, the N level spaces being importance level spaces, and N being an integer greater than 1; step S220 of sequentially performing grid distribution based on the constraint of standard grid size according to the order of level spaces, and saving the distributed grid as a frozen grid; and step S230 of performing traversal segmentation of each level space sequentially under the constraint of the frozen grid to establish an initial grid distribution result.

[0030] Preferably, by hierarchical progressive mesh partitioning, high importance areas are ensured to obtain high-precision mesh resources in priority, while strictly avoiding mesh conflicts and computational redundancy. Specifically, according to the importance rating, the device failure impact range is typically divided into N levels, and N level spaces are established, wherein the N level spaces are importance level spaces, and N is an integer greater than 1, such as N = 3-5; then, according to the order of the level space, mesh distribution is performed based on the standard mesh size constraint, including using the adaptive finite element mesh algorithm to dynamically adjust the mesh density based on the configured standard mesh size, determining the independent mesh precision parameters corresponding to each level, forming the level space mesh distribution parameter mapping relationship, and saving the distributed mesh as a frozen mesh for marking, that is, when the mesh of a certain level space is completed, it is immediately marked as frozen, and the boundary of the frozen mesh cannot be modified in subsequent level space partitioning; then, under the constraint of the frozen mesh, the traversal segmentation of each level space is performed in sequence, that is, the area segmentation processing is performed level by level from the highest level to the lowest level, the non-uniform mesh distribution with level labels is generated, and the mesh-equipment association index is established as the initial mesh distribution result, to ensure the monitoring accuracy of the key equipment and optimize the resource configuration.

[0031] In step S300, the initial mesh distribution result is used to associate the original perception data set, and abnormal analysis is performed with the mesh as the minimum analysis unit to construct the power state abnormality identification.

[0032] Preferably, the physical space is discretized into standard analysis meshes, and the initial mesh distribution result is used to associate the original perception data set to achieve accurate abnormal positioning through multi-source data fusion, wherein the original perception data set includes three types of core data, i.e., nameplate data, design life, and topological connection relationship, etc. static parameters of power equipment, real-time voltage / current, partial discharge / temperature, and inspection records, etc. dynamic monitoring data, and environmental parameters such as temperature and humidity, geographic coordinates, etc. Specifically, the power grid perception data is spatially and attributively associated with each mesh unit, that is, each mesh is used as an independent data container, and different sources of sensor data are mapped to the initial mesh distribution result by spatial interpolation or nearest neighbor matching algorithm; then, abnormal analysis is performed with the mesh as the minimum analysis unit, that is, the mean, variance, gradient, etc. statistical characteristics, harmonic content, etc. frequency domain characteristics, and mutation trend, etc. time series characteristics are calculated in each mesh, a multi-dimensional feature vector is constructed, and an isolation forest or a long short-term memory autoencoder is used to compare the current features of the mesh with the historical normal mode, output the abnormal probability score, and finally mark the mesh exceeding the threshold as an abnormal unit, that is, construct the power state abnormality identification, including recording the abnormal type and the impact range, forming the abnormal identification atlas with topological relationship, to ensure accurate perception of the power state.

[0033] Step S400, after the region clustering of the power state anomaly identifier, a first shrinkage constraint is established according to the region clustering result, and a second shrinkage constraint of the grid is established according to the power state anomaly identifier.

[0034] Step S400 further comprises step S410 of establishing a comprehensive decision distance of the anomaly, the comprehensive decision distance comprising a first distance and a second distance, the first distance being a position Euclidean distance, and the second distance being a similarity distance of the anomaly type; and step S420 of performing clustering recognition of the power state anomaly identifier according to the comprehensive decision distance to establish a region clustering result.

