Photovoltaic power station operation and maintenance management level improving method based on micrometeorological monitoring

Abstract

The invention discloses a method for improving the operation and maintenance management level of a photovoltaic power station based on micro-meteorological monitoring. The method comprises the following steps: S1, arranging a high-precision micro-meteorological monitoring network and acquiring data, wherein the high-precision micro-meteorological monitoring network is used for acquiring accurate and reliable real micro-meteorological data synchronous with the operation state of the photovoltaic power station; s2, data fusion, cleaning and benchmark establishment: fusing multi-source data into high-quality data assets which can be used for analysis, and establishing a performance evaluation benchmark; s3, intelligent operation and maintenance application based on the micrometeorological data, wherein the intelligent operation and maintenance application is used for converting the processed data into a specific operation and maintenance action instruction, and active early warning and precise operation and maintenance are achieved; s4, continuous evaluation and closed-loop optimization: forming a'monitoring-analysis-decision-execution-evaluation 'closed loop, and continuously managing the level; the method has the advantages of accurate sensing, accurate quantification, active early warning and data driving, and provides data support for core operation and maintenance targets of maximizing power generation benefits, minimizing operation and maintenance cost and guaranteeing asset safety.

Description

Technical Field

[0001] This invention belongs to the field of photovoltaic power plant operation and maintenance management technology, specifically involving a method for improving the operation and maintenance management level of photovoltaic power plants based on micro-meteorological monitoring. Background Technology

[0002] The level of operation and maintenance management of photovoltaic power plants is related to their own efficient operation and their ability to play a highly economical auxiliary role in regional power dispatching.

[0003] The existing photovoltaic power plants still have the following problems in operation and maintenance management: 1. Using irradiance data estimated by remote weather stations or satellites to calculate theoretical power generation results in a large difference from actual power generation, leading to distortion of the core performance ratio (PR), which cannot truly reflect the health of the power station. Furthermore, treating the entire photovoltaic power station as a homogeneous whole makes it impossible to perceive the subtle differences in irradiance, temperature, and wind speed within the station caused by terrain, local shading, and different array spacing, which is not conducive to refined operation and maintenance management. 2. It is impossible to quantify the power generation loss caused by dust. Cleaning decisions rely on fixed cycles or visual observation. Either excessive cleaning wastes water resources, or insufficient cleaning results in a loss of power generation revenue. Alarms only indicate abnormal branch current, but the cause may be instantaneous fluctuations in irradiance, cloud cover, or a real fault, making it difficult to quickly determine and resulting in extremely low troubleshooting efficiency. 3. Relying on macro weather forecasts, it is not sensitive to local sudden cloud clusters, fog and other micro-meteorological phenomena, resulting in inaccurate short-term / ultra-short-term power forecasts and frequent fines from the power grid. At the same time, equipment such as inverters and transformer substations may be derated or damaged due to overheating in high temperature and low wind speed environments, and traditional operation and maintenance cannot detect this risk in advance. 4. Cleaning, inspection, and equipment checks are all carried out according to a fixed plan, which lacks specificity and results in low resource utilization efficiency. It is difficult to accurately locate and analyze the impact of newly added obstructions (such as new trees or new buildings) on power generation during the operation period. To address the aforementioned issues, it is essential to develop a method for improving the operation and maintenance management of photovoltaic power plants based on micrometeorological monitoring. Summary of the Invention

[0004] The purpose of this invention is to overcome the shortcomings of the prior art and provide a method for improving the operation and maintenance management of photovoltaic power plants based on micro-meteorological monitoring, which is characterized by precise perception, accurate quantification, proactive early warning, and data-driven approach. This method can not only ensure the efficient operation of the photovoltaic power plant itself, but also effectively contribute to the development of the regional power grid.

[0005] The objective of this invention is achieved as follows: a method for improving the operation and maintenance management level of photovoltaic power plants based on micrometeorological monitoring, comprising the following steps: S1, High-precision micro-meteorological monitoring network deployment and data acquisition: used to obtain accurate, reliable, and real micro-meteorological data that is synchronized with the operating status of photovoltaic power plants; S2, Data Fusion, Cleaning and Benchmarking: Used to fuse multi-source data into high-quality, analytically usable data assets and to establish performance evaluation benchmarks; S3, an intelligent operation and maintenance application based on micro-meteorological data: used to transform processed data into specific operation and maintenance action instructions to achieve proactive early warning and precise operation and maintenance; S4, Continuous Evaluation and Closed-Loop Optimization: Used to form a complete closed loop of "monitoring-analysis-decision-execution-evaluation" to continuously improve the level of operation and maintenance management.

