A data operation and maintenance method and system applied to a photovoltaic power station

Abstract

The application relates to the technical field of photovoltaic power station data operation and maintenance, and provides a data operation and maintenance method and system applied to a photovoltaic power station, wherein an evaluation input set is constructed, a unified index system is loaded, equipment-level evaluation calculation, power station-level evaluation calculation and multi-time granularity evaluation calculation are performed, transverse comparative analysis and longitudinal trend analysis are performed, potential abnormality identification data are generated by combining preset abnormality judgment conditions, and a risk dynamic level is determined, report information, analysis information and operation action information are generated and written back to a historical memory file, and a complete data operation and maintenance closed loop is formed. The application can effectively solve the problems in photovoltaic power station operation and maintenance analysis, such as inconsistent index caliber, insufficient multi-granularity evaluation, difficulty in combined analysis of transverse difference and longitudinal trend, lack of unified framework for health degree scoring and risk grading, disconnection between analysis results and operation actions, and difficulty in reuse of historical analysis results, and has good practicability, expansibility and operation value.

Description

Technical Field

[0001] This application relates to the field of photovoltaic power plant data operation and maintenance technology, and more specifically, to a data operation and maintenance method and system applied to photovoltaic power plants. Background Technology

[0002] As the installed capacity of photovoltaic power generation continues to expand, the operation and maintenance (O&M) model of photovoltaic power plants is gradually evolving from single-site on-site monitoring to centralized, remote, digital, and intelligent approaches. In the traditional O&M model, O&M personnel primarily rely on monitoring systems to view operational data such as inverter operating points, meter readings, and meteorological data, and discover equipment faults through alarm systems or manual inspections. While this approach can meet the basic monitoring and single-point anomaly detection needs to a certain extent, it falls short of the comprehensive analysis requirements of the asset operation phase, which includes understanding equipment health status, overall plant performance, cross-equipment horizontal differences, cross-cycle trend changes, and risk classification management.

[0003] Photovoltaic power plants continuously generate a large amount of data during operation, including but not limited to power plant operation data, equipment operation data, weather data, external auxiliary data, equipment alarm data, power curtailment records, maintenance plans, historical report results, anomaly comparison records, and manual maintenance records. These data sources differ, as do their formats, sampling frequencies, and business semantics, and also vary in terms of data quality, temporal continuity, completeness, and availability. For example, equipment operation data typically comes from inverters or power plant monitoring systems, and while its sampling frequency is high, it may suffer from issues such as equipment offline, measurement point anomalies, communication interruptions, instantaneous spikes, and missing values. Weather data typically includes irradiance, ambient temperature, module temperature, and wind speed, which are closely related to equipment power generation performance, but time alignment and validity assessment issues exist between different data sources. External auxiliary data can include power curtailment records, maintenance plans, and external interface query results, which have a significant impact on power plant power generation performance, equipment operating status, and anomaly detection, but traditional monitoring systems typically do not incorporate them into a unified evaluation process.

[0004] It is evident that establishing a unified indicator system to overcome existing problems in photovoltaic power plant operation and maintenance analysis, such as inconsistent indicator definitions, insufficient multi-granularity assessment capabilities, difficulty in jointly analyzing horizontal differences and vertical trends, difficulty in forming comprehensive health scores and dynamic risk levels, difficulty in translating analysis results into operational actions, and difficulty in accumulating and reusing historical analysis results, in order to achieve the standardization, continuity, and closed-loop operation and maintenance of photovoltaic power plants, is a pressing technical issue that needs to be addressed. Summary of the Invention

[0005] The purpose of this invention is to provide a data operation and maintenance method and system for photovoltaic power plants, in order to solve the problems existing in the operation and maintenance analysis of photovoltaic power plants, such as inconsistent indicator standards, insufficient multi-granularity assessment capabilities, difficulty in jointly analyzing horizontal differences and vertical trends, difficulty in forming comprehensive health scores and dynamic risk levels, difficulty in converting analysis results into operational actions, and difficulty in accumulating and reusing historical analysis results.

[0006] Firstly, a data operation and maintenance method for photovoltaic power plants is provided, including the following steps:

[0007] Obtain the operational data of the photovoltaic power plants in the current evaluation batch, and read the historical memory files to construct the evaluation input set;

[0008] Load the unified indicator system and obtain the set of indicators to participate in this round of evaluation calculations;

[0009] Based on the indicator set, equipment-level evaluation calculation, station-level evaluation calculation, and multi-time granularity evaluation calculation are performed on the evaluation input set to obtain sub-item evaluation data, which includes equipment-level evaluation data, station-level evaluation data, and time granularity evaluation data.

[0010] Based on the sub-item evaluation data, we perform horizontal comparative analysis and vertical trend analysis to obtain horizontal comparative analysis data and vertical trend analysis data.

[0011] Based on horizontal comparative analysis data and vertical trend analysis data, and combined with preset anomaly judgment conditions, potential anomaly identification data is generated. The potential anomaly identification data is used to characterize at least one of equipment differences, trend changes and potential anomalies.

[0012] Based on the sub-item evaluation data, horizontal comparative analysis data, and vertical trend analysis data, calculate the comprehensive health score of equipment and the comprehensive health score of the site;

[0013] By combining comprehensive equipment health scores, comprehensive site health scores, longitudinal trend analysis data, and potential anomaly identification data, the dynamic risk level is determined.

[0014] Based on the comprehensive health score of equipment, the comprehensive health score of the site, and the dynamic risk level, generate report information, analysis information, and operational action information;

[0015] The sub-assessment data, horizontal comparative analysis data, vertical trend analysis data, potential anomaly identification data, comprehensive equipment health score, comprehensive site health score, dynamic risk level, report information, analysis information, and operational action information generated in this round will be written back to the historical memory file.

[0016] Furthermore, the photovoltaic power station operation data includes power station operation data, equipment operation data, weather data, and external auxiliary data; historical memory files include equipment operation memory, equipment trend memory, historical report specifications, and anomaly comparison records.

[0017] Furthermore, a unified indicator system is loaded to obtain the set of indicators participating in this round of evaluation calculations, including:

[0018] Read the indicator registration and maintenance results;

[0019] Read the indicator classification results;

[0020] Read the indicator lifecycle status and indicator approval status;

[0021] Based on the current assessment objectives and input data conditions, construct a set of indicators to participate in this round of assessment calculations.

[0022] Furthermore, the indicator set is filtered based on indicator activation status, approval status, application level, computability, assessment task type, time granularity adaptability, and data source availability; among them, the indicator set used for equipment-level assessment calculation is the equipment-level indicator set, the indicator set used for site-level assessment calculation is the site-level indicator set, and the indicator set used for multi-time granularity assessment calculation is the indicator set adapted to the current statistical period.

[0023] Furthermore, the equipment-level evaluation calculation is used to generate equipment-level evaluation data for a single device within the current statistical period. The equipment-level evaluation data includes equipment power performance data, power generation performance data, stability performance data, and quality performance data. The station-level evaluation calculation is used to generate station-level evaluation data corresponding to the overall station operation performance based on the equipment-level evaluation data. The multi-time-granularity evaluation calculation is used to generate daily, weekly, and monthly reports, as well as time-granularity evaluation data corresponding to custom time granularities.

[0024] Furthermore, the horizontal comparative analysis includes comparisons of equipment of the same model, equipment of the same site, and sites in the same region. The data from the horizontal comparative analysis is used to characterize the degree of operational deviation of the target object relative to similar objects. The vertical trend analysis includes health change trend analysis, continuous degradation analysis, and abnormal persistence analysis. The data from the vertical trend analysis is used to characterize the status changes of the target object in different statistical periods.

[0025] Furthermore, the comprehensive health score of the equipment is calculated based on multiple dimensions of data, including power performance, operational stability, data quality, alarm status, and trend changes; the comprehensive health score of the site is obtained by summarizing, weighting, or hierarchically summarizing the comprehensive health scores of the equipment.

[0026] Furthermore, the dynamic risk levels include normal, attention, warning, and high risk; potential anomaly identification data includes at least one of single indicator anomaly identification data, continuous period anomaly identification data, multi-indicator superimposed anomaly identification data, and horizontal and vertical joint anomaly identification data; among them, single indicator anomaly identification data is used to characterize whether a single indicator hits the anomaly threshold, continuous period anomaly identification data is used to characterize whether the same indicator meets the anomaly conditions for multiple consecutive periods, multi-indicator superimposed anomaly identification data is used to characterize whether multiple different indicators simultaneously meet the anomaly conditions in the same period, and horizontal and vertical joint anomaly identification data is used to characterize whether horizontal comparison anomalies and vertical trend anomalies occur simultaneously.

