A distributed photovoltaic power station operation power generation efficiency analysis method

Abstract

The application belongs to the technical field of intelligent operation and maintenance of photovoltaic power stations, and relates to a distributed photovoltaic station operation power generation efficiency analysis method, which divides photovoltaic strings in the station into multiple comparable analysis groups based on the consistency of effective light receiving conditions and the consistency of component configuration parameters, carries out operation power generation efficiency calculation on each photovoltaic string in each comparable analysis group, identifies abnormal photovoltaic strings deviating from the average level of the group by comparing the operation power generation efficiency of each photovoltaic string in the group, extracts the operation parameter sequence of the abnormal photovoltaic string and calculates the correlation compliance index relative to a pre-established correlation rule library, comprehensively diagnoses the cause of low efficiency of the abnormal photovoltaic string by combining the power generation efficiency deviation degree and the correlation compliance index, and outputs the diagnosis result. The strategy of grouping comparison and parameter correlation compliance collaborative analysis can realize accurate identification of local abnormal equipment and deep positioning of the reason for low efficiency.

Description

TECHNICAL FIELD

[0001] The application belongs to the technical field of intelligent operation and maintenance of photovoltaic power stations, and relates to a distributed photovoltaic station operation and power generation efficiency analysis method. BACKGROUND

[0002] As an important form of new energy power generation, a distributed photovoltaic station is usually composed of a large number of photovoltaic strings, inverters and other devices, and its power generation efficiency is directly related to the investment return rate and energy output benefit. Therefore, real-time and accurate analysis of the operation and power generation efficiency of the distributed photovoltaic station, and timely discovery and processing of faults affecting the efficiency, are the core links of intelligent operation and maintenance management of photovoltaic power stations. This usually depends on the collection, processing and analysis of power station operation data.

[0003] At present, the evaluation and analysis of the power generation efficiency of photovoltaic stations in the industry mainly relies on computer systems to compare the total actual power generation and theoretical power generation of the whole station at a macro level. Although some advanced systems introduce data processing methods to monitor and sort the output power of inverters or photovoltaic strings, and make preliminary judgments on the running state of the equipment based on this, the essence of their technology is still limited to statistical analysis of a single performance indicator such as power generation power or efficiency. Specifically, such methods usually identify units with poor performance by setting fixed performance thresholds or simply comparing them horizontally among similar devices.

[0004] However, the above-mentioned existing technical means has obvious technical defects in actual application: the existing macro comparison method is too rough and cannot reveal and locate the subtle efficiency loss caused by device aging or component failure and other factors, and the monitoring and sorting based on a single indicator are easily affected by complex and variable environmental factors, leading to false positives or false negatives, and failing to make full use of the inherent physical correlation between operating parameters, thus failing to give a fundamental cause diagnosis to the identified performance deviation, making the subsequent operation and maintenance investigation work still face great blindness. SUMMARY

[0005] In view of this, in order to solve the problems raised in the background art, a distributed photovoltaic station operation and power generation efficiency analysis method is proposed.

[0006] The purpose of the application can be achieved by the following technical solutions: The application provides a distributed photovoltaic station operation and power generation efficiency analysis method, comprising: dividing photovoltaic strings in the station into a plurality of comparable analysis groups based on the consistency of effective light receiving conditions and the consistency of component configuration parameters. The consistency of effective light receiving conditions is determined according to the comprehensive difference degree of real-time effective light receiving area and real-time effective light intensity.

[0007] For each comparable analysis group, based on real-time environmental data, the theoretical power generation of each photovoltaic string in the group is calculated according to the photovoltaic power generation physical model, and the actual power generation is obtained to calculate the operating power generation efficiency.

[0008] By comparing the power generation efficiency of each photovoltaic string within the same group, abnormal photovoltaic strings whose power generation efficiency deviates from the average level within the group are identified.

[0009] Extract the sequence of operating parameters of the abnormal photovoltaic string and calculate its correlation conformity index relative to the pre-established association rule base. The association rule base is constructed based on the physical principles of photovoltaic power generation and the historical set of operating parameters, and is used to define the expected correlation relationship between key operating parameters based on primary and secondary priorities.