[0035] Preferably, the spatial-feature joint clustering of the power anomaly is realized by a two-dimensional distance measurement, the physical position relationship and the similarity relationship of the anomaly type between the anomaly points are quantified, that is, a comprehensive decision distance of the anomaly is established, comprising a first distance and a second distance, specifically, the first distance is a position Euclidean distance, that is, the Euclidean distance of the center points of two anomaly grids is calculated, and the second distance is a similarity distance of the anomaly type, wherein each anomaly identifier contains a feature vector, the similarity is calculated by using an improved Gower similarity coefficient to obtain the similarity distance; then the first distance and the second distance are fused by weighting to obtain the comprehensive decision distance of the anomaly, wherein the weight coefficient proportion is automatically adjusted according to the anomaly severity, for example, in the power equipment intensive area, the weight of the position distance and the anomaly similarity distance is 0.7 and 0.3 respectively; in the power equipment dispersed area, the weight of the position distance and the anomaly similarity distance is 0.4 and 0.6 respectively. Then, the clustering recognition of the power state anomaly identifier is performed according to the comprehensive decision distance, that is, the adaptive density clustering algorithm is used to take the comprehensive distance as the measurement standard, the grids that are spatially adjacent and similar in anomaly type are aggregated into the same cluster, wherein the membership of the boundary area is determined by fuzzy membership, and finally the region clustering result with clear physical boundary and unified anomaly feature is output, realizing the dual consistency judgment of the physical position and the fault characteristics, for example, the cable terminal anomaly of insulation deterioration which is far away but belongs to the same type in the substation, and the mechanical vibration anomaly in close proximity can be correctly distinguished into different clustering areas.

[0036] Further, step 400 further comprises step S430 of extracting the boundary identifier of the clustering cluster according to the region clustering result to define the region contour; step S440 of calculating the tightness factor of each clustering cluster by using the region contour, the tightness factor representing the concentration degree of the anomaly region; and step S450 of establishing the first shrinkage constraint by using the tightness factor and the region clustering result.

[0037] Preferably, the boundary markers of the clustering clusters are extracted from the regional clustering results by using an alpha-shape algorithm, hierarchical boundary division is performed on the multi-connected regions to form closed polygonal region contours, and the region contours can adaptively identify the geometric features of the clustering clusters. For compact anomalies such as partial discharge point clusters, small convex hulls are generated, and for diffusion anomalies such as overheat areas along the cable, irregular concave boundaries are retained. Then, the tightness factor of each clustering cluster is calculated using the region contour, that is, by calculating the ratio of the contour area to the minimum circumscribed circle area, a standardized tightness factor of 0~1 is defined, wherein the tightness factor represents the concentration degree of the abnormal region, and close to 1 indicates high concentration, such as a circular fault point; close to 0 indicates dispersed distribution, such as a dendritic deterioration region. Then, the tightness factor is mapped to the grid contraction strength, the high-tightness region (tightness factor > 0.7) performs strong contraction, and the encryption grid of 30% size of the original grid is generated within the contour; the medium-tightness region (tightness factor 0.3~0.7) maintains the existing grid density; and the low-tightness region (tightness factor < 0.3) triggers boundary expansion, extending 2 layers of buffer grids to the periphery. Finally, the first contraction constraint is established according to the tightness factor and the regional clustering result, including multiple spatial contraction instructions, such as compressing the grid granularity to 0.5 times the original size within the rectangular domain of coordinates (x1, y1) and (x2, y2), so as to guarantee the monitoring accuracy of the core fault area and improve the accuracy of the power state evaluation.

[0038] Preferably, the second contraction constraint of the grid is established according to the power state anomaly identifier, that is, by deeply analyzing the inherent characteristics of the anomaly itself, establishing fine grid adjustment rules different from the spatial clustering features, including constructing anomaly strength oriented constraints, anomaly evolution trend constraints, and multi-anomaly coupling constraints, wherein the anomaly strength oriented constraints are dynamically set according to the deviation degree of the anomaly index, such as the temperature exceeding percentage, the discharge amount standard deviation multiple, for example, when the partial discharge amount > 20pC, the corresponding grid is prohibited from merging and triggers a 10% radial contraction of the surrounding grid; the anomaly evolution trend constraint is to judge the development trend of the anomaly through time series prediction, for example, a rapidly spreading anomaly generates a negative contraction constraint that expands outward, and a stable anomaly allows moderate merging to generate a centripetal contraction constraint; and the multi-anomaly coupling constraint is to use a weighted fusion strategy when electrical-mechanical composite anomalies occur in the same grid, trigger a freeze and contraction dual constraint, and finally fuse to generate the second contraction constraint of the grid, which is complementary to the first contraction constraint, and realizes the optimal dynamic configuration of the monitoring resources.

