Insulation resistance full-chain detection system and method for fish-light complementary photovoltaic power station

Abstract

The application discloses a full-chain insulation resistance detection system and method for fish-light complementary photovoltaic power stations, and relates to the technical field of insulation detection of photovoltaic power stations. Insulation detection units are arranged at each bus box, and the positive and negative parallel equivalent insulation resistances of each bus box are calculated. When the corrected insulation resistance is lower than a preset proportion of the lower limit of the early warning or shows a continuous downward trend, the corresponding bus box is marked as a suspicious abnormality. For the suspicious abnormality bus box and each group of strings under the jurisdiction of the bus box, a decay vector including a decay rate, a correlation between adjacent points and a diurnal fluctuation amplitude is constructed, and the Euclidean distance between the bus box decay vector and the decay vectors of each group of strings is calculated. Through the full-chain layered detection and intelligent analysis mechanism, the application effectively adapts to the water area operation environment of the fish-light complementary photovoltaic power station, and greatly improves the accuracy and comprehensiveness of insulation detection.

Description

Technical Field

[0001] This invention relates to the field of insulation testing technology for photovoltaic power plants, and in particular to a full-chain testing system and method for insulation resistance of solar-fishery complementary photovoltaic power plants. Background Technology

[0002] Photovoltaic power stations that integrate fisheries and solar power are built in water areas. The photovoltaic modules, connecting cables, and combiner equipment are constantly exposed to a complex outdoor environment with high humidity and dynamic water level changes. Compared to conventional onshore photovoltaic power stations, the insulation performance of the equipment is more prone to degradation. Factors such as water vapor erosion, water level fluctuations, and alternating day and night temperature and humidity changes can continuously cause damage to the insulation layer of photovoltaic cables, aging of module backsheets, and decline in node insulation performance. These factors can easily lead to DC leakage, equipment tripping, module burnout, and even power station safety accidents, directly affecting the stability and safety of the photovoltaic power station's operation.

[0003] Currently, the photovoltaic insulation testing methods in the industry are relatively simple, mostly involving fixed-period single-point insulation resistance sampling and testing, which can only achieve simple early warning of insulation exceeding the standard and has many technical defects.

[0004] First, the test data was not corrected for key influencing factors such as ambient temperature and humidity and water level, resulting in large errors in the test results. It was impossible to distinguish between insulation value fluctuations caused by environmental interference and actual insulation deterioration of the equipment, leading to a high probability of false alarms and missed alarms.

[0005] Secondly, existing detection technologies can only provide overall insulation anomaly alerts, but cannot accurately distinguish whether the fault occurs inside the string, inside the combiner box, or in the common connection node. The fault location accuracy is low, making it difficult to support precise operation and maintenance.

[0006] Third, it cannot effectively identify the type of insulation fault, and cannot distinguish between cable-to-ground insulation degradation and module backsheet insulation aging faults, resulting in insufficient targeted operation and maintenance. Fourth, the early warning mechanism has not been optimized in conjunction with the seasonal water level change characteristics of the fishery-solar hybrid power station's water area. During the high-water season, when environmental interference is strong, the fixed early warning threshold has poor adaptability, further reducing the reliability of detection and early warning. At the same time, existing technologies lack a full-chain data perception, intelligent analysis, dynamic control, and operation and maintenance feedback optimization system, resulting in low levels of intelligent and systematic detection, which cannot meet the detection requirements for the long-term safe and stable operation of fishery-solar hybrid photovoltaic power stations. Summary of the Invention

[0007] To address the aforementioned technical problems, this invention provides a full-chain detection system and method for insulation resistance in a solar-fishery hybrid photovoltaic power station. The technical solution adopted is as follows: The method for full-chain testing of insulation resistance in a solar-fishery hybrid photovoltaic power station includes the following steps: Step 1: Deploy insulation detection units at each junction box and collect data on positive-to-ground insulation resistance, negative-to-ground insulation resistance, leakage current, ambient temperature, ambient humidity, and water level according to the first sampling cycle. Step 2: Calculate the equivalent insulation resistance of the positive and negative terminals of each combiner box in parallel. After temperature and humidity normalization correction, establish a dynamic baseline and warning lower limit for the corrected insulation resistance based on the sliding time window. Step 3: When the corrected insulation resistance is lower than the preset ratio of the lower warning limit or shows a continuous downward trend, the corresponding combiner box is marked as an abnormality suspect, and the sampling period of the abnormality suspect combiner box and adjacent combiner boxes is automatically increased to the second sampling period. Step 4: For the suspected abnormal combiner box and each of its subordinate strings, construct attenuation vectors that include attenuation rate, correlation between adjacent points and day and night fluctuation amplitude. Calculate the Euclidean distance between the combiner box attenuation vector and the attenuation vector of each string. Based on the distance and the similarity between strings, determine whether the fault is located inside the string, inside the combiner box, or at the T-connected common node. Step 5: Simultaneously collect the leakage current and DC bus voltage ripple of the suspected abnormal combiner box, extract the phase difference of the inverter switching frequency band component, and determine the fault as a decrease in cable insulation to ground or aging of component backplane insulation based on the phase difference range. Step 6: Establish the relationship curve between the corrected insulation resistance and water level at each detection point, distinguish between reversible fluctuations caused by water level changes and insulation degradation trends, and automatically adjust the early warning threshold before the high water season.

[0008] Optionally, in step 2, the equivalent insulation resistance of the positive and negative poles in parallel is... Calculate using the following formula: ; in , These are the measured values ​​of the positive and negative insulation resistances of the i-th junction box at time t.

[0009] Optionally, in step 3, the abnormality suspicion marking conditions are: the current corrected insulation resistance is lower than 50% of the warning lower limit, or the corrected insulation resistance of 10 consecutive sampling points shows a monotonically decreasing trend and the cumulative decrease exceeds 30%; The second sampling period is 10 seconds, and the first sampling period is resumed after the anomaly is eliminated. The condition for eliminating the anomaly is that the corrected insulation resistance is restored to above the lower limit of the warning and the trend remains stable for 1 hour.

[0010] Optionally, in step 4, the attenuation vector , where k is the slope of the linear fit of the corrected insulation resistance sequence over the past 24 hours. The Pearson correlation coefficient of the insulation resistance sequence between the detection point and adjacent upstream and downstream detection points is corrected; It is the absolute value of the difference between the average corrected insulation resistance during the daytime and the average corrected insulation resistance during the nighttime.

[0011] Optional, combiner box attenuation vector With string decay vector Euclidean distance for: ; The fault location determination method is as follows: if there exists a string j that satisfies... Less than the preset similarity threshold And the decay rate of the string If the rate of decay is greater than η times the average decay rate of all other strings, where η is a preset multiplier and η > 1, then the fault is determined to be located within the corresponding string; if the rate of decay of all strings is greater than η times the average decay rate of all other strings, then the fault is determined to be located within the corresponding string. All greater than The fault is determined to be located inside the combiner box or in a common cable section; if the Euclidean distance between the attenuation vectors of two adjacent strings is less than... And the distances between each and the attenuation vector of the combiner box are all greater than The fault was determined to be located at the common node of the T-connection of the two strings.

