Photovoltaic power station abnormality monitoring method and system based on large model data analysis

Abstract

The present invention discloses a photovoltaic power station anomaly monitoring method and system based on large-scale model data analysis, which relates to the field of photovoltaic power station monitoring technology; the method includes: obtaining historical power generation data of each photovoltaic component, performing time alignment, generating a historical data group, constructing a plan view according to the power station layout, mapping component positions, configuring power generation performance modules, and constructing a power generation performance model; using historical data to drive the model, generating historical power generation performance; extracting intra-regional and inter-regional expression features in the historical performance, including module power generation vector lines and regional power generation vector lines; constructing real-time power generation performance, extracting corresponding features, matching historical performance through equivalence screening, and judging component anomalies based on feature comparison and vector line analysis. The technical solution of the present invention efficiently processes spatiotemporal data, accurately captures abnormal patterns, and realizes real-time and robust anomaly detection and positioning. It is applicable to a variety of photovoltaic power station scenarios and greatly improves operation and maintenance efficiency and system reliability.

Description

Technical Field

[0001] The present invention relates to the technical field of photovoltaic power station monitoring, and in particular to a photovoltaic power station abnormality monitoring method and system based on large model data analysis. Background Art

[0002] With the rapid development of photovoltaic power generation technology, the scale and complexity of photovoltaic power plants are constantly increasing, posing numerous challenges to their operation and maintenance. PV modules can experience reduced power generation efficiency due to factors such as shadowing, surface contamination, aging, or failure, severely impacting the overall performance of the power plant. Therefore, real-time monitoring and precise location of PV module anomalies have become pressing technical challenges.

[0003] Traditional anomaly monitoring methods rely primarily on rule-based threshold judgments or simple statistical analysis, such as identifying anomalies by monitoring power output below a fixed value. However, these methods suffer from the following shortcomings: First, PV power plant power generation data is high-dimensional and time-varying, making it difficult for traditional methods to effectively integrate time series data and spatial layout information, resulting in insufficient anomaly detection accuracy. Second, anomaly patterns (such as localized occlusion or regional failures) have complex spatiotemporal correlations, making simple feature extraction methods unable to capture dynamic relationships within and between regions. Furthermore, real-time data is subject to interference from factors such as environmental noise and weather changes, making traditional methods prone to false positives and false negatives, lacking robustness and interpretability. Finally, after detecting an anomaly, existing methods struggle to accurately locate the abnormal component and its cause, limiting improvements in operational efficiency. Summary of the Invention

[0004] The object of the present invention is to provide a monitoring method and system for effectively determining abnormalities in a photovoltaic power station.

[0005] The present invention discloses a photovoltaic power station abnormality monitoring method based on large model data analysis, comprising:

[0006] Step S100, obtaining historical power generation data of each photovoltaic module, and aligning each historical power generation data in time to obtain a historical power generation data group;

[0007] Step S200: Determine a ground plan of the photovoltaic power station layout and construct the photovoltaic power station plan. Based on the location node of each photovoltaic module, construct a corresponding photovoltaic module mapping point on the photovoltaic power station plan. Configure a power generation performance module for each photovoltaic module mapping point to obtain a photovoltaic power station power generation performance model. The power generation performance module is used to express the power generation power of the photovoltaic module.

[0008] Step S300 , based on the historical power generation data set, driving each power generation performance module in the photovoltaic power station power generation performance model to obtain the historical power generation performance state of the photovoltaic power station power generation performance model;

[0009] Step S400, determining intra-regional joint expression features between power generation performance modules in the historical power generation performance state, and determining inter-regional joint expression features between different regions;

[0010] Step S500: construct a real-time power generation performance state, and determine the intra-regional and inter-regional expression features of the real-time power generation performance state. Use the equality between the power generation performance states as a screening condition to screen out the corresponding historical power generation performance states, and use the intra-regional and inter-regional expression features of the historical power generation performance states as abnormality analysis conditions to analyze the real-time power generation performance state and determine whether there is any abnormality in the photovoltaic module.

[0011] In some embodiments disclosed herein, a method for driving each power generation performance module in a photovoltaic power station power generation performance model includes:

[0012] Step S301: Analyze the historical power generation data set to determine the historical power generation data corresponding to each photovoltaic module, construct a time reference axis for all photovoltaic modules, and determine the power generation at different time points on the time reference axis based on the historical power generation data corresponding to each module;

[0013] In step S302, the power generation performance model is a dynamic area change component. Based on the power generation power of the photovoltaic component at different time nodes on the time reference axis, the corresponding power generation performance model is adjusted to obtain the historical sub-power generation performance state of the power generation performance model. The combination of the historical sub-power generation performance states of all power generation performance models is recorded as the historical power generation performance state of the photovoltaic power station power generation performance model.

[0014] In some embodiments disclosed herein, a method for determining the expression characteristics of regions between power generation performance modules in a historical power generation performance state includes:

[0015] Step S401: Divide the photovoltaic power station plan into power generation areas, and group power generation performance modules with similar locations into the same power generation area;

[0016] In step S402, module power generation vector lines are constructed between adjacent power generation performance modules within the same power generation area. The directions of the module power generation vector lines are determined based on the power generation power between the power generation performance modules. The module power generation vector lines between the power generation performance modules and the changes of the power generation performance modules over time are identified as the combined expression features within the area.

