Distributed photovoltaic power station intelligent safety operation and maintenance management method and system

Abstract

The invention discloses an intelligent safety operation and maintenance management method and system for a distributed photovoltaic power station, and relates to the technical field of intelligent safety operation and maintenance, and the method comprises the steps: collecting the operation data of a photovoltaic string, carrying out the feature parameter extraction, and carrying out the disturbance risk recognition according to the extracted feature parameter, so as to mark a risk photovoltaic string. The method comprises the following steps: establishing a risk photovoltaic string, calculating the potential leakage probability # imgabs0 # of the risk photovoltaic string, further analyzing the electrical insulation performance degradation degree of the risk photovoltaic string, evaluating the risk level of the risk photovoltaic string, and then carrying out trend aggregation analysis on the risk photovoltaic string of a first risk level and a second risk level to identify a space aggregation risk and trigger an early warning mechanism. The potential leakage risk accumulation process is accurately quantified, faults caused by the fact that hidden dangers such as micro-cracks, water accumulation and aging are not found in time are avoided, the prospective early warning and regional linkage capacity of an intelligent operation and maintenance system is improved, and more efficient and finer-grained safety management support is provided for a photovoltaic power station.

Description

Technical Field

[0001] The present invention relates to the field of intelligent safety operation and maintenance technology, and specifically to an intelligent safety operation and maintenance management method and system for distributed photovoltaic power stations. Background Art

[0002] In distributed photovoltaic power stations, the string system constitutes the basic unit for collecting and transmitting electricity. The operating status of each string directly affects the power generation safety and stability of the entire site. During the operation of these strings, the long-term degradation of the "insulation status" has become one of the key hidden factors in current safety management. Especially in hot and humid environments with frequent rain and fog or high salt fog on the coast, the insulation protection layer of the components and the junction box can easily be gradually weakened due to problems such as material aging, micro-cracks, dust accumulation and moisture, eventually developing into potential leakage or breakdown accidents. Therefore, how to establish a dynamic identification and early warning mechanism for the "insulation degradation trend" at the string level without adding new hardware has become one of the core issues to ensure the safe operation of distributed photovoltaic power stations.

[0003] In the Chinese invention patent application with publication number CN118552055A, a distributed photovoltaic intelligent health evaluation method and system are proposed. By using a remote monitoring platform to obtain the operating big data of distributed photovoltaic power stations, unified monitoring and evaluation of power stations across the country can be achieved, improving management and operation efficiency; multiple indicators related to health are selected, and the quantitative value of each indicator is determined. In combination with an intelligent algorithm, dynamic weight allocation is performed on the influencing factors of the distributed photovoltaic power station to more accurately reflect the health status of the power station; and a multi-objective intelligent evaluation of the health of distributed photovoltaic power stations across the country is performed through a unified monitoring platform to generate an evaluation report.

[0004] The above methods can promptly identify and resolve problems, ensuring the safe and stable operation of power stations. However, in addition to these, in existing intelligent and safe operation and maintenance management methods for distributed photovoltaic power stations, the detection of string insulation performance mainly relies on "periodic insulation resistance testing" and "manual infrared / grounding inspection".

[0005] However, these two methods have defects: first, the detection frequency is low and the interval time is long, which makes it difficult to track the gradual change trend of the insulation status and easily misses the early hidden danger window; second, the insulation resistance test requires manual power-off operation, which is not suitable for frequent execution and is very likely to cause missed detection; third, the detection only reflects the status at a certain moment, and has poor dynamic response capabilities to sudden weather, drastic changes in climate humidity, etc.; fourth, there is a lack of modeling and analysis methods for the hidden leakage channels in the systematic structure of "component body-combiner box-ground wire coupling path", resulting in the existing distributed photovoltaic power station intelligent safety operation and maintenance management methods "unable to see, measure, and predict" the insulation degradation status, seriously affecting the intelligent scheduling and automatic prevention and control capabilities of the overall safety strategy, and also weakening the power generation efficiency of the power station, triggering inverter failures, increasing fire risks, and directly endangering the safety of people and equipment.

[0006] To this end, the present invention proposes a distributed photovoltaic power station intelligent safety operation and maintenance management method and system. Summary of the Invention

[0007] In response to the shortcomings of the existing technology, the present invention provides a distributed photovoltaic power station intelligent safety operation and maintenance management method and system. By collecting photovoltaic string operation data, characteristic parameters are extracted, and disturbance risk identification is performed based on the extracted characteristic parameters to mark risk photovoltaic strings and calculate the potential leakage probability of risk photovoltaic strings. , further analyze the degree of degradation of the electrical insulation performance of the risky photovoltaic strings, evaluate the risk level of the risky photovoltaic strings, and then further conduct trend aggregation analysis on the risky photovoltaic strings of the first risk level and the second risk level to identify spatial clustering risks, trigger the early warning mechanism, and accurately quantify the accumulation process of potential leakage risks, avoid failures caused by hidden dangers such as microcracks, water accumulation, and aging that are not discovered in time, improve the forward-looking warning and regional linkage capabilities of the intelligent operation and maintenance system, provide more efficient and fine-grained safety management support for photovoltaic power stations, and solve the problems in the above-mentioned background technology.

[0008] To achieve the above objectives, the present invention is implemented through the following technical solutions: a distributed photovoltaic power station intelligent safety operation and maintenance management method, comprising the following steps: S1. Deploy smart sensor groups on photovoltaic strings of a distributed photovoltaic power station and use the smart sensor groups to collect photovoltaic string operation data in real time to construct a photovoltaic string operation data set S. Extract characteristic parameters based on the photovoltaic string operation data set S, and construct a characteristic parameter set F based on the extracted characteristic parameters. S2. Based on the characteristic parameter set F, perform characteristic parameter coupling calculation to obtain the physical disturbance risk score , and the physical disturbance risk score and physical disturbance risk thresholds Conduct comparative analysis to identify risky PV strings; S3. For risky PV strings, the risk score of the physical disturbance of the risky PV strings is calculated based on the PV string operation data set S, the characteristic parameter set F and the risky PV strings. , perform coupled calculation of physical disturbance and humidity parameters to obtain the potential leakage probability , and further analyze the degree of electrical insulation degradation of risky PV strings to obtain the leakage risk accumulation index LRAI; S4. Compare the leakage risk accumulation index LRAI with the first leakage risk threshold and the second leakage risk threshold Conduct comparative analysis to assess the risk level of risky PV strings; S5. Based on the risk level assessment results, construct a risky PV string set and perform trend aggregation analysis on the risky PV string set to obtain the trend aggregation factor (STCI), identify spatial aggregation risks, and issue early warnings.

[0009] Preferably, the specific steps of S1 include: S11, using each photovoltaic string of the distributed photovoltaic power station as a data acquisition unit, deploying an intelligent sensor group on each photovoltaic string to obtain photovoltaic string operation data, wherein the photovoltaic string operation data includes the photovoltaic string output current value I, voltage V, photovoltaic module surface temperature Tsurf, air temperature Tair and relative humidity H; A photovoltaic string refers to the smallest power generation and operation unit constructed by connecting multiple photovoltaic modules in a DC series manner; The intelligent sensor group includes a combiner box current transformer, a combiner box voltage divider, a thermocouple temperature sensor, and a humidity sensor; S12. Preprocessing the collected photovoltaic string operation data, including data cleaning, data smoothing and normalization, and constructing a photovoltaic string operation data set S based on the preprocessed photovoltaic string operation data.