[0012] Optionally, in step 5, the phase difference between the leakage current and the DC bus voltage ripple... Calculate using the following formula: ; and These are the sampling sequences of leakage current and DC bus voltage after bandpass filtering, respectively. for The conjugate of , N is the number of sampling points, and the passband of the bandpass filter is taken as the twice harmonic component of the inverter switching frequency; The fault nature determination method is as follows: when |φ|<30°, it is determined that the cable insulation to ground has deteriorated; when |φ|>60°, it is determined that the component backplane insulation has aged; when 30°≤|φ|≤60°, it is determined that it is a mixed fault.

[0013] Optionally, the modified insulation resistance versus water level curve is as follows: ; in To correct the insulation resistance for the weekly average, The weekly average water level The lowest water level within the statistical period. This is the insulation resistance value corresponding to the lowest water level. The water level influence coefficient is obtained through least squares fitting. The condition for reversible fluctuation is: the absolute value of the difference between the current corrected insulation resistance and the expected value calculated based on the current water level according to the relationship curve is less than the preset tolerance ε. The method for adjusting the early warning threshold before the high water season is as follows: lower the lower limit of the warning by δ%, with δ set at 10%-20%.

[0014] Optionally, the leakage current range acquired by the insulation detection unit is 0mA-500mA; the sampling rate of the DC bus voltage ripple is not less than 1kHz, and the continuous acquisition time is not less than 1 minute; the passband of the bandpass filter is 100Hz-300Hz.

[0015] Optionally, a maintenance result feedback step is also included: comparing the actual fault type and location confirmed on-site with the judgment results of the steps, calculating the judgment accuracy, and optimizing the attenuation vector similarity threshold accordingly. And phase difference discrimination threshold.

[0016] The insulation resistance full-chain detection system for fishery-solar complementary photovoltaic power stations is used to realize the full-chain detection method of insulation resistance in fishery-solar complementary photovoltaic power stations. The system includes a data perception layer, a data aggregation layer, an analysis and decision-making layer, and an execution feedback layer. The data sensing layer consists of insulation detection units deployed at each combiner box. Each detection unit integrates a DC unbalanced bridge insulation resistance detection module, a leakage current acquisition module, a temperature and humidity sensor, a water level sensor, and a communication module. It is used to collect data on the positive-to-ground insulation resistance, the negative-to-ground insulation resistance, leakage current, ambient temperature, ambient humidity, and water level. The data aggregation layer includes local data concentrators, which are connected to each detection unit and to the power plant monitoring backend, for summarizing and forwarding detection data; The analysis and decision-making layer is deployed on the power plant monitoring backend or cloud platform to perform the following steps: The equivalent insulation resistance of the positive and negative poles in parallel is calculated and normalized for temperature and humidity. A dynamic baseline and early warning lower limit are established based on a sliding window. The sampling period is switched according to the abnormal suspicion marking conditions. The attenuation vector of the combiner box and the string is constructed and the Euclidean distance is calculated to determine the fault location. The phase difference between the leakage current and the DC bus voltage ripple is analyzed simultaneously to determine the fault nature. A correlation curve between water level and insulation resistance is established to distinguish between reversible fluctuations and insulation degradation, and seasonal early warning threshold adjustments are performed. The execution feedback layer is a mobile operation and maintenance terminal used to receive fault location and fault nature judgment results and generate operation and maintenance work orders, as well as to send back maintenance results to optimize the judgment threshold in the analysis and decision layer.

[0017] In summary, the present invention has at least one of the following beneficial technical effects: This invention provides a full-chain insulation resistance testing system and method for solar-aquaculture hybrid power plants. Through a multi-layered testing and intelligent analysis mechanism, it effectively adapts to the aquatic operating environment of solar-aquaculture hybrid power plants, significantly improving the accuracy and comprehensiveness of insulation testing. By collecting multi-dimensional environmental parameters and applying temperature and humidity normalization corrections, combined with a sliding time window to construct a dynamic insulation baseline, it avoids interference from environmental factors on the test results, effectively distinguishes between reversible numerical changes caused by water level fluctuations and continuous insulation degradation of equipment, and reduces the probability of false positives.

[0018] Through an adaptive sampling cycle switching mechanism under abnormal conditions, low-power monitoring under normal conditions and refined data acquisition under abnormal conditions are achieved, balancing detection efficiency and equipment operating energy consumption. Relying on attenuation vector and Euclidean distance comparison algorithms, the specific location of insulation faults can be accurately pinpointed, enabling hierarchical fault location at the string, combiner box, and common node levels. Combined with leakage current and bus voltage ripple phase difference analysis, the specific type of insulation fault can be effectively identified, providing accurate data for operation and maintenance.

[0019] By establishing a correlation model between water level and insulation resistance and implementing adaptive adjustment of thresholds during the high-water season, the adaptability of the early warning mechanism in complex aquatic environments is improved. Simultaneously, by relying on feedback from operation and maintenance results to optimize discrimination parameters, continuous iterative optimization of the detection model is achieved, gradually improving the accuracy of fault identification. The entire technology forms a closed-loop system of perception, analysis, early warning, location, operation and maintenance, and optimization, effectively reducing the safety risks caused by insulation faults in solar-fishery complementary power plants, reducing operation and maintenance costs, and improving the overall operational stability and intelligent operation and maintenance level of the power plant. Attached Figure Description

[0020] Figure 1 This is a flowchart illustrating the whole-chain detection method for insulation resistance of a solar-fishery complementary photovoltaic power station according to the present invention. Figure 2 This is a schematic diagram of the relationship between water level and modified insulation resistance in a specific embodiment of the present invention; Figure 3 This is a schematic diagram of the architecture of the insulation resistance full-chain detection system for the solar-fishery complementary photovoltaic power station of the present invention. Detailed Implementation

[0021] The present invention will be further described in detail below with reference to the accompanying drawings.

[0022] This invention discloses a full-chain detection system and method for insulation resistance of a solar-fishery complementary photovoltaic power station.