[0017] In some embodiments disclosed herein, a method for determining inter-regional expression characteristics between different regions includes:

[0018] In step S403, regional power generation vector lines are constructed between adjacent power generation areas in the photovoltaic power station plan. The average power generation corresponding to different power generation areas in the photovoltaic power station plan is calculated. Based on the average power generation, the direction of the regional power generation vector line is determined. The average power generation of the power generation area and the change of the power generation vector line over time are identified as inter-regional joint expression features.

[0019] In some embodiments disclosed herein, a method for analyzing real-time power generation performance includes:

[0020] Step S501, constructing module power generation vector lines for adjacent power generation performance modules in the real-time power generation performance state, and determining the change of each module power generation vector line as the real-time power generation performance state changes;

[0021] Step S502: Compare the module power generation vector lines of each frame in the real-time power generation performance state with the corresponding historical power generation performance state, determine the module power generation vector lines that do not match in each frame, and record them as abnormal module power generation vector lines. Determine the area difference between the power generation performance modules corresponding to the abnormal module power generation vector lines. If the area difference is less than or equal to a preset value, the abnormal module power generation vector line is steered and modified, and the abnormal record is cancelled.

[0022] Step S503: Compare the regional power generation vector lines of each frame in the real-time power generation performance with the corresponding historical power generation performance, determine the regional power generation vector lines in each frame that do not match, and record them as abnormal regional power generation vector lines. Determine the power difference between the average power generation powers corresponding to the abnormal regional power generation vector lines. If the power difference is less than or equal to a preset value, the abnormal regional power generation vector line is steered and modified, and the abnormal record is cancelled.

[0023] Step S504: Count the number of consecutive occurrences of each abnormal module power generation vector line. If the number of consecutive occurrences is greater than or equal to a preset value, then mark the abnormal module power generation vector line. Count the number of consecutive occurrences of each abnormal region power generation vector line. If the number of consecutive occurrences is greater than or equal to a preset value, then mark the abnormal region power generation vector line.

[0024] Step S505 : determining whether the associated photovoltaic assembly has an abnormality based on the marked abnormal module power generation vector line and the abnormal region power generation vector line.

[0025] In some embodiments disclosed herein, the method of using the equality between power generation states as a screening condition to screen out corresponding historical power generation states includes:

[0026] Step S506: Determine the average power generation power of each power generation area in each frame of the real-time power generation state, construct an average power generation power sequence for each power generation area, and record the combination of the average power generation power sequences corresponding to all power generation areas as an average power generation power sequence group;

[0027] Step S507: Using the average power generation sequence as a screening condition, a number of historical power generation states are screened out, and each historical power generation state is compared with the real-time power generation state. The comparison method includes aligning the time axis and performing an overlap comparison on each frame of the historical power generation state and the real-time power generation state. The overlap comparison includes determining the area difference between each power generation state module, recording power generation state modules with an area difference less than or equal to a preset value as equivalent power generation state modules, and canceling the equivalent record of the equivalent power generation state module if the area difference is greater than or equal to the preset value after comparing consecutive frames.

[0028] Step S508 calculates the proportion of equivalent power generation performance modules relative to all power generation performance modules, and determines the degree of similarity between the historical power generation performance and the real-time power generation performance based on the number of frames involved in the comparison. If the degree of similarity is greater than or equal to a preset value, the corresponding historical power generation performance is used as the final selected historical power generation performance.

[0029] In some embodiments disclosed herein, the expression for determining the degree of equality between historical power generation performance and real-time power generation performance is:

[0030]

[0031] Where D is the degree of equality, α(i) is the area difference judgment function of the i-th power generation performance module. If the area difference is less than or equal to the preset value, α(i) outputs 1, otherwise it outputs 0. n is the total number of power generation performance modules, L is the impact adjustment coefficient of the proportion of equivalent power generation performance modules, b is the impact adjustment constant of the proportion of equivalent power generation performance modules, t is the number of frames involved in the comparison between the historical power generation performance state and the real-time power generation performance state, R is the frame number impact adjustment coefficient, and c is the frame number impact adjustment constant.

[0032] Some embodiments disclosed in the present invention also disclose a photovoltaic power station abnormality monitoring system based on large model data analysis, including:

[0033] The first module is used to obtain the historical power generation data of each photovoltaic module and align each historical power generation data in time to obtain a historical power generation data group;

[0034] The second module is used to determine the ground plan of the photovoltaic power station layout and construct the photovoltaic power station plan. Based on the location node of each photovoltaic module, the corresponding photovoltaic module mapping point is constructed on the photovoltaic power station plan. A power generation performance module is configured for each photovoltaic module mapping point to obtain a photovoltaic power generation performance model of the photovoltaic power station. The power generation performance module is used to express the power generation power of the photovoltaic module.

[0035] The third module is configured to drive each power generation performance module in the photovoltaic power station power generation performance model based on the historical power generation data set to obtain a historical power generation performance state of the photovoltaic power station power generation performance model;

[0036] The fourth module is used to determine the intra-regional joint expression characteristics between power generation performance modules in the historical power generation performance state, and to determine the inter-regional joint expression characteristics between different regions;

[0037] The fifth module is used to construct real-time power generation performance and determine the intra-regional and inter-regional expression characteristics of the real-time power generation performance. The equality between the power generation performances is used as a screening condition to screen out the corresponding historical power generation performances. The intra-regional and inter-regional expression characteristics of the historical power generation performances are used as abnormality analysis conditions to analyze the real-time power generation performance and determine whether there is any abnormality in the photovoltaic components.