[0010] By combining the degree of deviation of power generation efficiency with the correlation conformity index, the cause of the low efficiency of the abnormal photovoltaic string is diagnosed and the diagnosis result is output.

[0011] Compared with the prior art, the beneficial effects of the present invention are as follows: (1) The present invention scientifically groups the equipment based on the physical configuration attributes and effective light conditions, establishes a fair benchmark for comparing the power generation efficiency, effectively eliminates the interference of inherent differences in equipment and external environmental fluctuations on the analysis results, and combines statistical analysis of the consistency of power generation efficiency within the same group, which can accurately and quickly screen out local abnormal photovoltaic strings with potential performance problems from a large number of equipment in the station, greatly improving the sensitivity and accuracy of abnormal detection.

[0012] (2) This invention establishes a rule base for association between key operating parameters, deeply integrates the physical principles of photovoltaic power generation with the historical operating parameter set of the station, and constructs a dynamic and adaptive equipment health behavior reference system. By calculating the correlation conformity index between the actual operating parameters of abnormal photovoltaic strings and the rule base, it achieves in-depth diagnosis of the root cause behind the low efficiency, and elevates the analysis capability from phenomenon identification to causal location, providing clear guidance for subsequent operation and maintenance.

[0013] (3) This invention effectively combines data analysis with operation and maintenance management processes. Through automated cause diagnosis, work order generation and maintenance effect verification, a closed-loop intelligent operation and maintenance management system is formed. Furthermore, by introducing a machine learning mechanism to continuously optimize and update the association rule base, the analysis method is ensured to be effective and adaptable throughout the entire life cycle of the photovoltaic power station, thereby fundamentally improving the overall operation efficiency of the photovoltaic power station. Attached Figure Description

[0014] To more clearly illustrate the technical solutions of the embodiments of the present invention, the accompanying drawings used in the description of the embodiments will be briefly introduced below. Obviously, the drawings described below are only some embodiments of the present invention. For those skilled in the art, other drawings can be obtained based on these drawings without creative effort.

[0015] Figure 1 This is a flowchart illustrating the implementation steps of the method of the present invention.

[0016] Figure 2 This is a schematic diagram of the logic for dividing the comparable analysis groups in this invention.

[0017] Figure 3 This is a flowchart illustrating the logic of quantifying the correlation compliance index in this invention. Detailed Implementation

[0018] The technical solutions of the embodiments of the present invention will be clearly and completely described below with reference to the accompanying drawings. Obviously, the described embodiments are only some embodiments of the present invention, and not all embodiments. Based on the embodiments of the present invention, all other embodiments obtained by those skilled in the art without creative effort are within the scope of protection of the present invention.

[0019] Please see Figure 1 As shown, this invention provides a method for analyzing the power generation efficiency of a distributed photovoltaic (PV) power station, including: S11. Dividing the PV strings within the power station into multiple comparable analysis groups based on the consistency of effective light-receiving conditions and the consistency of component configuration parameters. The consistency of effective light-receiving conditions is determined based on the comprehensive difference between the real-time effective light-receiving area and the real-time effective light-receiving intensity.

[0020] See Figure 2 As shown in a preferred embodiment of the present invention, the comparable analysis group division process includes: acquiring images of the surrounding environment of the location of the photovoltaic string through an image acquisition device, analyzing the shadow distribution on the surface of the photovoltaic module based on image recognition, and determining the real-time effective light-receiving area of ​​each photovoltaic string.

[0021] It should be noted that the process of acquiring the real-time effective light-receiving area of ​​each photovoltaic string is as follows: by using an image acquisition device deployed on the station tower or drone, images of the photovoltaic string from a top-down or oblique angle are periodically acquired.

[0022] The image is preprocessed, including geometric correction, photometric normalization, and component region segmentation, to eliminate perspective distortion and locate the precise outline of each photovoltaic string.

[0023] Pixel-level analysis is performed based on differences in brightness and color to distinguish and extract the light-receiving and shadow-covered areas on the surface of photovoltaic modules.

[0024] The ratio of the area of ​​the bright area not covered by shadow to the total area of ​​the string within the outline region of each photovoltaic string is calculated and defined as the real-time effective light-receiving area ratio of the string.