[0006] Preferably, step S1 includes the following steps: S11, Precise Site Modeling: Three-dimensional digital modeling of photovoltaic power plants to identify key areas for the deployment of micro-meteorological monitoring equipment; S12, Scientific Selection and Layout of Equipment: S121. A main meteorological station is established upwind of the photovoltaic power station. The monitoring equipment includes a total radiation meter for measuring total solar irradiance, an anemometer for measuring horizontal wind speed and wind direction, a temperature and humidity sensor for measuring ambient temperature and humidity, and an atmospheric pressure sensor. S122, inside a photovoltaic power station, multiple low-cost, highly consistent irradiance meters and backplane temperature sensors are deployed according to different inverter units or array blocks to collect irradiance and temperature differences inside the photovoltaic power station in a distributed manner. S123, a plane irradiance meter is installed in a key location to directly measure the effective irradiance incident on the surface of the tilted photovoltaic module; S13, Data Synchronization and Transmission: Ensure that the data acquisition frequency of all micro-meteorological monitoring equipment is synchronized with the acquisition frequency of the power station SCADA, and establish a stable and reliable data transmission link to aggregate all micro-meteorological data into a unified database or big data platform in real time.

[0007] Preferably, in step S11, the key areas include edge areas, areas prone to dust accumulation, inverter unit boundary points, and terrain undulation points.

[0008] Preferably, step S2 includes the following steps: S21, Data Fusion and Alignment: On the data platform, micro-meteorological data and power plant operation data are accurately aligned and correlated in terms of timestamps; S22, Data Quality Control: Establish data cleaning rules, automatically identify and remove outliers, and use data from distributed sensors to cross-verify data to improve data reliability; S23, Establish a theoretical power generation model: Based on real-time monitored effective irradiance, ambient temperature, and wind speed data, and combined with the technical parameters of photovoltaic modules, including the power temperature coefficient and nominal efficiency, establish a theoretical power generation model for each inverter unit or branch. The formula is: Theoretical power = Effective irradiance × Module area × Overall efficiency × [1 + Power temperature coefficient × (Module temperature - Standard temperature)], where the overall efficiency = [80%, 90%] × Nominal efficiency, and the module temperature is estimated from the ambient temperature, irradiance, and wind speed.

[0009] S24, Calculate the key performance indicator: Performance Ratio PR = Actual Power Generation / Theoretical Power Generation, used to assess the health of photovoltaic power plants.

[0010] Preferably, in step S23, the component temperature = ambient temperature + 0.32 × [effective irradiance / (1 + 0.041 × wind speed)].

[0011] Preferably, step S3 includes the following steps: S31, accurate power generation prediction: S311, Short-term forecast (0-72 hours): Combines micro-meteorological data of photovoltaic power plants with numerical weather prediction model NWP, and corrects it through machine learning algorithm to improve the prediction accuracy of irradiance, cloud cover and temperature. Then, based on theoretical power generation model, it gives a more reliable short-term power generation forecast. S312, Ultra-short-term forecast (0-4 hours): Utilizing the real-time trend of photovoltaic power station irradiance and combining it with satellite cloud image data, the system tracks and forecasts upcoming cloud clusters. Then, based on a theoretical power generation model, it achieves accurate prediction of sharp rises and falls in ultra-short-term power generation. S32, Precise Fault Diagnosis and Early Warning: S321, Alarm based on difference analysis: Real-time comparison of PR values ​​of different branches. If the PR value of a certain branch is consistently significantly lower than that of the adjacent branches, an alarm is immediately triggered to indicate that there may be a fault. S322, Dust Loss Assessment and Cleaning Reminder: By comparing the total radiation meter and the irradiance meter on the photovoltaic module on site, the difference between the current and voltage curves is analyzed to calculate the degree of dust's impact on light transmittance. When the calculated dust loss exceeds the preset threshold, a photovoltaic module cleaning work order is automatically generated, and the best cleaning time is recommended based on the weather forecast. S323, Hot Spot and Connection Fault Warning: When using an infrared thermal imager for inspection, combined with the irradiance and ambient wind speed at the time, it is determined whether the temperature rise of the equipment is within a reasonable range, and abnormal heat points are accurately located. S33, Operation and Maintenance Strategy Optimization: S331, Component Cleaning Optimization: Based on dust loss data and weather forecasts, develop the most economical cleaning plan; S332, Equipment Thermal Stress Management: When the ambient temperature is high and the wind speed is low, the system can issue an early warning of "inverter derating risk" and remind maintenance personnel to check the inverter's ventilation and heat dissipation in a timely manner. S333, Shading Analysis: By analyzing sudden drops in power generation of certain branches during specific time periods each day, combined with the sun's position trajectory, new obstructions can be accurately located and addressed.