[0038] The present invention discloses a photovoltaic power station anomaly monitoring method and system based on large-scale model data analysis, which relates to the field of photovoltaic power station monitoring technology; the method includes: obtaining historical power generation data of each photovoltaic component, performing time alignment, generating a historical data group, constructing a plan view according to the power station layout, mapping component positions, configuring power generation performance modules, and constructing a power generation performance model; using historical data to drive the model, generating historical power generation performance; extracting intra-regional and inter-regional expression features in the historical performance, including module power generation vector lines and regional power generation vector lines; constructing real-time power generation performance, extracting corresponding features, matching historical performance through equivalence screening, and judging component anomalies based on feature comparison and vector line analysis. The technical solution of the present invention efficiently processes spatiotemporal data, accurately captures abnormal patterns, and realizes real-time and robust anomaly detection and positioning. It is applicable to a variety of photovoltaic power station scenarios and greatly improves operation and maintenance efficiency and system reliability.

[0039] The technical solution of the present invention is further described in detail below through the accompanying drawings and embodiments. BRIEF DESCRIPTION OF THE DRAWINGS

[0040] Figure 1 This is a method step diagram of a photovoltaic power station abnormality monitoring method based on big data analysis disclosed in an embodiment of the present invention. DETAILED DESCRIPTION

[0041] The technical solution of the present invention is further described below with reference to the accompanying drawings and embodiments.

[0042] The following will be combined with the accompanying drawings and specific embodiments to clearly and completely describe the technical solutions of the present invention. It should be understood that the preferred embodiments described herein are only used to illustrate and explain the present invention and cannot be understood as limiting the scope of protection of the present invention. Those skilled in the art in this field can make some non-essential improvements and adjustments based on the content of the present invention described below. In the present invention, unless otherwise clearly specified and limited, the technical terms used in the present invention should have the common meanings understood by those skilled in the art of the present invention.

[0043] Example:

[0044] The object of the present invention is to provide a monitoring method and system for effectively determining abnormalities in a photovoltaic power station.

[0045] The present invention discloses a photovoltaic power station abnormality monitoring method based on large model data analysis, see Figure 1 ,include:

[0046] Step S100 : acquiring historical power generation data of each photovoltaic module, and performing time alignment on each historical power generation data to obtain a historical power generation data group.

[0047] The purpose of step S100 is to provide a reliable data basis for abnormal monitoring of photovoltaic power stations. A photovoltaic power station is composed of multiple photovoltaic modules, and each module will generate power generation data during operation. The collection of this data is usually completed by sensors or monitoring systems. Since the operating environment, installation time or data recording method of different photovoltaic modules may be different, the collection time points of their power generation data may not be completely consistent. In order to ensure the accuracy of subsequent analysis, it is necessary to align the historical power generation data of each photovoltaic module in the time dimension, that is, to organize all data according to a unified time axis (such as every hour or every minute) to form a time-synchronized historical power generation data group. This data group contains the power generation value of each photovoltaic module at the same time point, providing standardized input data for subsequent spatial modeling and anomaly detection. The process of time alignment is similar to time series standardization in data preprocessing, which can eliminate the impact of time deviation on the analysis results and ensure that the model can be compared and modeled based on a consistent time reference.

[0048] In step S200, a ground plan of the photovoltaic power station layout is determined and a photovoltaic power station plan is constructed. Based on the location node of each photovoltaic module, a corresponding photovoltaic module mapping point is constructed on the photovoltaic power station plan. A power generation performance module is configured for each photovoltaic module mapping point to obtain a photovoltaic power generation performance model of the photovoltaic power station. The power generation performance module is used to express the power generation power of the photovoltaic module.

[0049] The purpose of step S200 is to convert the physical layout of the photovoltaic power plant into a digital model to facilitate analysis of the spatial relationships between components and their power generation performance. A photovoltaic power plant typically consists of multiple photovoltaic modules installed on the ground in a specific layout (such as a matrix or trapezoidal arrangement). This layout information can be obtained from design drawings or field measurements. First, a digital photovoltaic power plant plan is constructed based on the plant's ground plan, reflecting the actual spatial distribution of the modules. Next, a location node (i.e., a mapping point) is assigned to each photovoltaic module on the plan. These nodes correspond to the physical location of the module. Each mapping point is then assigned a power generation performance module, an abstract digital unit that represents the power generation characteristics of the corresponding photovoltaic module. The power generation performance module can be viewed as a dynamic mathematical function or visualization component that dynamically adjusts its representation (such as area size or color depth) based on input power generation data (such as historical or real-time data). By integrating all mapping points and their power generation performance modules, a power generation performance model of the photovoltaic power plant is obtained. This model not only reflects the spatial structure of the power plant but also dynamically displays the power generation performance of each module, laying the foundation for subsequent regional analysis and anomaly detection.

[0050] Step S300 : Based on the historical power generation data set, each power generation performance module in the photovoltaic power station power generation performance model is driven to obtain the historical power generation performance state of the photovoltaic power station power generation performance model.

[0051] The purpose of step S300 is to use the historical power generation data set to drive the power generation performance model and generate a power generation performance state that reflects the historical operating status of the power plant. The historical power generation data set contains the power generation values ​​of each photovoltaic module at different time points. This data is input into the power generation performance model constructed in step S200, specifically driving each power generation performance module. Each power generation performance module adjusts its presentation based on the corresponding historical power data, for example, by changing the module's area, color, or other visual attributes to reflect the power level or changing trend. This process is similar to mapping time series data onto a spatial model, allowing the model to dynamically reproduce the power plant's historical power generation status. By integrating the performance of all power generation performance modules at different time points, the historical power generation performance state of the photovoltaic power plant power generation performance model is formed. This state is a multidimensional digital representation that includes the power generation characteristics of all components within the power plant over time and space. The historical power generation performance state provides baseline data for subsequent feature extraction and anomaly analysis, and can reflect the behavioral patterns of the power plant during normal operation.