[0025] By combining the pre-stored physical size parameters of the photovoltaic module string, the effective light-receiving area ratio is converted into the real-time effective light-receiving physical area.

[0026] The real-time effective light intensity of each photovoltaic string is collected synchronously through a distributed irradiance sensor network deployed within the site.

[0027] The process of obtaining the real-time effective light intensity at the location of each photovoltaic string is as follows: The planar irradiance at different spatial locations within the site is measured simultaneously using multiple irradiance sensors arranged in a grid pattern or following the string layout.

[0028] An irradiance distribution map of the site area based on sensor coordinates was established, and a spatial interpolation algorithm was used to calculate the theoretical irradiance value at the center point of each photovoltaic string.

[0029] The real-time effective light-receiving area ratio of each photovoltaic string obtained based on image recognition is used as an attenuation factor. The product of the attenuation factor and the theoretical irradiance value of its location is used as the real-time effective light-receiving intensity of each photovoltaic string under the influence of shadow.

[0030] Based on the real-time effective light-receiving area and real-time effective light-receiving intensity, the matching degree of light-receiving conditions between each photovoltaic string is calculated.

[0031] It should be noted that the specific calculation process for the matching degree of light reception conditions between photovoltaic strings is as follows: For any two photovoltaic strings to be compared, their real-time effective light-receiving area and real-time effective light-receiving intensity are obtained respectively, which together constitute their real-time effective light-receiving parameter set.

[0032] Calculate the absolute difference between two photovoltaic strings on the same type of light-receiving parameters.

[0033] Obtain the arithmetic mean of all photovoltaic strings in the power station at the current moment for this type of light-receiving parameter, and use it as a reference benchmark.

[0034] The absolute difference is compared with the arithmetic mean to obtain a dimensionless parameter of difference that characterizes the relative difference.

[0035] The difference between the effective light-receiving area and the effective light-receiving intensity parameters is obtained separately, and the reciprocal of the sum of the differences between the two parameters is taken as the matching degree of the light-receiving conditions.

[0036] Obtain the module configuration parameters for each photovoltaic string, including cell technology type, module series topology, and inverter electrical characteristics.

[0037] Photovoltaic strings with a matching degree of light reception conditions higher than the preset matching threshold and consistent component configuration parameters are classified into the same comparable analysis group.

[0038] The aforementioned matching degree of light-receiving conditions is higher than the preset matching threshold, which is considered consistent representation data of effective light-receiving conditions.

[0039] This invention establishes a fair benchmark for comparing power generation efficiency by scientifically grouping devices based on their physical configuration attributes and effective light reception conditions. This effectively eliminates the interference of inherent differences in equipment and external environmental fluctuations on the analysis results. Combined with statistical analysis of the consistency of power generation efficiency within the same group, it can accurately and quickly screen out local abnormal photovoltaic strings with potential performance problems from a large number of devices in the site, greatly improving the sensitivity and accuracy of anomaly detection.

[0040] S12. For each of the comparable analysis groups, based on real-time environmental data, calculate the theoretical power generation of each photovoltaic string in the group according to the photovoltaic power generation physical model, and obtain the actual power generation to calculate the operating power generation efficiency.

[0041] In a preferred embodiment of the present invention, the theoretical power generation calculation process for each photovoltaic string in the group includes: obtaining real-time environmental parameters including real-time irradiance and ambient temperature, as well as the rated parameters of the photovoltaic string.

[0042] The real-time environmental parameters and equipment rated parameters are input into the photovoltaic power generation physical model, which calculates the theoretical power generation by coupling the effects of irradiance and temperature on power generation.

[0043] The model outputs the theoretical power generation value under the current environmental conditions, and the calculation process includes an estimation step of the component operating temperature based on the ambient temperature and real-time irradiance.

[0044] It should be noted that the core calculation functions of the above photovoltaic power generation physical model are as follows: .

