Intelligent photovoltaic operation and maintenance method and system based on power generation difference
Patent Information
- Authority / Receiving Office
- CN · China
- Patent Type
- Applications(China)
- Current Assignee / Owner
- SICHUAN TAILONG CONSTR GRP CO LTD
- Filing Date
- 2026-05-19
- Publication Date
- 2026-08-07
AI Technical Summary
近年来，物联网和机器学习技术被引入光伏运维管理，例如使用传感器数据训练异常检测模型，对单一数据源分析，导致光伏运维管理决策缺乏针对性，运维效率低
本申请的基于发电量差的智能光伏运维方法，智能分析模型基于历史数据训练能够识别常见故障模式，通过智能分析模型对光伏发电系统的实际发电量数据和预期发电量数据的发电量差数据进行精准诊断分析生成诊断数据，准确得到发电量下降的原因，大幅提升故障诊断准确率，显著减少误报和漏报，取代了传统简单阈值报警，使维护策略更具针对性，提升运维效率，降低运维成本。定期汇总用户反馈数据重新训练智能分析模型，智能分析模型具有自学习能力，智能分析模型持续优化训练，随使用时间持续优化诊断的准确率和生成维护策略的有效性，提升光伏发电系统长期适应性、稳定性和智能化水平。智能分析模型对发电量差数据诊断分析，能够在故障完全显现或造成重大损失之前，早期识别光伏发电系统潜在问题，并提前预警或安排维护，实现了从被动维护到预测性维护的转变，有效延长了光伏发电设备的使用寿命，保障了光伏发电系统的长期稳定运行。可视化交互式平台设计，用户直接观察实际发电数据、发电量差数据、诊断结果、故障原因、维护策略，并进行反馈操作，用户可随系统自动决策，也可基于诊断结果人工干预，增强了用户体验，使运维过程透明化和智能化。
Smart Images
Figure CN122529697A_ABST
Abstract
Description
Technical Field
[0001] This invention belongs to the field of photovoltaic power generation technology, and more specifically, relates to an intelligent photovoltaic operation and maintenance method and system based on power generation difference. Background Technology
[0002] In the power generation sector, photovoltaic (PV) power generation, as a crucial component of clean energy, has seen rapid industry development, making operation and maintenance (O&M) efficiency a key factor influencing power generation efficiency. O&M management directly impacts the lifespan and power output of power generation systems. Traditional O&M management methods primarily rely on periodic manual inspections or automatic alarm systems based on simple thresholds. For example, triggering alarms by monitoring a decrease in power generation is insufficient to accurately identify the cause of the decline, such as whether it's due to shading, component aging, dirt accumulation, or environmental changes. Existing technologies, such as statistical analysis or rule engines, employ diagnostics, but limited by data quality and algorithm capabilities, often result in high false alarm rates and a lack of predictability. In recent years, IoT and machine learning technologies have been introduced into PV O&M management, such as using sensor data to train anomaly detection models and analyzing single data sources. However, this approach leads to a lack of targeted decision-making in PV O&M management and low O&M efficiency. Summary of the Invention
[0003] The present invention provides an intelligent photovoltaic operation and maintenance method and system based on the difference in power generation. By accurately diagnosing and analyzing the reasons for the decline in power generation by comparing the actual power generation with the expected power generation, the method reduces false alarms and missed alarms, provides targeted maintenance strategies, and improves operation and maintenance efficiency.
[0004] To achieve the above objectives, the technical solution adopted by the present invention is as follows: This invention discloses an intelligent photovoltaic operation and maintenance method based on power generation difference, comprising the following steps: S1. Obtain the expected power generation data and actual power generation data of the photovoltaic power generation system, and calculate the power generation difference data; S2. Input the power generation difference data into the intelligent analysis model. The intelligent analysis model is trained based on historical data to identify common fault modes. The intelligent analysis model analyzes the power generation difference data to generate diagnostic data. S3. Generate a maintenance strategy based on the diagnostic data, and generate control commands based on the maintenance strategy and send them to the field equipment; S4. Visualize operation and maintenance through an interactive platform and receive user feedback; S5. Generate feedback data based on user feedback, input the feedback data into the intelligent analysis model and optimize and train it, and periodically summarize user feedback data to retrain the intelligent analysis model to achieve self-learning.