[0012] Preferably, step S311 includes the following steps: S3111, Data preparation: Obtain high-resolution NWP data from meteorological service providers, including irradiance, cloud cover, ambient temperature, relative humidity, wind speed, wind direction and aerosol concentration, as well as historical data of photovoltaic power plants, including historical power data and historical micrometeorological data; S3112, NWP data downscaling and interpolation: Using digital elevation models and local geographic features, NWP data is downscaled to the specific location of the photovoltaic power station to obtain the future meteorological data sequence of the power station. S3113, Physical Model Conversion: The downscaled future meteorological data is input into the theoretical power generation model to obtain a preliminary power prediction curve based on pure physics. S3114, Machine Learning Correction: Training dataset: Paired data of historical NWP predictions and actual power generation of photovoltaic power plants over several months; Model selection: Random forest, gradient boosting decision tree, or neural network; Model training: Let the machine learn a correction function, and the corrected power = ML_Model(NWP predicted irradiance, NWP predicted temperature, NWP predicted cloud cover, NWP predicted wind speed, NWP predicted relative humidity, month, hour, day type, historical actual power_lag 1 day, historical actual power_lag 2 days, theoretical maximum power of the power plant). Online forecasting: New NWP data is input into a trained machine learning model, which automatically outputs a predicted power generation value. S3115, Output and Reporting: The system automatically generates power generation forecast curves for the next 72 hours at 15-minute or 1-hour intervals and reports them to the power grid dispatch center through a standard interface.

[0013] Preferably, step S312 includes the following steps: S3121, Data foundation preparation: real-time data stream from the power plant and real-time satellite cloud data; S3122, Cloud cluster motion vector calculation: By analyzing several consecutive satellite cloud images, the speed and direction of cloud cluster movement are calculated using cross-correlation algorithm or optical flow method, resulting in a cloud cluster motion vector map; S3123, Cloud Extrapolation Prediction: Based on the calculated motion vector, the mechanical energy of the latest satellite cloud image is linearly extrapolated to obtain the predicted cloud image for the next 1-4 hours; S3124, Cloud-Irradiance-Power Conversion: Cloud obstruction model: Based on predicted cloud formations, it determines whether the photovoltaic power station will be covered by clear skies, thin clouds, or thick clouds at each point in the future; Irradiance prediction: If it is clear sky, the theoretical maximum irradiance at that moment is calculated using the clear sky model; if it is blocked by clouds, the clear sky irradiance is attenuated according to the optical thickness of the clouds, and the predicted irradiance = clear sky irradiance × transmittance. Power prediction: Input the predicted irradiance sequence into the theoretical power generation model and output the predicted power generation value for the next 4 hours; S3125, Real-time Rolling Update and Fusion: The system runs once every minute or every 5 minutes, rolling up the prediction results and fusing them with short-term prediction results to form a seamless and complete prediction curve.

[0014] Preferably, in step S322, the preset threshold is 3%.

[0015] Preferably, step S4 includes the following steps: S41, Quantitative Evaluation of Effectiveness: Regularly evaluate the application effectiveness, including mean time to repair faults, power generation loss due to dust, and power prediction deviation rate; S42, Model Iterative Optimization: Based on long-term accumulated data, continuously revise the theoretical power generation model and various early warning thresholds to improve the accuracy of operation and maintenance management. S43, Knowledge Base Construction: Link typical fault modes, handling solutions, and corresponding micro-meteorological characteristics to build a dedicated intelligent operation and maintenance knowledge base for photovoltaic power plants, which can be used to assist in the training of new employees and the rapid and accurate judgment of future faults.