[0045] in, This represents the theoretical power generation to be calculated. This indicates the rated peak power of the photovoltaic module under standard test conditions. Indicates real-time irradiance. , These represent the irradiance and module temperature under standard test conditions, respectively, and are typically taken as 1000. 25 , This indicates the component's preset power temperature coefficient, reflecting the impact of temperature changes on power generation. and These are all fixed parameters obtained from the equipment manufacturer's specifications. This indicates the actual operating temperature of the photovoltaic module, which cannot be measured directly and is usually estimated using empirical formulas based on ambient temperature and irradiance.

[0046] The meaning, basis for setting, and function of each monomial in the formula are explained below: This is the ideal power generation under current irradiance conditions, considering only the light intensity factor. Its physical meaning is that the short-circuit current of the photovoltaic module is approximately proportional to the irradiance. Therefore, its maximum output power is also proportional to the irradiance in the initial estimate. This single-term setting is directly based on the basic physical principle of photovoltaic power generation: by linearly scaling the rated power according to the ratio of real-time irradiance to standard irradiance, the core function of transferring the performance benchmark under laboratory standard conditions to the actual operating environment is realized, thereby separating the independent influence of irradiance, the main environmental variable, on power generation.

[0047] The temperature correction factor is a dimensionless coefficient used to correct for the aforementioned ideal power generation. It is a negative value, characterizing the characteristic that the output power of a photovoltaic module decreases as the temperature rises. This singleton is set based on the physical characteristics of semiconductors: the open-circuit voltage of a photovoltaic module decreases significantly as the temperature rises, resulting in a reduction in the maximum output power. This correction factor quantifies the reverse effect of operating temperature on power generation in a linear approximation, compensating for the power calculation deviation caused by the temperature deviating from the standard conditions, and making the theoretical model more consistent with the actual physical characteristics of the module.

[0048] This photovoltaic power generation physical model is constructed as a product of a baseline power and a temperature correction factor. The underlying basis is that irradiance mainly affects the photocurrent and temperature mainly affects the open-circuit voltage. The two have relatively independent effects on the output power, and their combined effect is multiplicative. While ensuring the physical meaning is clear, the model couples the synergistic effects of the two key environmental variables, irradiance and temperature, on the power generation. This allows the theoretical power generation capacity of the equipment to be extracted from the complex actual operating environment, providing a scientific, reliable, and unified benchmark that eliminates major environmental interference for subsequent calculations of operating power generation efficiency.

[0049] Furthermore, an empirical formula for the actual operating temperature of photovoltaic modules can be exemplified as follows: .

[0050] in Indicates ambient temperature. This indicates the nominal operating temperature of the photovoltaic module, which is also a fixed parameter in the equipment development specifications.

[0051] This indicates the temperature rise of the component caused by solar irradiance. This represents the difference between the nominal operating temperature and the reference ambient temperature, reflecting the component's temperature rise capability under standard test conditions. The denominator 800 represents the reference irradiance. The entire formula is essentially a linear scaling factor that converts real-time irradiance into the corresponding temperature rise.

[0052] This empirical formula is based on the thermal balance principle of photovoltaic modules. When the module is working, part of the solar radiation energy absorbed is converted into electrical energy, and the rest is converted into heat energy, resulting in an increase in temperature. The temperature rise is approximately proportional to the irradiance. It ignores secondary factors such as wind speed and installation method, and focuses on the two dominant variables of irradiance and ambient temperature. This is a widely adopted simplified calculation method for engineering, providing key input parameters for the physical model of photovoltaic power generation.

[0053] In a preferred embodiment of the present invention, the operating power generation efficiency is defined as the result of a calculation of the ratio of actual power generation to theoretical power generation.

[0054] S13. By comparing the operating power generation efficiency of each photovoltaic string in the same group, abnormal photovoltaic strings whose power generation efficiency deviates from the average level of the group are identified.

[0055] In a preferred embodiment of the present invention, the process for identifying abnormal photovoltaic strings whose power generation efficiency deviates from the average level within the group includes: for each comparable analysis group, calculating the statistical average and standard deviation of the power generation efficiency of the photovoltaic strings within the group.

[0056] Based on the statistical mean and standard deviation, and combined with the efficiency distribution characteristics of similar groups in historical operating efficiency data, the confidence level is analyzed to determine the dynamic efficiency deviation threshold for each comparable analysis group.