[0005] Furthermore, the actual power generation data P1 is acquired in real time through IoT sensors, and the expected power generation data P2 is calculated using the following physical model: P2 = η A I (1-β) (T1-T2) cos(θ), where η is the photovoltaic module efficiency, A is the photovoltaic module area, I is the light intensity, β is the temperature coefficient, T1 is the actual operating temperature of the photovoltaic module, T2 is the reference temperature, θ is the solar incidence angle, and the power generation difference data ΔP=P1–P2.
[0006] Furthermore, the actual power generation data is collected in real time using IoT sensors such as light sensors, temperature sensors, and current and voltage sensors, with a sampling frequency of once per minute. The collected actual power generation data is wirelessly transmitted to the data processing center of the interactive platform, where the data processing center performs data cleaning and data calibration on the collected actual power generation data.
[0007] Furthermore, the data cleaning includes removing outliers, and the data calibration includes compensating for sensor errors.
[0008] Furthermore, the intelligent analysis model employs a convolutional neural network model or a long short-term memory network deep learning model, and the intelligent analysis model integrates multi-source data recognition patterns.
[0009] Furthermore, the common failure modes include linear degradation caused by dirt accumulation in photovoltaic modules, abrupt changes caused by shading, and linear degradation and temperature rise caused by heat dissipation problems.
[0010] Furthermore, the maintenance strategy includes triggering automatic cleaning of photovoltaic modules and adjusting the inverter's maximum power point tracking parameters.
[0011] Furthermore, based on the degree of impact of the fault modes corresponding to the diagnostic data, fault modes with a higher degree of impact are prioritized for processing.
[0012] Furthermore, the interactive platform adopts a web interface, which displays a curve composed of real-time power generation difference data ΔP, diagnostic results corresponding to diagnostic data, and maintenance suggestions corresponding to maintenance strategies.
[0013] This invention provides an intelligent photovoltaic operation and maintenance system based on power generation difference, used to implement the aforementioned intelligent photovoltaic operation and maintenance method based on power generation difference, including: The data acquisition and processing module is used to acquire the expected power generation data and actual power generation data of the photovoltaic power generation system, and to calculate and generate power generation difference data based on the expected power generation data and actual power generation data; The intelligent analysis module is communicatively connected to the data acquisition and processing module. It is used to receive the power generation difference data and use the intelligent analysis model to analyze the power generation difference data to generate diagnostic data. The intelligent analysis model is trained based on historical data to identify common fault modes. The maintenance strategy generation and execution module is communicatively connected to the intelligent analysis module. It is used to generate maintenance strategies based on the diagnostic data and generate control commands based on the maintenance strategies to send to the field equipment. An interactive display and feedback module is communicatively connected to the intelligent analysis module and the maintenance strategy generation and execution module. It is used to visually display the power generation difference data, diagnostic data and maintenance strategies to the user, and to receive user feedback. The model self-learning optimization module is communicatively connected to the interactive display and feedback module and the intelligent analysis module. It is used to input the feedback data generated by user feedback into the intelligent analysis model and optimize and train it. It also achieves self-learning by periodically summarizing user feedback data and retraining the intelligent analysis model.