[0010] Preferably, the specific step S1 further includes: S13. Extract characteristic parameters based on the photovoltaic string operation data set S to obtain a current fluctuation index DL, a voltage edge drift index Dv, a capacitive coupling residual index Dr, and a condensation risk index JL; The current fluctuation index DL refers to the volatility of the PV string output current value. By taking the difference between the PV string output current values ​​I at adjacent time points, the current change rate per unit time is obtained, a current change rate time series is constructed, and the standard deviation of the current change rate time series is calculated as the current fluctuation index. The voltage edge drift index Dv refers to the deviation of the waveform slope of the rising or falling edge of the PV string voltage waveform relative to the standard PV string voltage waveform; The capacitive coupling residual index Dr refers to the total amount of fitting residuals between the actual recovery curve of the PV string and the standard RC model when the voltage recovers from a low level to a steady state during the operation of the PV string; The condensation risk index JL refers to the relative critical proximity between the surface temperature Tsurf of the PV module and the air dew point temperature, where the air dew point temperature refers to the critical temperature at which water vapor in the air condenses into liquid water; S14. Construct a characteristic parameter set F based on the obtained current fluctuation index DL, voltage edge drift index Dv, capacitive coupling residual index Dr, and condensation risk index JL.

[0011] Preferably, the specific steps of S2 include: S21. Perform dimensionless processing on the characteristic parameters in the characteristic parameter set F to eliminate the physical dimension, and analyze the risk intensity level of the photovoltaic string under the condition of coupling multiple characteristic parameters based on the dimensionless processed characteristic parameters to obtain the physical disturbance risk score. , where the physical disturbance risk score The specific way to obtain it is: ; Where, 、 and They represent the current fluctuation index DL, voltage edge drift index Dv and capacitive coupling residual index Dr of the PV string at the current time point t, respectively, and ln represents the natural logarithmic constant; S22. Collect the operating data of several photovoltaic strings when they are in a healthy state, so as to calculate the physical disturbance risk scores of several photovoltaic strings when they are in a healthy state. , and score the risk of all physical disturbances in a healthy state Perform statistical distribution analysis and select the 95th percentile physical disturbance risk score from the statistical distribution analysis results As a physical disturbance risk threshold ; S23. Set the physical disturbance risk threshold and physical disturbance risk scoring Conduct comparative analysis to assess the disturbance risk status of PV strings. If the physical disturbance risk score Greater than or equal to the physical disturbance risk threshold , then the PV string is judged to be in an abnormally enhanced state, and the PV string is marked as a risky PV string. If the physical disturbance risk score Less than the physical disturbance risk threshold , it is determined that the PV string is in a normal state and no processing is required.

[0012] Preferably, the specific steps of S3 include: S31. For the marked risky PV strings, based on the PV string operation data set S, characteristic parameter set F and physical disturbance risk score , comprehensively analyze the probability of potential leakage of risk PV strings under the influence of current physical disturbance and environmental humidity to obtain the potential leakage probability , where the potential leakage probability The specific manifestations are: ; Where, represents the leakage trend coupling coefficient, exp represents the exponential function, represents the relative humidity of the risky PV string at the current time point t, Indicates the condensation risk index of the risky PV string at the current time point t.

[0013] Preferably, the specific step S3 further includes: S32, set time sampling window { , t}, and according to the photovoltaic string operation data set S, obtain the time sampling window { , the potential leakage probability at each data acquisition time point within t} The sliding integral method is used to analyze the degree of degradation of the electrical insulation performance of the risky PV strings to obtain the leakage risk accumulation index LRAI. The specific method for obtaining the leakage risk accumulation index LRAI is as follows: ; Where, Indicates the risk of PV strings at the data collection time point The potential leakage probability when represents the integration variable, Indicates the length of the historical time window, Represents the attenuation factor.

[0014] Preferably, the specific step of S4 includes: S41. Preset the first leakage risk threshold and the second leakage risk threshold and the leakage risk accumulation index LRAI and the first leakage risk threshold and the second leakage risk threshold A comparative analysis is conducted to assess the risk level of risky PV strings. The specific assessment contents are as follows: If the leakage risk accumulation index LRAI is less than or equal to the first leakage risk threshold , the risk level of the risky PV string is judged to be the first risk level. At this time, the normal operation mode of the PV power station is maintained, and the data monitoring frequency is maintained at normal level; If the leakage risk accumulation index LRAI is greater than the first leakage risk threshold , and is less than or equal to the second leakage risk threshold , the risk level of the risky PV string is determined to be the second risk level. At this time, there is abnormal fluctuation in the risky PV string, which needs to trigger the early warning mechanism, generate an early warning tag, and send the early warning tag to the safety monitoring platform. At the same time, the data monitoring frequency is adjusted from once every 30 minutes to once every 10 minutes. If the leakage risk accumulation index LRAI is greater than or equal to the second leakage risk threshold , the risk level of the risky PV string is determined to be the third risk level. At this time, a PV string isolation instruction is immediately generated, and the isolation switch corresponding to the PV string is closed to disconnect the PV string from the main circuit. An isolation log and an execution report are generated and continuously sent to the safety monitoring platform. An alarm is issued until the safety monitoring platform administrator responds.

[0015] Preferably, the specific steps of S5 include: S51, gather risky photovoltaic strings at the first risk level and the second risk level, construct a risky photovoltaic string set, extract the leakage risk accumulation index LRAI of each risky photovoltaic string in the risky photovoltaic string set, and perform data averaging to obtain the average value of the leakage risk accumulation index ; S52. Calculate the leakage risk accumulation index LRAI and the average value of the leakage risk accumulation index of each risky PV string in the risky PV string set and divide the difference by the average value of the leakage risk accumulation index The sum of the trend deviation rate BIAS of each risky PV string is obtained by adding the non-zero small constant ε.

[0016] Preferably, the specific step S5 further includes: S53, based on the trend deviation rate of each risky PV string and the average value of leakage risk accumulation index , use the variance calculation formula to obtain the trend deviation rate of the risk photovoltaic string set Variance, that is, the trend clustering factor STCI; S54. Extract the historical photovoltaic string operation data of the risk photovoltaic string set to continuously calculate the trend aggregation factor STCI of the risk photovoltaic string set over a period of time, construct the trend aggregation factor time series, and obtain the historical mean and historical standard deviation of the trend aggregation factor STCI based on the trend aggregation factor time series, and obtain the dynamic trend risk threshold through the standard deviation boundary model. ; S55. Dynamic trend risk threshold A comparative analysis is conducted with the trend clustering factor (STCI) to assess the risk of degradation trend fluctuations among risky PV strings. The specific assessment contents are as follows: If the trend concentration factor STCI is greater than or equal to the dynamic trend risk threshold , it is determined that the degradation trend fluctuations between risky PV strings are in an abnormal range, and there is a phenomenon of synchronous enhancement of degradation trends between risky PV strings, and there is a spatial clustering risk. At this time, the trend deviation rate in the risky PV string set is screened out. Greater than 0 and higher than the trend deviation rate at multiple consecutive time points The system then determines the risky PV strings based on the average value and, based on the PV power station equipment distribution map, determines whether the risky PV strings belong to a physically adjacent area. If so, it indicates that the risky PV strings in the physically adjacent area have a regional potential leakage risk, immediately triggering an early warning mechanism, generating a manual maintenance task, and executing a voltage reduction operation strategy. If not, no action is required. If the trend concentration factor STCI is less than the dynamic trend risk threshold , it is determined that the degradation trend fluctuation between risky PV strings is within the normal range and no processing is required.