[0016] Due to the adoption of the above technical solution, the beneficial effects of the present invention are: 1. This invention uses a high-precision radiation meter installed in the station to directly measure the incident radiation, providing a unique and accurate benchmark for calculating theoretical power generation, making the PR value a true standard for measuring the status of the power station. At the same time, through a distributed sensor network, it captures the micro-meteorological conditions of different areas in real time, identifies advantageous and weak power generation areas, and provides map support for refined operation and maintenance management and fault diagnosis. 2. This invention adopts quantitative dust blockage loss. By comparing the irradiance / current output of a clean radiation meter with that of the component surface, it accurately calculates the percentage decrease in light transmittance caused by dust and the corresponding power generation loss, so as to achieve on-demand cleaning. At the same time, based on the comparison under the same operating conditions, under the same irradiance and temperature conditions, the output power of different branches is compared in real time. The one that is significantly lower is the real fault point, which improves the fault location accuracy. 3. This invention uses real-time micro-meteorological data to correct NWP data and combines it with cloud map tracking to significantly reduce prediction deviation, reduce assessment costs, and improve the economic benefits of power plants. At the same time, it monitors ambient temperature, wind speed and equipment temperature in real time. When unfavorable micro-meteorological conditions such as high temperature and calm wind are detected, it issues equipment derating or heat dissipation alarms in advance to remind personnel to intervene and prevent power generation loss and equipment damage. 4. This invention adopts dynamic optimization of operation and maintenance strategies to improve operation and maintenance efficiency. At the same time, by analyzing the sudden drop in power generation of specific branches at fixed time periods each day (corresponding to the solar azimuth angle), and combining it with the solar trajectory model, the location and duration of the obstruction can be accurately deduced, providing data support for processing decisions. In summary, this invention has the advantages of accurate sensing, precise quantification, proactive early warning, and data-driven approach. It provides data support for the core operation and maintenance objectives of maximizing power generation efficiency, minimizing operation and maintenance costs, and ensuring asset security. While ensuring the efficient operation of the photovoltaic power station itself, it also effectively contributes to the development of the regional power grid. Attached Figure Description

[0017] Figure 1 This is a flowchart of the method steps of the present invention. Detailed Implementation

[0018] The technical solution of the present invention will be further described in detail below through embodiments and in conjunction with the accompanying drawings.

[0019] like Figure 1 As shown, the present invention provides a method for improving the operation and maintenance management level of photovoltaic power plants based on micrometeorological monitoring, including the following steps: S1, High-precision micro-meteorological monitoring network deployment and data acquisition: used to obtain accurate, reliable, and real-time micro-meteorological data that is synchronized with the operating status of photovoltaic power plants.

[0020] S11, Precise Site Modeling: Perform 3D digital modeling of photovoltaic power plants to identify key areas (such as edge areas, areas prone to dust accumulation, inverter unit boundary points, and terrain undulation points, etc.). These key points are the preferred locations for deploying micro-meteorological monitoring equipment.

[0021] S12, Scientific Selection and Layout of Equipment: S121. Establish a main meteorological station upwind of the photovoltaic power station (the direction of the prevailing wind). The monitoring equipment includes a total radiation meter for measuring total solar irradiance, an anemometer for measuring horizontal wind speed and wind direction, a temperature and humidity sensor for measuring ambient temperature and humidity, and an atmospheric pressure sensor. S122, inside a photovoltaic power station, multiple low-cost, highly consistent irradiance meters and backplane temperature sensors are deployed according to different inverter units or array blocks to collect irradiance and temperature differences inside the photovoltaic power station in a distributed manner. S123, a plane irradiance meter is installed in a key location to directly measure the effective irradiance incident on the surface of the tilted photovoltaic module.

[0022] S13, Data Synchronization and Transmission: Ensure that the data acquisition frequency of all micro-meteorological monitoring equipment (e.g., 1 minute / time) is synchronized with the acquisition frequency of the power station SCADA (i.e., data acquisition and monitoring system), and establish a stable and reliable data transmission link to aggregate all micro-meteorological data into a unified database or big data platform in real time.

[0023] S2, Data Fusion, Cleaning and Benchmarking: Used to fuse multi-source data into high-quality, analytically usable data assets and to establish benchmarks for performance evaluation.

[0024] S21, Data Fusion and Alignment: On the data platform, micro-meteorological data (irradiance, temperature, wind speed, etc.) and power plant operation data (current, voltage, power, inverter temperature, etc. of each branch) are accurately aligned and correlated on the timestamp.

[0025] S22, Data Quality Control: Establish data cleaning rules to automatically identify and remove outliers (such as non-zero irradiance at night), and at the same time use data from distributed sensors to cross-verify and improve data reliability.

[0026] S23, Establish a theoretical power generation model: Based on real-time monitored effective irradiance, ambient temperature, and wind speed data, combined with the technical parameters of photovoltaic modules (power temperature coefficient, nominal efficiency, etc.), establish a theoretical power generation model for each inverter unit or branch.