[0039] In step S500, the first contraction constraint, the second contraction constraint, and the initial grid distribution result are input into the power grid model to perform adaptive grid granularity reconstruction of the equipment features, and a reconstructed grid distribution result is established.

[0040] The step S500 further comprises a step S510 of, after importing the initial grid distribution result into the power grid model, performing shrink grid positioning using the second shrink constraint to establish a shrink grid positioning result; a step S520 of activating the shrink analysis layer of the power grid model, performing shrink constraint fusion after receiving the first shrink constraint and the second shrink constraint to establish a joint shrink constraint; and a step S530 of, after calling the device characteristics according to the shrink grid positioning result, performing grid granularity reconstruction of the shrink grid positioning result under the device characteristics self-adaption based on the joint shrink constraint to establish a reconstructed grid distribution result.

[0041] Preferably, the initial grid distribution result is imported into the power grid model, wherein the power grid model is used to decompose the power system into a plurality of small-scale, regional microgrids or subgrids, each grid can be independently operated or connected in parallel with the main grid, contains a plurality of power distribution equipment and protection devices, and the second shrink constraint is used for shrink grid positioning, that is, based on the abnormal characteristic analysis of the second constraint, the region to be adjusted is marked in the initial grid, including marking the strong abnormal grid as the core shrink area, which needs to keep or improve the grid distribution density; the associated influence grid is marked as the gradient shrink area, and the adjustment amplitude is attenuated according to the distance; and the grid topology with the shrink priority label is output as the shrink grid positioning result.

[0042] Preferably, the shrink analysis layer of the power grid model is reactivated, wherein the shrink analysis layer is a special computing layer of the power grid model, located between the basic grid layer and the optimization execution layer, receives the first shrink constraint from the spatial clustering analysis and the second shrink constraint from the abnormal identification, and performs shrink constraint fusion, specifically, the physical dimensions of the two types of constraints are unified into a grid deformation scale factor, including a tightness factor of the first shrink constraint and an abnormal intensity coefficient of the second shrink constraint, wherein the sum of the abnormal intensity coefficient and the tightness factor is 1, and then the weighted output of the fused joint shrink constraint is output, for example, the tightness factor of the new device area is kept at 0.7, focusing on the spatial clustering constraint; when the device is in the aging stage, the tightness factor is automatically reduced to 0.3, focusing on the abnormal feature constraint.

[0043] Preferably, the device characteristics are called according to the shrinkage grid positioning result, including the structural characteristics, functional characteristics and historical failure characteristics of similar devices of the power device, the grid granularity reconstruction of the shrinkage grid positioning result under the adaptive device characteristics is performed based on the joint shrinkage constraint, the first shrinkage constraint and the second shrinkage constraint are superimposed by tensor, the double-constraint coupled model is established, and the device characteristic weight coefficient is introduced, the constraint strength of the key device area is automatically improved, the grid granularity reconstruction of the shrinkage grid positioning result is performed by using the double-constraint coupled model, including performing device structure characteristic adaptation, that is, performing non-uniform shrinkage on the composite device to maintain the connection relationship of the components; performing monitoring demand adaptation to form a grid density protection ring at key monitoring points; and then minimizing the monitoring accuracy loss and the calculation overhead as the target, performing dynamic balance reconstruction through a particle swarm optimization algorithm, the constraint condition being that the joint shrinkage field requirement and the device minimum grid size limit are met, and finally obtaining the reconstruction grid distribution result, realizing dynamic optimal matching of the grid granularity and the monitoring demand, and ensuring the grid resolution to be improved.