[0017] Preferably, a distributed photovoltaic power station intelligent safety operation and maintenance management system includes a data acquisition and feature extraction module, a disturbance analysis module, a potential leakage probability calculation module, a risk assessment module and a spatial aggregation analysis module; The data acquisition and feature extraction module is used to deploy an intelligent sensor group on the photovoltaic strings of the distributed photovoltaic power station, and use the intelligent sensor group to collect photovoltaic string operation data in real time to construct a photovoltaic string operation data set S, and extract feature parameters based on the photovoltaic string operation data set S, and construct a feature parameter set F based on the extracted feature parameters; The disturbance analysis module is used to perform characteristic parameter coupling calculation based on the characteristic parameter set F to obtain the physical disturbance risk score. , and the physical disturbance risk score and physical disturbance risk thresholds Conduct comparative analysis to identify risky PV strings

[0018] The potential leakage probability calculation module is used for risky photovoltaic strings, based on the photovoltaic string operation data set S, characteristic parameter set F and the physical disturbance risk score of the risky photovoltaic strings. , perform coupled calculation of physical disturbance and humidity parameters to obtain the potential leakage probability , and further analyze the degree of electrical insulation degradation of risky PV strings to obtain the leakage risk accumulation index LRAI; The risk assessment module is used to compare the leakage risk accumulation index LRAI with the first leakage risk threshold and the second leakage risk threshold Conduct comparative analysis to assess the risk level of risky PV strings; The spatial aggregation analysis module is used to construct a risky photovoltaic string set based on the risk level assessment results, and perform trend aggregation analysis on the risky photovoltaic string set to obtain the trend aggregation factor (STCI), identify spatial aggregation risks, and issue early warnings.

[0019] The present invention provides a distributed photovoltaic power station intelligent safety operation and maintenance management method and system, which has the following beneficial effects: (1) By constructing the "potential leakage probability" and "leakage risk accumulation index LRAI", easily overlooked parameters such as humidity, condensation trend, and capacitive coupling residual are introduced into the quantitative modeling of insulation performance degradation. Compared with the traditional method of relying on "periodic insulation resistance detection" or "manual patrol detection", this method can dynamically obtain risk indicators during the operation of photovoltaic modules, capture the gradual evolution characteristics of "insulation trend", and identify potential but not yet prominent hidden dangers. Especially in humid environments such as thunderstorms, high humidity, and coastal areas, it can significantly improve the early identification ability of problems such as microcracks, water vapor erosion, and packaging aging, solve the pain points of traditional methods such as "low detection frequency, poor timeliness, and high risk of misjudgment", and improve the foresight and intelligence level of operation and maintenance.

[0020] (2) By comparing the leakage risk accumulation index LRAI with the trend deviation degree of photovoltaic strings, the "trend deviation rate BIAS" and "trend clustering factor STCI" are proposed to identify the "degradation trend spatial clustering" phenomenon among photovoltaic modules. When multiple physically adjacent photovoltaic strings show a synchronous trend of high leakage risk accumulation index LRAI, such as batch defects of junction boxes, regional water vapor accumulation, abnormal potential path of module series connection, etc., they may cause "hidden regional risk clustering". It can complete the identification of zoning hidden dangers only through trend data analysis without destroying the on-site structure and without the need for external instruments. It is a powerful supplement to the "individual module risk assessment".

[0021] (3) By comparing the leakage risk accumulation index LRAI with the dynamic threshold, the three-level risk level determination is completed, and the "abnormal spatial aggregation" event is further identified in combination with the trend aggregation factor STCI. Finally, the linkage system triggers specific strategies including "early warning reporting", "voltage reduction operation" and "string isolation", thereby realizing closed-loop control from "risk identification" to "safe disposal". Compared with traditional early warning means, this method introduces the trend evolution dimension and spatial distribution evaluation mechanism, which can actively adjust the inspection frequency, voltage control logic and maintenance task issuance granularity according to the risk level and trend direction, so that the safety strategy is no longer static or manually set, but dynamically adapted as the data evolves, effectively improving the adaptive ability and rapid fault suppression efficiency of the intelligent operation and maintenance system of the photovoltaic power station, and enhancing the long-term safety and stability of the overall power station. BRIEF DESCRIPTION OF THE DRAWINGS

[0022] Figure 1 This is a flow chart of a method for intelligent and safe operation and maintenance management of distributed photovoltaic power stations according to the present invention; Figure 2 This is a block diagram of a distributed photovoltaic power station intelligent safety operation and maintenance management system of the present invention; Figure 3 This is a schematic diagram of the process of collecting photovoltaic string operation data and pre-setting characteristic parameters according to the present invention; Figure 4 Schematic diagram of the risk level assessment process of risky photovoltaic strings according to the present invention. DETAILED DESCRIPTION

[0023] The following will clearly and completely describe the technical solutions in the embodiments of the present invention in conjunction with the accompanying drawings. Obviously, the described embodiments are only part of the embodiments of the present invention, not all of the embodiments. Based on the embodiments of the present invention, all other embodiments obtained by ordinary technicians in this field without making creative efforts are within the scope of protection of the present invention.

[0024] Example 1 See also Figure 1 The present invention provides a distributed photovoltaic power station intelligent safety operation and maintenance management method, comprising the following steps: S1. Deploy smart sensor groups on photovoltaic strings of a distributed photovoltaic power station and use the smart sensor groups to collect photovoltaic string operation data in real time to construct a photovoltaic string operation data set S. Extract characteristic parameters based on the photovoltaic string operation data set S, and construct a characteristic parameter set F based on the extracted characteristic parameters. S2. Based on the characteristic parameter set F, perform characteristic parameter coupling calculation to obtain the physical disturbance risk score , and the physical disturbance risk score and physical disturbance risk thresholds Conduct comparative analysis to identify risky PV strings; S3. For risky PV strings, the risk score of the physical disturbance of the risky PV strings is calculated based on the PV string operation data set S, the characteristic parameter set F and the risky PV strings. , perform coupled calculation of physical disturbance and humidity parameters to obtain the potential leakage probability , and further analyze the degree of electrical insulation degradation of risky PV strings to obtain the leakage risk accumulation index LRAI; S4. Compare the leakage risk accumulation index LRAI with the first leakage risk threshold and the second leakage risk threshold Conduct comparative analysis to assess the risk level of risky PV strings; S5. Based on the risk level assessment results, construct a risky PV string set and perform trend aggregation analysis on the risky PV string set to obtain the trend aggregation factor (STCI), identify spatial aggregation risks, and issue early warnings.

[0025] In the embodiment, by constructing a closed-loop process from data collection, disturbance risk identification, trend evolution tracking to spatial aggregation warning, the technical shortcomings of "invisible, unmeasurable and unpredictable" in the existing distributed photovoltaic power station insulation monitoring methods are effectively broken through. Compared with the traditional mode that relies on manual detection, this method can dynamically identify the insulation degradation trend based on component operation data, accurately quantify the leakage risk accumulation process, and avoid failures caused by hidden dangers such as microcracks, water accumulation, and aging that are not discovered in time. In particular, through the calculation of trend aggregation factors, regional synchronous degradation phenomena can be accurately identified, realizing the transformation from single-point risk identification to systemic risk control, and significantly improving the forward-looking warning and regional linkage capabilities of the intelligent operation and maintenance system, providing more efficient and fine-grained safety management support for photovoltaic power stations.