[0027] Furthermore, based on the comprehensive health score of equipment, the comprehensive health score of the site, and the dynamic risk level, report information, analysis information, and operational action information are generated, including:

[0028] Generate daily, weekly, and monthly reports for the site as reporting information;

[0029] Generate device-level analysis results and custom analysis results as analysis information;

[0030] Generate maintenance suggestions, proactive early warning information, and push notification information as operational action information;

[0031] Specifically, when the overall health score of equipment or site is lower than the preset level threshold, or the dynamic risk level reaches the warning level or above, maintenance suggestions are generated based on sub-item assessment data, horizontal comparative analysis data, vertical trend analysis data, potential anomaly identification data, and historical memory files, and proactive warning information is generated.

[0032] Secondly, a data operation and maintenance system for photovoltaic power plants is provided, including:

[0033] The data and memory file layer is used to obtain the photovoltaic power plant operation data of the current evaluation batch and read historical memory files to construct the evaluation input set;

[0034] The business logic layer is used to load a unified indicator system, obtain the set of indicators participating in this round of evaluation calculations, perform equipment-level evaluation calculations, site-level evaluation calculations, and multi-time-granularity evaluation calculations on the evaluation input set based on the indicator set, obtain sub-item evaluation data, perform horizontal comparative analysis and vertical trend analysis based on the sub-item evaluation data, obtain horizontal comparative analysis data and vertical trend analysis data, generate potential anomaly identification data based on the horizontal comparative analysis data and vertical trend analysis data and combined with preset anomaly judgment conditions, calculate the comprehensive equipment health score and the comprehensive site health score based on the sub-item evaluation data, the horizontal comparative analysis data, and the vertical trend analysis data, and determine the dynamic risk level by combining the comprehensive equipment health score, the comprehensive site health score, the vertical trend analysis data, and the potential anomaly identification data.

[0035] The business application layer is used to generate report information, analysis information and operational action information based on the comprehensive health score of equipment, comprehensive health score of site and dynamic risk level, and write back the sub-evaluation data, horizontal comparison analysis data, vertical trend analysis data, potential anomaly identification data, comprehensive health score of equipment, comprehensive health score of site, dynamic risk level, report information, analysis information and operational action information generated in this round to the historical memory file;

[0036] The business logic layer includes a basic indicator system, an evaluation and calculation module, and an auxiliary indicator calling module. The business application layer includes a report output module, an asset operation action module, and a memory write-back module.

[0037] Compared with existing technologies, the data operation and maintenance method and system for photovoltaic power plants provided in this application have the following advantages: This application constructs an evaluation input set by acquiring the operating data of the photovoltaic power plant in the current evaluation batch and reading historical memory files; it obtains the set of indicators participating in the current round of evaluation calculation by loading a unified indicator system; it obtains sub-item evaluation data through equipment-level evaluation calculation, power plant-level evaluation calculation, and multi-time-granularity evaluation calculation; it obtains horizontal comparative analysis data and vertical trend analysis data through horizontal comparative analysis and vertical trend analysis; it generates potential anomaly identification data by pre-setting anomaly judgment conditions; it obtains a comprehensive health score for equipment and a comprehensive health score for power plant through comprehensive calculation; it determines the dynamic risk level by combining the comprehensive health score, vertical trend analysis data, and potential anomaly identification data; it realizes the conversion of analysis results into operational actions by generating report information, analysis information, and operational action information; and it realizes subsequent evaluation calls and continuous optimization by writing the data generated in this round back to the historical memory file. It solves the problems existing in the operation and maintenance analysis of photovoltaic power plants, such as inconsistent indicator standards, insufficient multi-granularity assessment capabilities, difficulty in jointly analyzing horizontal differences and vertical trends, difficulty in forming comprehensive health scores and dynamic risk levels, difficulty in translating analysis results into operational actions, and difficulty in accumulating and reusing historical analysis results. Attached Figure Description

[0038] Figure 1 This is a flowchart illustrating a data operation and maintenance method for photovoltaic power plants provided in this application;

[0039] Figure 2 This is a schematic diagram of a combined data and memory file input process provided in this application;

[0040] Figure 3 This is a schematic diagram of a multi-granularity evaluation calculation process provided in this application;

[0041] Figure 4 This application provides a schematic diagram of a comprehensive health score and dynamic risk classification process.

[0042] Figure 5 This application provides a schematic diagram of a closed loop for report output and asset operation.

[0043] Figure 6 This application provides a schematic diagram illustrating the relationship between memory file write-back and continuous optimization.

[0044] Figure 7 This application provides a lifecycle management diagram for indicators;

[0045] Figure 8 This is a schematic diagram of the overall architecture of a data operation and maintenance system for photovoltaic power plants provided in this application. Detailed Implementation

[0046] The technical solution of the present invention will be further described below with reference to the accompanying drawings and specific embodiments. It should be understood that the following embodiments are for illustrative purposes only and are not intended to limit the scope of protection of the present invention. Without departing from the technical concept of the present invention, those skilled in the art can make equivalent substitutions or appropriate adjustments to the data sources, deployment methods, indicator filtering methods, scoring and summarizing methods, message push methods, etc., in the embodiments.

[0047] The data operation and maintenance method and system described in this invention are applicable to scenarios such as daily operation and maintenance of photovoltaic power plants, regional photovoltaic asset analysis, group-level photovoltaic asset operation management, equipment health status assessment, power plant operation risk identification, automatic report generation, maintenance suggestion generation, proactive early warning triggering, and reuse of historical analysis results. Through this invention, photovoltaic power plant operation data, equipment operation data, weather data, external auxiliary data, historical memory files, indicator systems, analysis results, and operational action results can be incorporated into a unified data operation and maintenance process, thereby achieving multi-granular, continuous, standardized, and closed-loop management of the photovoltaic power plant's operational status.

[0048] like Figure 1As shown, this application provides a data operation and maintenance method for photovoltaic power plants, including the following steps:

[0049] Step S100: Obtain the photovoltaic power station operation data of the current evaluation batch, and read the historical memory file to construct the evaluation input set;

[0050] Step S200: Load the unified indicator system and obtain the set of indicators participating in this round of evaluation calculation;

[0051] Step S300: Based on the indicator set, perform equipment-level evaluation calculation, site-level evaluation calculation, and multi-time-granularity evaluation calculation on the evaluation input set to obtain sub-item evaluation data, which includes equipment-level evaluation data, site-level evaluation data, and time-granularity evaluation data.

[0052] Step S400: Perform horizontal comparative analysis and vertical trend analysis based on the sub-item evaluation data to obtain horizontal comparative analysis data and vertical trend analysis data;

[0053] Step S500: Based on the horizontal comparison analysis data and the vertical trend analysis data, and combined with the preset anomaly judgment conditions, potential anomaly identification data is generated. The potential anomaly identification data is used to characterize at least one of equipment differences, trend changes and potential anomalies.

[0054] Step S600: Based on the sub-item evaluation data, horizontal comparative analysis data, and vertical trend analysis data, calculate the comprehensive health score of the equipment and the comprehensive health score of the site.

[0055] Step S700: Combine the comprehensive health score of equipment, the comprehensive health score of the site, the longitudinal trend analysis data, and the potential anomaly identification data to determine the dynamic risk level;

[0056] Step S800: Generate report information, analysis information, and operational action information based on the comprehensive health score of equipment, the comprehensive health score of the site, and the dynamic risk level.

[0057] Step S900: Write back the sub-evaluation data, horizontal comparison analysis data, vertical trend analysis data, potential anomaly identification data, equipment health comprehensive score, site health comprehensive score, risk dynamic level, report information, analysis information and operational action information generated in this round to the historical memory file.

[0058] In this embodiment, the method can be executed by a photovoltaic power station data operation and maintenance platform, a regional-level new energy asset operation platform, a group-level photovoltaic asset management platform, or other computing systems with data acquisition, indicator calculation, evaluation and analysis, and result output capabilities. The method generally includes steps such as constructing an evaluation input set, loading a unified indicator system, multi-granularity evaluation calculation, horizontal comparative analysis, vertical trend analysis, potential anomaly identification, comprehensive health scoring, dynamic risk grading, business information generation, and historical memory file write-back.

[0059] In one embodiment, the photovoltaic power station operation data includes power station operation data, equipment operation data, weather data, and external auxiliary data; the historical memory file includes equipment operation memory, equipment trend memory, historical report specifications, and anomaly comparison records.

[0060] In one embodiment, the indicator set is filtered based on indicator activation status, approval status, application level, computability, evaluation task type, time granularity adaptability, and data source availability; wherein, the indicator set used for equipment-level evaluation calculation is the equipment-level indicator set, the indicator set used for site-level evaluation calculation is the site-level indicator set, and the indicator set used for multi-time granularity evaluation calculation is the indicator set adapted to the current statistical period.

[0061] In one embodiment, equipment-level evaluation calculation is used to generate equipment-level evaluation data for a single device within the current statistical period. The equipment-level evaluation data includes equipment power performance data, power generation performance data, stability performance data, and quality performance data. Station-level evaluation calculation is used to generate station-level evaluation data corresponding to the overall station operation performance based on the equipment-level evaluation data. Multi-time granularity evaluation calculation is used to generate daily, weekly, monthly reports, and time granularity evaluation data corresponding to custom time granularities.