[0044] Further, step S530 further includes step S531 of establishing a characteristic space of the inner grid of each shrinkage grid positioning result according to the analysis result; step S532 of performing grid granularity reconstruction under the joint shrinkage constraint after mapping the joint shrinkage constraint to the characteristic space to establish a benchmark reconstruction result; step S533 of performing adaptation decision of the benchmark reconstruction result and the characteristics according to the characteristic space to establish a compensation reconstruction result by using the adaptation decision; and step S534 of establishing a reconstruction grid distribution result according to the compensation reconstruction result.

[0045] Preferably, a device three-dimensional feature vector is analyzed from each shrinkage grid positioning result according to the analysis result, including a structural characteristic, a functional characteristic and a state characteristic, and is converted into a standardized grid characteristic space, such as that the key device grid is located in a high-dimensional region of the characteristic space and the ordinary device grid is distributed in a low-dimensional region. Then the joint shrinkage constraint is mapped to the characteristic space, including converting the first shrinkage constraint into a radial constraint in the characteristic space and converting the second shrinkage constraint into a tangential constraint in the characteristic space, and a synthetic constraint vector of each grid is generated by tensor operation, a preliminary grid granularity reconstruction is performed according to the synthetic constraint vector, for example, in a high characteristic value region, a center most dense and edge gradual change refinement is implemented, and in a low characteristic value region, a key monitoring point reservation and merging is performed, and then the benchmark reconstruction result is determined.

[0046] Preferably, the adaptation decision of the benchmark reconstruction result and the feature is performed according to the feature space, that is, a feature-granularity matching degree evaluation matrix is established to construct a compensation reconstruction result. Specifically, for the structural feature, the matching index is the equipment size matching error, and anisotropic mesh deformation is triggered; for the functional feature, the matching index is the control sensitivity loss degree, and an auxiliary mesh layer is added for adjustment; for the state feature, the matching index is the monitoring blind area coverage rate, and a micro sentinel mesh is inserted for adjustment. Finally, the reconstruction mesh distribution result is established according to the compensation reconstruction result, that is, the benchmark result is fine-tuned in a feature-oriented manner, including increasing the mesh density by 10%-15% along the weak insulation direction of the equipment, setting a vibration perception enhancement mesh at the spatial position corresponding to the mechanical vibration main frequency, and finally outputting the reconstruction mesh with a feature label as the reconstruction mesh distribution result, thereby realizing dynamic optimal matching of the mesh granularity and the monitoring demand and ensuring accurate evaluation of the power state.

[0047] Step S600, using the reconstruction mesh distribution result to perform power state anomaly identification verification, and generating a state evaluation result.

[0048] Step S600 further includes step S610, labeling the perception data of each reconstruction mesh based on the reconstruction mesh distribution result; step S620, establishing a multi-dimensional state vector set of the reconstruction mesh based on the perception data labeling result, generating a reconstruction mesh state feature cluster; step S630, using the reconstruction mesh state feature cluster to perform multi-scale cross-validation of power state anomaly identification, and generating a state evaluation result based on the multi-scale cross-validation result.

[0049] Preferably, for each grid cell in the reconstructed grid distribution result, multi-source perception data such as voltage, current, power, device temperature, etc. are collected and structuredly labeled, and the labeling content can include timestamp, spatial position, device type, data quality label, etc., thereby obtaining a perception data labeling result. Then, based on the perception data labeling result, key state indicators of each grid are extracted, such as voltage deviation rate, load rate, harmonic distortion rate, etc., which are then combined into a feature vector containing electrical, environmental, and device health degree, i.e. a multi-dimensional state vector of the reconstructed grid, and then the vectors of all grids are aggregated to form a reconstructed grid state feature cluster, so as to map the physical power grid to a computable high-dimensional feature space. Then, the reconstructed grid state feature cluster is used for multi-scale cross-validation of power state anomaly identification, including cross-analysis of the reconstructed grid state feature cluster in multiple scales such as time, space, and electrical parameters, wherein, in the time scale, historical data or trend prediction is compared to detect instantaneous mutations or long-term degradation; in the spatial scale, the state difference of adjacent grids is compared to identify local anomalies; in the electrical scale, the data rationality is verified through association rules; through multi-dimensional consistency verification, false positives are excluded and real anomalies are accurately located, and finally the state evaluation result is output, i.e. the grid-based hierarchical evaluation conclusion such as normal / early warning / fault, and possibly with root cause speculation or repair suggestions. Through multi-dimensional correlation analysis driven by data, the accuracy and reliability of power state evaluation are improved.