[0023] Reference Figures 1-3 Example 1: A method for full-chain testing of insulation resistance in a solar-fishery hybrid photovoltaic power station, comprising the following steps: Step 1: Deploy insulation detection units at each junction box and collect data on positive-to-ground insulation resistance, negative-to-ground insulation resistance, leakage current, ambient temperature, ambient humidity, and water level according to the first sampling cycle. Step 2: Calculate the equivalent insulation resistance of the positive and negative terminals of each combiner box in parallel. After temperature and humidity normalization correction, establish a dynamic baseline and warning lower limit for the corrected insulation resistance based on the sliding time window. Step 3: When the corrected insulation resistance is lower than the preset ratio of the lower warning limit or shows a continuous downward trend, the corresponding combiner box is marked as an abnormality suspect, and the sampling period of the abnormality suspect combiner box and adjacent combiner boxes is automatically increased to the second sampling period. Step 4: For the suspected abnormal combiner box and each of its subordinate strings, construct attenuation vectors that include attenuation rate, correlation between adjacent points and day and night fluctuation amplitude. Calculate the Euclidean distance between the combiner box attenuation vector and the attenuation vector of each string. Based on the distance and the similarity between strings, determine whether the fault is located inside the string, inside the combiner box, or at the T-connected common node. Step 5: Simultaneously collect the leakage current and DC bus voltage ripple of the suspected abnormal combiner box, extract the phase difference of the inverter switching frequency band component, and determine the fault as a decrease in cable insulation to ground or aging of component backplane insulation based on the phase difference range. Step 6: Establish the relationship curve between the corrected insulation resistance and water level at each detection point, distinguish between reversible fluctuations caused by water level changes and insulation degradation trends, and automatically adjust the early warning threshold before the high water season.

[0024] By adopting the above technical solution, insulation detection units are deployed at each combiner box to collect data on the positive-to-ground insulation resistance, negative-to-ground insulation resistance, leakage current, ambient temperature, ambient humidity, and water level according to the first sampling cycle. The insulation detection units continuously collect these parameters during the normal operation of the photovoltaic power station, forming multi-dimensional time-series data on the insulation status of each combiner box.

[0025] The equivalent insulation resistance of the positive and negative terminals of each combiner box is calculated. After temperature and humidity normalization correction, a dynamic baseline and warning lower limit for the corrected insulation resistance are established based on a sliding time window. This step eliminates the influence of temperature and humidity changes on the insulation resistance measurement by correcting for environmental factors and performing statistical processing on the original collected data, providing a normalized comparison benchmark for subsequent anomaly judgment.

[0026] When the corrected insulation resistance falls below the preset lower limit of the warning threshold or shows a continuous downward trend, the corresponding combiner box is marked as potentially abnormal, and the sampling period of the suspected abnormal combiner box and adjacent combiner boxes is automatically increased to the second sampling period. This step achieves dynamic adjustment of the detection frequency. Under normal conditions, a low sampling frequency is maintained to save communication resources and storage space, while under abnormal conditions, the sampling frequency is automatically increased to obtain more dense insulation resistance change data. At the same time, synchronous encrypted acquisition of adjacent combiner boxes is activated to provide upstream and downstream reference information for subsequent attenuation vector analysis.

[0027] For suspected faulty combiner boxes and their constituent strings, attenuation vectors are constructed, incorporating attenuation rate, correlation between adjacent points, and diurnal fluctuation amplitude. The Euclidean distance between the combiner box attenuation vector and the attenuation vectors of each string is calculated. Based on the distance and the similarity between strings, the fault is determined to be located within a string, inside the combiner box, or at a T-connector common node. This step utilizes the differentiated characteristics of insulation resistance attenuation behavior at different fault locations to achieve hierarchical localization: when a fault occurs within a string, only the attenuation vector of that string is highly similar to the combiner box attenuation vector, while other strings are dissimilar; when a fault occurs on the combiner box common bus, the attenuation vectors of all strings show significant differences from the combiner box attenuation vector; when a fault occurs at a T-connector node, the attenuation vectors of adjacent strings are highly similar to each other but dissimilar to the combiner box attenuation vector.

[0028] Simultaneously collect leakage current and DC bus voltage ripple from suspected abnormal combiner boxes, extract the phase difference of the inverter switching frequency band components, and determine whether the fault is due to decreased cable-to-ground insulation or aging of module backplane insulation based on the phase difference range. This step utilizes the inherent switching frequency ripple of the DC bus voltage during inverter operation as a natural excitation source, and distinguishes the nature of the fault by the phase relationship between the leakage current and voltage ripple on the fault impedance: decreased cable-to-ground insulation exhibits resistive or weak capacitive characteristics, and the leakage current and voltage ripple are nearly in phase; aging module backplane insulation exhibits strong capacitive characteristics, and the leakage current and voltage ripple are nearly orthogonal.

[0029] A modified insulation resistance-water level relationship curve is established for each monitoring point to distinguish between reversible fluctuations caused by water level changes and insulation degradation trends. The warning threshold is automatically adjusted before the high-water season. This step is based on the physical fact that there is a correlation between seasonal water level changes and insulation resistance measurements in the waters of the solar-aquaculture hybrid power station. By accumulating long-term data, the relationship curves between water level and insulation resistance at each monitoring point are fitted. Insulation resistance fluctuations caused by water level changes are identified as reversible fluctuations, and false alarms are suppressed. Simultaneously, before the high-water season, the warning threshold is automatically lowered based on historical hydrological data to adapt to the overall lower insulation resistance caused by rising water levels.

[0030] Example 2, in step 2, the equivalent insulation resistance of the positive and negative electrodes in parallel. Calculate using the following formula: ; in , These are the measured values ​​of the positive and negative insulation resistances of the i-th junction box at time t.

[0031] By adopting the above technical solution, the insulation resistances of the positive and negative electrodes to ground in the photovoltaic DC system are electrically connected in parallel, and the total insulation resistance is determined by the parallel connection of the positive and negative electrode insulation resistances to ground. When the insulation resistance values ​​of the positive and negative electrodes to ground differ significantly, the parallel equivalent insulation resistance tends to be closer to the smaller resistance value, reflecting the actual situation on the weaker side of the insulation.

[0032] Temperature and humidity normalization correction is performed using the following formula: ; in To correct the insulation resistance, This refers to the ambient temperature, expressed in degrees Celsius. This refers to ambient humidity, expressed as a percentage of relative humidity. This is a temperature correction factor, expressed in degrees Celsius. This is a humidity correction factor, dimensionless.

[0033] and The insulation resistance was obtained through laboratory calibration. The calibration method was as follows: the insulation testing unit was placed in a constant temperature and humidity test chamber, a standard insulation resistor was connected, and the insulation resistance measurements were recorded at multiple temperature and humidity points. The results were then obtained by least squares fitting. and The optimal estimate.

[0034] The sliding time window has a length of 7 days. The dynamic baseline is the moving median of the corrected insulation resistance sequence within this window, and the warning lower bound is the 5th percentile of the corrected insulation resistance sequence within this window. The advantage of using the moving median as the dynamic baseline is its inherent robustness to occasional outliers, while using the 5th percentile as the warning lower bound provides a reasonable statistical deviation criterion during the slow degradation of insulation resistance.

[0035] In Example 3, step 3, the abnormality suspicion marking conditions are: the current corrected insulation resistance is lower than 50% of the warning lower limit, or the corrected insulation resistance of 10 consecutive sampling points shows a monotonically decreasing trend and the cumulative decrease exceeds 30%; The second sampling period is 10 seconds, and the first sampling period is resumed after the anomaly is eliminated. The condition for eliminating the anomaly is that the corrected insulation resistance is restored to above the lower limit of the warning and the trend remains stable for 1 hour.