[0062] In one embodiment, the horizontal comparative analysis includes comparison of equipment of the same model, comparison of equipment at the same site, and comparison of sites in the same region. The horizontal comparative analysis data is used to characterize the degree of operational deviation of the target object relative to similar objects. The vertical trend analysis includes health change trend analysis, continuous degradation analysis, and abnormal persistence analysis. The vertical trend analysis data is used to characterize the status changes of the target object in different statistical periods.

[0063] In one embodiment, the overall equipment health score is calculated based on multiple dimensions of data, including power performance, operational stability, data quality, alarm status, and trend changes; the overall site health score is obtained by summarizing, weighting, or hierarchically summarizing the overall equipment health score.

[0064] In one embodiment, the dynamic risk levels include normal, attention, warning, and high risk; the potential anomaly identification data includes at least one of single indicator anomaly identification data, continuous period anomaly identification data, multi-indicator superimposed anomaly identification data, and horizontal and vertical joint anomaly identification data; wherein, single indicator anomaly identification data is used to characterize whether a single indicator hits an anomaly threshold, continuous period anomaly identification data is used to characterize whether the same indicator meets the anomaly condition for multiple consecutive periods, multi-indicator superimposed anomaly identification data is used to characterize whether multiple different indicators simultaneously meet the anomaly condition in the same period, and horizontal and vertical joint anomaly identification data is used to characterize whether horizontal comparison anomalies and vertical trend anomalies occur simultaneously.

[0065] In one embodiment, based on the comprehensive health score of equipment, the comprehensive health score of the site, and the dynamic risk level, report information, analysis information, and operational action information are generated, including:

[0066] Generate daily, weekly, and monthly reports for the site as reporting information;

[0067] Generate device-level analysis results and custom analysis results as analysis information;

[0068] Generate maintenance suggestions, proactive early warning information, and push notification information as operational action information;

[0069] Specifically, when the overall health score of equipment or site is lower than the preset level threshold, or the dynamic risk level reaches the warning level or above, maintenance suggestions are generated based on sub-item assessment data, horizontal comparative analysis data, vertical trend analysis data, potential anomaly identification data, and historical memory files, and proactive warning information is generated.

[0070] Specifically, such as Figures 1 to 7 As shown, step S100: Obtain the photovoltaic power station operation data of the current evaluation batch, and read the historical memory file to construct the evaluation input set.

[0071] In this step, the system acquires the operational data of the photovoltaic power plants in the current evaluation batch. The current evaluation batch refers to the statistical period or analysis task scope corresponding to this round of data operation and maintenance analysis. This statistical period can be daily, weekly, monthly, or a user-specified custom time granularity, such as three consecutive days, ten consecutive days, a specified natural week, a specified assessment period, or any start and end date range selected manually.

[0072] Photovoltaic power plant operation data includes plant operation data, equipment operation data, weather data, and external auxiliary data. Plant operation data reflects the overall operating status of the photovoltaic power plant and may include total active power, grid-connected electricity, power consumption, planned power generation, and theoretical power generation. Equipment operation data reflects the operating status of individual devices within the photovoltaic power plant and may include device number, active power, output voltage, output current, cumulative power generation, daily power generation, power factor, status code, and fault alarm code. Weather data reflects the environmental conditions of the power plant and may include irradiance, ambient temperature, module temperature, and wind speed. External auxiliary data reflects external business factors affecting the plant's operating status and analysis, and may include power curtailment records, maintenance plans, and external interface query results.

[0073] While acquiring the operational data of the photovoltaic power plants in the current evaluation batch, the system also reads historical memory files. These historical memory files include equipment operation memory, equipment trend memory, historical report specifications, and anomaly comparison records. Equipment operation memory stores data such as equipment performance, overall health score, and dynamic risk level over historical periods. Equipment trend memory stores trend changes in equipment over multiple periods, such as health trends, continuous degradation, and persistence of anomalies. Historical report specifications store information such as historical report structure, field configuration, output format, and statistical granularity. Anomaly comparison records store information such as historical anomaly objects, anomaly types, anomaly detection conditions, processing results, and review status.

[0074] The system compiles the operational data of the photovoltaic power plants in the current evaluation batch with historical memory files into an evaluation input set. This evaluation input set is not simply a collection of data, but rather a unified foundation for subsequent equipment-level evaluation calculations, power plant-level evaluation calculations, multi-time-granularity evaluation calculations, horizontal comparative analysis, vertical trend analysis, potential anomaly identification, comprehensive health scoring, and dynamic risk grading. Through this step, the system can simultaneously consider the current operating status and historical operating patterns in this round of analysis, improving the continuity and interpretability of the evaluation results.

[0075] For example, for a given inverter, the system not only acquires the inverter's active power, daily power generation, output voltage, output current, power factor, and fault alarm codes within the current statistical period, but also reads the inverter's comprehensive health score, historical anomaly records, and trend changes over multiple historical statistical periods. In this way, during subsequent evaluations, the system can determine whether the inverter currently deviates from similar devices and whether its performance has continuously deteriorated compared to its historical performance.

[0076] Step S200: Load the unified indicator system and obtain the set of indicators participating in this round of evaluation calculation.

[0077] In this step, the system loads a unified indicator system and retrieves the set of indicators used in this round of evaluation calculations from the unified indicator system. The unified indicator system is used to uniformly manage the indicators used in the operation and maintenance of photovoltaic power plant data, ensuring consistency and traceability in the definition, classification, status, approval, and retrieval processes of the indicators.

[0078] Specifically, loading a unified indicator system and obtaining the set of indicators to participate in this round of evaluation calculations may include the following sub-steps.

[0079] Retrieves the indicator registration and maintenance results. The indicator registration and maintenance results record the registration definition and maintenance information for each indicator. Indicator definitions may include indicator identifier, indicator name, input mapping, formula identifier, threshold configuration, application level, and version number. The indicator identifier uniquely identifies the indicator; the indicator name describes the indicator's business meaning; the input mapping indicates the data source and fields required for indicator calculation; the formula identifier indicates the indicator calculation rules; the threshold configuration indicates the thresholds required for anomaly detection, grading, or scoring calculation; the application level indicates whether the indicator is applicable to equipment level, site level, or other levels; and the version number distinguishes changes in indicator definition at different stages.

[0080] Retrieve indicator classification results. Indicator classification results are used for categorizing and managing indicators. Indicators can be categorized into equipment-level indicators and site-level indicators based on their application object, or into basic indicators and composite indicators based on their attributes. Equipment-level indicators describe the operational performance of a single piece of equipment, while site-level indicators describe the overall operational performance of the entire site. Basic indicators are typically derived directly from a single measuring point or simple calculation results, while composite indicators are usually obtained by combining multiple basic indicators, multiple object data, or multiple time period data.

[0081] The system reads the indicator lifecycle status and indicator approval status. The indicator lifecycle status indicates whether the indicator is in an activated, pending evaluation, or decommissioned state. An activated status means the indicator can participate in the formal evaluation calculation; a pending evaluation status means the indicator is still in the candidate or verification stage and generally cannot be directly used in the formal calculation; a decommissioned status means the indicator will no longer participate in the formal calculation. The indicator approval status determines whether the indicator is allowed to participate in this round of formal calculation. By controlling the lifecycle status and approval status, indicators that have not been approved or have already been decommissioned can be prevented from directly entering the formal evaluation process.

[0082] Based on the current assessment task objectives and input data conditions, a set of indicators for this round of assessment calculations is constructed. The current assessment task objectives can be equipment-level assessment, site-level assessment, daily report assessment, weekly report assessment, monthly report assessment, custom time-granularity assessment, horizontal comparative analysis, vertical trend analysis, or comprehensive score calculation, etc. Input data conditions can include whether the data source is available, whether the fields are complete, whether the time range meets the requirements, whether the equipment objects match, and whether the statistical period is suitable. The system filters indicators based on their activation status, approval status, application level, computability, assessment task type, time granularity suitability, and data source availability to construct the set of indicators required for this round of assessment.

[0083] For equipment-level assessment calculations, the system calls upon a set of equipment-level indicators. This set can include indicators of equipment power performance, power generation performance, stability, quality performance, alarm status, and trend changes. For site-level assessment calculations, the system calls upon a set of site-level indicators. This set can include indicators of overall site power generation performance, average health status, percentage of abnormal equipment, risk distribution, and overall operational status. For multi-time-granularity assessment calculations, the system calls upon an indicator set appropriate for the current statistical period. For example, daily report analysis calls indicators suitable for daily granularity, weekly report analysis calls indicators suitable for weekly granularity, monthly report analysis calls indicators suitable for monthly granularity, and custom time-granularity analysis calls indicators matching the custom time interval.