[0026] Example 2 Please refer to Figure 1 and Figure 3 ,Specifically: S1 specific steps include, S11, using each photovoltaic string of the distributed photovoltaic power station as a data acquisition unit, deploying an intelligent sensor group on each photovoltaic string to obtain photovoltaic string operation data, wherein the photovoltaic string operation data includes the photovoltaic string output current value I, voltage V, photovoltaic module surface temperature Tsurf, air temperature Tair and relative humidity H; A photovoltaic string refers to the smallest power generation and operation unit constructed by connecting multiple photovoltaic modules in a DC series manner; The intelligent sensor group includes a combiner box current transformer, a combiner box voltage divider, a thermocouple temperature sensor, and a humidity sensor; The PV string output current value I is acquired by using the combiner box current transformer, the voltage V is acquired by using the combiner box voltage divider, the PV module surface temperature Tsurf and the air temperature Tair are acquired by using thermocouple temperature sensors, and the relative humidity H is acquired by using a humidity sensor; S12. Preprocessing the collected photovoltaic string operation data, including data cleaning, data smoothing and normalization, and constructing a photovoltaic string operation data set S based on the preprocessed photovoltaic string operation data.

[0027] S13. Extract characteristic parameters based on the photovoltaic string operation data set S to obtain a current fluctuation index DL, a voltage edge drift index Dv, a capacitive coupling residual index Dr, and a condensation risk index JL; The current fluctuation index DL refers to the rate of change of the short-circuit current over time. It is obtained by subtracting the PV string output current value I at a certain data collection time point from the PV string output current value I at the previous data collection time point, and dividing the result by the data sampling time interval. The voltage edge drift index Dv refers to the deviation of the slope of the rising or falling edge of the PV string voltage waveform relative to the standard PV string voltage waveform, and is used to reflect whether the module potential establishment speed is abnormal; The capacitive coupling residual index, Dr, refers to the total residual error between the actual recovery curve of a PV string and the standard RC model when the voltage recovers from a low level to a steady state during PV string operation. It reflects the asymmetry, degradation, or charge leakage of the capacitive coupling path between components within the string. The standard RC model refers to a voltage recovery model under ideal conditions, i.e., without performance degradation or additional losses. Voltage recovery from low level to steady state refers to the dynamic voltage recovery process during the operation of PV strings, when the voltage temporarily drops or is in a low voltage state due to certain reasons such as shading, load dump and rapid switching, and the system gradually recovers to its normal stable operating voltage. The condensation risk index JL refers to the critical proximity between the PV module surface temperature Tsurf and the air dew point temperature. The air dew point temperature refers to the critical temperature at which water vapor in the air condenses into liquid water. Liquid water can form conductive water film pathways in areas with micro-cracks on the backsheet, poorly sealed junction boxes, and aged welds, posing a safety hazard to PV power plants. The air dew point temperature is calculated using the PV panel surface temperature Tsurf and relative humidity H, combined with the Magnus formula. The Magnus formula is an empirical formula used to estimate the relationship between water vapor saturation pressure and temperature and humidity. S14. Construct a characteristic parameter set F based on the obtained current fluctuation index DL, voltage edge drift index Dv, capacitive coupling residual index Dr, and condensation risk index JL.

[0028] In the embodiment, by deploying an intelligent sensor group in each photovoltaic string, basic operating data including short-circuit current, voltage, temperature and humidity are collected, and on this basis, a set of refined feature parameter sets F is constructed, including current fluctuation index DL, voltage edge drift index Dv, capacitance coupling residual index Dr and condensation risk index JL. This breaks through the passive method of traditional photovoltaic insulation status assessment that relies on manual infrared detection or periodic resistance measurement. This method can not only accurately capture the operating dynamics at the component level, but also complete pre-judgment through feature offset trends before the component undergoes potential aging, microcracks or packaging degradation. In particular, the introduction of the condensation risk index can effectively identify environmentally induced hidden dangers such as backplane water vapor penetration. Overall, this embodiment improves the response sensitivity to early signals of insulation degradation, lays a high-precision foundation for subsequent potential leakage modeling and trend identification, and enhances the foresight and technical effectiveness of system-level safety warnings.

[0029] Example 3 Please refer to Figure 1 and Figure 4 ,Specifically: S2 specific steps include, S21. Perform dimensionless processing on the characteristic parameters in the characteristic parameter set F to eliminate the physical dimension, and analyze the risk intensity level of the photovoltaic string under the condition of coupling multiple characteristic parameters based on the dimensionless processed characteristic parameters to obtain the physical disturbance risk score. , where the physical disturbance risk score The specific way to obtain it is: ; Where, 、 and They represent the current fluctuation index DL, voltage edge drift index Dv and capacitive coupling residual index Dr of the PV string at the current time point t, respectively, and ln represents the natural logarithmic constant; Formula derivation process and physical significance: The formula performs energy-type synthesis on the three dimensionless characteristic parameters, including the current fluctuation index DL, the voltage edge drift index Dv, and the capacitive coupling residual index Dr, through square operations to construct a disturbance intensity term. This synthesis method is equivalent to the sum of squares of the Euclidean norm, reflecting the overall disturbance intensity. The constructed disturbance intensity term is then raised to a power by a power function to obtain a disturbance amplification term, and then the constant term 1+ is added to avoid zero input. Finally, the whole is logarithmically transformed to enhance scale discrimination and compress extreme value fluctuations to form the final physical disturbance risk score. The formula integrates disturbance signals, nonlinear enhancement, and logarithmic mapping mechanisms to construct a quantifiable and evaluable physical disturbance risk score, which effectively supports subsequent risk level classification, early warning identification, and trend aggregation analysis, and has a clear physical basis and mathematical completeness.

[0030] S22. Collect the operating data of several photovoltaic strings when they are in a healthy state, so as to calculate the physical disturbance risk scores of several photovoltaic strings when they are in a healthy state. , and score the risk of all physical disturbances in a healthy state Perform statistical distribution analysis and select the 95th percentile physical disturbance risk score from the statistical distribution analysis results As a physical disturbance risk threshold ; Statistical distribution analysis refers to the process of dividing several data samples, i.e. physical disturbance risk scores, into , using the quantile extraction method, quantitatively characterize the overall volatility and boundary characteristics of the sample, thereby extracting the upper tolerance value of the physical disturbance risk; S23. Set the physical disturbance risk threshold and physical disturbance risk scoring Conduct comparative analysis to assess the disturbance risk status of PV strings. If the physical disturbance risk score Greater than or equal to the physical disturbance risk threshold , then the PV string is judged to be in an abnormally enhanced state and marked as a risky PV string. If the physical disturbance risk score Less than the physical disturbance risk threshold , it is determined that the PV string is in a normal state and no processing is required.