[0036] By adopting the above technical solution, the specific conditions for marking abnormal suspects are as follows: Condition 1: The current corrected insulation resistance is below 50% of the warning lower limit. This condition corresponds to operating conditions where the insulation resistance suddenly drops significantly, such as instantaneous water ingress into the cable joint or insulation material breakdown. Condition 2: The corrected insulation resistance at ten consecutive sampling points shows a monotonically decreasing trend with a cumulative decrease exceeding 30%. This condition corresponds to a condition where the insulation resistance deteriorates slowly and continuously, such as when the module backsheet is exposed to moisture for a long time and gradually ages.

[0037] An anomaly suspicion flag is triggered if either of the following two conditions is met. The first sampling period is 60 seconds by default, and the second sampling period is 10 seconds. The condition for eliminating anomaly suspicion is that the corrected insulation resistance recovers to above the warning lower limit and the trend remains stable for one hour. The one-hour stable recovery observation period ensures that the anomaly recovery is not an accidental data fluctuation, but a genuine improvement in insulation condition.

[0038] Example 4, in step 4, the attenuation vector , where k is the slope of the linear fit of the corrected insulation resistance sequence over the past 24 hours. The Pearson correlation coefficient of the insulation resistance sequence between the detection point and adjacent upstream and downstream detection points is corrected; It is the absolute value of the difference between the average corrected insulation resistance during the daytime and the average corrected insulation resistance during the nighttime.

[0039] By adopting the above technical solution, k represents the decay rate, specifically the slope of the linear fit of the corrected insulation resistance sequence over the past 24 hours, expressed in kiloohms per hour, reflecting the rate at which the insulation resistance decreases. The linear fit employs the least squares method, performing a univariate linear regression with time as the independent variable and corrected insulation resistance as the dependent variable; the resulting regression coefficient is k.

[0040] The correlation coefficient between adjacent points is specifically the Pearson correlation coefficient between the detection point and its upstream and downstream adjacent detection points, representing the corrected insulation resistance sequences. Upstream and downstream adjacent detection points refer to the detection points corresponding to the previous and next nodes directly electrically connected to the combiner box in the single-line electrical topology of the power station. The Pearson correlation coefficient characterizes the consistency of the changing trends of the two sequences, with a value range of [-1, 1]. The closer it is to 1, the more consistent the rising and falling trends of the two sequences.

[0041] The diurnal fluctuation range is specifically the absolute value of the difference between the average corrected insulation resistance during the daytime and the average corrected insulation resistance during the nighttime, expressed in kiloohms. The daytime period is from 6:00 AM to 6:00 PM, and the nighttime period is from 6:00 PM to 6:00 AM the following day. The diurnal fluctuation range reflects the degree to which insulation resistance is affected by periodic changes in temperature and humidity. Backsheet insulation aging faults in modules are usually accompanied by significant diurnal fluctuations because aging insulation materials are more sensitive to temperature changes.

[0042] Example 5, Combiner box attenuation vector With string decay vector Euclidean distance for: ; The fault location determination method is as follows: if there exists a string j that satisfies... Less than the preset similarity threshold And the decay rate of the string If the rate of decay is greater than η times the average decay rate of all other strings, where η is a preset multiplier and η > 1, then the fault is determined to be located within the corresponding string; if the rate of decay of all strings is greater than η times the average decay rate of all other strings, then the fault is determined to be located within the corresponding string. All greater than The fault is determined to be located inside the combiner box or in a common cable section; if the Euclidean distance between the attenuation vectors of two adjacent strings is less than... And the distances between each and the attenuation vector of the combiner box are all greater than The fault was determined to be located at the common node of the T-connection of the two strings.

[0043] By adopting the above technical solution, rule one states that if there exists a string j that satisfies... Less than the preset similarity threshold And the decay rate of this string If the attenuation rate is greater than η times the average attenuation rate of all other strings, where η is a preset multiplier and η is greater than 1, then the fault is determined to be located within that string. The physical logic of this rule is as follows: when an insulation fault occurs within a specific string, the insulation resistance of that string will decrease at a significantly faster rate than other normal strings in the same combiner box. This results in a string attenuation vector that is highly similar to the overall attenuation vector of the combiner box, but with a significantly larger attenuation rate.

[0044] Rule 2, if all strings All are greater than the preset similarity threshold If the fault is located inside the combiner box or in the common cable segment from the combiner box to each string, the physical logic of this rule is as follows: when the fault is located in the common part of the combiner box, the insulation resistance change detected at each string end is transmitted from the same fault source through the common bus. Therefore, the attenuation vector of each string is significantly different from the overall attenuation vector of the combiner box.

[0045] Rule 3: If there are two adjacent strings, the Euclidean distance between their decay vectors is less than 1. Half of it, and each of them has an Euclidean distance greater than the junction box attenuation vector. If the fault is located at the T-junction common node of the two strings, the physical logic of this rule is as follows: the T-junction is located at the junction of the branches of the two strings. When an insulation fault occurs at this point, the detection circuits of the two strings share the leakage path of the fault point. Therefore, the attenuation behavior of the two strings is highly similar. However, the fault point is at a certain distance from the common bus of the combiner box in terms of electrical topology. Therefore, the attenuation vectors of the two strings are not similar to the attenuation vector of the combiner box.

[0046] Example 6, in step 5, the phase difference between the leakage current and the DC bus voltage ripple. Calculate using the following formula: ; and These are the sampling sequences of leakage current and DC bus voltage after bandpass filtering, respectively. for The conjugate of , N is the number of sampling points, and the passband of the bandpass filter is taken as the twice harmonic component of the inverter switching frequency;

[0047] The fault nature determination method is as follows: when |φ|<30°, it is determined that the cable insulation to ground has deteriorated; when |φ|>60°, it is determined that the component backplane insulation has aged; when 30°≤|φ|≤60°, it is determined that it is a mixed fault.

[0048] By adopting the above technical solution, the passband of the bandpass filter extracts a harmonic component of the inverter switching frequency, typically ranging from 100Hz to 300Hz. The reason for using this frequency band is that the inverter switching operation generates a voltage ripple of the corresponding frequency on the DC bus. This ripple is coupled to ground through insulation resistance and distributed capacitance, generating a component of the same frequency in the leakage current. This frequency band offers stable energy, is easy to extract, and effectively avoids DC components and power frequency interference.

[0049] Phase difference The physical meaning of φ is the phase lag angle of the AC component of the leakage current relative to the DC bus voltage ripple, determined by the resistive-capacitive nature of the fault point's impedance to ground. The fault characteristic discrimination rule is as follows: when the absolute value of φ is less than 30 degrees, the fault impedance is resistive or weakly capacitive, and the leakage current and voltage ripple are nearly in phase, consistent with the electrical characteristics of a leakage path dominated by resistance when the cable insulation is damp or damaged, thus indicating a decrease in cable insulation to ground; when... When the absolute value is greater than 60 degrees, the fault impedance exhibits strong capacitiveness, and the leakage current and voltage ripple are nearly orthogonal, consistent with the electrical characteristics of increased dielectric constant and distributed capacitance of polymer materials during module backsheet insulation aging, thus indicating module backsheet insulation aging; when When the absolute value is between 30 and 60 degrees, the fault impedance has both resistive and capacitive characteristics, and is judged as a mixed fault.