[0084] This step helps the system avoid inconsistencies in the definition of indicators across different reports, devices, sites, or scripts, ensuring that all subsequent analyses are based on a unified indicator system.

[0085] Step S300: Perform equipment-level evaluation calculation, site-level evaluation calculation, and multi-time granularity evaluation calculation based on the indicator set to obtain sub-item evaluation data.

[0086] In this step, the system performs equipment-level evaluation calculations, site-level evaluation calculations, and multi-time-granularity evaluation calculations on the evaluation input set based on the indicator set, to obtain sub-item evaluation data. The sub-item evaluation data includes equipment-level evaluation data, site-level evaluation data, and time-granularity evaluation data.

[0087] Equipment-level assessment calculations are used to generate equipment-level assessment data for a single device within the current statistical period. This data includes power performance data, power generation performance data, stability performance data, and quality performance data. Power performance data reflects the difference between the device's current output level and that of similar devices, historical levels, or theoretical levels. Power generation performance data reflects the device's power generation contribution within the statistical period. Stability performance data reflects the device's fluctuations, continuous operating status, and operational stability within the statistical period. Quality performance data reflects the completeness, validity, and availability of the device's measurement data, and can also reflect issues such as missing data, abnormal spikes, offline status, and time asynchrony.

[0088] The station-level assessment calculation is used to generate station-level assessment data corresponding to the overall station operation performance based on equipment-level assessment data. Station-level assessment data can be obtained by aggregating equipment-level assessment data, such as statistics on the number of online devices, average health level, risk distribution, anomaly summaries, overall power generation performance, percentage of abnormal devices, and overall station operation status. The station-level assessment calculation reflects not only the status of individual devices but also the overall operational health status of the entire station within the current statistical period.

[0089] Multi-time-granularity evaluation calculations are used to generate daily, weekly, and monthly reports, as well as time-granularity evaluation data corresponding to custom time granularities. Daily reports correspond to a single-day statistical period, weekly reports to a weekly statistical period, monthly reports to a monthly statistical period, and custom time granularities correspond to any time interval specified by the user or business system. Custom time granularities can be used for scenarios such as ad-hoc special analysis, periodic assessment analysis, pre- and post-fault comparison analysis, and analysis of non-fixed report cycles. For example, when a device malfunctions on a certain day, maintenance personnel can choose the three days before and after the malfunction as custom time granularities to compare the device's power performance, power generation performance, stability performance, and health status trends before and after the malfunction.

[0090] Through this step, the sub-assessment data obtained by the system provides the foundation for subsequent horizontal comparative analysis, vertical trend analysis, potential anomaly identification, comprehensive health scoring, and dynamic risk grading. Because the sub-assessment data clearly distinguishes between equipment-level assessment data, site-level assessment data, and time-granularity assessment data, the data sources for subsequent steps are clear.

[0091] Step S400: Perform horizontal comparative analysis and vertical trend analysis based on the sub-item evaluation data.

[0092] In this step, the system performs horizontal comparative analysis and vertical trend analysis based on the sub-item evaluation data to obtain horizontal comparative analysis data and vertical trend analysis data.

[0093] Horizontal comparative analysis includes comparisons of equipment of the same model, equipment at the same photovoltaic plant, and plants within the same region. Horizontal comparative analysis data is used to characterize the degree of operational deviation of the target device relative to similar devices. The target device can be a single unit of equipment or a photovoltaic plant. Comparisons of equipment of the same model involve comparing the target device with other equipment of the same model to determine whether the target device deviates under similar equipment conditions. Comparisons of equipment at the same photovoltaic plant involve comparing the target device with other equipment within the same plant to determine whether the target device differs from similar equipment within the plant. Comparisons of plants within the same region involve comparing the target plant with other plants within the same region to determine whether the target plant deviates under similar regional conditions.

[0094] For example, if an inverter consistently performs worse than the average for similar models in terms of active power, daily power generation, or overall health score under similar irradiance and operating conditions, the system can record the degree of operational deviation of that inverter relative to similar devices in the horizontal comparative analysis data. As another example, if a power station's overall power generation performance is lower than other stations in the same region, and the proportion of abnormal equipment is higher, the system can record the degree of operational deviation of that station relative to other stations in the same region in the horizontal comparative analysis data.

[0095] Longitudinal trend analysis includes health status change trend analysis, continuous degradation analysis, and anomaly persistence analysis. Longitudinal trend analysis data is used to characterize the status changes of the target object within different statistical periods. Health status change trend analysis is used to determine whether the overall health score of the target object increases, decreases, remains stable, or fluctuates over multiple consecutive periods. Continuous degradation analysis is used to determine whether the power performance, power generation performance, stability performance, quality performance, or other indicators of the target object continue to deteriorate. Anomaly persistence analysis is used to determine whether the target object repeatedly exhibits the same type of anomaly within multiple statistical periods.

[0096] For example, if a piece of equipment experiences a continuous decline in its overall health score over several statistical periods, consistently lower than historical levels in power generation, and a gradual increase in the number of anomalies, the system can record in the longitudinal trend analysis data that the equipment is experiencing continuous degradation or persistent anomalies. As another example, if a power plant experiences a decline in its overall health score and a continuous increase in the proportion of abnormal equipment over several consecutive weekly reporting periods, the system can record in the longitudinal trend analysis data that the power plant exhibits an anomaly in its trend changes at the plant level.

[0097] Horizontal comparative analysis and vertical trend analysis are not independent of each other. Horizontal comparative analysis emphasizes the relative differences between the target object and similar objects, while vertical trend analysis emphasizes the target object's own state over time. Combining the two can provide a more comprehensive assessment of potential anomalies. For example, if a device's current power performance is lower than that of other devices in the same group, and its overall health score has been declining continuously over several recent periods, then this device exhibits both horizontal deviation and vertical degradation, and can receive greater attention in subsequent potential anomaly identification and dynamic risk grading.

[0098] Step S500: Generate potential anomaly identification data based on horizontal comparative analysis data and vertical trend analysis data.

[0099] In this step, the system generates potential anomaly identification data based on horizontal comparative analysis data and vertical trend analysis data, combined with preset anomaly judgment conditions. The potential anomaly identification data is used to characterize at least one of equipment differences, trend changes, and potential anomalies.

[0100] Preset anomaly detection criteria can be determined based on threshold configurations, indicator rules, historical statistical results, business experience, or current assessment task requirements within a unified indicator system. These criteria can be used to determine whether a single indicator is abnormal, whether the same indicator is continuously abnormal, whether multiple indicators are simultaneously abnormal, and whether horizontal comparison anomalies and vertical trend anomalies occur simultaneously.

[0101] Potential anomaly identification data includes at least one of the following: single-indicator anomaly identification data, continuous periodic anomaly identification data, multi-indicator superimposed anomaly identification data, and horizontal and vertical joint anomaly identification data.

[0102] Individual indicator anomaly identification data is used to characterize whether an individual indicator hits an anomaly threshold. For example, if a device's power performance is lower than a preset threshold or its data quality is lower than a preset threshold during the current statistical period, the system can generate individual indicator anomaly identification data.

[0103] Continuous periodic anomaly identification data is used to characterize whether the same indicator meets the abnormal conditions for multiple consecutive periods. For example, if the overall health score of a device declines for three consecutive statistical periods, or if the power generation performance of a device is lower than the average level of similar devices for multiple consecutive periods, the system can generate continuous periodic anomaly identification data.

[0104] Multi-indicator overlay anomaly identification data is used to characterize whether multiple different indicators simultaneously meet abnormal conditions within the same period. For example, if a device simultaneously exhibits abnormal power performance, abnormal operational stability, abnormal data quality, and abnormal alarm status within the same statistical period, the system can generate multi-indicator overlay anomaly identification data.

[0105] Combined horizontal and vertical anomaly identification data is used to characterize whether horizontal comparative anomalies and vertical trend anomalies occur simultaneously. For example, if a device deviates significantly from other devices of the same model, and its overall health score declines over multiple consecutive periods, the system can generate combined horizontal and vertical anomaly identification data. This type of data reflects that the target object not only exhibits current abnormal performance but also carries a risk of continuous degradation, thus possessing high value in subsequent dynamic risk grading.

[0106] Through this step, the present invention transforms the analysis results of "equipment differences," "trend changes," and "potential anomalies" into potential anomaly identification data with clear sources and types. This data originates from both horizontal comparative analysis data and vertical trend analysis data, and is also constrained by preset anomaly judgment conditions, thus possessing a relatively clear logical foundation and referential relationships.

[0107] Step S600: Calculate the overall health score of the equipment and the overall health score of the site.

[0108] In this step, the system calculates the comprehensive health score of equipment and the comprehensive health score of the site based on the sub-item evaluation data, the horizontal comparative analysis data, and the vertical trend analysis data.