[0031] In the embodiment, by introducing physical disturbance risk scoring, a comprehensive evaluation is performed on the current, voltage and capacitance behaviors exhibited during the operation of photovoltaic strings, and a dynamic 95% quantile risk threshold is constructed using historical samples in a healthy state, thereby achieving sensitive capture and judgment of early characteristics of insulation degradation. Compared with the traditional method that relies on manual insulation resistance testing, the present invention can continuously and non-invasively identify the cumulative trend of weak disturbances, effectively avoiding the problems of low detection frequency and delayed response. At the same time, through unified physical dimension processing, the comparability and data adaptability between different indicators are improved, so that the identification mechanism has stronger generalization ability and field adaptability, which helps to improve the efficiency of early identification of insulation degradation trends in photovoltaic power stations, provide accurate basis for subsequent risk warnings and graded disposal of photovoltaic strings, and enhance the active perception and dynamic response capabilities of the intelligent operation and maintenance system.

[0032] Example 4 Please refer to Figure 1 and Figure 4 ,Specific: S3 specific steps include, S31. For the marked risky PV strings, based on the PV string operation data set S, characteristic parameter set F and physical disturbance risk score , comprehensively analyze the probability of potential leakage of risk PV strings under the influence of current physical disturbance and environmental humidity to obtain the potential leakage probability , where the potential leakage probability The specific manifestations are: ; Where, represents the leakage trend coupling coefficient, exp represents the exponential function, represents the relative humidity of the risky PV string at the current time point t, Indicates the condensation risk index of the risky PV string at the current time point t.

[0033] Leakage trend coupling coefficient Indicates the risk score of PV strings in physical disturbance and relative humidity Condensation risk index Under the joint action, the nonlinear coupling adjustment factor used to adjust the growth amplitude of the potential leakage probability and the leakage trend coupling coefficient The larger the value, the more sensitive the PV string is to the leakage response of the external disturbance conditions and the stronger the risk amplification trend is. It is used to reflect the sensitivity of the PV string to the leakage response of the coupling conditions. Among them, the leakage trend coupling coefficient The specific value of is obtained by fitting the historical operating data of the photovoltaic strings using the least squares method; Formula derivation process and physical significance: First, the formula introduces the electron breakdown probability model as a source of inspiration for physical modeling. The core structure of the formula is an exponential transition expression. This structure is analogous to the Weibull-type nonlinear mutation mechanism in the electron breakdown probability model. When the physical disturbance continues to increase and the environmental conditions approach the critical state, the leakage risk does not increase linearly, but rather rapidly approaches 1 in an explosive manner. Therefore, an exponential function is used to handle this. The exponential function has a risk expression form that is smooth before the critical state and sharply accelerates after the critical state. Among them, the physical disturbance risk score Indicates the abnormal electrical dynamics and physical disturbance risk score of the PV string during operation. The stronger it is, the more intense the potential change is, and the easier it is to trigger latent discharge; Represents humidity. The higher the humidity, the easier it is for the insulating surface or gap to form a conductive bridge due to water vapor accumulation. It is a wetting approach amplification factor, reflecting the critical trend of the surface water film form. The smaller the value, the closer the component surface temperature is to the air dew point, and condensation is very likely to occur. It represents the environmental factor adjustment term, which is a nonlinear function structure. It ensures the release of real risk signals under the conditions of "high humidity + thermal criticality" and constructs a thermal-humidity linkage assessment model with physical response sensitivity and risk filtering capabilities. It represents the equivalent leakage excitation factor, which is a coupling term of disturbance, humidity and wetting. It means that at a certain moment, if the electrical fluctuation of the PV string is abnormal, the humidity is too high, and the surface of the component is close to the critical point of condensation, then the system is in a highly sensitive state, which can easily lead to the formation of potential leakage paths, thereby causing ground faults, arc discharges or potential imbalance.

[0034] The following are some examples: Assume that the parameters are as shown in Table 1:

[0035] Table 1 According to Table 1, calculate the potential leakage probability : ; The specific steps of S3 also include: S32, set time sampling window { , t}, and according to the photovoltaic string operation data set S, obtain the time sampling window { , the potential leakage probability at each data acquisition time point within t} The sliding integral method is used to analyze the degree of degradation of the electrical insulation performance of the risky PV strings to obtain the leakage risk accumulation index LRAI. The specific method for obtaining the leakage risk accumulation index LRAI is as follows: ; Where, Indicates the risk of PV strings at the data collection time point The potential leakage probability when represents the integration variable, Indicates the length of the historical time window, represents the attenuation factor, where the attenuation factor The specific value of is given by .

[0036] During the degradation of the insulation performance of photovoltaic modules, the degradation of electrical insulation performance does not occur instantaneously, but is a long-term gradual degradation process affected by environmental loads, module aging, and micro-leakage accumulation. Therefore, the evaluation of the current insulation degradation degree of a photovoltaic string should not rely solely on the potential leakage probability at a certain point in time. , but it is necessary to comprehensively consider whether it continues to present a high-risk leakage situation over a period of time.

[0037] Formula derivation process and physical meaning: The formula is at any current time point t, looking back at the historical time window { , t}, and the potential leakage probability of each historical time point is calculated based on the time interval t- The physical attenuation effects are sorted and accumulated to obtain the past period of time The potential leakage accumulation level of It is an exponential decay function. During the operation and maintenance of photovoltaic strings, insulation performance degradation does not occur suddenly, but develops gradually with the continuous accumulation of environmental pressures such as humidity, thermal stress and electrical breakdown trends. However, the potential anomalies occurring at different historical moments have uneven impacts on the current state. Therefore, an exponential decay function is needed to represent the physical evolution law that the influence of historical leakage probability on the current moment decreases with time. Through the exponential decay function, the potential anomaly behavior closer to the current moment in the historical time window can be evaluated, thereby forming a time integral indicator that more truly reflects the insulation degradation trend of the string, avoiding misjudgment due to long-term interference from distant disturbances.

[0038] In the embodiment, by introducing the potential leakage probability and sliding integration mechanism, a dynamic quantitative assessment of the electrical insulation performance of risky photovoltaic strings is achieved. This method not only combines easily overlooked factors such as ambient humidity and condensation trend, but also sets time windows and attenuation factors to conduct trend accumulation analysis on the risk evolution process in different time periods, effectively depicting the cumulative change path of potential hidden dangers. Compared with the existing method that relies solely on manual inspections or single-point resistance measurements, this method has stronger continuity, sensitivity and early warning, and can identify early signs of potential leakage risks before there are any obvious abnormalities on the system surface, significantly improving the active identification and operation and maintenance response capabilities of "structural hidden degradation", and providing core support for improving the intelligent safety protection capabilities of distributed photovoltaic power stations in complex climate and aging environments.

[0039] Example 5 Please refer to Figure 1 and Figure 4 ,Specifically: S4 specific steps include, S41. Preset the first leakage risk threshold and the second leakage risk threshold and the leakage risk accumulation index LRAI and the first leakage risk threshold and the second leakage risk threshold A comparative analysis is conducted to assess the risk level of risky PV strings. The specific assessment contents are as follows: If the leakage risk accumulation index LRAI is less than or equal to the first leakage risk threshold , the risk level of the risky PV string is judged to be the first risk level. At this time, the normal operation mode of the PV power station is maintained, and the data monitoring frequency is maintained at normal level; If the leakage risk accumulation index LRAI is greater than the first leakage risk threshold , and is less than or equal to the second leakage risk threshold , the risk level of the risky PV string is determined to be the second risk level. At this time, there is abnormal fluctuation in the risky PV string, which needs to trigger the early warning mechanism, generate an early warning tag, and send the early warning tag to the safety monitoring platform. At the same time, the data monitoring frequency is adjusted from once every 30 minutes to once every 10 minutes. If the leakage risk accumulation index LRAI is greater than or equal to the second leakage risk threshold , the risk level of the risky PV string is determined to be the third risk level. At this time, a PV string isolation instruction is immediately generated, and the isolation switch corresponding to the PV string is closed to disconnect the PV string from the main circuit. An isolation log and an execution report are generated and continuously sent to the safety monitoring platform. An alarm is issued until the safety monitoring platform administrator responds.