[0050] Example 7, the modified insulation resistance versus water level curve is as follows: ; in To correct the insulation resistance for the weekly average, The weekly average water level The lowest water level within the statistical period. This is the insulation resistance value corresponding to the lowest water level. The water level influence coefficient is obtained through least squares fitting. The condition for reversible fluctuation is: the absolute value of the difference between the current corrected insulation resistance and the expected value calculated based on the current water level according to the relationship curve is less than the preset tolerance ε. The method for adjusting the early warning threshold before the high water season is as follows: lower the lower limit of the warning by δ%, with δ set at 10%-20%.

[0051] By adopting the above technical solution The weekly average corrected insulation resistance, in kiloohms, is the arithmetic mean of all corrected insulation resistance samples within each week. The weekly average water level is expressed in meters and is calculated as the arithmetic mean of all water level samples taken within each week. The lowest water level within the statistical period, in meters; The value of insulation resistance corresponding to the lowest water level is expressed in kiloohms, derived from the fitted curve. equal The value at the location is determined; γ is the water level influence coefficient, with the unit being kilohms per meter, which characterizes the rate at which the corrected insulation resistance decreases with increasing water level.

[0052] γ and The data is obtained through least squares fitting, and the fitting data needs to accumulate weekly average corrected insulation resistance and weekly average water level data pairs for at least one complete hydrological year.

[0053] The mechanism of the water level influence coefficient γ is as follows: When the water level in the water area of ​​the solar-aquaculture hybrid power station rises, the distance between the water surface and the cable trays and combiner boxes shortens, the local ambient humidity increases, and the surface leakage current of cable joints and insulation surfaces increases due to condensation or water vapor penetration, resulting in a corresponding decrease in the equivalent insulation resistance. This effect is reversible within a certain range, and the insulation resistance can be partially recovered when the water level falls back.

[0054] The condition for reversible fluctuation is: the absolute value of the difference between the current corrected insulation resistance and the expected value calculated based on the relationship curve according to the current water level is less than the preset tolerance ε. If this condition is met, it indicates that the current decrease in insulation resistance matches the increase in water level, which falls within the scope of reversible fluctuation, and the system suppresses the alarm output; if this condition is not met, it indicates that the decrease in insulation resistance exceeds the range affected by the water level, and there may be actual insulation degradation, and the system outputs an alarm normally.

[0055] The method for adjusting the warning threshold before the high water season is as follows: Before the start of the high water season, the highest water level of the current year is predicted based on historical hydrological data, and the lower limit of the warning is lowered by δ percentage points. The value of δ ranges from 10% to 20%, so that the overall low insulation resistance caused by the rise in water level will not trigger false alarms.

[0056] In Example 8, the leakage current range acquired by the insulation detection unit is 0mA-500mA; the sampling rate of the DC bus voltage ripple is not less than 1kHz, and the continuous acquisition time is not less than 1 minute; the passband of the bandpass filter is 100Hz-300Hz.

[0057] By adopting the above technical solution, the measurement range covers the typical leakage current values ​​of photovoltaic power station combiner boxes under normal insulation and slight degradation conditions, while leaving a margin for severe degradation conditions.

[0058] The sampling rate of the DC bus voltage ripple is no less than 1 kHz, and the continuous acquisition time is no less than 1 minute. The 1 kHz sampling rate can completely capture the second harmonic component of the inverter switching frequency from 100 Hz to 300 Hz, which meets the requirements of the Nyquist sampling theorem; the 1 minute acquisition time can ensure that enough signal periods are acquired for stable phase difference calculation.

[0059] The passband of the bandpass filter is from 100Hz to 300Hz. This frequency band covers the range of the second harmonic component corresponding to the switching frequency of common photovoltaic inverters from 50Hz to 150Hz, and is suitable for various application scenarios of inverter models.

[0060] Example 9 also includes a maintenance result feedback step: comparing the actual fault type and fault location confirmed on-site with the judgment results of the steps, calculating the judgment accuracy, and optimizing the attenuation vector similarity threshold accordingly. And phase difference discrimination threshold.

[0061] By adopting the above technical solution, after completing on-site maintenance, the maintenance personnel transmit the actual fault type and actual fault location via a mobile maintenance terminal. The system compares the actual fault type with the fault nature determined in step 5, and the actual fault location with the fault location determined in step 4, respectively calculating the accuracy rate of fault nature determination and fault location positioning. When the fault nature determination accuracy rate is lower than the preset target value, the phase difference determination threshold in step 5 is adjusted, i.e., the boundary angle value between 30 degrees and 60 degrees is adjusted; when the fault location positioning accuracy rate is lower than the preset target value, the attenuation vector similarity threshold in step 5 is adjusted. The value of the multiplier η is determined. This feedback adjustment mechanism enables the system's discrimination parameters to be continuously optimized as the running time increases and the number of failure cases accumulates, forming a closed-loop self-learning mechanism.

[0062] Example 10: Insulation resistance full-chain detection system for fishery-solar complementary photovoltaic power station. This system is used to realize the full-chain detection method for insulation resistance of fishery-solar complementary photovoltaic power station. The system includes a data sensing layer, a data aggregation layer, an analysis and decision layer, and an execution feedback layer. The data sensing layer consists of insulation detection units deployed at each combiner box. Each detection unit integrates a DC unbalanced bridge insulation resistance detection module, a leakage current acquisition module, a temperature and humidity sensor, a water level sensor, and a communication module. It is used to collect data on the positive-to-ground insulation resistance, the negative-to-ground insulation resistance, leakage current, ambient temperature, ambient humidity, and water level. The data aggregation layer includes local data concentrators, which are connected to each detection unit and to the power plant monitoring backend, for summarizing and forwarding detection data; The analysis and decision-making layer is deployed on the power plant monitoring backend or cloud platform to perform the following steps: The equivalent insulation resistance of the positive and negative poles in parallel is calculated and normalized for temperature and humidity. A dynamic baseline and early warning lower limit are established based on a sliding window. The sampling period is switched according to the abnormal suspicion marking conditions. The attenuation vector of the combiner box and the string is constructed and the Euclidean distance is calculated to determine the fault location. The phase difference between the leakage current and the DC bus voltage ripple is analyzed simultaneously to determine the fault nature. A correlation curve between water level and insulation resistance is established to distinguish between reversible fluctuations and insulation degradation, and seasonal early warning threshold adjustments are performed. The execution feedback layer is a mobile operation and maintenance terminal used to receive fault location and fault nature judgment results and generate operation and maintenance work orders, as well as to send back maintenance results to optimize the judgment threshold in the analysis and decision layer.