[0109] The overall equipment health score characterizes the comprehensive health status of a single piece of equipment within the current statistical period. This score is calculated based on multiple dimensions of data, including power performance, operational stability, data quality, alarm status, and trend changes. Power performance reflects the difference between the equipment's current output level and that of similar devices or historical levels; operational stability reflects the degree of fluctuation and continuous operation of the equipment within the statistical period; data quality reflects the completeness, validity, and availability of the equipment's measurement data; alarm status reflects whether the equipment frequently generates abnormal or fault alarms within the statistical period; and trend changes reflect whether the equipment's operating status shows continuous improvement or degradation over multiple consecutive periods.

[0110] In practical implementation, the system can assign different weights to the aforementioned dimensions based on different business priorities and perform comprehensive aggregation according to preset rules. For example, for scenarios where power generation capacity is the primary focus, the weights of dimensions related to power performance and power generation can be increased; for scenarios where risk identification is the primary focus, the weights of dimensions related to alarm status and trend changes can be increased; and for scenarios where data reliability is the primary focus, the weights of dimensions related to data quality can be increased. It should be understood that the weight setting method, scoring range, and aggregation rules can be configured according to business needs, as long as they are based on comprehensive calculations of sub-item evaluation data, horizontal comparative analysis data, and vertical trend analysis data, they fall within the scope of this invention's implementation.

[0111] The overall health score of a photovoltaic power plant is used to characterize the comprehensive health status of the entire plant within the current statistical period. The overall health score can be obtained based on the overall equipment health scores through average aggregation, weighted aggregation, or stratified aggregation. Average aggregation can be used to calculate the arithmetic average of the overall equipment health scores within the plant; weighted aggregation can aggregate the overall equipment health scores according to preset weights; stratified aggregation can first aggregate the overall equipment health scores of different equipment types or different areas, and then form the overall plant score.

[0112] For example, if a power station contains multiple inverters, the system calculates the overall health score for each inverter separately, and then obtains the overall health score for the entire power station based on average aggregation, weighted aggregation, or stratified aggregation. If the health scores of some devices are significantly low, the overall health score of the power station can reflect the overall operational health status of the power station accordingly.

[0113] Through this step, the system transforms data from multiple dimensions, multiple objects, and multiple periods into a comprehensive health score for equipment and a comprehensive health score for the site, enabling maintenance personnel to understand the overall health status of equipment and sites in an intuitive way.

[0114] Step S700: Determine the dynamic risk level.

[0115] In this step, the system combines comprehensive equipment health scores, comprehensive site health scores, longitudinal trend analysis data, and potential anomaly identification data to determine the dynamic risk level. The dynamic risk levels include normal, attention level, warning level, and high risk.

[0116] When a target object has a high overall score, its longitudinal trend analysis data indicates a stable state, and the potential anomaly identification data does not indicate obvious anomalies, the system can determine the target object's dynamic risk level as normal. When a target object's score declines, local indicators show anomalies, or the trend shows a slight deterioration, the system can determine the target object as a focus. When a target object shows anomalies in multiple key indicators, the trend continues to deteriorate, or problems recur for multiple consecutive periods, the system can determine the target object as a warning. When a target object's overall score is significantly low, key indicators are persistently abnormal, or there is a significant deterioration trend, the system can determine the target object as high-risk.

[0117] The "dynamic" aspect of the risk dynamic rating is reflected in the fact that the system does not make static judgments based solely on a single abnormal result, but rather combines comprehensive assessments of equipment health, site health, longitudinal trend analysis data, and potential anomaly identification data. For example, if a single indicator of a piece of equipment slightly exceeds its limit within a single period, the system can designate it as a matter of concern; if the same indicator of the equipment meets the abnormal conditions for multiple consecutive periods, the system can raise its risk level; and if the equipment exhibits multiple overlapping anomalies within the same period, and both horizontal comparison anomalies and longitudinal trend anomalies occur simultaneously, the system can further raise its dynamic risk rating.

[0118] For equipment-level objects, the dynamic risk level reflects the operational risk of a single device. For site-level objects, the dynamic risk level reflects the overall operational risk of the entire photovoltaic power plant. For example, when the overall health of a power plant declines for several consecutive periods, the proportion of abnormal devices continues to rise, and multiple devices have joint horizontal and vertical anomaly identification data, the system can identify the power plant as under warning or high-risk.

[0119] Through this step, the present invention realizes dynamic risk classification driven by comprehensive health score, trend change and potential anomaly identification, which is more suitable for photovoltaic power station asset operation scenarios than traditional fixed threshold alarm methods.

[0120] Step S800: Generate report information, analysis information, and operational action information.

[0121] In this step, the system generates report information, analysis information, and operational action information based on the comprehensive health score of equipment, the comprehensive health score of the site, and the dynamic risk level.

[0122] The reports can include daily, weekly, and monthly reports at the facility level. Daily reports can include information such as the number of online devices, average health status, risk distribution, anomaly summaries, and overall operational status. Weekly and monthly reports can add information such as periodic trends, ongoing anomalies, and changes in key risk targets to the daily reports. Through these reports, facility maintenance personnel, regional operations personnel, and management personnel can quickly obtain the facility's operational status at different time granularities.

[0123] The analysis information can include device-level analysis results and custom analysis results. Device-level analysis results can include overall device health scores, trend changes, detailed anomaly indicators, horizontal comparison results, vertical trend results, and potential anomaly identification results. Custom analysis results can output results for specific device models, specific sites, specific time intervals, or specific combinations of indicators according to business needs. For example, maintenance personnel can select a certain device type and a custom time interval to generate horizontal comparison analysis results, vertical trend analysis results, and health comparison results for that type of device within that time interval.

[0124] Operational action information can include maintenance recommendations, proactive early warning information, and push notifications. When the overall health score of equipment or a site falls below a preset threshold, or the dynamic risk level reaches or exceeds the warning level, the system generates maintenance recommendations and proactive early warning information based on sub-assessment data, horizontal comparative analysis data, vertical trend analysis data, potential anomaly identification data, and historical memory files. Maintenance recommendations may include component cleaning recommendations, inverter inspection recommendations, wiring inspection recommendations, and trend observation recommendations. Proactive early warning information may include the risk object, risk level, anomaly type, recommended handling measures, and trigger time. Push notifications can be used to send reminders to relevant personnel via Lark messages, emails, work order systems, or third-party business platforms.

[0125] For example, if a device's overall health score falls below a preset threshold, and potential anomaly identification data indicates a combined horizontal and vertical anomaly, the system can generate maintenance suggestions, prompting maintenance personnel to focus on checking the device's inverter operating status, wiring, or component cleanliness. Simultaneously, the system can generate proactive warning messages and notify relevant personnel via push notifications.

[0126] Through this step, the present invention realizes a direct mapping from evaluation calculation results to business output and operational actions, enabling photovoltaic power plant data operation and maintenance results to not only remain at the level of report display, but also further support actual maintenance and risk response.

[0127] Step S900: Write back the data generated in this round to the history memory file.

[0128] In this step, the system writes back the sub-evaluation data, horizontal comparative analysis data, vertical trend analysis data, potential anomaly identification data, comprehensive equipment health score, comprehensive site health score, dynamic risk level, report information, analysis information, and operational action information generated in this round to the historical memory file.

[0129] Specifically, the system can write equipment-level assessment data, site-level assessment data, and time-granularity assessment data into a historical memory file to form historical sub-assessment records required for subsequent assessments. The system can also write horizontal comparative analysis data and vertical trend analysis data into the historical memory file to subsequently determine relative deviations and trend changes in equipment. Furthermore, the system can write potential anomaly identification data into the historical memory file to create records of anomaly objects, anomaly types, anomaly judgment conditions, and anomaly occurrence cycles. Finally, the system can write comprehensive equipment health scores, comprehensive site health scores, and dynamic risk levels into the historical memory file for subsequent trend analysis and risk review. Finally, the system can write report information, analysis information, and operational action information into the historical memory file to save information such as output paths, output times, report structures, maintenance recommendations, proactive warning records, message push records, and execution status.

[0130] By writing back the data from the current analysis, the results can become historical input for the next round of evaluation. For example, in the next evaluation, the system can read the comprehensive health score, dynamic risk level, and potential anomaly identification data from the previous round or multiple historical cycles to determine whether the equipment is continuously degrading, repeatedly encountering the same type of anomaly, or whether the status has been upgraded from concern to warning or high risk. This mechanism enables the invention to form a continuously optimized data operation and maintenance closed loop.

[0131] Furthermore, this application also provides a data operation and maintenance system for photovoltaic power plants. For example... Figure 8 As shown, this system can be used to execute the data operation and maintenance method of Embodiment 1. The system includes a data and memory file layer, a business logic layer, and a business application layer.

[0132] The data and memory file layer is used to obtain the photovoltaic power plant operation data of the current evaluation batch and read historical memory files to construct the evaluation input set.