[0027] The theoretical power generation model formula is: Theoretical power = Effective irradiance × Module area × Overall efficiency × [1 + Power temperature coefficient × (Module temperature - Standard temperature)]; where, the overall efficiency = [80%, 90%] × Nominal efficiency, that is, the overall efficiency is reduced to 80 to 90% of the nominal efficiency after considering other losses; where, the module temperature is estimated by ambient temperature, irradiance and wind speed, and the calculation formula is: Module temperature = Ambient temperature + 0.32 × [Effective irradiance / (1 + 0.041 × Wind speed)].

[0028] S24, calculate the key performance indicator: Performance Ratio PR = Actual Power Generation / Theoretical Power Generation, used to assess the health of photovoltaic power plants and can truly reflect the operating status of photovoltaic power plants.

[0029] S3, an intelligent operation and maintenance application based on micro-meteorological data: It is used to transform processed data into specific operation and maintenance action instructions to achieve proactive early warning and precise operation and maintenance.

[0030] S31, accurate power generation prediction: S311, Short-Term Forecast (0-72 hours): This method combines micro-meteorological data from photovoltaic power plants with the Numerical Weather Prediction (NWP) model, and uses machine learning algorithms for correction to significantly improve the prediction accuracy of irradiance, cloud cover, and temperature. Then, based on a theoretical power generation model, it provides a more reliable short-term power generation forecast, reducing grid assessment costs. The specific process is as follows: S3111, Data preparation: Obtain high-resolution NWP data from meteorological service providers (such as the China Meteorological Administration), including irradiance, cloud cover, ambient temperature, relative humidity, wind speed, wind direction and aerosol concentration, as well as historical data of photovoltaic power plants, including historical power data and historical micrometeorological data; S3112, NWP data downscaling and interpolation: Using digital elevation models and local geographic features, NWP data is downscaled to the specific location of the photovoltaic power station to obtain the future meteorological data sequence of the power station. S3113, Physical Model Conversion: The downscaled future meteorological data is input into the theoretical power generation model to obtain a preliminary power prediction curve based on pure physics. S3114, Machine Learning Correction: Training dataset: Paired data of historical NWP predictions and actual power generation of photovoltaic power plants over several months; Model selection: Use machine learning algorithms such as random forest, gradient boosting decision tree, or neural network; Model training: Let the machine learn a correction function, and the corrected power = ML_Model(NWP predicted irradiance, NWP predicted temperature, NWP predicted cloud cover, NWP predicted wind speed, NWP predicted relative humidity, month, hour, day type, historical actual power_lag 1 day, historical actual power_lag 2 days, theoretical maximum power of the power plant). Online forecasting: New NWP data is input into a trained machine learning model, which automatically outputs a predicted power generation value. S3115, Output and Reporting: The system automatically generates power generation forecast curves for the next 72 hours at 15-minute or 1-hour intervals and reports them to the power grid dispatch center through a standard interface.

[0031] S312, Ultra-Short-Term Forecasting (0-4 hours): Utilizing the real-time trend of photovoltaic power plant irradiance changes, combined with satellite cloud image data, it tracks and predicts upcoming cloud formations. Similarly, based on theoretical power generation models, it achieves accurate prediction of sharp rises and falls in ultra-short-term power generation, assisting the power grid in real-time dispatch. The specific process is as follows: S3121, Data foundation preparation: real-time data stream from the power plant and real-time satellite cloud data; S3122, Cloud cluster motion vector calculation: By analyzing several consecutive satellite cloud images, the speed and direction of cloud cluster movement are calculated using cross-correlation algorithm or optical flow method, resulting in a cloud cluster motion vector map; S3123, Cloud Extrapolation Prediction: Based on the calculated motion vector, the mechanical energy of the latest satellite cloud image is linearly extrapolated to obtain the predicted cloud image for the next 1-4 hours; S3124, Cloud-Irradiance-Power Conversion: Cloud obstruction model: Based on predicted cloud formations, it determines whether the photovoltaic power station will be covered by clear skies, thin clouds, or thick clouds at each point in the future; Irradiance prediction: If it is clear sky, the theoretical maximum irradiance at that moment is calculated using the clear sky model; if it is blocked by clouds, the clear sky irradiance is attenuated according to the optical thickness of the clouds, and the predicted irradiance = clear sky irradiance × transmittance. Power prediction: Input the predicted irradiance sequence into the theoretical power generation model and output the predicted power generation value for the next 4 hours; S3125, Real-time Rolling Update and Fusion: The system runs once every minute or every 5 minutes, rolling up the prediction results and fusing them with short-term prediction results. For example, the ultra-short-term prediction is responsible for the fine fluctuations of 0-2 hours, while the 2-4 hour prediction is gradually weighted and transitioned to the short-term prediction results, forming a seamless and complete prediction curve.