[0052] Step S400 : determining intra-regional joint expression features between power generation performance modules in the historical power generation performance state, and determining inter-regional joint expression features between different regions.

[0053] The purpose of step S400 is to extract spatial relationship features from historical power generation performance to provide an analytical basis for anomaly detection. Historical power generation performance encompasses the temporal and spatial performance of all power generation modules within the power plant. To capture the correlation between modules, the power plant plan is first divided into regions, with modules located in close proximity grouped together (e.g., based on geographic proximity or functional zoning). Within each region, the relationships between adjacent power generation modules are analyzed, and module power generation vector lines are constructed. The direction of these vector lines is determined by the power generation between modules (e.g., a high-power module points to a low-power module). Combined with the temporal changes in the modules, regional joint expression features are formed. These features reflect the power distribution and dynamic correlation patterns between modules within the region. Similarly, between different regions, the average power generation of each region is calculated, and regional power generation vector lines are constructed. The direction is determined by the difference in average power between regions. Combined with temporal changes, inter-regional joint expression features are formed. These features capture the power flow and coordinated behavior between regions. Together, the intra-regional and inter-regional joint expression features constitute the "fingerprint" of power plant operation, providing a key reference for subsequent real-time status comparison and anomaly identification.

[0054] Step S500: construct a real-time power generation performance state, and determine the intra-regional and inter-regional expression features of the real-time power generation performance state. Use the equality between the power generation performance states as a screening condition to screen out the corresponding historical power generation performance states, and use the intra-regional and inter-regional expression features of the historical power generation performance states as abnormality analysis conditions to analyze the real-time power generation performance state and determine whether there is any abnormality in the photovoltaic module.

[0055] The purpose of step S500 is to detect abnormalities in PV modules by comparing real-time and historical power generation performance. First, using real-time power data, the power generation performance model is driven to generate real-time power generation performance. This process is similar to step S300, but based on current data. Next, using the same method as step S400, the intra-regional and inter-regional expression features (module power generation vector lines and their variations) of the real-time power generation performance are extracted. To determine whether the real-time state is abnormal, it is necessary to compare it with the historical state. To do this, using the equality of the power generation performance as a screening criterion, the historical power generation performance that is most similar to the real-time state is selected by calculating the similarity between the real-time and historical states (e.g., matching of regional average power sequences or inter-frame area differences). The intra-regional and inter-regional expression features of the selected historical power generation performance are then used as a reference for normal operation, and the corresponding features of the real-time state are compared frame by frame. If the vector line direction, power difference, or area change in the real-time state differs significantly from the historical state (e.g., exceeding a preset threshold and persisting), the relevant PV module is determined to be abnormal. This process combines time series analysis and spatial feature comparison to accurately locate abnormal components and improve the reliability and efficiency of monitoring.

[0056] In some embodiments disclosed herein, a method for driving each power generation performance module in a photovoltaic power station power generation performance model includes:

[0057] Step S301: Analyze the historical power generation data set to determine the historical power generation data corresponding to each photovoltaic module, build a time reference axis for all photovoltaic modules, and determine the power generation at different time nodes on the time reference axis based on the corresponding historical power generation data.

[0058] In step S302, the power generation performance model is a dynamic area change component. Based on the power generation power of the photovoltaic component at different time nodes on the time reference axis, the corresponding power generation performance model is adjusted to obtain the historical sub-power generation performance state of the power generation performance model. The combination of the historical sub-power generation performance states of all power generation performance models is recorded as the historical power generation performance state of the photovoltaic power station power generation performance model.

[0059] In some embodiments disclosed herein, a method for determining the expression characteristics of regions between power generation performance modules in a historical power generation performance state includes:

[0060] Step S401 : dividing the photovoltaic power station plan into power generation areas, and dividing power generation performance modules with similar locations into the same power generation area.

[0061] In step S402, module power generation vector lines are constructed between adjacent power generation performance modules within the same power generation area. The directions of the module power generation vector lines are determined based on the power generation power between the power generation performance modules. The module power generation vector lines between the power generation performance modules and the changes of the power generation performance modules over time are identified as the combined expression features within the area.

[0062] In some embodiments disclosed herein, a method for determining inter-regional expression characteristics between different regions includes:

[0063] In step S403, regional power generation vector lines are constructed between adjacent power generation areas in the photovoltaic power station plan. The average power generation corresponding to different power generation areas in the photovoltaic power station plan is calculated. Based on the average power generation, the direction of the regional power generation vector line is determined. The average power generation of the power generation area and the change of the power generation vector line over time are identified as inter-regional joint expression features.

[0064] In some embodiments disclosed herein, a method for analyzing real-time power generation performance includes:

[0065] Step S501 : constructing module power generation vector lines for adjacent power generation performance modules in the real-time power generation performance state, and determining the change of each module power generation vector line as the real-time power generation performance state changes.