[0133] The data and memory file layer can acquire photovoltaic (PV) power plant operation data from the power plant monitoring system, equipment interfaces, data gateways, databases, external interfaces, message queues, or other sources where PV power plant operation data is available. PV power plant operation data includes power plant operation data, equipment operation data, weather data, and external auxiliary data. The data and memory file layer also reads historical memory files, which include equipment operation memories, equipment trend memories, historical report specifications, and anomaly comparison records.

[0134] The data and memory file layer serves to provide the business logic layer with a unified, complete, and reusable set of evaluation inputs. Through this layer, the system can combine data from the current evaluation batch with historical accumulated data as input, avoiding subsequent calculations relying solely on single real-time data.

[0135] The business logic layer is used to load a unified indicator system, obtain the set of indicators participating in this round of evaluation calculations, and perform equipment-level evaluation calculations, site-level evaluation calculations, and multi-time-granularity evaluation calculations on the evaluation input set based on the indicator set to obtain sub-item evaluation data. Based on the sub-item evaluation data, horizontal comparative analysis and vertical trend analysis are performed to obtain horizontal comparative analysis data and vertical trend analysis data. Based on the horizontal comparative analysis data and vertical trend analysis data, and combined with preset anomaly judgment conditions, potential anomaly identification data is generated. Based on the sub-item evaluation data, horizontal comparative analysis data, and vertical trend analysis data, the comprehensive equipment health score and the comprehensive site health score are calculated. Finally, the dynamic risk level is determined by combining the comprehensive equipment health score, the comprehensive site health score, the vertical trend analysis data, and the potential anomaly identification data.

[0136] The business logic layer includes a basic indicator system, an evaluation and calculation module, and an auxiliary indicator calling module.

[0137] The basic indicator system is used to read indicator registration and maintenance results, indicator classification results, indicator lifecycle status, and indicator approval status, and provides a unified, standardized, and reusable set of indicators for subsequent evaluation calculations. The basic indicator system supports indicator definition, registration, and maintenance; indicator classification management; indicator activation status management; pending evaluation status management; decommissioning status management; and approval status management. Through the basic indicator system, the system can ensure that indicators participating in formal evaluation calculations have consistent definitions and traceable sources.

[0138] The assessment and calculation module performs equipment-level assessment calculations, site-level assessment calculations, multi-time-granularity assessment calculations, horizontal comparative analysis, vertical trend analysis, potential anomaly identification, comprehensive health score calculation, and dynamic risk grading. It is the core processing module in the business logic layer. Its inputs are the assessment input set and the indicator set, and its outputs include sub-assessment data, horizontal comparative analysis data, vertical trend analysis data, potential anomaly identification data, comprehensive equipment health score, comprehensive site health score, and dynamic risk level.

[0139] The auxiliary indicator retrieval module is used to retrieve registered and approved indicators, and to reference historically accumulated indicators as supplementary input when needed. This module does not rediscover or arbitrarily generate indicators; rather, it retrieves registered, approved indicators that are applicable to the current task within a unified indicator system. Furthermore, the auxiliary indicator retrieval module can assist the evaluation and calculation module in selecting appropriate indicators based on the current task objectives, time granularity, data source availability, and indicator suitability.

[0140] The business application layer is used to generate report information, analysis information and operational action information based on the comprehensive health score of equipment, comprehensive health score of site and dynamic risk level, and write back the sub-evaluation data, horizontal comparison analysis data, vertical trend analysis data, potential anomaly identification data, comprehensive health score of equipment, comprehensive health score of site, dynamic risk level, report information, analysis information and operational action information generated in this round to the historical memory file.

[0141] The business application layer includes a report output module, an asset operation action module, and a memory write-back module.

[0142] The report output module generates daily, weekly, and monthly reports at the site level, and can also generate equipment-level analysis results and custom analysis results. The module can generate reports based on historical report specifications and current evaluation data, ensuring a consistent output format for analysis results across different time granularities, object levels, and management needs.

[0143] The asset operation action module generates maintenance suggestions, proactive early warning information, and push notifications as operational action information. When the overall health score of equipment or site falls below a preset threshold, or the dynamic risk level reaches or exceeds the warning level, the asset operation action module generates maintenance suggestions and proactive early warning information based on sub-assessment data, horizontal comparative analysis data, vertical trend analysis data, potential anomaly identification data, and historical memory files. The asset operation action module can send proactive early warning information to relevant personnel or business systems via Lark messages, emails, work order systems, or third-party business platforms.

[0144] The memory write-back module is used to write various types of data generated in the current round back to the historical memory file. This module can uniformly write evaluation data, analysis data, scoring data, risk levels, report information, analytical information, and operational action information into the historical memory file, enabling subsequent evaluations to access the results of this round. Through the memory write-back module, the system forms a closed loop from data input to operational output and then to historical memory updates.

[0145] The following is a further explanation of the unified indicator system and indicator lifecycle management:

[0146] In photovoltaic power plant data operation and maintenance scenarios, indicators are key elements connecting raw operational data with evaluation results. If indicators lack unified management, different analysis tasks may use different indicator definitions, leading to inconsistent results. Therefore, this invention manages indicator registration and maintenance results, indicator classification results, indicator lifecycle status, and indicator approval status through a unified indicator system.

[0147] The indicator registration and maintenance results record the basic definitions of the indicators, including indicator identifier, indicator name, input mapping, formula identifier, threshold configuration, application level, and version number. The indicator classification results categorize indicators into equipment-level indicators, site-level indicators, basic indicators, and composite indicators. The indicator lifecycle status records the indicator's status at different stages, including activation, pending evaluation, and decommissioning. The indicator approval status determines whether the indicator is allowed to participate in the current round of formal evaluation calculations.

[0148] like Figure 7 As shown, in one implementation, indicator lifecycle management may include processes such as candidate indicator generation, pending approval status, manual review, approval, writing to the official indicator registry, entry into the official indicator pool, participation in official evaluation calculations, feedback on operating results and usage effects, continuous monitoring and review, remaining active, decommissioning, ceasing participation in official calculations, and entering the decommissioned indicator pool. When a candidate indicator is generated, it is in a pending approval status. If the review is rejected, the approval record is retained, and it does not enter the official indicator pool; if the review is approved, it is written to the official indicator registry, enters the official indicator pool, and can participate in official evaluation calculations. After an indicator participates in official evaluation calculations, the system continuously monitors its operating and usage effects. If the indicator remains valid, it remains active; if the indicator is no longer applicable, decommissioning is performed, causing it to stop participating in official calculations and enter the decommissioned indicator pool.

[0149] Through the above lifecycle management, this invention can achieve full-process governance of indicator definition, calculation, approval, release, application, traceability and decommissioning, avoiding candidate indicators or unapproved indicators from entering the formal analysis process in an uncontrolled manner, and improving the standardization of the indicator system.

[0150] The following is a further explanation of the multi-time granularity evaluation calculation:

[0151] Multi-time-granularity assessment calculations are used to generate daily, weekly, and monthly reports, as well as time-granularity assessment data corresponding to custom time granularities. In the daily report scenario, the system assesses equipment and sites based on daily data, generating equipment-level assessment data, site-level assessment data, and related analysis information for that day. In the weekly report scenario, the system summarizes equipment and site data according to the weekly statistical cycle, generating cycle change trends, persistent anomalies, and changes in key risk objects. In the monthly report scenario, the system generates monthly operational performance, health changes, risk distribution, and anomaly summaries according to the monthly statistical cycle.

[0152] Custom time-granularity evaluation calculations refer to the system's ability to perform evaluations on target objects within any predefined statistical time interval, in addition to supporting fixed statistical periods of daily, weekly, and monthly evaluations. For example, the system can perform unified indicator aggregation, trend judgment, and anomaly identification for equipment or sites based on any start and end date interval, such as three consecutive days, ten consecutive days, a specified natural week, a specified assessment period, or manually selected intervals. Upon receiving the custom time-granularity parameters, the system filters the operating data and historical memory files within the corresponding time range and calls upon the indicator set adapted to that time granularity for evaluation calculations.

[0153] By using customizable time-granularity assessment calculations, the system can adapt to scenarios such as ad-hoc special analyses, periodic performance evaluations, pre- and post-failure comparative analyses, and analyses with non-fixed reporting cycles. For example, after an equipment failure occurs, maintenance personnel can set a certain number of days before and after the failure date as a custom time interval. Based on the data within this interval, the system calculates sub-assessment data, horizontal comparative analysis data, vertical trend analysis data, and a comprehensive health score, thereby helping to determine whether there was a degradation trend before the failure and the recovery effect after the failure.

[0154] The following is a further explanation of potential anomaly identification and dynamic risk classification:

[0155] Potential anomaly identification data is generated jointly from horizontal comparative analysis data, vertical trend analysis data, and preset anomaly judgment conditions. Potential anomaly identification data can characterize at least one of equipment differences, trend changes, and potential anomalies, and can specifically include single-indicator anomaly identification data, continuous periodic anomaly identification data, multi-indicator superimposed anomaly identification data, and horizontal and vertical joint anomaly identification data.