[0014] The beneficial effects of this invention are: This application presents an intelligent photovoltaic (PV) operation and maintenance (O&M) method based on power generation difference. The intelligent analysis model, trained on historical data, can identify common fault modes. Through precise diagnostic analysis of the difference between actual and expected power generation data of the PV system, the model generates diagnostic data, accurately identifying the causes of power generation decline. This significantly improves fault diagnosis accuracy, reduces false alarms and missed alarms, and replaces traditional simple threshold alarms, making maintenance strategies more targeted, improving O&M efficiency, and reducing O&M costs. The intelligent analysis model is retrained regularly using user feedback data. It possesses self-learning capabilities, continuously optimizing its training and improving diagnostic accuracy and the effectiveness of generated maintenance strategies over time, thus enhancing the long-term adaptability, stability, and intelligence level of the PV system. The intelligent analysis model's diagnostic analysis of power generation difference data can identify potential problems in the PV system before faults fully manifest or cause significant losses, providing early warnings or scheduling maintenance. This achieves a shift from passive to predictive maintenance, effectively extending the service life of PV equipment and ensuring the long-term stable operation of the PV system. The visual interactive platform design allows users to directly observe actual power generation data, power generation difference data, diagnostic results, fault causes, and maintenance strategies, and provide feedback. Users can make decisions automatically with the system or intervene manually based on diagnostic results, which enhances the user experience and makes the operation and maintenance process more transparent and intelligent. Attached Figure Description
[0015] To more clearly illustrate the technical solutions in the embodiments of the present invention or the prior art, the drawings used in the description of the embodiments or the prior art will be briefly introduced below. Obviously, the drawings described below are only some embodiments of the present invention. For those skilled in the art, other drawings can be obtained based on these drawings without creative effort.
[0016] Figure 1 This is a flowchart of an intelligent photovoltaic operation and maintenance method based on power generation difference provided by an embodiment of the present invention; Figure 2 This is a block diagram of an intelligent photovoltaic operation and maintenance system based on power generation difference provided in an embodiment of the present invention; Detailed Implementation
[0017] To make the objectives, technical solutions, and advantages of this invention clearer, the technical solutions of this invention will be described in detail below. Obviously, the described embodiments are merely some embodiments of this invention, and not all embodiments. Based on the embodiments of this invention, all other implementation methods obtained by those skilled in the art without creative effort are within the scope of protection of this invention.
[0018] The present invention will be further described below with reference to the accompanying drawings and embodiments.
[0019] like Figure 1 As shown, the present invention provides an intelligent photovoltaic operation and maintenance method based on power generation difference, comprising the following steps: S1. Obtain the expected power generation data and actual power generation data of the photovoltaic power generation system, and calculate the power generation difference data; The actual power generation data P1 is collected in real time by IoT sensors. For example, at the photovoltaic power station site, IoT sensors such as light sensors, temperature sensors, and current and voltage sensors are installed and deployed in the existing way. The actual power generation data is collected in real time by the light sensors, temperature sensors, and current and voltage sensors at a sampling frequency of 1 time / minute. The collected actual power generation data is wirelessly transmitted to the data processing center of the interactive platform. The data processing center performs data cleaning and data calibration on the collected actual power generation data. The expected power generation data P2 is calculated using the following physical model: P2 = η A I (1-β) (T1-T2) cos(θ), where η is the photovoltaic module efficiency, A is the photovoltaic module area in m², I is the light intensity in W / m², β is the temperature coefficient in 1 / ℃, T1 is the actual operating temperature of the photovoltaic module in ℃, T2 is the reference temperature in ℃, θ is the solar incidence angle in degrees, and the power generation difference data ΔP=P1–P2; S2. Input the power generation difference data into the intelligent analysis model. The intelligent analysis model is trained based on historical data to identify common fault modes. The intelligent analysis model analyzes the power generation difference data to generate diagnostic data. The common failure modes include linear degradation caused by dirt accumulation in photovoltaic modules, abrupt changes caused by shading, and linear degradation and temperature rise caused by heat dissipation problems. S3. Generate a maintenance strategy based on the diagnostic data, and generate control commands based on the maintenance strategy and send them to the field equipment. The maintenance strategy includes triggering automatic cleaning of photovoltaic modules and adjusting the inverter's maximum power point tracking (MPPT) parameters. The strategy library can be implemented