[0063] By adopting the above technical solution, the data sensing layer consists of insulation detection units deployed at each combiner box. Each insulation detection unit integrates a DC unbalanced bridge insulation resistance detection module, a leakage current acquisition module, a temperature and humidity sensor, a water level sensor, and a communication module. The DC unbalanced bridge insulation resistance detection module connects measuring resistors of known resistance values ​​between the DC positive terminal and ground, and between the DC negative terminal and ground, switching the bridge to a balanced state. It calculates the insulation resistance between the positive and negative terminals by measuring the unbalanced current of the bridge. The leakage current acquisition module measures the total leakage current flowing from the DC bus to ground through the insulation impedance using a high-precision Hall sensor or a shunt resistor connected in series in the grounding loop. The temperature and humidity sensor collects the ambient temperature and humidity inside the combiner box, and the water level sensor collects the real-time water level at the combiner box installation location. The communication module supports wired or LoRa wireless communication, uploading the collected data to the data aggregation layer.

[0064] The data aggregation layer includes a local data concentrator, which communicates with each insulation detection unit via bus or LoRa wireless connection, and with the power plant monitoring backend via fiber optic or 4G / 5G networks. The local data concentrator sequentially reads the latest collected data from each insulation detection unit in a polling manner, verifies and temporarily stores the data, and packages and uploads the summarized data to the analysis and decision-making layer according to a preset reporting cycle.

[0065] The analysis and decision-making layer is deployed on the power plant monitoring backend or cloud platform, and is a server with computing and storage capabilities. The analysis and decision-making layer performs the following functions: calculates the equivalent insulation resistance of the positive and negative parallel terminals and performs temperature and humidity normalization correction; establishes a dynamic baseline and early warning lower limit based on a sliding window; switches the sampling period according to the anomaly suspicion marker conditions, constructs the attenuation vector of the combiner box and string, and calculates the Euclidean distance to determine the fault location; simultaneously analyzes the phase difference between leakage current and DC bus voltage ripple to determine the fault nature; establishes a correlation curve between water level and insulation resistance, distinguishes between reversible fluctuations and insulation degradation, and performs seasonal early warning threshold adjustments.

[0066] The execution feedback layer consists of mobile maintenance terminals that connect to the power plant monitoring backend or cloud platform via a wireless network. These mobile terminals receive fault location and fault nature determination results from the analysis and decision-making layer, automatically generate maintenance work orders containing fault location, fault nature, urgency level, and suggested handling measures, and push them to maintenance personnel. After completing on-site handling, maintenance personnel use the mobile terminals to transmit the actual fault type and location back to the system. The analysis and decision-making layer then optimizes parameters such as attenuation vector similarity threshold and phase difference discrimination threshold based on this information.

[0067] The following describes the implementation principle of the present invention using specific embodiments: This study uses a 50 MW solar-fishery hybrid photovoltaic power station as an example. The station employs a solar-fishery hybrid model, generating electricity on the rooftops and raising fish underneath. The water area is approximately 1200 mu (about 80 hectares), with an average annual water level fluctuation of 1.5 meters. The high-water season is from June to September, and the low-water season is from December to February of the following year. The power station is equipped with 25 2 MW centralized inverters. Each inverter is connected to 16 DC combiner boxes with 16 inputs each. Each combiner box connects to 16 photovoltaic strings, each string consisting of 28 540 watt monocrystalline silicon photovoltaic modules connected in series. The maximum DC voltage of the system is 1500 volts.

[0068] Step 1: Deploy insulation detection units at each combiner box. Select an IP68-rated insulation detection unit with an integrated DC unbalanced bridge insulation resistance detection module, a leakage current acquisition module with a range of 0 mA to 500 mA, a temperature and humidity sensor with a measurement range of -20°C to 80°C, a water level sensor with a range of 0 m to 5 m, and a LoRa communication module. Install this unit inside each combiner box. The insulation detection unit collects data on the positive-to-ground insulation resistance, negative-to-ground insulation resistance, leakage current, ambient temperature, ambient humidity, and water level every 60 seconds in the first sampling cycle. The data is then uploaded via LoRa wireless communication to a local data concentrator installed on the inverter platform. The local data concentrator then forwards the data via a fiber optic network to the power station monitoring backend server deployed in the substation control room.

[0069] At 8:30 AM on August 15, 2024, the insulation detection unit of the No. 3 combiner box under the No. 5 inverter collected the following data: the insulation resistance R₊ of the positive terminal to ground was 320 kΩ, the insulation resistance R₋ of the negative terminal to ground was 680 kΩ, the leakage current was 42 mA, the ambient temperature was 38 degrees Celsius, the ambient humidity was 92%, and the real-time water level was 3.2 meters.

[0070] Step 2: After receiving the data collected from combiner box No. 3, the power plant monitoring backend server first calculates the equivalent insulation resistance of the positive and negative poles in parallel. , Substituting the data, the calculated value is 217.6 kΩ. Then, temperature and humidity normalization corrections are performed. The temperature correction factor α for the insulation testing unit of this power station, calibrated in the laboratory, is 0.008 degrees Celsius, and the humidity correction factor β is 0.005. The corrected insulation resistance is then calculated. Substituting the data, the value is calculated to be 274.9 kilo-euros.

[0071] The median moving value of the corrected insulation resistance within the 7-day sliding time window of the junction box is 485 kΩ, and the 5th percentile is 412 kΩ. Therefore, the dynamic baseline is 485 kΩ and the lower warning limit is 412 kΩ.

[0072] Step 3: The current corrected insulation resistance of 274.9 kΩ is below 50% of the warning threshold of 412 kΩ, i.e., 206 kΩ, which does not meet the condition. However, the system reviews the corrected insulation resistance sequence of the most recent 10 sampling points for this combiner box and finds that the 10 sampling points from 8:21 AM to 8:30 AM on the same day were 386, 375, 361, 348, 337, 325, 314, 303, 291, and 275 kΩ respectively, showing a monotonically decreasing trend, with a cumulative decrease of 28.8%, close to the 30% threshold. At 8:31 AM, the corrected insulation resistance of the next sampling point dropped to 268 kΩ, with a cumulative decrease of 30.6%, exceeding the 30% threshold. The system then marks combiner box number 3 as a suspected anomaly.

[0073] After marking the suspected anomaly, the system automatically increases the sampling period of the insulation detection unit of the No. 3 combiner box from 60 seconds to 10 seconds, and at the same time increases the sampling period of the adjacent No. 2 and No. 4 combiner boxes under the No. 5 inverter to 10 seconds, activating encrypted data acquisition.

[0074] Step 4: As of 8:30 AM on August 16th, the anomaly suspicion marker had been in place for 24 hours. The system constructed attenuation vectors for combiner box No. 3 and its 16 sub-connectors.