[0066] The purpose of step S501 is to extract the spatial relationship features between modules from the real-time power generation performance state, providing basic data for subsequent anomaly detection. The real-time power generation performance state is generated by the power generation performance model driven by the currently collected photovoltaic module power generation data, reflecting the current operating status of the power station. In this step, module power generation vector lines are constructed for adjacent power generation performance modules within the same power generation area. These vector lines represent the power generation relationship between adjacent modules, and their direction is generally determined by the power level (for example, from the higher power module to the lower power module). As the real-time power generation performance state changes over time (for example, one frame is updated every second or every minute), the power generation of the module will fluctuate, causing the direction or length of the vector line to change accordingly. By tracking the dynamic changes of these vector lines (for example, the direction flips or the length changes significantly), the real-time characteristics of the power distribution between modules can be captured. These change data provide key input for subsequent comparison with historical status, and can reflect whether the power generation behavior of a single module or local area deviates from the normal mode.

[0067] In step S502, the module power generation vector lines of each frame corresponding to the real-time power generation performance are compared with the historical power generation performance, and the module power generation vector lines that do not match in each frame are determined and recorded as abnormal module power generation vector lines. The area difference between the power generation performance modules corresponding to the abnormal module power generation vector lines is determined. If the area difference is less than or equal to a preset value, the abnormal module power generation vector line is steered and modified, and the abnormal record is cancelled.

[0068] The purpose of step S502 is to preliminarily identify possible abnormal modules by comparing the module power generation vector lines of the real-time and historical power generation performance. Each frame of the real-time and historical power generation performance contains module power generation vector lines, which reflect the power relationships between adjacent modules. During the comparison, each frame of the real-time state is compared one by one with the corresponding frame of the selected similar historical state to check whether the direction and length of the vector lines are consistent. If the real-time vector line of a frame does not match the historical vector line (for example, the direction is opposite or the length difference is too large), it is marked as an abnormal module power generation vector line, indicating that the relevant module may be abnormal. To avoid misjudgment, the area difference between the power generation modules corresponding to these abnormal vector lines is further calculated (the area difference may reflect the visual changes in the module power performance, such as the size difference of the dynamic area component). If the area difference is less than or equal to a preset threshold, it indicates that the deviation may be caused by normal fluctuations (such as changes in light). Therefore, the abnormal vector line is redirected (for example, adjusted to a direction consistent with the historical state) and the abnormal record is cancelled. This process improves the accuracy of anomaly detection and reduces false positives through dual verification (vector line and area difference).

[0069] In step S503, the regional power generation vector lines of each frame corresponding to the real-time power generation performance are compared with the historical power generation performance, and the regional power generation vector lines that do not match in each frame are determined and recorded as abnormal regional power generation vector lines. The power difference between the average power generation powers corresponding to the abnormal regional power generation vector lines is determined. If the power difference is less than or equal to a preset value, the abnormal regional power generation vector line is steered and modified, and the abnormal record is cancelled.

[0070] The purpose of step S503 is to identify regional-level abnormal behavior by comparing regional power generation vector lines between real-time and historical power generation performance, thus supplementing module-level analysis. Regional power generation vector lines are constructed based on the average power generation between different power generation regions and reflect the power flow relationship between regions (e.g., from high-power regions to low-power regions). In this step, the regional power generation vector line for each frame of the real-time power generation performance is compared with the vector line of the corresponding frame of the historical power generation performance to check for differences in direction and length. If the real-time vector line does not match the historical vector line (e.g., the direction is reversed or the length is significantly different), it is marked as an abnormal regional power generation vector line, indicating that the overall performance of the relevant region may be abnormal. For further verification, the difference in average power generation between the two regions corresponding to the abnormal vector line is calculated. If the power difference is less than or equal to a preset threshold, it indicates that the deviation is likely caused by normal factors (such as inter-regional sunlight differences). Therefore, the abnormal regional power generation vector line is redirected (e.g., adjusted to a direction consistent with the historical state) and the abnormal record is cancelled. Regional-level analysis can capture group anomalies that may be overlooked by module-level analysis, while power difference verification reduces the risk of misjudgment.

[0071] Step S504: Count the number of consecutive appearances of the power generation vector line of each abnormal module. If the number of consecutive appearances is greater than or equal to a preset value, mark the abnormal module power generation vector line. Count the number of consecutive appearances of the power generation vector line of each abnormal area. If the number of consecutive appearances is greater than or equal to a preset value, mark the abnormal area power generation vector line.

[0072] The purpose of step S504 is to confirm the real anomaly and filter out short-term non-abnormal fluctuations by analyzing the persistence of the abnormal vector line. In steps S502 and S503, the abnormal module power generation vector line and the abnormal area power generation vector line are preliminarily identified, but these anomalies may be caused by transient factors (such as cloud cover or sensor noise). In order to improve the reliability of detection, the number of occurrences of each abnormal module power generation vector line in consecutive frames is counted. If the number of consecutive occurrences reaches or exceeds the preset value (for example, 5 consecutive frames), it means that the anomaly is persistent and may be caused by component failure or system problems. Therefore, the vector line is marked as a highly suspicious anomaly. Similarly, the number of consecutive occurrences of each abnormal area power generation vector line is counted. If it reaches or exceeds the preset value, it is marked as a highly suspicious anomaly, indicating that the regional level anomaly is persistent. By introducing the continuity analysis of the time dimension, this step effectively distinguishes between short-term fluctuations and persistent anomalies, further improving the accuracy and robustness of anomaly detection, and providing a reliable basis for the final anomaly location.

[0073] Step S505 : determining whether the associated photovoltaic assembly has an abnormality based on the marked abnormal module power generation vector line and the abnormal region power generation vector line.