[0156] In one scenario, if the system detects that a device's power performance data is below a preset anomaly threshold, it generates single-index anomaly identification data. In another scenario, if the system detects that a device's overall health score has been continuously declining over multiple statistical periods, it generates continuous-period anomaly identification data. In yet another scenario, if the system detects that a device exhibits anomalies in power performance, operational stability, and alarm status simultaneously within the same statistical period, it generates multi-index superimposed anomaly identification data. In yet another scenario, if the system detects that a device deviates significantly from other devices of the same model, and its historical trends also show continuous degradation, it generates combined horizontal and vertical anomaly identification data.

[0157] The dynamic risk level is determined based on the comprehensive health score of equipment, the comprehensive health score of the site, longitudinal trend analysis data, and potential anomaly identification data. The dynamic risk levels include Normal, Attention, Warning, and High Risk. The Normal level indicates that the object's current health is high, the trend is stable, and no obvious potential anomalies have been found. The Attention level indicates that the object has localized anomalies or a slight deterioration in the trend. The Warning level indicates that the object has multiple abnormal key indicators, a continuously deteriorating trend, or continuous periodic anomalies. The High Risk level indicates that the object's comprehensive score is significantly low, key indicators are persistently abnormal, or there is a significant deterioration trend.

[0158] During the dynamic risk grading process, the system can adjust the risk level based on the data type and severity of potential anomalies. For example, a slight anomaly in a single indicator can correspond to a level of concern; continuous periodic anomalies can raise the risk level; multiple anomalies overlapping can further raise the risk level; and combined horizontal and vertical anomalies usually indicate that the object has both relative deviation and continuous degradation, thus serving as an important basis for raising the dynamic risk level.

[0159] In this way, the present invention can identify risks from multiple dimensions, multiple periods, and multiple object comparison perspectives, making the risk classification results more consistent with the actual operation of photovoltaic power stations.

[0160] The following is a further explanation of the generation of report information, analysis information, and operational action information:

[0161] The reports include daily, weekly, and monthly reports at the plant level. Daily reports may include the number of online devices, average health status, risk distribution, anomaly summaries, and overall operational status. Weekly reports can add weekly health status trends, persistently abnormal devices, risk level changes, and a summary of maintenance recommendations. Monthly reports can add monthly power generation performance, monthly risk distribution, key equipment trends, and operational action execution status to the weekly reports.

[0162] The analysis information includes equipment-level analysis results and custom analysis results. Equipment-level analysis results can be output for individual devices, including equipment-level assessment data, horizontal comparative analysis data, vertical trend analysis data, potential anomaly identification data, comprehensive equipment health score, and dynamic risk level. Custom analysis results can be generated based on the user-selected equipment type, site scope, time interval, or combination of indicators. For example, a user can query the health results, horizontal deviation, trend changes, and potential anomalies of a specific inverter model within a site over a specified period.

[0163] Operational action information includes maintenance recommendations, proactive early warning information, and push notifications. When the overall health score of equipment or a site falls below a preset threshold, or the dynamic risk level reaches or exceeds the warning level, the system generates maintenance recommendations and proactive early warning information based on sub-item assessment data, horizontal comparative analysis data, vertical trend analysis data, potential anomaly identification data, and historical memory files. Maintenance recommendations can vary depending on the type of anomaly. For example, for power performance anomalies, recommendations may include checking component cleanliness, inverter operating status, or wiring; for data quality anomalies, recommendations may include checking communication links, measurement point configuration, or data acquisition devices; for continuous degradation anomalies, recommendations may include scheduling key inspections or continuous observation; and for frequent alarm anomalies, recommendations may include troubleshooting equipment faults by combining alarm codes.

[0164] Proactive early warning information can include the object name, dynamic risk level, triggering reason, anomaly type, relevant indicators, suggested actions, and warning time. Push notifications can be sent via Lark messages, emails, ticketing systems, or third-party business platforms. It should be noted that push notification channels can be configured according to the actual business system; their purpose is to promptly deliver operational action information to relevant personnel or systems.

[0165] The following is a further explanation of memory file write-back and continuous optimization:

[0166] This invention generates sub-item assessment data, horizontal comparative analysis data, vertical trend analysis data, potential anomaly identification data, comprehensive equipment health score, comprehensive site health score, dynamic risk level, report information, analysis information, and operational action information, and then writes the above data back to a historical memory file. The historical memory file is thus updated and can be retrieved for subsequent assessments.

[0167] Memory file write-back can include the following:

[0168] Equipment-level assessment data, site-level assessment data, and time-granularity assessment data are written to a history memory file to save the basic assessment data of each object in each statistical period.

[0169] The data from the horizontal comparative analysis is written into a historical memory file to save the degree of deviation of the target object from similar objects in operation;

[0170] Write the longitudinal trend analysis data into a historical memory file to save the status changes of the target object in different statistical periods;

[0171] Potential anomaly identification data is written to a history memory file to save identification results such as single indicator anomalies, continuous period anomalies, multi-indicator superposition anomalies, and horizontal and vertical joint anomalies.

[0172] The comprehensive health scores of equipment and sites are written into a historical memory file for subsequent scoring trend analysis and risk review.

[0173] The dynamic risk level is written to a historical memory file for subsequent judgment on whether the risk level continues, escalates, or decreases.

[0174] Write report information, analysis information, and operational action information into a history memory file to save output results, analysis conclusions, maintenance suggestions, proactive warning records, message push records, and their execution status.

[0175] In the next round of evaluation, the system can read the written-back historical memory files and use the historical results as part of the new evaluation input set. For example, if the next round of evaluation finds that a device exhibits the same anomaly again, the system can combine the historical memory files to determine whether the device has a continuous cycle of anomalies; when the risk level of a device changes from concern to warning, the system can combine historical trends to determine whether it has experienced continuous degradation; when a maintenance suggestion has been generated but not yet processed, the system can continue to display the relevant processing status in the new operational action information.

[0176] Through the above methods, the present invention achieves continuous optimization of the data operation and maintenance process, enabling the results of each round of analysis to become the input basis for subsequent analyses, avoiding the loss of analysis results after one-time use, and improving the ability of photovoltaic power station asset operation knowledge accumulation.

[0177] Without departing from the technical concept of this invention, some implementations of this invention can be equivalently replaced by the following embodiments:

[0178] Regarding data acquisition methods, photovoltaic power plant operation data can be acquired through direct inverter interface acquisition, forwarding acquisition by the power plant monitoring system, unified acquisition by the data gateway, offline database retrieval, synchronous acquisition via API interface, or message queue subscription. Any method that can obtain the power plant operation data, equipment operation data, weather data, and external auxiliary data needed for evaluation calculations, and construct the evaluation input set together with historical memory files, can be used as the data acquisition method of this invention.

[0179] Regarding the implementation of the indicator system, a unified indicator system can be implemented through a registry configuration file, database configuration table, metadata management platform, rule engine configuration center, or visual indicator configuration interface. Indicator approval and lifecycle management can be implemented through a manual review interface, approval platform, workflow engine, or system configuration process. Any method that can achieve indicator registration and maintenance, indicator classification, lifecycle status and approval status management, and accordingly construct the set of indicators participating in this round of evaluation calculations, falls under the implementation methods of this invention.

[0180] Regarding the scoring calculation method, the comprehensive health score of equipment can be obtained using weighted summation, rule-based models, scoring models, or other methods that comprehensively consider power performance, operational stability, data quality, alarm status, and trend changes. The comprehensive health score of the site can be obtained using average summation, weighted summation, or hierarchical summation. As long as it is based on comprehensive calculation of sub-item evaluation data, horizontal comparative analysis data, and vertical trend analysis data, it will not affect the implementation of the technical solution of this invention.

[0181] Regarding anomaly detection methods, preset anomaly detection conditions can be derived from manual experience configuration, historical statistical derivation, fixed thresholds, quantile intervals, dynamic adaptive rules, or model learning results. Potential anomaly identification can be achieved through methods such as single-indicator anomalies, continuous periodic anomalies, multi-indicator superposition anomalies, and horizontal and vertical joint anomalies. Any method that generates potential anomaly identification data based on horizontal comparative analysis data and vertical trend analysis data falls under the scope of this invention.

[0182] Regarding output and operational actions, report information can be output as daily, weekly, monthly, structured JSON, dashboard interfaces, or other readable formats; operational action information can be generated through message notifications, email pushes, work order systems, mobile applications, or third-party business platforms. As long as report information, analysis information, and operational action information can be generated based on the comprehensive equipment health score, the comprehensive site health score, and the dynamic risk level, and the relevant data can be written back to the historical memory file, the implementation of the technical solution of this invention will not be affected.