[0032] S32, Precise Fault Diagnosis and Early Warning: S321, alarm based on difference analysis: compare the PR values ​​of different branches in real time. If the PR value of a certain branch is significantly lower than that of the adjacent branches, an alarm will be triggered immediately, indicating that there may be faults such as string failure, obstruction, or fuse failure. S322, Dust Loss Assessment and Cleaning Reminder: By comparing the total radiation meter (measured on clean glass, the baseline for calculating dust loss) and the irradiance meter on the photovoltaic module on site, the difference in current and voltage curves is analyzed to calculate the degree of dust's impact on light transmittance, i.e., dust loss. When the calculated dust loss exceeds a preset threshold (usually 3%), a photovoltaic module cleaning work order is automatically generated, and the best cleaning time is recommended based on the weather forecast (e.g., the weather is predicted to be sunny in the next few days, which is suitable for cleaning). S323, Hot Spot and Connection Fault Warning: When using an infrared thermal imager (which can be mounted on a drone) for inspection, combined with the irradiance and ambient wind speed at the time, it scientifically judges whether the temperature rise of the equipment is within a reasonable range, avoids misjudgment, and accurately locates abnormal heat points.

[0033] S33, Operation and Maintenance Strategy Optimization: S331, Component Cleaning Optimization: Based on dust loss data and weather forecasts, develop the most economical cleaning plan, such as not cleaning before rain and cleaning promptly after a sandstorm; S332, Equipment Thermal Stress Management: When the ambient temperature is high and the wind speed is low, the system can issue an early warning of "inverter derating risk" and remind maintenance personnel to check the inverter's ventilation and heat dissipation in a timely manner. S333, Shading Analysis: By analyzing sudden drops in power generation of certain branches during specific time periods each day, combined with the sun's position trajectory, it accurately locates and processes newly added obstructions, such as newly grown trees and new buildings.

[0034] S4, Continuous Evaluation and Closed-Loop Optimization: Used to form a complete closed loop of "monitoring-analysis-decision-execution-evaluation" to continuously improve the level of operation and maintenance management.

[0035] S41, Quantitative evaluation of effectiveness: Regularly (monthly / quarterly) evaluate the application effect of the method provided by this invention, such as how much the mean time to repair faults is reduced, how much the power generation loss caused by dust is reduced, and how much the deviation rate of power prediction is reduced.

[0036] S42, Model Iterative Optimization: Based on long-term accumulated data, continuously revise the theoretical power generation model and various early warning thresholds to improve the accuracy of operation and maintenance management.

[0037] S43, Knowledge Base Construction: Link typical fault modes, handling solutions, and corresponding micro-meteorological characteristics to build a dedicated intelligent operation and maintenance knowledge base for photovoltaic power plants, which can be used to assist in the training of new employees and the rapid and accurate judgment of future faults.

[0038] Finally, it should be noted that the above embodiments are only used to illustrate the technical solutions of the present invention, and not to limit it. Although the present invention has been described in detail with reference to the above embodiments, those skilled in the art should understand that modifications or equivalent substitutions can still be made to the specific implementation of the present invention. Any modifications or equivalent substitutions that do not depart from the spirit and scope of the present invention should be covered within the scope of the claims of the present invention.

Claims

1. A method for improving the operation and maintenance management level of photovoltaic power plants based on micrometeorological monitoring, characterized in that... It includes the following steps: S1, High-precision micro-meteorological monitoring network deployment and data acquisition: used to obtain accurate, reliable, and real micro-meteorological data that is synchronized with the operating status of photovoltaic power plants; S2, Data Fusion, Cleaning and Benchmarking: Used to fuse multi-source data into high-quality, analytically usable data assets and to establish performance evaluation benchmarks; S3, an intelligent operation and maintenance application based on micro-meteorological data: used to transform processed data into specific operation and maintenance action instructions to achieve proactive early warning and precise operation and maintenance; S4, Continuous Evaluation and Closed-Loop Optimization: Used to form a complete closed loop of "monitoring-analysis-decision-execution-evaluation" to continuously improve the level of operation and maintenance management.