using the Drools rule engine. The strategy library stores predefined diagnostic rules expressed in if-then form. Rule 1: If the fault mode is "dirt accumulation" and the confidence level is greater than 0.8, the generated maintenance strategy is "trigger automatic cleaning of photovoltaic modules," generating a "clean" control command. Rule 2: If the fault mode is "shading" and the confidence level is greater than 0.7, the generated maintenance strategy is "adjust the inverter's maximum power point tracking (MPPT) parameters to reduce operating voltage and decrease shading losses," generating a "adjust inverter MPPT parameters" control command. Rule 3: If the fault mode is "module aging" and the confidence level is greater than 0.9, the generated maintenance strategy is "send a manual inspection command to mark the location of aging modules," generating a "manual inspection" control command. Rule 4: If the fault mode is "inverter fault" and the confidence level is greater than 0.85, the generated maintenance strategy is "send an inverter restart command," generating a "restart inverter" control command. Rule 5: If the fault mode is "normal" or the confidence level is less than the corresponding threshold, the maintenance policy is "No maintenance required, continue monitoring," and a "normal" control command is generated. Control commands can be in JSON format. Diagnostic data is input into the rule engine. The rule engine iterates through all rules in the policy library, determining whether the diagnostic data meets the condition of a rule. If it does, the action part of that rule is executed, generating the corresponding maintenance policy. Control commands can be sent to the controller of the field device via HTTP API. After receiving the control command, the field device controller calls the corresponding device driver to execute the control action.
[0020] S4. Visualize the operation and maintenance through an interactive platform and receive user feedback. The interactive platform adopts a web interface, which displays the curve composed of real-time power generation difference data ΔP, the diagnostic results corresponding to the diagnostic data, and the maintenance suggestions corresponding to the maintenance strategy. S5. Generate feedback data based on user feedback, input the feedback data into the intelligent analysis model and optimize and train it, and periodically summarize user feedback data to retrain the intelligent analysis model to achieve self-learning.
[0021] Based on the aforementioned intelligent photovoltaic (PV) operation and maintenance (O&M) method based on power generation difference, the intelligent analysis model, trained on historical data, can identify common fault modes. Through precise diagnostic analysis of the difference between actual and expected power generation data of the PV system, the model generates diagnostic data, accurately identifying the causes of power generation decline. This significantly improves fault diagnosis accuracy, reduces false alarms and missed alarms, and replaces traditional simple threshold alarms, making maintenance strategies more targeted, improving O&M efficiency, and reducing O&M costs. Regularly summarizing user feedback data and retraining the intelligent analysis model, which possesses self-learning capabilities, allows for continuous optimization and training. Over time, the model continuously improves diagnostic accuracy and the effectiveness of generated maintenance strategies, enhancing the long-term adaptability, stability, and intelligence level of the PV system. The intelligent analysis model's diagnostic analysis of power generation difference data can identify potential problems in the PV system before faults fully manifest or cause significant losses, providing early warnings or scheduling maintenance. This achieves a shift from passive to predictive maintenance, effectively extending the service life of PV equipment and ensuring the long-term stable operation of the PV system. The visual interactive platform design allows users to directly observe actual power generation data, power generation difference data, diagnostic results, fault causes, and maintenance strategies, and provide feedback. Users can make decisions automatically with the system or intervene manually based on diagnostic results, which enhances the user experience and makes the operation and maintenance process more transparent and intelligent.
[0022] As one possible implementation, in step S1, the data cleaning includes removing outliers, and the data calibration includes compensating for sensor errors.
[0023] The collected actual power generation data undergoes data cleaning to remove outliers, and interpolation is used to process missing values, ensuring data quality and integrity and providing reliable data for subsequent intelligent analysis models. For example, data points with illuminance I less than 0 or greater than 1100 W / m² are marked as outliers and removed; the standard operating temperature range of photovoltaic modules is typically between -40℃ and 85℃, so data points with temperature T1 less than -40℃ or greater than 85℃ are marked as outliers and removed; and data points with actual power generation P1 less than 0 or greater than 1.2 times the system's rated power are marked as outliers and removed. Sensor errors are compensated by preset calibration coefficients. These calibration coefficients are obtained during installation and deployment by comparing measurements with standard instruments. The calibration coefficient range for illuminance sensors is 0.95 to 1.05, and for temperature sensors, it is 0.98 to 1.02. Based on these calibration coefficient ranges, the sensor readings are reverse-calibrated to restore a more accurate representation of power generation.