[0074] The purpose of step S505 is to integrate module- and region-level anomaly markers to accurately locate abnormal PV modules. In the previous steps, continuity analysis identified highly suspicious abnormal module power generation vector lines and abnormal region power generation vector lines. These vector lines point to abnormal relationships between modules and regions, respectively. In this step, these identified vector lines are analyzed to trace their corresponding power generation performance modules and PV modules. For example, an abnormal module power generation vector line directly linked to two adjacent power generation modules indicates a potential problem with at least one of the corresponding PV modules; an abnormal region power generation vector line indicates overall power anomalies in certain areas, potentially involving multiple modules. By comprehensively analyzing module- and region-level anomaly information, specific abnormal PV modules can be identified (for example, by locating modules that appear multiple times in the abnormal vector line). Furthermore, by combining the plant floor plan and power generation performance model, the spatial location and impact range of the abnormal module can be further confirmed. This comprehensive analysis process utilizes multi-dimensional spatial and temporal information to efficiently and accurately determine the abnormal state of PV modules, providing clear targets for subsequent maintenance and repair.

[0075] In some embodiments disclosed herein, the method of using the equality between power generation states as a screening condition to screen out corresponding historical power generation states includes:

[0076] In step S505, the average power generation power of each power generation area in each frame of the real-time power generation state is determined, and an average power generation power sequence is constructed for each power generation area. The combination of the average power generation power sequences corresponding to all power generation areas is recorded as an average power generation power sequence group.

[0077] The purpose of step S505 is to extract regional features from the real-time power generation performance, providing preliminary screening criteria for filtering historical power generation performance. Real-time power generation performance is generated based on the current PV module power generation data and reflects the current operating status of the power plant. In this step, the real-time power generation performance for each frame (i.e., a specific point in time) is first analyzed to calculate the average power generation of all power generation modules within each power generation area. A power generation area is a geographical or functional unit defined by the PV power plant plan. The average power generation reflects the overall power generation performance of the area. Next, for each power generation area, the average power generation at different time points (i.e., multiple frames) is arranged in chronological order to form an average power generation sequence, which describes the power variation trend of the area over time. Finally, the average power generation sequences of all power generation areas are combined to form an average power generation sequence group. This sequence group is a multidimensional feature set that summarizes the power generation behavior patterns of each area within the power plant. It can be used as a screening criterion to quickly identify power generation performance similar to the real-time status from historical data, reducing the computational complexity of subsequent comparisons.

[0078] In step S506, the average power generation sequence is used as a screening condition to select several historical power generation states. Each historical power generation state is then compared with the real-time power generation state. The comparison method includes aligning the time axis and performing an overlap comparison between each frame of the historical power generation state and the real-time power generation state. The overlap comparison includes determining the area difference between each power generation state module. Power generation state modules with an area difference less than or equal to a preset value are recorded as equivalent power generation state modules. If the area difference is greater than or equal to the preset value after comparing consecutive frames, the equivalent record of the equivalent power generation state module is cancelled.

[0079] The purpose of step S506 is to identify historical power generation states that are highly similar to the real-time power generation state through a combination of coarse and fine screening. First, using the average power generation sequence group generated in step S505 as a screening criterion, the historical power generation state database is searched for historical states that are similar to the real-time state sequence group. Specifically, the average power generation sequence groups of each region are compared (for example, using a sequence similarity algorithm such as dynamic time warping or Euclidean distance) to identify candidate historical power generation states whose regional power trends are similar to the real-time state. Next, a detailed comparison is performed between each candidate historical power generation state and the real-time power generation state. During the comparison, the time axes of the two are first aligned to ensure that the time points of each frame are consistent. Next, the performance of each power generation module in the historical and real-time power generation states is compared frame by frame, and the area difference between the modules is calculated. (The area difference may reflect the difference in the visual representation of module power, such as the size change of dynamic area components.) If the area difference of a module is less than or equal to a preset threshold, the module is considered to have an equivalent power generation performance module in that frame, and the equivalence is recorded. However, if the area difference of a module in subsequent consecutive frame comparisons is greater than or equal to a preset value, indicating that its equality is unstable (possibly affected by transient fluctuations), the module's equality record is canceled. This combination of coarse screening (sequence group matching) and fine screening (inter-frame module alignment) ensures the accuracy and robustness of the screening results.

[0080] Step S507 calculates the proportion of equivalent power generation performance modules relative to all power generation performance modules, and determines the degree of similarity between the historical power generation performance and the real-time power generation performance based on the number of frames involved in the comparison. If the degree of similarity is greater than or equal to a preset value, the corresponding historical power generation performance is used as the final selected historical power generation performance.

[0081] In some embodiments disclosed herein, the expression for determining the degree of equality between historical power generation performance and real-time power generation performance is:

[0082]

[0083] Where D is the degree of equality, α(i) is the area difference judgment function of the i-th power generation performance module. If the area difference is less than or equal to the preset value, α(i) outputs 1, otherwise it outputs 0. n is the total number of power generation performance modules, L is the impact adjustment coefficient of the proportion of equivalent power generation performance modules, b is the impact adjustment constant of the proportion of equivalent power generation performance modules, t is the number of frames involved in the comparison between the historical power generation performance state and the real-time power generation performance state, R is the frame number impact adjustment coefficient, and c is the frame number impact adjustment constant.