2. The method for improving the operation and maintenance management level of photovoltaic power plants based on micrometeorological monitoring according to claim 1, characterized in that, Step S1 includes the following steps: S11, Precise Site Modeling: Three-dimensional digital modeling of photovoltaic power plants to identify key areas for the deployment of micro-meteorological monitoring equipment; S12, Scientific Selection and Layout of Equipment: S121. A main meteorological station is established upwind of the photovoltaic power station. The monitoring equipment includes a total radiation meter for measuring total solar irradiance, an anemometer for measuring horizontal wind speed and wind direction, a temperature and humidity sensor for measuring ambient temperature and humidity, and an atmospheric pressure sensor. S122, inside a photovoltaic power station, multiple low-cost, highly consistent irradiance meters and backplane temperature sensors are deployed according to different inverter units or array blocks to collect irradiance and temperature differences inside the photovoltaic power station in a distributed manner. S123, a plane irradiance meter is installed in a key location to directly measure the effective irradiance incident on the surface of the tilted photovoltaic module; S13, Data Synchronization and Transmission: Ensure that the data acquisition frequency of all micro-meteorological monitoring equipment is synchronized with the acquisition frequency of the power station SCADA, and establish a stable and reliable data transmission link to aggregate all micro-meteorological data into a unified database or big data platform in real time.

3. The method for improving the operation and maintenance management level of photovoltaic power plants based on micrometeorological monitoring according to claim 2, characterized in that: In step S11, the key areas include edge areas, areas prone to dust accumulation, inverter unit boundary points, and terrain undulation points.

4. The method for improving the operation and maintenance management level of photovoltaic power plants based on micrometeorological monitoring according to claim 1, characterized in that, Step S2 includes the following steps: S21, Data Fusion and Alignment: On the data platform, micro-meteorological data and power plant operation data are accurately aligned and correlated in terms of timestamps; S22, Data Quality Control: Establish data cleaning rules, automatically identify and remove outliers, and use data from distributed sensors to cross-verify data to improve data reliability; S23, Establish a theoretical power generation model: Based on real-time monitored effective irradiance, ambient temperature, and wind speed data, and combined with the technical parameters of photovoltaic modules, including the power temperature coefficient and nominal efficiency, establish a theoretical power generation model for each inverter unit or branch. The formula is: Theoretical power = Effective irradiance × Module area × Overall efficiency × [1 + Power temperature coefficient × (Module temperature - Standard temperature)], where the overall efficiency = [80%, 90%] × Nominal efficiency, and the module temperature is estimated from the ambient temperature, irradiance, and wind speed. S24, Calculate the key performance indicator: Performance Ratio PR = Actual Power Generation / Theoretical Power Generation, used to assess the health of photovoltaic power plants.

5. The method for improving the operation and maintenance management level of photovoltaic power plants based on micrometeorological monitoring according to claim 4, characterized in that: In step S23, the component temperature = ambient temperature + 0.32 × [effective irradiance / (1 + 0.041 × wind speed)].

6. The method for improving the operation and maintenance management level of photovoltaic power plants based on micrometeorological monitoring according to claim 1, characterized in that, Step S3 includes the following steps: S31, accurate power generation prediction: S311, Short-term forecast (0-72 hours): Combines micro-meteorological data of photovoltaic power plants with numerical weather prediction model NWP, and corrects it through machine learning algorithm to improve the prediction accuracy of irradiance, cloud cover and temperature. Then, based on theoretical power generation model, it gives a more reliable short-term power generation forecast. S312, Ultra-short-term forecast (0-4 hours): Utilizing the real-time trend of photovoltaic power station irradiance and combining it with satellite cloud image data, the system tracks and forecasts upcoming cloud clusters. Then, based on a theoretical power generation model, it achieves accurate prediction of sharp rises and falls in ultra-short-term power generation. S32, Precise Fault Diagnosis and Early Warning: S321, Alarm based on difference analysis: Real-time comparison of PR values ​​of different branches. If the PR value of a certain branch is consistently significantly lower than that of the adjacent branches, an alarm is immediately triggered to indicate that there may be a fault. S322, Dust Loss Assessment and Cleaning Reminder: By comparing the total radiation meter and the irradiance meter on the photovoltaic module on site, the difference between the current and voltage curves is analyzed to calculate the degree of dust's impact on light transmittance. When the calculated dust loss exceeds the preset threshold, a photovoltaic module cleaning work order is automatically generated, and the best cleaning time is recommended based on the weather forecast. S323, Hot Spot and Connection Fault Warning: When using an infrared thermal imager for inspection, combined with the irradiance and ambient wind speed at the time, it is determined whether the temperature rise of the equipment is within a reasonable range, and abnormal heat points are accurately located. S33, Operation and Maintenance Strategy Optimization: S331, Component Cleaning Optimization: Based on dust loss data and weather forecasts, develop the most economical cleaning plan; S332, Equipment Thermal Stress Management: When the ambient temperature is high and the wind speed is low, the system can issue an early warning of "inverter derating risk" and remind maintenance personnel to check the inverter's ventilation and heat dissipation in a timely manner. S333, Shading Analysis: By analyzing sudden drops in power generation of certain branches during specific time periods each day, combined with the sun's position trajectory, new obstructions can be accurately located and addressed.