[0040] In the embodiment, by setting a two-level leakage risk threshold and constructing a three-stage risk level division mechanism, a fine-grained dynamic assessment of the insulation degradation status of photovoltaic strings is achieved. Compared with the traditional method of relying only on single-point insulation resistance detection, it can identify "trend degradation" earlier rather than just responding to "faults that have occurred", especially for chronic hidden dangers such as microcracks, moisture, and packaging aging. It has significant advantages. According to the assessment results, the response strategy is automatically matched, such as dynamically adjusting the monitoring frequency, generating early warning labels, triggering isolation instructions, etc., to form a closed-loop link from identification to disposal, thereby improving the safety response capability of distributed photovoltaic power stations in complex climates, avoiding the spread of accidents caused by manual delayed response or untimely judgment, taking into account both early warning accuracy and operational stability, and solving the problem of "not being able to see, measure, or predict" mentioned in the background technology, and enhancing the monitoring granularity and control initiative of the insulation degradation process.

[0041] Example 6 Please refer to Figure 1 ,Specifically: S5 specific steps include, S51, gather risky photovoltaic strings at the first risk level and the second risk level, construct a risky photovoltaic string set, extract the leakage risk accumulation index LRAI of each risky photovoltaic string in the risky photovoltaic string set, and perform data averaging to obtain the average value of the leakage risk accumulation index ; S52. Calculate the leakage risk accumulation index LRAI and the average value of the leakage risk accumulation index of each risky PV string in the risky PV string set and divide the difference by the average value of the leakage risk accumulation index The sum of the non-zero small constant ε is used to obtain the trend deviation rate of each risky PV string. , where the trend deviation rate Refers to the leakage risk accumulation index LRAI of the risky PV string and the average value of the leakage risk accumulation index relative deviation.

[0042] The specific steps of S5 also include: S53, based on the trend deviation rate of each risky PV string and the average value of leakage risk accumulation index , use the variance calculation formula to obtain the trend deviation rate of the risk photovoltaic string set Variance, that is, the trend clustering factor STCI; By calculating the trend deviation rate of each risk photovoltaic string Variance can reflect the synchronization and deviation of trend performance between risky photovoltaic strings, thereby achieving quantitative characterization of trend aggregation. The smaller the variance, the more concentrated the trend performance, and the higher the spatial synchronization risk characteristics. It is suitable for subsequent cluster identification and risk attribution analysis, ensuring that this indicator has operability and discriminant value in engineering judgment and dynamic evolution modeling.

[0043] S54. Extract the historical photovoltaic string operation data of the risk photovoltaic string set to continuously calculate the trend aggregation factor STCI of the risk photovoltaic string set over a period of time, construct the trend aggregation factor time series, and obtain the historical mean and historical standard deviation of the trend aggregation factor STCI based on the trend aggregation factor time series, and obtain the dynamic trend risk threshold through the standard deviation boundary model. ; The trend aggregation factor time series refers to the set sequence of trend aggregation factors STCI of the risk photovoltaic string set extracted in time order within continuous time; The standard deviation boundary model is a dynamic threshold setting model based on statistical distribution laws. Its core idea is to construct upper and lower boundaries reflecting the normal fluctuation range by calculating the mean and standard deviation of historical samples in the trend aggregation factor time series, and thus define the dynamic threshold range for risk judgment. S55. Dynamic trend risk threshold A comparative analysis is conducted with the trend clustering factor (STCI) to assess the risk of degradation trend fluctuations among risky PV strings. The specific assessment contents are as follows: If the trend concentration factor STCI is greater than or equal to the dynamic trend risk threshold , it is determined that the degradation trend fluctuations between risky PV strings are in an abnormal range, and there is a phenomenon of synchronous enhancement of degradation trends between risky PV strings, and there is a spatial clustering risk. At this time, the trend deviation rate in the risky PV string set is screened out. Greater than 0 and higher than the trend deviation rate at multiple consecutive time points The system then determines the risky PV strings based on the average value and, based on the PV power station equipment distribution map, determines whether the risky PV strings belong to a physically adjacent area. If so, it indicates that the risky PV strings in the physically adjacent area have a regional potential leakage risk, immediately triggering an early warning mechanism, generating a manual maintenance task, and executing a voltage reduction operation strategy. If not, no action is required. The current fluctuation, voltage offset, and increased leakage probability of a single PV string may be just occasional fluctuations or the aging of individual equipment. However, when multiple geographically adjacent PV strings simultaneously show consistent trend deviations in a similar time period, such as increased leakage probability and voltage drop, it may indicate that there are environmental coupling risks such as systemic thermal imbalance, aging of the same batch of components, and regional condensation in the system. In a specific example, in a distributed PV power station, from early morning to 10 a.m. on a certain day, the collected data showed that the trend aggregation factors STCI of 30 PV strings in area A increased simultaneously, from 0.03 to 0.09. At this time, the voltage curve fluctuated slightly, but none of them reached the single-point alarm threshold. The system will automatically track the trend deviation rate BIAS of these PV strings in time series and calculate the trend aggregation factor STCI within the time window. When the trend aggregation factor STCI exceeds the dynamic trend risk threshold When the PV strings in Area A are judged to be "trending towards synchronous degradation", it is determined to be a potential condensation-type group risk. Subsequently, it is identified that the area has a large temperature difference between warm and cold at night, the box seals are aging, and there is repeated moisture retention. In this case, the ventilation device is forced to open to remove moisture, and an early warning is issued to notify maintenance personnel to perform manual maintenance. If the trend concentration factor STCI is less than the dynamic trend risk threshold , it is determined that the degradation trend fluctuation between risky PV strings is within the normal range and no processing is required.

[0044] In an embodiment, by constructing the trend deviation rate Together with the trend aggregation factor STCI, it effectively identifies the hidden spatial aggregation phenomenon caused by the "enhanced trend consistency" between risk strings in photovoltaic power stations, making up for the defect that traditional insulation risk assessment only focuses on single-point status and is difficult to discover systematic degradation coordination. Compared with the existing method that relies on regular inspections and manual inspections, this solution can realize intelligent identification driven by trend evolution, automatically trigger regional hidden danger warnings, improve early perception capabilities and response timeliness, and at the same time, introduce equipment distribution information to judge "physical proximity", improve strategy accuracy, effectively avoid false alarms or mishandling, and significantly enhance the system's dynamic tracking capabilities for hidden degradation paths such as component aging, water vapor erosion and potential unevenness, promote the transformation of photovoltaic power stations from "regular inspections" to "trend-driven active prevention and control", and ensure the operation safety and maintenance economy of large-scale power stations.