[0024] As one possible implementation, in step S2, the intelligent analysis model adopts a convolutional neural network (CNN) model or a long short-term memory network (LSTM) deep learning model, and the intelligent analysis model integrates multi-source data recognition patterns.
[0025] For example, the intelligent analysis model employs a Long Short-Term Memory (LSTM) deep learning model. This model can better learn the temporal variation patterns of power generation difference and identify the characteristics of different fault modes. For instance, the characteristic of a fouling accumulation fault mode is a persistently negative power generation difference with a slow decreasing trend, as fouling gradually reduces the effective utilization rate of sunlight intensity. The characteristic of a shading fault mode is a sudden change in power generation difference, such as a sharp drop in power generation difference within a specific time period, as shading causes a rapid decrease in the output power of some components. The characteristic of a component aging fault mode is a persistently negative power generation difference that slowly increases over time, as component efficiency continuously decreases with increasing aging. The characteristic of an inverter fault mode is irregular fluctuations in power generation difference or a sudden drop to zero, as abnormal inverter operation leads to unstable or completely interrupted output power. By learning these pattern characteristics from historical fault data, the LSM deep learning model can identify similar patterns in new data and output the corresponding fault category.
[0026] As one possible implementation, in step S2, fault modes with a greater impact are processed first based on the degree of impact of the fault modes corresponding to the diagnostic data.
[0027] If multiple rules are met simultaneously, the fault mode with the greatest impact will be prioritized to ensure the power generation of the photovoltaic system. The severity of the fault mode's impact is ranked as follows: inverter failure > module aging > shading > dirt accumulation.
[0028] As one possible implementation, in step S4, the user can interact via a mobile app or desktop application, such as confirming alarms or manually adjusting policies.
[0029] The mobile app displays content consistent with the web interface, but adds push notification functionality. When a high-risk fault is detected, a notification is sent to the user's mobile phone, including the fault category, occurrence time, and suggested actions. Users can click the "Confirm Execution" button to approve the automatic maintenance strategy. After clicking, the front-end sends a confirmation request to the back-end, which records the user's confirmation and sends control commands to the field equipment. Users can click the "Ignore" button to reject the strategy. Clicking this button will bring up a dialog box asking the user to fill in the reason for rejection. Reasons for rejection include options such as "Fault already manually handled," "Strategy unreasonable," and "Equipment under maintenance." After the user selects a reason, the front-end sends an ignore request to the back-end, which records the user's ignore action and the reason for rejection. Users can manually adjust strategy parameters. For example, for the automatic cleaning strategy, users can adjust the cleaning equipment's working time, ranging from 5 to 30 minutes, with a default value of 10 minutes. After adjustment, the front-end sends a parameter update request to the back-end, which regenerates control commands based on the updated parameters.
[0030] User feedback is collected as follows: the backend records user actions, including confirmation, ignoring, and parameter adjustments. Records include the action type, action time, corresponding diagnostic data (fault category and confidence level), and the user's reason for rejection or the adjusted parameters. User feedback data is stored in a feedback database implemented using MongoDB, a document-oriented database suitable for storing unstructured user feedback data. User feedback data is stored in document format, with each document corresponding to one user action. Each document includes fields such as action type, action time, diagnostic data, and user input.
[0031] As one possible implementation, in step S5, if user feedback data is collected monthly for retraining, the intelligent analysis model retraining task is automatically triggered at 1:00 AM on the 1st of each month. The retraining task first reads the feedback data from the past month from the feedback database, filters out the confirmed diagnostic results from users, and uses these diagnostic results as new labeled samples. The new labeled samples are merged with the original training data to form an expanded training dataset. Retraining uses incremental learning, continuing training on the existing LSTM model rather than training a new model from scratch. Incremental learning can retain the knowledge already learned by the model while learning patterns in the new data. The learning rate during retraining is reduced to 0.0001; this low learning rate prevents new data from corrupting the learned knowledge. The retraining rounds are 10 epochs. During training, the model performance is evaluated on the validation set, and the model with the highest accuracy on the validation set is selected as the updated model. The updated model is shared to the intelligent analysis module through the file system. The intelligent analysis module loads the updated model parameters during the next inference, completing the model update. The execution log of the retraining task is recorded to a log file. The log content includes information such as retraining start time, end time, number of new samples, training loss, and validation accuracy, which is used to monitor the trend of model performance changes.