[0084] Some embodiments disclosed in the present invention also disclose a photovoltaic power station abnormality monitoring system based on large model data analysis, including:

[0085] The first module is used to obtain the historical power generation data of each photovoltaic module and align each historical power generation data in time to obtain a historical power generation data group;

[0086] The second module is used to determine the ground plan of the photovoltaic power station layout and construct the photovoltaic power station plan. Based on the location node of each photovoltaic module, the corresponding photovoltaic module mapping point is constructed on the photovoltaic power station plan. A power generation performance module is configured for each photovoltaic module mapping point to obtain a photovoltaic power generation performance model of the photovoltaic power station. The power generation performance module is used to express the power generation power of the photovoltaic module.

[0087] The third module is configured to drive each power generation performance module in the photovoltaic power station power generation performance model based on the historical power generation data set to obtain a historical power generation performance state of the photovoltaic power station power generation performance model;

[0088] The fourth module is used to determine the intra-regional joint expression characteristics between power generation performance modules in the historical power generation performance state, and to determine the inter-regional joint expression characteristics between different regions;

[0089] The fifth module is used to construct real-time power generation performance and determine the intra-regional and inter-regional expression characteristics of the real-time power generation performance. The equality between the power generation performances is used as a screening condition to screen out the corresponding historical power generation performances. The intra-regional and inter-regional expression characteristics of the historical power generation performances are used as abnormality analysis conditions to analyze the real-time power generation performance and determine whether there is any abnormality in the photovoltaic components.

[0090] The present invention discloses a photovoltaic power station anomaly monitoring method and system based on large-scale model data analysis, which relates to the field of photovoltaic power station monitoring technology; the method includes: obtaining historical power generation data of each photovoltaic component, performing time alignment, generating a historical data group, constructing a plan view according to the power station layout, mapping component positions, configuring power generation performance modules, and constructing a power generation performance model; using historical data to drive the model, generating historical power generation performance; extracting intra-regional and inter-regional expression features in the historical performance, including module power generation vector lines and regional power generation vector lines; constructing real-time power generation performance, extracting corresponding features, matching historical performance through equivalence screening, and judging component anomalies based on feature comparison and vector line analysis. The technical solution of the present invention efficiently processes spatiotemporal data, accurately captures abnormal patterns, and realizes real-time and robust anomaly detection and positioning. It is applicable to a variety of photovoltaic power station scenarios and greatly improves operation and maintenance efficiency and system reliability.

[0091] Through the description of the above embodiments, those skilled in the art can clearly understand that the present invention can be implemented through hardware or by using software plus the necessary general hardware platform. Based on this understanding, the technical solution of the present invention can be embodied in the form of a software product, which can be stored in a non-volatile storage medium (such as a CD-ROM, a USB flash drive, a mobile hard disk, etc.) and includes a number of instructions for enabling a computer device (such as a personal computer, a server, or a network device) to execute the methods described in various implementation scenarios of the present invention.

[0092] Finally, it should be noted that the above embodiments are only used to illustrate the technical solutions of the present invention rather than to limit the same. Although the present invention has been described in detail with reference to the preferred embodiments, those skilled in the art should understand that they can still modify or replace the technical solutions of the present invention with equivalents, and these modifications or equivalent replacements cannot cause the modified technical solutions to deviate from the spirit and scope of the technical solutions of the present invention.