[0045] Example 7 Please refer to Figure 2 ,Specifically: A distributed photovoltaic power station intelligent safety operation and maintenance management system, including a data acquisition and feature extraction module, a disturbance analysis module, a potential leakage probability calculation module, a risk assessment module and a spatial aggregation analysis module; The data acquisition and feature extraction module is used to deploy an intelligent sensor group on the photovoltaic strings of the distributed photovoltaic power station, and use the intelligent sensor group to collect photovoltaic string operation data in real time to construct a photovoltaic string operation data set S, and extract feature parameters based on the photovoltaic string operation data set S, and construct a feature parameter set F based on the extracted feature parameters; The disturbance analysis module is used to perform characteristic parameter coupling calculation based on the characteristic parameter set F to obtain the physical disturbance risk score. , and the physical disturbance risk score and physical disturbance risk thresholds Conduct comparative analysis to identify risky PV strings

[0046] The potential leakage probability calculation module is used for risky photovoltaic strings, based on the photovoltaic string operation data set S, characteristic parameter set F and the physical disturbance risk score of the risky photovoltaic strings. , perform coupled calculation of physical disturbance and humidity parameters to obtain the potential leakage probability , and further analyze the degree of electrical insulation degradation of risky PV strings to obtain the leakage risk accumulation index LRAI; The risk assessment module is used to compare the leakage risk accumulation index LRAI with the first leakage risk threshold and the second leakage risk threshold Conduct comparative analysis to assess the risk level of risky PV strings; The spatial aggregation analysis module is used to construct a risky photovoltaic string set based on the risk level assessment results, and perform trend aggregation analysis on the risky photovoltaic string set to obtain the trend aggregation factor (STCI), identify spatial aggregation risks, and issue early warnings.

[0047] In the embodiment, by constructing a multi-module linked intelligent operation and maintenance system, in order to address the problem that traditional distributed photovoltaic power stations are difficult to identify insulation degradation trends and monitor regional hidden danger aggregation in real time, a closed-loop technical path from "data collection - risk scoring - trend identification - level assessment - aggregation judgment" is proposed. The dynamic quantification of potential leakage trends at the component level, spatial identification of trend evolution and graded response are achieved. The system does not rely on traditional intermittent detection methods and has continuous adaptive assessment capabilities, which significantly improves the overall risk perception granularity of photovoltaic power stations and the timeliness and accuracy of hidden danger disposal, providing effective support for the construction of a more forward-looking intelligent safety assurance system for distributed photovoltaic power stations.

[0048] While embodiments of the present invention have been shown and described, it will be appreciated by those skilled in the art that various changes, modifications, substitutions, and variations may be made to these embodiments without departing from the principles and spirit of the invention, and that the scope of the invention is defined by the appended claims and their equivalents.

Claims

1. A method for intelligent and safe operation and maintenance management of a distributed photovoltaic power station, characterized by: The following steps are included: S1. Deploy smart sensor groups on photovoltaic strings of a distributed photovoltaic power station and use the smart sensor groups to collect photovoltaic string operation data in real time to construct a photovoltaic string operation data set S. Extract characteristic parameters based on the photovoltaic string operation data set S, and construct a characteristic parameter set F based on the extracted characteristic parameters. S2. Based on the characteristic parameter set F, perform characteristic parameter coupling calculation to obtain the physical disturbance risk score , and the physical disturbance risk score and physical disturbance risk thresholds Conduct comparative analysis to identify risky PV strings; S3. For risky PV strings, the risk score of the physical disturbance of the risky PV strings is calculated based on the PV string operation data set S, the characteristic parameter set F and the risky PV strings. , perform coupled calculation of physical disturbance and humidity parameters to obtain the potential leakage probability , and further analyze the degree of electrical insulation degradation of risky PV strings to obtain the leakage risk accumulation index LRAI; S4. Compare the leakage risk accumulation index LRAI with the first leakage risk threshold and the second leakage risk threshold Conduct comparative analysis to assess the risk level of risky PV strings; S5. Based on the risk level assessment results, construct a risky PV string set and perform trend aggregation analysis on the risky PV string set to obtain the trend aggregation factor (STCI), identify spatial aggregation risks, and issue early warnings.

2. The method for intelligent and safe operation and maintenance of a distributed photovoltaic power station according to claim 1, characterized in that: The specific steps of S1 include: S11, using each photovoltaic string of the distributed photovoltaic power station as a data acquisition unit, deploying an intelligent sensor group on each photovoltaic string to obtain photovoltaic string operation data, wherein the photovoltaic string operation data includes the photovoltaic string output current value I, voltage V, photovoltaic module surface temperature Tsurf, air temperature Tair and relative humidity H; A photovoltaic string refers to the smallest power generation and operation unit constructed by connecting multiple photovoltaic modules in a DC series manner; The intelligent sensor group includes a combiner box current transformer, a combiner box voltage divider, a thermocouple temperature sensor, and a humidity sensor; S12. Preprocessing the collected photovoltaic string operation data, including data cleaning, data smoothing and normalization, and constructing a photovoltaic string operation data set S based on the preprocessed photovoltaic string operation data.

3. The method for intelligent and safe operation and maintenance management of a distributed photovoltaic power station according to claim 2, characterized in that: The specific steps of S1 also include: S13. Extract characteristic parameters based on the photovoltaic string operation data set S to obtain a current fluctuation index DL, a voltage edge drift index Dv, a capacitive coupling residual index Dr, and a condensation risk index JL; The current fluctuation index DL refers to the volatility of the PV string output current value. By taking the difference between the PV string output current values ​​I at adjacent time points, the current change rate per unit time is obtained, a current change rate time series is constructed, and the standard deviation of the current change rate time series is calculated as the current fluctuation index. The voltage edge drift index Dv refers to the deviation of the waveform slope of the rising or falling edge of the PV string voltage waveform relative to the standard PV string voltage waveform; The capacitive coupling residual index Dr refers to the total amount of fitting residuals between the actual recovery curve of the PV string and the standard RC model when the voltage recovers from a low level to a steady state during the operation of the PV string; The condensation risk index JL refers to the relative critical proximity between the surface temperature Tsurf of the PV module and the air dew point temperature, where the air dew point temperature refers to the critical temperature at which water vapor in the air condenses into liquid water; S14. Construct a characteristic parameter set F based on the obtained current fluctuation index DL, voltage edge drift index Dv, capacitive coupling residual index Dr, and condensation risk index JL.

4. The method for intelligent and safe operation and maintenance of a distributed photovoltaic power station according to claim 3, characterized in that: The specific steps of S2 include: S21. Perform dimensionless processing on the characteristic parameters in the characteristic parameter set F to eliminate the physical dimension, and analyze the risk intensity level of the photovoltaic string under the condition of coupling multiple characteristic parameters based on the dimensionless processed characteristic parameters to obtain the physical disturbance risk score. , where the physical disturbance risk score The specific way to obtain it is: ； Where, 、 and They represent the current fluctuation index DL, voltage edge drift index Dv and capacitive coupling residual index Dr of the PV string at the current time point t, respectively, and ln represents the natural logarithmic constant; S22. Collect the operating data of several photovoltaic strings when they are in a healthy state, so as to calculate the physical disturbance risk scores of several photovoltaic strings when they are in a healthy state. , and score the risk of all physical disturbances in a healthy state Perform statistical distribution analysis and select the 95th percentile physical disturbance risk score from the statistical distribution analysis results As a physical disturbance risk threshold ; S23. Physical disturbance risk threshold and physical disturbance risk scoring Conduct comparative analysis to assess the disturbance risk status of PV strings. If the physical disturbance risk score Greater than or equal to the physical disturbance risk threshold , then the PV string is judged to be in an abnormally enhanced state, and the PV string is marked as a risky PV string. If the physical disturbance risk score Less than the physical disturbance risk threshold , it is determined that the PV string is in a normal state and no processing is required.