[0032] As one possible implementation, latency can be reduced by using edge computing and data transmission reliability can be improved by combining it with 5G technology.
[0033] like Figure 1 , Figure 2 As shown, the present invention provides an intelligent photovoltaic operation and maintenance system based on power generation difference, used to implement the above-mentioned intelligent photovoltaic operation and maintenance method based on power generation difference, comprising: The data acquisition and processing module is used to acquire the expected power generation data and actual power generation data of the photovoltaic power generation system, and to calculate and generate power generation difference data based on the expected power generation data and actual power generation data; The intelligent analysis module is communicatively connected to the data acquisition and processing module. It is used to receive the power generation difference data and use the intelligent analysis model to analyze the power generation difference data to generate diagnostic data. The intelligent analysis model is trained based on historical data to identify common fault modes. The maintenance strategy generation and execution module is communicatively connected to the intelligent analysis module. It is used to generate maintenance strategies based on the diagnostic data and generate control commands based on the maintenance strategies to send to the field equipment. An interactive display and feedback module is communicatively connected to the intelligent analysis module and the maintenance strategy generation and execution module. It is used to visually display the power generation difference data, diagnostic data and maintenance strategies to the user, and to receive user feedback. The model self-learning optimization module is communicatively connected to the interactive display and feedback module and the intelligent analysis module. It is used to input the feedback data generated by user feedback into the intelligent analysis model and optimize and train it. It also achieves self-learning by periodically summarizing user feedback data and retraining the intelligent analysis model.
[0034] The aforementioned intelligent photovoltaic (PV) operation and maintenance (O&M) system based on power generation difference is used to implement an intelligent PV O&M method based on power generation difference. By accurately diagnosing and analyzing the difference between actual and expected power generation, it reduces false alarms and missed alarms, improving O&M efficiency. The intelligent analysis module diagnoses and analyzes power generation difference data, enabling early identification of potential problems in the PV system before faults fully manifest or cause significant losses. It provides early warnings or schedules maintenance, shifting from passive to predictive maintenance, effectively extending the lifespan of PV equipment and ensuring the long-term stable operation of the PV system. The model self-learning optimization module periodically summarizes user feedback data to retrain the intelligent analysis module. The intelligent analysis module has self-learning capabilities and continuously optimizes its training, constantly improving diagnostic accuracy and the effectiveness of generated maintenance strategies over time, thus enhancing the long-term adaptability, stability, and intelligence level of the PV system.
[0035] The intelligent photovoltaic operation and maintenance method and system based on power generation difference are not only applicable to large-scale photovoltaic power plants, but can also be extended to distributed household photovoltaic power generation systems.
[0036] The above description is merely a specific embodiment of the present invention, but the scope of protection of the present invention is not limited thereto. Any changes or substitutions that can be easily conceived by those skilled in the art within the scope of the technology disclosed in the present invention should be included within the scope of protection of the present invention.
Claims
1. A smart photovoltaic operation and maintenance method based on power generation difference, characterized in that, Includes the following steps: S1. Obtain the expected power generation data and actual power generation data of the photovoltaic power generation system, and calculate the power generation difference data; S2. Input the power generation difference data into the intelligent analysis model. The intelligent analysis model is trained based on historical data to identify common fault modes. The intelligent analysis model analyzes the power generation difference data to generate diagnostic data. S3. Generate a maintenance strategy based on the diagnostic data, and generate control commands based on the maintenance strategy and send them to the field equipment; S4. Visualize operation and maintenance through an interactive platform and receive user feedback; S5. Generate feedback data based on user feedback, input the feedback data into the intelligent analysis model and optimize and train it, and periodically summarize user feedback data to retrain the intelligent analysis model to achieve self-learning.