[0050] Further, step S610 further includes step S611 of identifying data richness of the perception data labeling result to establish a first perception defect; step S612 of performing multi-dimensional data verification on the perception data labeling result to establish a second perception defect; and step S613 of collecting additional data according to the first perception defect and the second perception defect, and using the additional data collection result to compensate the perception data labeling result.

[0051] Preferably, through data quality assessment and dynamic compensation mechanism, the incomplete or unreliable perception data caused by sensor error, communication interruption or environmental interference is solved, specifically, the data richness of the perception data annotation result is identified through time series continuity check and spatial distribution uniformity detection, the grid area, time segment or parameter type with data missing or sparsity is identified, and the first perception defect, i.e. the blank or low confidence area caused by insufficient data or incomplete coverage, is output. The multi-dimensional data verification is performed on the perception data annotation result, specifically, the physical rule verification is performed, the data is checked whether it conforms to the basic law of power system, such as power balance, voltage and current phase relationship; the cross-source comparison is performed, the associated parameters of different sensors in the same grid are compared, such as whether the transformer temperature matches the load current; the abnormal mode detection is performed by using the machine learning model, the outlier deviating from the normal mode is identified; and then the second perception defect, i.e. the abnormal point with contradictory data, noise or physical untrustworthiness, is output. Finally, the additional data acquisition is performed according to the first perception defect and the second perception defect, i.e. for the first perception defect, the backup sensor, unmanned aerial vehicle inspection or manual meter reading is triggered; for the second perception defect, the interpolation, adjacent grid data fusion or digital twin simulation is used to generate a substitute value; and then the additional data acquisition result is obtained, and the perception data annotation result is compensated to ensure the accuracy of power fault positioning.

[0052] Further, step S600 further includes step S640 of performing early warning signal matching based on the state evaluation result to construct an early warning signal set; and step S650 of performing visualization identification of the monitoring area for the early warning signal set.

[0053] Preferably, based on the state evaluation result, such as that a certain grid is marked as overload, voltage sag or device aging, the early warning signal matching is performed, i.e. the potential risk mode is identified according to the historical fault library, and a structured early warning signal set is output, including early warning type, level, location, timestamp and confidence; and then the early warning signal set is visualized and identified in the monitoring area, i.e. it is converted into an intuitive space-time-level multi-dimensional view, including marking the abnormal grid in the power grid topology, highlighting according to the early warning level, and clicking the early warning area to view detailed data, thereby shortening the fault response time and improving the operation stability of the power equipment.

[0054] In the foregoing, the power state evaluation method based on the power grid model is described in detail according to the embodiments of the present application. Next, the power state evaluation system based on the power grid model according to the embodiments of the present application will be described with reference to the accompanying drawings. Figure 1 The power state evaluation system based on the power grid model according to the embodiments of the present application is described in detail. Figure 2 The power state evaluation system based on the power grid model according to the embodiments of the present application is described in detail.

[0055] The power state evaluation system based on the power grid model according to the embodiment of the present application is used to solve the technical problems of fixed grid division of the power state evaluation, difficulty in adapting to the importance difference of equipment, and lack of dynamic optimization after abnormality detection in the prior art, and thus the technical effects of improving the accuracy, real-time performance and reliability of the power state evaluation are achieved. Figure 2 As shown in the figure, the power state evaluation system based on the power grid model comprises: a power equipment modeling unit 10, a grid distribution unit 20, an abnormality analysis unit 30, a contraction constraint establishment unit 40, a grid granularity reconstruction unit 50, and a state evaluation result generation unit 60.