7. The method for improving the operation and maintenance management level of photovoltaic power plants based on micrometeorological monitoring according to claim 6, characterized in that, Step S311 includes the following steps: S3111, Data preparation: Obtain high-resolution NWP data from meteorological service providers, including irradiance, cloud cover, ambient temperature, relative humidity, wind speed, wind direction and aerosol concentration, as well as historical data of photovoltaic power plants, including historical power data and historical micrometeorological data; S3112, NWP data downscaling and interpolation: Using digital elevation models and local geographic features, NWP data is downscaled to the specific location of the photovoltaic power station to obtain the future meteorological data sequence of the power station. S3113, Physical Model Conversion: The downscaled future meteorological data is input into the theoretical power generation model to obtain a preliminary power prediction curve based on pure physics. S3114, Machine Learning Correction: Training dataset: Paired data of historical NWP predictions and actual power generation of photovoltaic power plants over several months; Model selection: Random forest, gradient boosting decision tree, or neural network; Model training: Let the machine learn a correction function, and the corrected power = ML_Model(NWP predicted irradiance, NWP predicted temperature, NWP predicted cloud cover, NWP predicted wind speed, NWP predicted relative humidity, month, hour, day type, historical actual power_lag 1 day, historical actual power_lag 2 days, theoretical maximum power of the power plant). Online forecasting: New NWP data is input into a trained machine learning model, which automatically outputs a predicted power generation value. S3115, Output and Reporting: The system automatically generates power generation forecast curves for the next 72 hours at 15-minute or 1-hour intervals and reports them to the power grid dispatch center through a standard interface.

8. The method for improving the operation and maintenance management level of photovoltaic power plants based on micrometeorological monitoring according to claim 6, characterized in that, Step S312 includes the following steps: S3121, Data foundation preparation: real-time data stream from the power plant and real-time satellite cloud data; S3122, Cloud cluster motion vector calculation: By analyzing several consecutive satellite cloud images, the speed and direction of cloud cluster movement are calculated using cross-correlation algorithm or optical flow method, resulting in a cloud cluster motion vector map; S3123, Cloud Extrapolation Prediction: Based on the calculated motion vector, the mechanical energy of the latest satellite cloud image is linearly extrapolated to obtain the predicted cloud image for the next 1-4 hours; S3124, Cloud-Irradiance-Power Conversion: Cloud obstruction model: Based on predicted cloud formations, it determines whether the photovoltaic power station will be covered by clear skies, thin clouds, or thick clouds at each point in the future; Irradiance prediction: If it is clear sky, use the clear sky model to calculate the theoretical maximum irradiance at that moment; If the irradiance is blocked by clouds, the clear sky irradiance is attenuated based on the optical thickness of the clouds, and the predicted irradiance is calculated as: clear sky irradiance × transmittance. Power prediction: Input the predicted irradiance sequence into the theoretical power generation model and output the predicted power generation value for the next 4 hours; S3125, Real-time Rolling Update and Fusion: The system runs once every minute or every 5 minutes, rolling up the prediction results and fusing them with short-term prediction results to form a seamless and complete prediction curve.

9. The method for improving the operation and maintenance management level of photovoltaic power plants based on micrometeorological monitoring according to claim 6, characterized in that, In step S322, the preset threshold is 3%.

10. The method for improving the operation and maintenance management level of photovoltaic power plants based on micrometeorological monitoring according to claim 1, characterized in that, Step S4 includes the following steps: S41, Quantitative Evaluation of Effectiveness: Regularly evaluate the application effectiveness, including mean time to repair faults, power generation loss due to dust, and power prediction deviation rate; S42, Model Iterative Optimization: Based on long-term accumulated data, continuously revise the theoretical power generation model and various early warning thresholds to improve the accuracy of operation and maintenance management. S43, Knowledge Base Construction: Link typical fault modes, handling solutions, and corresponding micro-meteorological characteristics to build a dedicated intelligent operation and maintenance knowledge base for photovoltaic power plants, which can be used to assist in the training of new employees and the rapid and accurate judgment of future faults.

Images