Claims

1. A photovoltaic power station abnormality monitoring method based on large model data analysis is characterized by: include: Step S100, obtaining historical power generation data of each photovoltaic module, and aligning each historical power generation data in time to obtain a historical power generation data group; Step S200: Determine a ground plan of the photovoltaic power station layout and construct the photovoltaic power station plan. Based on the location node of each photovoltaic module, construct a corresponding photovoltaic module mapping point on the photovoltaic power station plan. Configure a power generation performance module for each photovoltaic module mapping point to obtain a photovoltaic power station power generation performance model. The power generation performance module is used to express the power generation power of the photovoltaic module. Step S300 , based on the historical power generation data set, driving each power generation performance module in the photovoltaic power station power generation performance model to obtain the historical power generation performance state of the photovoltaic power station power generation performance model; Step S400, determining intra-regional joint expression features between power generation performance modules in the historical power generation performance state, and determining inter-regional joint expression features between different regions; Step S500: Constructing a real-time power generation performance state, determining the intra-regional and inter-regional combined expression features of the real-time power generation performance state, using the equality between power generation performance states as a screening condition, screening out the corresponding historical power generation performance states, and using the intra-regional and inter-regional combined expression features of the historical power generation performance states as abnormality analysis conditions. The real-time power generation performance state is analyzed to determine whether there are any abnormalities in the photovoltaic modules. The method for determining the expression characteristics of the region between power generation performance modules in the historical power generation performance includes: Step S401: Divide the photovoltaic power station plan into power generation areas, and group power generation performance modules with similar locations into the same power generation area; Step S402: constructing module power generation vector lines between adjacent power generation performance modules within the same power generation area, determining the direction of the module power generation vector lines based on the power generation power between the power generation performance modules, and identifying the module power generation vector lines between the power generation performance modules and the changes in the power generation performance modules over time as joint expression features within the area; Methods for determining inter-regional expression characteristics between different regions include: Step S403: construct regional power generation vector lines between adjacent power generation areas in the photovoltaic power station plan. Calculate the average power generation corresponding to different power generation areas in the photovoltaic power station plan. Determine the direction of the regional power generation vector lines based on the average power generation. The average power generation of the power generation area and the change of the power generation vector lines over time are identified as inter-regional joint expression features. Methods for analyzing real-time power generation performance include: Step S501, constructing module power generation vector lines for adjacent power generation performance modules in the real-time power generation performance state, and determining the change of each module power generation vector line as the real-time power generation performance state changes; Step S502: Compare the module power generation vector lines of each frame in the real-time power generation performance state with the corresponding historical power generation performance state, determine the module power generation vector lines that do not match in each frame, and record them as abnormal module power generation vector lines. Determine the area difference between the power generation performance modules corresponding to the abnormal module power generation vector lines. If the area difference is less than or equal to a preset value, the abnormal module power generation vector line is steered and modified, and the abnormal record is cancelled. Step S503: Compare the regional power generation vector lines of each frame in the real-time power generation performance with the corresponding historical power generation performance, determine the regional power generation vector lines in each frame that do not match, and record them as abnormal regional power generation vector lines. Determine the power difference between the average power generation powers corresponding to the abnormal regional power generation vector lines. If the power difference is less than or equal to a preset value, the abnormal regional power generation vector line is steered and modified, and the abnormal record is cancelled. Step S504: Count the number of consecutive occurrences of each abnormal module power generation vector line. If the number of consecutive occurrences is greater than or equal to a preset value, then mark the abnormal module power generation vector line. Count the number of consecutive occurrences of each abnormal region power generation vector line. If the number of consecutive occurrences is greater than or equal to a preset value, then mark the abnormal region power generation vector line. Step S505, determining whether the associated photovoltaic assembly has an abnormality based on the marked abnormal module power generation vector line and the abnormal area power generation vector line; Using the equality between power generation performance states as a screening condition, methods for screening out corresponding historical power generation performance states include: Step S506: Determine the average power generation power of each power generation area in each frame of the real-time power generation state, construct an average power generation power sequence for each power generation area, and record the combination of the average power generation power sequences corresponding to all power generation areas as an average power generation power sequence group; Step S507: Using the average power generation sequence as a screening condition, a number of historical power generation states are screened out, and each historical power generation state is compared with the real-time power generation state. The comparison method includes aligning the time axis and performing an overlap comparison on each frame of the historical power generation state and the real-time power generation state. The overlap comparison includes determining the area difference between each power generation state module, recording power generation state modules with an area difference less than or equal to a preset value as equivalent power generation state modules, and canceling the equivalent record of the equivalent power generation state module if the area difference is greater than or equal to the preset value after comparing consecutive frames. Step S507 calculates the proportion of equivalent power generation performance modules relative to all power generation performance modules, and determines the degree of similarity between the historical power generation performance and the real-time power generation performance based on the number of frames involved in the comparison. If the degree of similarity is greater than or equal to a preset value, the corresponding historical power generation performance is used as the final selected historical power generation performance.

2. The photovoltaic power station abnormality monitoring method based on large model data analysis according to claim 1 is characterized in that: The method for driving each power generation performance module in the photovoltaic power station power generation performance model includes: Step S301: Analyze the historical power generation data set to determine the historical power generation data corresponding to each photovoltaic module, construct a time reference axis for all photovoltaic modules, and determine the power generation at different time points on the time reference axis based on the historical power generation data corresponding to each module; In step S302, the power generation performance model is a dynamic area change component. Based on the power generation power of the photovoltaic component at different time nodes on the time reference axis, the corresponding power generation performance model is adjusted to obtain the historical sub-power generation performance state of the power generation performance model. The combination of the historical sub-power generation performance states of all power generation performance models is recorded as the historical power generation performance state of the photovoltaic power station power generation performance model.

3. The photovoltaic power station abnormality monitoring method based on large model data analysis according to claim 2 is characterized in that: The expression for determining the degree of equality between historical power generation performance and real-time power generation performance is: ； in, To the same extent, For the The area difference judgment function of the power generation performance module, if the area difference is less than or equal to the preset value, then Output 1, otherwise output 0. is the total number of power generation performance modules, The adjustment coefficient for the impact of the module ratio on equal power generation performance is: Equivalent power generation performance module ratio affects the adjustment constant, is the number of frames involved in the comparison between historical power generation performance and real-time power generation performance, is the frame rate adjustment coefficient, Adjust the constant for frame rate effect.

4. Photovoltaic power station abnormality monitoring system based on large model data analysis, characterized by: The method for monitoring abnormality of a photovoltaic power station according to any one of claims 1 to 3 comprises: The first module is used to obtain the historical power generation data of each photovoltaic module and align each historical power generation data in time to obtain a historical power generation data group; The second module is used to determine the ground plan of the photovoltaic power station layout and construct the photovoltaic power station plan. Based on the location node of each photovoltaic module, the corresponding photovoltaic module mapping point is constructed on the photovoltaic power station plan. A power generation performance module is configured for each photovoltaic module mapping point to obtain a photovoltaic power generation performance model of the photovoltaic power station. The power generation performance module is used to express the power generation power of the photovoltaic module. The third module is configured to drive each power generation performance module in the photovoltaic power station power generation performance model based on the historical power generation data set to obtain a historical power generation performance state of the photovoltaic power station power generation performance model; The fourth module is used to determine the intra-regional joint expression characteristics between power generation performance modules in the historical power generation performance state, and to determine the inter-regional joint expression characteristics between different regions; The fifth module is used to construct real-time power generation performance and determine the intra-regional and inter-regional expression characteristics of the real-time power generation performance. The equality between the power generation performances is used as a screening condition to screen out the corresponding historical power generation performances. The intra-regional and inter-regional expression characteristics of the historical power generation performances are used as abnormality analysis conditions to analyze the real-time power generation performance and determine whether there is any abnormality in the photovoltaic components.

Images