[0056] The power equipment modeling unit 10 is configured to perform power equipment modeling of a monitoring area, and configure a standard grid size based on the monitoring accuracy of the power state; the grid distribution unit 20 is configured to perform equipment structure feature and function feature analysis based on the power equipment modeling result, perform adaptive grid distribution under the standard grid size according to the importance rating order after importance rating using the analysis result, and establish an initial grid distribution result; the abnormality analysis unit 30 is configured to perform abnormality analysis using the initial grid distribution result after associating the original perception data set, and construct a power state abnormality identifier using the grid as the minimum analysis unit; the contraction constraint establishment unit 40 is configured to perform regional clustering on the power state abnormality identifier, establish a first contraction constraint according to the regional clustering result, and establish a second contraction constraint of the grid according to the power state abnormality identifier; the grid granularity reconstruction unit 50 is configured to input the first contraction constraint, the second contraction constraint, and the initial grid distribution result into the power grid model, perform equipment feature adaptive grid granularity reconstruction, and establish a reconstructed grid distribution result; and the state evaluation result generation unit 60 is configured to verify the power state abnormality identifier using the reconstructed grid distribution result, and generate a state evaluation result.

[0057] In the following, the specific configuration of the grid granularity reconstruction unit 50 will be described in detail. The grid granularity reconstruction unit 50 further comprises: after importing the initial grid distribution result into the power grid model, performing contraction grid positioning using the second contraction constraint, and establishing a contraction grid positioning result; activating the contraction analysis layer of the power grid model, performing contraction constraint fusion after receiving the first contraction constraint and the second contraction constraint, and establishing a joint contraction constraint; calling the equipment feature according to the contraction grid positioning result, and performing grid granularity reconstruction of the contraction grid positioning result under the equipment feature adaptation based on the joint contraction constraint, and establishing a reconstructed grid distribution result.

[0058] The specific configuration of the mesh granularity reconstruction unit 50 will be described in detail below. The mesh granularity reconstruction unit 50 further includes: establishing a feature space for each shrinkage mesh positioning result based on the analytical results; mapping the joint shrinkage constraint to the feature space, performing mesh granularity reconstruction under the joint shrinkage constraint, and establishing a baseline reconstruction result; performing an adaptation decision between the baseline reconstruction result and the features based on the feature space, and using the adaptation decision to establish a compensated reconstruction result; and establishing a reconstructed mesh distribution result based on the compensated reconstruction result.

[0059] The specific configuration of the grid distribution unit 20 will be described in detail below. The grid distribution unit 20 further includes: classifying the importance rating into levels, establishing N level spaces, where N level spaces are importance level spaces and N is an integer greater than 1; distributing the grid according to the level space order based on the standard grid size constraint, and saving the distributed grid as a frozen grid; performing sequential traversal and segmentation of each level space under the frozen grid constraint to establish the initial grid distribution result.

[0060] The specific configuration of the state assessment result generation unit 60 will be described in detail below. The state assessment result generation unit 60 further includes: performing perceptual data annotation on the reconstructed grid distribution results for each reconstructed grid; establishing a multi-dimensional state vector set for the reconstructed grid based on the perceptual data annotation results, generating a reconstructed grid state feature cluster; performing multi-scale cross-validation of power state anomaly identification using the reconstructed grid state feature clusters, and generating a state assessment result based on the multi-scale cross-validation results.

[0061] The specific configuration of the state assessment result generation unit 60 will be described in detail below. The state assessment result generation unit 60 further includes: identifying the data richness of the perception data annotation results to establish a first perception defect; performing multi-dimensional data verification on the perception data annotation results to establish a second perception defect; collecting additional data based on the first perception defect and the second perception defect, and using the additional data collection results to compensate for the perception data annotation results.

[0062] The specific configuration of the contraction constraint establishment unit 40 will be described in detail below. The contraction constraint establishment unit 40 further includes: establishing a comprehensive judgment distance for anomalies, wherein the comprehensive judgment distance includes a first distance and a second distance, wherein the first distance is the Euclidean distance of the location and the second distance is the similarity metric distance of the anomaly type; and performing clustering identification of power status anomaly identifiers based on the comprehensive judgment distance to establish regional clustering results.