2. The intelligent photovoltaic operation and maintenance method based on power generation difference according to claim 1, characterized in that, The actual power generation data P1 is collected in real time by IoT sensors, and the expected power generation data P2 is calculated using the following physical model: P2 = η A I (1-β) (T1-T2) cos(θ), where η is the photovoltaic module efficiency, A is the photovoltaic module area, I is the light intensity, β is the temperature coefficient, T1 is the actual operating temperature of the photovoltaic module, T2 is the reference temperature, θ is the solar incidence angle, and the power generation difference data ΔP=P1–P2.
3. The intelligent photovoltaic operation and maintenance method based on power generation difference according to claim 2, characterized in that, The actual power generation data is collected in real time by IoT sensors such as light sensors, temperature sensors, current and voltage sensors, with a sampling frequency of 1 time per minute. The collected actual power generation data is wirelessly transmitted to the data processing center of the interactive platform, where the data processing center performs data cleaning and data calibration on the collected actual power generation data.
4. The intelligent photovoltaic operation and maintenance method based on power generation difference according to claim 3, characterized in that, The data cleaning includes removing outliers, and the data calibration includes compensating for sensor errors.
5. The intelligent photovoltaic operation and maintenance method based on power generation difference according to claim 1, characterized in that, The intelligent analysis model employs a convolutional neural network model or a long short-term memory network deep learning model, and integrates multi-source data recognition patterns.
6. The intelligent photovoltaic operation and maintenance method based on power generation difference according to claim 1, characterized in that, The common failure modes include linear degradation caused by dirt accumulation in photovoltaic modules, abrupt changes caused by shading, and linear degradation and temperature rise caused by heat dissipation problems.
7. The intelligent photovoltaic operation and maintenance method based on power generation difference according to claim 1, characterized in that, The maintenance strategy includes triggering automatic cleaning of photovoltaic modules and adjusting the inverter's maximum power point tracking parameters.
8. The intelligent photovoltaic operation and maintenance method based on power generation difference according to claim 1, characterized in that, Based on the degree of impact of the fault modes corresponding to the diagnostic data, fault modes with a higher degree of impact are prioritized for processing.
9. The intelligent photovoltaic operation and maintenance method based on power generation difference according to claim 1, characterized in that, The interactive platform uses a web interface, which displays a curve composed of real-time power generation difference data ΔP, diagnostic results corresponding to diagnostic data, and maintenance suggestions corresponding to maintenance strategies.
10. An intelligent photovoltaic operation and maintenance system based on power generation difference, used to implement the intelligent photovoltaic operation and maintenance method based on power generation difference as described in any one of claims 1-9, characterized in that, include: The data acquisition and processing module is used to acquire the expected power generation data and actual power generation data of the photovoltaic power generation system, and to calculate and generate power generation difference data based on the expected power generation data and actual power generation data; The intelligent analysis module is communicatively connected to the data acquisition and processing module. It is used to receive the power generation difference data and use the intelligent analysis model to analyze the power generation difference data to generate diagnostic data. The intelligent analysis model is trained based on historical data to identify common fault modes. The maintenance strategy generation and execution module is communicatively connected to the intelligent analysis module. It is used to generate maintenance strategies based on the diagnostic data and generate control commands based on the maintenance strategies to send to the field equipment. An interactive display and feedback module is communicatively connected to the intelligent analysis module and the maintenance strategy generation and execution module. It is used to visually display the power generation difference data, diagnostic data and maintenance strategies to the user, and to receive user feedback. The model self-learning optimization module is communicatively connected to the interactive display and feedback module and the intelligent analysis module. It is used to input the feedback data generated by user feedback into the intelligent analysis model and optimize and train it. It also achieves self-learning by periodically summarizing user feedback data and retraining the intelligent analysis model.