[0183] Regarding system deployment, the data operation and maintenance system of this invention can be deployed on local servers, private clouds, public clouds, edge computing nodes, or hybrid architecture environments. The various modules of the system can be deployed collaboratively using monolithic deployment, microservice deployment, message bus invocation, or batch processing scheduling. Changes in deployment form do not affect the achievement of the core technical objectives of this invention.

[0184] In summary, the present invention provides a data operation and maintenance method and system for photovoltaic power plants. This method obtains sub-item evaluation data by constructing an evaluation input set, loading a unified indicator system, performing equipment-level evaluation calculations, power plant-level evaluation calculations, and multi-time-granularity evaluation calculations; it obtains horizontal comparative analysis data and vertical trend analysis data through horizontal comparative analysis and vertical trend analysis; it generates potential anomaly identification data by combining preset anomaly judgment conditions; it determines the dynamic risk level by calculating the comprehensive health score of equipment and the comprehensive health score of the power plant; and it forms a complete data operation and maintenance closed loop by generating report information, analysis information, and operational action information and writing them back to historical memory files.

[0185] This application can effectively solve the problems in photovoltaic power plant operation and maintenance analysis, such as inconsistent indicator standards, insufficient multi-granularity assessment, difficulty in jointly analyzing horizontal differences and vertical trends, lack of a unified framework for health scoring and risk classification, disconnect between analysis results and operational actions, and difficulty in reusing historical analysis results. It has good practicality, scalability and operational value.

[0186] It should be noted that this application is not limited to a specific data access method, a specific algorithm model, a specific threshold generation method, or a specific output format. Any technical solution based on multi-source operational data from photovoltaic power plants, which achieves data standardization, indicator management, hierarchical analysis, anomaly identification, closed-loop output, and traceability archiving around a semantic model for power plant analysis, should fall within the protection scope of this invention.

Claims

1. A data operation and maintenance method applied to photovoltaic power plants, characterized in that, Includes the following steps: Obtain the operational data of the photovoltaic power plants in the current evaluation batch, and read the historical memory files to construct the evaluation input set; Load the unified indicator system and obtain the set of indicators to participate in this round of evaluation calculations; Based on the set of indicators, equipment-level evaluation calculation, station-level evaluation calculation, and multi-time-granularity evaluation calculation are performed on the evaluation input set to obtain sub-item evaluation data, which includes equipment-level evaluation data, station-level evaluation data, and time-granularity evaluation data. Based on the sub-evaluation data, perform horizontal comparative analysis and vertical trend analysis to obtain horizontal comparative analysis data and vertical trend analysis data; Based on the horizontal comparison analysis data and the vertical trend analysis data, and combined with the preset anomaly judgment conditions, potential anomaly identification data is generated. The potential anomaly identification data is used to characterize at least one of equipment differences, trend changes and potential anomalies. Based on the sub-item evaluation data, the horizontal comparison analysis data, and the vertical trend analysis data, calculate the comprehensive health score of the equipment and the comprehensive health score of the site; By combining the comprehensive health score of the equipment, the comprehensive health score of the site, the longitudinal trend analysis data, and the potential anomaly identification data, the dynamic risk level is determined; Based on the comprehensive health score of the equipment, the comprehensive health score of the site, and the dynamic risk level, generate report information, analysis information, and operational action information; The sub-evaluation data, horizontal comparative analysis data, vertical trend analysis data, potential anomaly identification data, comprehensive equipment health score, comprehensive site health score, dynamic risk level, report information, analysis information, and operational action information generated in this round will be written back to the historical memory file.

2. The method according to claim 1, characterized in that, The photovoltaic power station operation data includes power station operation data, equipment operation data, weather data, and external auxiliary data; the historical memory file includes equipment operation memory, equipment trend memory, historical report specifications, and anomaly comparison records.

3. The method according to claim 1, characterized in that, Load the unified indicator system to obtain the set of indicators participating in this round of evaluation calculations, including: Read the indicator registration and maintenance results; Read the indicator classification results; Read the indicator lifecycle status and indicator approval status; Based on the current assessment objectives and input data conditions, construct a set of indicators to participate in this round of assessment calculations.

4. The method according to claim 3, characterized in that, The indicator set is filtered based on indicator activation status, approval status, application level, computability, evaluation task type, time granularity adaptability, and data source availability; among them, the indicator set used for equipment-level evaluation calculation is the equipment-level indicator set, the indicator set used for site-level evaluation calculation is the site-level indicator set, and the indicator set used for multi-time granularity evaluation calculation is the indicator set adapted to the current statistical period.

5. The method according to claim 1, characterized in that, The equipment-level evaluation calculation is used to generate equipment-level evaluation data for a single device within the current statistical period. The equipment-level evaluation data includes equipment power performance data, power generation performance data, stability performance data, and quality performance data. The station-level evaluation calculation is used to generate station-level evaluation data corresponding to the overall station operation performance based on the equipment-level evaluation data. The multi-time-granularity evaluation calculation is used to generate daily, weekly, monthly reports, and time-granularity evaluation data corresponding to custom time granularities.

6. The method according to claim 1, characterized in that, The horizontal comparative analysis includes comparisons of equipment of the same model, equipment of the same site, and sites in the same region. The horizontal comparative analysis data is used to characterize the degree of operational deviation of the target object relative to similar objects. The vertical trend analysis includes health change trend analysis, continuous degradation analysis, and abnormal persistence analysis. The vertical trend analysis data is used to characterize the status changes of the target object in different statistical periods.

7. The method according to claim 1, characterized in that, The comprehensive health score of the equipment is calculated based on multiple dimensions of data, including power performance, operational stability, data quality, alarm status, and trend changes. The comprehensive health score of the site is obtained by summarizing, weighting, or hierarchically summarizing the comprehensive health scores of the equipment.

8. The method according to claim 1, characterized in that, The dynamic risk levels include normal, watch, warning, and high risk; the potential anomaly identification data includes at least one of single-indicator anomaly identification data, continuous-period anomaly identification data, multi-indicator superimposed anomaly identification data, and horizontal and vertical combined anomaly identification data; wherein, the single-indicator anomaly identification data is used to characterize whether a single indicator hits an anomaly threshold, the continuous-period anomaly identification data is used to characterize whether the same indicator meets anomaly conditions for multiple consecutive periods, the multi-indicator superimposed anomaly identification data is used to characterize whether multiple different indicators simultaneously meet anomaly conditions in the same period, and the horizontal and vertical combined anomaly identification data is used to characterize whether horizontal comparison anomalies and vertical trend anomalies occur simultaneously.

9. The method according to claim 1, characterized in that, Based on the comprehensive health score of the equipment, the comprehensive health score of the site, and the dynamic risk level, report information, analysis information, and operational action information are generated, including: Generate daily, weekly, and monthly reports at the site level as the aforementioned report information; Generate device-level analysis results and custom analysis results as the analysis information; The system generates maintenance suggestions, proactive early warning information, and push notification information as operational action information. Specifically, when the overall health score of equipment or site is lower than the preset level threshold, or the dynamic risk level reaches the warning level or above, maintenance suggestions are generated based on the sub-item evaluation data, the horizontal comparison analysis data, the vertical trend analysis data, the potential anomaly identification data, and the historical memory file, and proactive warning information is generated.

10. A data operation and maintenance system for photovoltaic power plants, characterized in that, include: The data and memory file layer is used to obtain the photovoltaic power plant operation data of the current evaluation batch and read historical memory files to construct the evaluation input set; The business logic layer is used to load a unified indicator system, obtain the set of indicators participating in this round of evaluation calculation, perform equipment-level evaluation calculation, site-level evaluation calculation, and multi-time-granularity evaluation calculation on the evaluation input set based on the indicator set, obtain sub-item evaluation data, perform horizontal comparison analysis and vertical trend analysis based on the sub-item evaluation data, obtain horizontal comparison analysis data and vertical trend analysis data, generate potential anomaly identification data based on the horizontal comparison analysis data and the vertical trend analysis data and combined with preset anomaly judgment conditions, calculate the comprehensive equipment health score and the comprehensive site health score based on the sub-item evaluation data, the horizontal comparison analysis data, and the vertical trend analysis data, and determine the dynamic risk level by combining the comprehensive equipment health score, the comprehensive site health score, the vertical trend analysis data, and the potential anomaly identification data. The business application layer is used to generate report information, analysis information, and operational action information based on the comprehensive health score of the equipment, the comprehensive health score of the site, and the dynamic risk level, and to write back the sub-evaluation data, the horizontal comparison analysis data, the vertical trend analysis data, the potential anomaly identification data, the comprehensive health score of the equipment, the comprehensive health score of the site, the dynamic risk level, the report information, the analysis information, and the operational action information generated in this round to the historical memory file; The business logic layer includes a basic indicator system, an evaluation calculation module, and an auxiliary indicator calling module, while the business application layer includes a report output module, an asset operation action module, and a memory write-back module.

Images