[0063] The specific configuration of the contraction constraint establishment unit 40 will be described in detail below. The contraction constraint establishment unit 40 further includes: extracting the boundary identifiers of clusters based on the region clustering results and defining the region contours; calculating the compactness factor of each cluster using the region contours, wherein the compactness factor characterizes the degree of concentration of abnormal regions; and establishing a first contraction constraint using the compactness factor and the region clustering results.

[0064] The specific configuration of the state assessment result generation unit 60 will be described in detail below. The state assessment result generation unit 60 further includes: matching early warning signals based on the state assessment results to construct an early warning signal set; and visually identifying the monitoring area using the early warning signal set.

[0065] The power state assessment system based on the power grid model provided in this embodiment of the invention can execute the power state assessment method based on the power grid model provided in any embodiment of the invention, and has the corresponding functional modules and beneficial effects of the method.

[0066] Although this application makes various references to certain modules in the system according to the embodiments of this application, any number of different modules can be used and run on user terminals and / or servers. The various units and modules included are only divided according to functional logic, but are not limited to the above division, as long as the corresponding functions can be achieved; in addition, the specific names of each functional unit are only for easy distinction between each other and are not used to limit the scope of protection of this invention.

[0067] The specific embodiments described above do not constitute a limitation on the scope of protection of this application. Those skilled in the art should understand that various modifications, combinations, and substitutions can be made according to design requirements and other factors. Any modifications, equivalent substitutions, and improvements made within the spirit and principles of this application should be included within the scope of protection of this application.

Claims

1. A power state assessment method based on a power grid model, characterized in that, The method includes: Perform power equipment modeling in the monitoring area and configure standard grid size based on the monitoring accuracy of power status; Based on the modeling results of power equipment, the structural and functional features of the equipment are analyzed. After the importance rating is performed using the analysis results, an adaptive grid distribution under the standard grid size constraint is performed according to the importance rating order to establish the initial grid distribution results. After associating the initial grid distribution results with the original sensing dataset, anomaly analysis is performed using the grid as the smallest analysis unit to construct power status anomaly identifiers. After performing regional clustering on the power status anomaly identifiers, a first contraction constraint is established based on the regional clustering results, and a second contraction constraint of the grid is established based on the power status anomaly identifiers. The first shrinkage constraint, the second shrinkage constraint, and the initial grid distribution result are input into the power grid model, and the device feature adaptive grid granularity reconstruction is performed to establish the reconstructed grid distribution result. The reconstructed grid distribution results are used to verify power state anomaly identification and generate state assessment results; The first shrinkage constraint, the second shrinkage constraint, and the initial grid distribution result are input into the power grid model. Adaptive grid granularity reconstruction based on equipment characteristics is performed to establish the reconstructed grid distribution result, including: After importing the initial grid distribution results into the power grid model, the shrinking grid is located using the second shrinking constraint to establish the shrinking grid location results. The contraction analysis layer of the power grid model is activated. After receiving the first contraction constraint and the second contraction constraint, the contraction constraint fusion is performed to establish a joint contraction constraint. After calling the device features based on the shrinking mesh positioning results, the mesh granularity of the shrinking mesh positioning results under the device feature adaptation is reconstructed based on the joint shrinking constraints, and the reconstructed mesh distribution results are established.

2. The power state assessment method based on a power grid model as described in claim 1, characterized in that, The mesh granularity reconstruction of the shrinkage mesh positioning result based on the joint shrinkage constraint under device feature adaptation, and the establishment of the reconstructed mesh distribution result, include: Based on the analytical results, establish the feature space of each shrinking grid positioning result within the grid; After mapping the joint contraction constraint to the feature space, perform mesh granularity reconstruction under the joint contraction constraint to establish a baseline reconstruction result; Based on the feature space, an adaptation decision is made between the baseline reconstruction result and the features, and the adaptation decision is used to establish a compensated reconstruction result. Based on the compensation and reconstruction results, a reconstruction grid distribution result is established.