5. The method for intelligent and safe operation and maintenance of a distributed photovoltaic power station according to claim 4, characterized in that: The specific steps of S3 include: S31. For the marked risky PV strings, based on the PV string operation data set S, characteristic parameter set F and physical disturbance risk score , comprehensively analyze the probability of potential leakage of risk PV strings under the influence of current physical disturbance and environmental humidity to obtain the potential leakage probability , where the potential leakage probability The specific manifestations are: ； Where, represents the leakage trend coupling coefficient, exp represents the exponential function, represents the relative humidity of the risky PV string at the current time point t, Indicates the condensation risk index of the risky PV string at the current time point t.

6. A distributed photovoltaic power station intelligent safety operation and maintenance management method according to claim 5, characterized in that: The specific steps of S3 also include: S32, set time sampling window { , t}, and according to the photovoltaic string operation data set S, obtain the time sampling window { , the potential leakage probability at each data acquisition time point within t} The sliding integral method is used to analyze the degree of degradation of the electrical insulation performance of the risky PV strings to obtain the leakage risk accumulation index LRAI. The specific method for obtaining the leakage risk accumulation index LRAI is as follows: ； Where, Indicates the risk of PV strings at the data collection time point The potential leakage probability when represents the integration variable, Indicates the length of the historical time window, Represents the attenuation factor.

7. The method for intelligent and safe operation and maintenance of a distributed photovoltaic power station according to claim 6, characterized in that: The specific steps of S4 include: S41. Preset the first leakage risk threshold and the second leakage risk threshold and the leakage risk accumulation index LRAI and the first leakage risk threshold and the second leakage risk threshold A comparative analysis is conducted to assess the risk level of risky PV strings. The specific assessment contents are as follows: If the leakage risk accumulation index LRAI is less than or equal to the first leakage risk threshold , the risk level of the risky PV string is judged to be the first risk level. At this time, the normal operation mode of the PV power station is maintained, and the data monitoring frequency is maintained at normal level; If the leakage risk accumulation index LRAI is greater than the first leakage risk threshold , and is less than or equal to the second leakage risk threshold , the risk level of the risky PV string is determined to be the second risk level. At this time, there is abnormal fluctuation in the risky PV string, which needs to trigger the early warning mechanism, generate an early warning tag, and send the early warning tag to the safety monitoring platform. At the same time, the data monitoring frequency is adjusted from once every 30 minutes to once every 10 minutes. If the leakage risk accumulation index LRAI is greater than or equal to the second leakage risk threshold , the risk level of the risky PV string is determined to be the third risk level. At this time, a PV string isolation instruction is immediately generated, and the isolation switch corresponding to the PV string is closed to disconnect the PV string from the main circuit. An isolation log and an execution report are generated and continuously sent to the safety monitoring platform. An alarm is issued until the safety monitoring platform administrator responds.

8. The method for intelligent and safe operation and maintenance of a distributed photovoltaic power station according to claim 7, characterized in that: The specific steps of S5 include: S51, gather risky photovoltaic strings at the first risk level and the second risk level, construct a risky photovoltaic string set, extract the leakage risk accumulation index LRAI of each risky photovoltaic string in the risky photovoltaic string set, and perform data averaging to obtain the average value of the leakage risk accumulation index ; S52. Calculate the leakage risk accumulation index LRAI and the average value of the leakage risk accumulation index of each risky PV string in the risky PV string set and divide the difference by the average value of the leakage risk accumulation index The sum of the trend deviation rate BIAS of each risky PV string is obtained by adding the non-zero small constant ε.

9. The method for intelligent and safe operation and maintenance of a distributed photovoltaic power station according to claim 8, characterized in that: The specific steps of S5 also include: S53, based on the trend deviation rate of each risky PV string and the average value of leakage risk accumulation index , use the variance calculation formula to obtain the trend deviation rate of the risk photovoltaic string set Variance, that is, the trend clustering factor STCI; S54. Extract the historical photovoltaic string operation data of the risk photovoltaic string set to continuously calculate the trend aggregation factor STCI of the risk photovoltaic string set over a period of time, construct the trend aggregation factor time series, and obtain the historical mean and historical standard deviation of the trend aggregation factor STCI based on the trend aggregation factor time series, and obtain the dynamic trend risk threshold through the standard deviation boundary model. ; S55. Dynamic trend risk threshold A comparative analysis is conducted with the trend clustering factor (STCI) to assess the risk of degradation trend fluctuations among risky PV strings. The specific assessment contents are as follows: If the trend concentration factor STCI is greater than or equal to the dynamic trend risk threshold , it is determined that the degradation trend fluctuations between risky PV strings are in an abnormal range, and there is a phenomenon of synchronous enhancement of degradation trends between risky PV strings, and there is a spatial clustering risk. At this time, the trend deviation rate in the risky PV string set is screened out. Greater than 0 and higher than the trend deviation rate at multiple consecutive time points The system then determines the risky PV strings based on the average value and, based on the PV power station equipment distribution map, determines whether the risky PV strings belong to a physically adjacent area. If so, it indicates that the risky PV strings in the physically adjacent area have a regional potential leakage risk, immediately triggering an early warning mechanism, generating a manual maintenance task, and executing a voltage reduction operation strategy. If not, no action is required. If the trend concentration factor STCI is less than the dynamic trend risk threshold , it is determined that the degradation trend fluctuation between risky PV strings is within the normal range and no processing is required.

10. A distributed photovoltaic power station intelligent safety operation and maintenance management system, used to implement the distributed photovoltaic power station intelligent safety operation and maintenance management method according to any one of claims 1 to 9, characterized in that: It includes data acquisition and feature extraction module, disturbance analysis module, potential leakage probability calculation module, risk assessment module and spatial aggregation analysis module; The data acquisition and feature extraction module is used to deploy an intelligent sensor group on the photovoltaic strings of the distributed photovoltaic power station, and use the intelligent sensor group to collect photovoltaic string operation data in real time to construct a photovoltaic string operation data set S, and extract feature parameters based on the photovoltaic string operation data set S, and construct a feature parameter set F based on the extracted feature parameters; The disturbance analysis module is used to perform characteristic parameter coupling calculation based on the characteristic parameter set F to obtain the physical disturbance risk score. , and the physical disturbance risk score and physical disturbance risk thresholds Conduct comparative analysis to identify risky PV strings; The potential leakage probability calculation module is used for risky photovoltaic strings, based on the photovoltaic string operation data set S, characteristic parameter set F and the physical disturbance risk score of the risky photovoltaic strings. , perform coupled calculation of physical disturbance and humidity parameters to obtain the potential leakage probability , and further analyze the degree of electrical insulation degradation of risky PV strings to obtain the leakage risk accumulation index LRAI; The risk assessment module is used to compare the leakage risk accumulation index LRAI with the first leakage risk threshold and the second leakage risk threshold Conduct comparative analysis to assess the risk level of risky PV strings; The spatial aggregation analysis module is used to construct a risky photovoltaic string set based on the risk level assessment results, and perform trend aggregation analysis on the risky photovoltaic string set to obtain the trend aggregation factor (STCI), identify spatial aggregation risks, and issue early warnings.

Images