[0075] The linear fitting slope of the corrected insulation resistance sequence of combiner box No. 3 over 24 hours is -2.35 kΩ per hour. The Pearson correlation coefficient between the corrected insulation resistance sequences of this combiner box and those of upstream combiner box No. 2 and downstream combiner box No. 4 is 0.42. The average corrected insulation resistance during the day is 285 kΩ, and the average at night is 312 kΩ, with a diurnal fluctuation of 27 kΩ. Therefore, the attenuation vector of combiner box No. 3... It equals [-2.35, 0.42, 27].

[0076] Of the 16 strings, the decay vector of the 7th string is [-2.18, 0.46, 25], the decay vector of the 12th string is [-0.31, 0.21, 8], and the decay vectors of the remaining 14 strings are similar to those of the 12th string. The decay rate is between -0.2 and -0.5 kΩ per hour, the correlation coefficient between adjacent points is between 0.2 and 0.3, and the diurnal fluctuation range is between 5 and 12 kΩ.

[0077] Calculate the Euclidean distance between each string and the attenuation vector of the combiner box. , The Euclidean distance between the 7th string and the attenuation vector of the combiner box The value is 2.01. The Euclidean distances of the remaining 15 strings are all between 8.5 and 12.3. (Preset similarity threshold) Take 4.0.

[0078] The 7th string 2.01 is less than The value is 4.0, and the attenuation rate of the 7th string (2.18 kΩ / h) is greater than η times the average attenuation rate of the other 15 strings. The average attenuation rate of the other 15 strings is 0.35 kΩ / h. η is taken as 1.5, 0.35 × 1.5 = 0.525, and 2.18 is greater than 0.525, satisfying rule one. The system determines that the fault is located within the 7th string under the 3rd combiner box.

[0079] Step 5: After the anomaly suspicion marker is triggered, the insulation detection unit of combiner box No. 3 simultaneously collects leakage current and DC bus voltage ripple. The inverter switching frequency of this power station is 75Hz, and its second harmonic component is 150Hz. The bandpass filter takes a passband of 100Hz to 300Hz, the sampling rate is 1KHz, and a total of 60,000 sampling points are obtained for 1 minute of continuous sampling.

[0080] Perform cross-correlation on the filtered leakage current sequence and voltage ripple sequence to calculate the phase difference. Calculation results The temperature is -18.7 degrees Celsius, and its absolute value is 18.7 degrees Celsius, which is less than 30 degrees Celsius. According to the fault nature judgment rules, the system determines the fault nature as a decrease in cable insulation to ground.

[0081] At this time, the system outputs a complete fault diagnosis result: the fault location is inside the 7th string of the 3rd combiner box of the 5th inverter, and the fault nature is a decrease in the insulation of the cable to ground.

[0082] Step 6: The water level versus corrected insulation resistance curve obtained by fitting a complete hydrological year's worth of data at this detection point is as follows: The relationship between water level and corrected insulation resistance is shown in the curve. Figure 2 As shown, where It is 1.8 meters. It is 562,000 euros. It is 52 kΩ per meter. The current water level is 3.2 meters. Based on the relationship curve, the expected corrected insulation resistance is... 489.2 kΩ. The current actual corrected insulation resistance is about 275 kΩ, which is 489.2-275=214.2 kΩ from the expected value. This is much greater than the preset tolerance ε of 30 kΩ. Therefore, it is determined that the decrease in insulation resistance is not a reversible fluctuation caused by water level change, and it is confirmed as a real insulation degradation. The alarm will be output normally.

[0083] It is currently August, the high-water season. The system automatically lowered the warning lower limit by 15% in early June. The original warning lower limit for combiner box No. 3 was 412 kΩ. After the adjustment, the actual warning lower limit is 412 × 0.85 = 350.2 kΩ. The current corrected insulation resistance of 275 kΩ is still lower than the lowered warning lower limit, so the alarm is valid.

[0084] Step 7: The system pushes a maintenance work order to the on-duty maintenance personnel via the mobile maintenance terminal. The work order content is as follows: Fault location: inside the 7th string of the 3rd combiner box of inverter 5; Fault nature: cable insulation to ground has deteriorated; Urgency level: to be handled within 48 hours; Suggested handling measures: board the vessel to the waters below the 3rd combiner box, check the integrity of the insulation layer of the DC cable of the 7th string, and focus on checking whether there is water ingress or insulation damage at the cable joints and the incoming terminals of the combiner box.

[0085] Maintenance personnel arrived at the site at 9:00 AM on August 17th and inspected the 7th string of cables. They found that the outer sheath of the positive cable in this string was worn and damaged due to long-term friction from water surface fluctuations near the combiner box inlet. Partial exposure of the core insulation layer was found, and the cable was susceptible to corrosion from water moisture. The maintenance personnel replaced the damaged section of cable and resealed the combiner box inlet. After this repair, the insulation resistance of combiner box 3 returned to 472 kΩ, exceeding the warning lower limit of 350.2 kΩ. The anomaly was eliminated, and the system resumed its 60-second sampling cycle.

[0086] Step 8: The maintenance personnel transmit the repair results via mobile maintenance terminal: the actual fault type is decreased cable insulation to ground, and the actual fault location is the positive cable inlet of the 7th string in combiner box No. 3. The system adds this fault case to the historical database and calculates the accuracy rate of fault nature identification and fault location. In this case, both the fault nature and fault location were correctly identified, and the identification accuracy rate remained high, so no adjustment of the similarity threshold is required. And phase difference discrimination threshold.

[0087] This implementation case fully demonstrates the complete workflow of the insulation resistance detection system for the fishery-solar complementary photovoltaic power station, from multi-parameter sensing, data preprocessing, anomaly screening, attenuation vector construction and string-level positioning, leakage current phase analysis to determine the nature of the fault, water level correlation to identify reversible fluctuations, maintenance work order generation to maintenance result feedback.

[0088] The above are all preferred embodiments of the present invention and are not intended to limit the scope of protection of the present invention. Therefore, all equivalent changes made in accordance with the structure, shape and principle of the present invention should be covered within the scope of protection of the present invention.

Claims

1. A method for full-chain testing of insulation resistance in a solar-fishery hybrid photovoltaic power station, characterized in that, Includes the following steps: Step 1: Deploy insulation detection units at each junction box and collect data on positive-to-ground insulation resistance, negative-to-ground insulation resistance, leakage current, ambient temperature, ambient humidity, and water level according to the first sampling cycle. Step 2: Calculate the equivalent insulation resistance of the positive and negative terminals of each combiner box in parallel. After temperature and humidity normalization correction, establish a dynamic baseline and warning lower limit for the corrected insulation resistance based on the sliding time window. Step 3: When the corrected insulation resistance is lower than the preset ratio of the lower warning limit or shows a continuous downward trend, the corresponding combiner box is marked as an abnormality suspect, and the sampling period of the abnormality suspect combiner box and adjacent combiner boxes is automatically increased to the second sampling period. Step 4: For the suspected abnormal combiner box and each of its subordinate strings, construct attenuation vectors that include attenuation rate, correlation between adjacent points and day and night fluctuation amplitude. Calculate the Euclidean distance between the combiner box attenuation vector and the attenuation vector of each string. Based on the distance and the similarity between strings, determine whether the fault is located inside the string, inside the combiner box, or at the T-connected common node. Step 5: Simultaneously collect the leakage current and DC bus voltage ripple of the suspected abnormal combiner box, extract the phase difference of the inverter switching frequency band component, and determine the fault as a decrease in cable insulation to ground or aging of component backplane insulation based on the phase difference range. Step 6: Establish the relationship curve between the corrected insulation resistance and water level at each detection point, distinguish between reversible fluctuations caused by water level changes and insulation degradation trends, and automatically adjust the early warning threshold before the high water season.