[0057] It should be noted that the confidence level analysis process described above is as follows: Based on historical operating efficiency data, photovoltaic string efficiency values ​​that are identical to those in the current comparable analysis group in terms of component configuration, installation tilt angle, and geographical location, and that have been operating during the same historical period, are selected to construct a historical efficiency sample set. Statistical analysis is then performed on this sample set to determine its distribution pattern, and the historical mean and historical standard deviation are calculated.

[0058] Obtain the number of photovoltaic strings in the current group, and use the ratio of the historical standard deviation to the square root of the number of photovoltaic strings in the current group as the standard error of the group's efficiency average. The standard error is used to characterize the expected fluctuation range of the sample average.

[0059] The confidence level is set according to the operation and maintenance strategy, such as 90%, and the corresponding critical value is queried based on the selected statistical distribution. The critical value is multiplied by the standard error to obtain the expected fluctuation range half-width of the group's efficiency statistical average relative to the historical level. This half-width is the quantitative representation of the confidence level, which comprehensively reflects the historical volatility, the current group size, and the statistical significance requirements, providing a basis for dynamically setting the efficiency deviation threshold.

[0060] The dynamic efficiency deviation threshold for each comparable analysis group is synthesized from two parts: i. The basic deviation based on the current dispersion within the group: usually taken as the product of a predefined multiplier factor and the standard deviation of the photovoltaic string operating power generation efficiency within the group, used to capture the normal dispersion range within the group.

[0061] ii. Adjustment based on historical confidence level: that is, the half-width of the confidence interval calculated above, used to take into account the reasonable fluctuation of the current group's average efficiency relative to the historical level.

[0062] The base deviation based on the current group dispersion is summed with the correction based on the historical confidence level to determine the dynamic efficiency deviation threshold for each comparable analysis group.

[0063] The operating power generation efficiency of each photovoltaic string in the group is compared with the statistical average value. When the efficiency deviation exceeds the corresponding dynamic efficiency deviation threshold, it is identified as an abnormal photovoltaic string.

[0064] See Figure 3 As shown, S14. Extract the operating parameter sequence of the abnormal photovoltaic string and calculate its correlation compliance index relative to the pre-established association rule base. The association rule base is constructed based on the physical principle of photovoltaic power generation and the historical operating parameter set, and is used to define the expected correlation relationship between key operating parameters.

[0065] In a preferred embodiment of the present invention, the process of establishing the association rule base includes: obtaining a set of historical operating parameters, wherein the set of parameters includes synchronous time-series data of irradiance, component temperature, DC current, DC voltage, inverter input and output power in multiple continuous operating cycles.

[0066] Based on the physical principles of photovoltaic power generation, the theoretical relationships between key operating parameters are predefined, and the primary and secondary priorities of parameter pairs are established to form physical rule constraints.

[0067] It should be noted that the theoretical correlation between the predefined key operating parameters includes, but is not limited to, the positive correlation between irradiance and DC current.

[0068] There is a negative correlation between component temperature and DC voltage.

[0069] Under maximum power point tracking, the inverse constraint relationship between DC voltage and DC current on the IV curve.

[0070] The conversion efficiency relationship between the inverter's input power and output power.

[0071] The primary and secondary priorities in the parameter pair are determined based on physical causal logic. Parameters that are the cause of change or independent variables are defined as primary parameters, while parameters that are the result of change or dependent variables are defined as secondary parameters. For example, in the parameter pair of irradiance and DC current, irradiance is defined as the primary parameter and DC current is defined as the secondary parameter.

[0072] The historical running dataset is analyzed using an association rule mining algorithm to identify frequently co-occurring parameter combinations. The correlation strength between parameters is calculated and verified based on the conditional probability distribution, and the primary and secondary priority relationships are verified.

[0073] It should be added that the above-mentioned pattern analysis of the historical operation dataset using the association rule mining algorithm specifically includes: dividing the continuous data in the historical operation parameter set into several intervals with fuzzy properties, such as discretizing the voltage value into normal range, high, low, etc. Each parameter and its discretized state constitute a data item, and the set of all data items constitutes a transaction database.