3. The power state assessment method based on a power grid model as described in claim 1, characterized in that, The adaptive grid distribution under standard grid size constraints, performed according to importance rating order, establishes the initial grid distribution results, including: The importance rating is divided into levels, and N level spaces are established. The N level spaces are importance level spaces, where N is an integer greater than 1. The grids are distributed sequentially according to the hierarchical spatial order based on the standard grid size constraints, and the distributed grids are saved as frozen grids. Under the constraint of frozen mesh, perform sequential traversal partitioning of spatial order at each level to establish the initial mesh distribution result.

4. The power state assessment method based on a power grid model as described in claim 1, characterized in that, The step of using the reconstructed grid distribution results to verify power state anomaly identification and generate state assessment results includes: The reconstructed grid distribution results are then labeled with perceptual data for each reconstructed grid. A multi-dimensional state vector set for the reconstructed mesh is established based on the annotation results of the perceptual data, and a cluster of state features of the reconstructed mesh is generated. The reconstructed grid state feature cluster is used to perform multi-scale cross-validation for power state anomaly identification, and state assessment results are generated based on the multi-scale cross-validation results.

5. The power state assessment method based on a power grid model as described in claim 4, characterized in that, The step of performing perceptual data annotation on each reconstructed grid in the reconstructed grid distribution result includes: Data richness identification is performed on the labeled results of the perception data to establish the first perception defect; The labeled sensory data is validated using multidimensional data to establish a second sensory defect. Additional data is collected based on the first and second perception defects, and the results of the additional data collection are used to compensate for the perception data annotation results.

6. The power state assessment method based on a power grid model as described in claim 1, characterized in that, After performing regional clustering on the power status anomaly identifiers, a first contraction constraint is established based on the regional clustering results, including: Establish a comprehensive judgment distance for anomalies, which includes a first distance and a second distance. The first distance is the Euclidean distance of the location, and the second distance is the similarity measurement distance of the anomaly type. Based on the comprehensive judgment distance, clustering identification of power status anomalies is performed to establish regional clustering results.

7. The power state assessment method based on a power grid model as described in claim 6, characterized in that, The establishment of the first contraction constraint based on the region clustering results includes: Extract the boundary identifiers of clusters based on the region clustering results and define the region outline; The compactness factor of each cluster is calculated using the region contour, and the compactness factor characterizes the degree of concentration of the anomalous region; The first contraction constraint is established using the tightness factor and the region clustering results.

8. The power state assessment method based on a power grid model as described in claim 1, characterized in that, The generated state evaluation results also include: Based on the state assessment results, early warning signals are matched to construct an early warning signal set; The warning signal set is used to visually identify the monitoring area.

9. A power state assessment system based on a power grid model, characterized in that, The system is used to implement the power state assessment method based on a power grid model as described in any one of claims 1 to 8, and the system comprises: The power equipment modeling unit is used to perform power equipment modeling in the monitoring area, and configures the standard grid size based on the monitoring accuracy of the power status. The grid distribution unit is used to analyze the structural and functional features of the equipment based on the modeling results of the power equipment. After the importance rating is performed using the analysis results, the adaptive grid distribution under the standard grid size constraint is performed according to the importance rating order to establish the initial grid distribution result. An anomaly analysis unit is used to perform anomaly analysis with the grid as the smallest analysis unit after associating the initial grid distribution results with the original sensing dataset, and to construct a power status anomaly identifier. The contraction constraint establishment unit is used to establish a first contraction constraint based on the region clustering result after performing region clustering on the power state anomaly identifier, and to establish a second contraction constraint on the grid based on the power state anomaly identifier. The mesh granularity reconstruction unit is used to input the first shrinkage constraint, the second shrinkage constraint, and the initial mesh distribution result into the power grid model, perform adaptive mesh granularity reconstruction based on equipment characteristics, and establish the reconstructed mesh distribution result. The status assessment result generation unit is used to verify the power status anomaly identification using the reconstructed grid distribution results and generate status assessment results.

Images