2. The method for full-chain detection of insulation resistance in a solar-fishery complementary photovoltaic power station according to claim 1, characterized in that, In step 2, the equivalent insulation resistance of the positive and negative terminals in parallel is... Calculate using the following formula: ； in , These are the measured values ​​of the positive and negative insulation resistances of the i-th junction box at time t.

3. The method for full-chain detection of insulation resistance in a solar-fishery complementary photovoltaic power station according to claim 2, characterized in that, In step 3, the abnormality suspicion marking conditions are: the current corrected insulation resistance is lower than 50% of the warning lower limit, or the corrected insulation resistance of 10 consecutive sampling points shows a monotonically decreasing trend and the cumulative decrease exceeds 30%; The second sampling period is 10 seconds, and the first sampling period is resumed after the anomaly is eliminated. The condition for eliminating the anomaly is that the corrected insulation resistance is restored to above the lower limit of the warning and the trend remains stable for 1 hour.

4. The method for full-chain detection of insulation resistance in a solar-fishery complementary photovoltaic power station according to claim 3, characterized in that, In step 4, the attenuation vector , where k is the slope of the linear fit of the corrected insulation resistance sequence over the past 24 hours. The Pearson correlation coefficient of the insulation resistance sequence between the detection point and adjacent upstream and downstream detection points is corrected; It is the absolute value of the difference between the average corrected insulation resistance during the daytime and the average corrected insulation resistance during the nighttime.

5. The method for full-chain detection of insulation resistance in a solar-fishery complementary photovoltaic power station according to claim 4, characterized in that, Combiner box attenuation vector With string decay vector Euclidean distance for: ； The fault location determination method is as follows: if there exists a string j that satisfies... Less than the preset similarity threshold And the decay rate of the string If the attenuation rate is greater than η times the average attenuation rate of all other strings, where η is a preset multiplier and η > 1, then the fault is determined to be located within the corresponding string; if the attenuation rate of all strings is greater than η times the average attenuation rate of all other strings, then the fault is determined to be located within the corresponding string. All greater than The fault is determined to be located inside the combiner box or in a common cable section; if the Euclidean distance between the attenuation vectors of two adjacent strings is less than... And the distances between each and the attenuation vector of the combiner box are all greater than The fault was determined to be located at the common node of the T-connection of the two strings.

6. The method for full-chain detection of insulation resistance in a solar-fishery complementary photovoltaic power station according to claim 5, characterized in that, In step 5, the phase difference between the leakage current and the DC bus voltage ripple is... Calculate using the following formula: ; and These are the sampling sequences of leakage current and DC bus voltage after bandpass filtering, respectively. for The conjugate of , N is the number of sampling points, and the passband of the bandpass filter is taken as the twice harmonic component of the inverter switching frequency; The fault nature determination method is as follows: when |φ|<30°, it is determined that the cable insulation to ground has deteriorated; when |φ|>60°, it is determined that the component backplane insulation has aged; when 30°≤|φ|≤60°, it is determined that it is a mixed fault.

7. The method for full-chain detection of insulation resistance in a solar-fishery complementary photovoltaic power station according to claim 6, characterized in that, The modified insulation resistance versus water level curve is as follows: ； in To correct the insulation resistance for the weekly average, The weekly average water level The lowest water level within the statistical period. This is the insulation resistance value corresponding to the lowest water level. The water level influence coefficient is obtained through least squares fitting. The condition for reversible fluctuation is: the absolute value of the difference between the current corrected insulation resistance and the expected value calculated based on the current water level according to the relationship curve is less than the preset tolerance ε. The method for adjusting the early warning threshold before the high water season is as follows: lower the lower limit of the warning by δ%, with δ set at 10%-20%.

8. The method for full-chain detection of insulation resistance in a solar-fishery complementary photovoltaic power station according to claim 7, characterized in that, The leakage current range acquired by the insulation detection unit is 0mA-500mA; the sampling rate of the DC bus voltage ripple is not less than 1kHz, and the continuous acquisition time is not less than 1 minute; the passband of the bandpass filter is 100Hz-300Hz.

9. The method for full-chain detection of insulation resistance in a solar-fishery complementary photovoltaic power station according to claim 8, characterized in that, It also includes a maintenance result feedback step: comparing the actual fault type and location confirmed on-site with the judgment results of the steps, calculating the judgment accuracy, and optimizing the attenuation vector similarity threshold accordingly. And phase difference discrimination threshold.

10. A full-chain testing system for insulation resistance of a solar-fishery complementary photovoltaic power station, characterized in that, The system for implementing the full-chain detection method for insulation resistance of the solar-fishery complementary photovoltaic power station as described in claim 9 includes a data sensing layer, a data aggregation layer, an analysis and decision-making layer, and an execution feedback layer. The data sensing layer consists of insulation detection units deployed at each combiner box. Each detection unit integrates a DC unbalanced bridge insulation resistance detection module, a leakage current acquisition module, a temperature and humidity sensor, a water level sensor, and a communication module. It is used to collect data on the positive-to-ground insulation resistance, the negative-to-ground insulation resistance, leakage current, ambient temperature, ambient humidity, and water level. The data aggregation layer includes local data concentrators, which are connected to each detection unit and to the power plant monitoring backend, for summarizing and forwarding detection data; The analysis and decision-making layer is deployed on the power plant monitoring backend or cloud platform to perform the following steps: The equivalent insulation resistance of the positive and negative poles in parallel is calculated and normalized for temperature and humidity. A dynamic baseline and early warning lower limit are established based on a sliding window. The sampling period is switched according to the abnormal suspicion marking conditions. The attenuation vector of the combiner box and the string is constructed and the Euclidean distance is calculated to determine the fault location. The phase difference between the leakage current and the DC bus voltage ripple is analyzed simultaneously to determine the fault nature. A correlation curve between water level and insulation resistance is established to distinguish between reversible fluctuations and insulation degradation, and seasonal early warning threshold adjustments are performed. The execution feedback layer is a mobile operation and maintenance terminal used to receive fault location and fault nature judgment results and generate operation and maintenance work orders, as well as to send back maintenance results to optimize the judgment threshold in the analysis and decision layer.

Images