[0074] Using existing The algorithm traverses the transaction database, constructs a frequent pattern tree, and recursively mines conditional pattern bases to retrieve all frequent itemsets with support higher than a preset minimum support threshold. Support refers to the frequency of parameter combinations in the entire dataset. Its calculation logic is the proportion of the number of times the parameter combination appears to the total number of data records. The higher the support, the more common the parameter combination is.

[0075] The above calculation of the correlation strength between parameters based on conditional probability distribution involves the following steps: For candidate association rules generated from the frequent itemset, calculate the confidence and lift of each rule. Confidence refers to the probability that, in an association rule, the result will also occur when the precondition occurs. Its value equals the proportion of the number of times the parameter combination occurs to the number of times each precondition occurs individually. A higher confidence indicates a stronger certainty that the result will occur when the precondition is true.

[0076] Lift measures the strength of the facilitation effect of the occurrence of a precondition on the occurrence of a result. Its value is equal to the ratio of the rule's confidence to the prevalence of the result spontaneously occurring in the dataset.

[0077] When the lift is greater than 1, it indicates that the premise has a positive promoting effect on the result, and the larger the value, the stronger the correlation.

[0078] When the lift is 1, it means that the premise and the result are independent of each other and there is no specific relationship between them.

[0079] When the lift is less than 1, it means that the presence of the premise may actually inhibit the occurrence of the result.

[0080] The association patterns obtained from data mining are compared with the physical rule constraints to retain the association rules that conform to physical principles and are statistically significant, thus generating the association rule library.

[0081] It should also be noted that the specific process of verifying the primary and secondary priority relationships mentioned above includes: comparing the strong association rules obtained from data mining that meet the statistical significance conditions, that is, the rules that simultaneously meet the minimum support, minimum confidence threshold and lift greater than 1, with the predefined physical rule constraints.

[0082] This is achieved by analyzing the directional asymmetry of the confidence levels for parameter pairs in the rules. Specifically, for a parameter pair defined in the physical principles, the primary parameter is denoted as P, and the secondary parameter is denoted as S. The calculation rules are then used. The confidence level, and compared with the reverse rule The confidence levels were compared.

[0083] Validation criterion: If and only if the rule The confidence level is higher than that of the rule. When the confidence level is preset, if the data mining results are considered to verify the predefined primary and secondary priority relationship, the rule will be retained; otherwise, it will be regarded as an invalid or abnormal rule and removed.

[0084] In a preferred embodiment of the present invention, the correlation compliance index calculation process includes: obtaining the actual operating parameter sequence of the abnormal photovoltaic string within a preset time window.

[0085] Based on the expected association relationships between parameters defined in the association rule base, identify each pair of associated parameters and their primary and secondary priorities.

[0086] Based on the actual change sequence of the primary parameter in the parameter pair, the expected change sequence of the secondary parameter is derived through the expected correlation.

[0087] It should be noted that the derivation process of the expected change sequence of the secondary parameters is specifically to call the quantitative transformation function pre-existing in the association rule base corresponding to the parameter pair, take the actual change sequence of the primary parameters as input, and calculate the expected change sequence of the secondary parameters as output.

[0088] Based on the comparison between the expected and actual change sequences of secondary parameters, a multi-dimensional quantitative analysis was conducted on the synchronicity of the change direction, the coordination of the change magnitude, and the statistical correlation of the sequences of primary and secondary parameters.

[0089] It should be added that the synchronicity of the direction of change is calculated by measuring the matching accuracy of the first-order difference between the expected change sequence and the actual change sequence at the same time point using secondary parameters.

[0090] The consistency of change magnitude is determined by calculating the mean relative error between the expected sequence and the actual change sequence of secondary parameters at each time point.

[0091] Sequence statistical correlation is calculated by measuring the Pearson correlation coefficient between the expected and actual changes in the sequence, which are minor parameters.

[0092] The results of the multi-dimensional quantitative analysis are standardized to generate correlation compliance indices for each pair of related parameters.

[0093] It should be noted that the quantification process of the correlation consistency index is as follows: the three sub-indicators—matching accuracy, Pearson correlation coefficient, and the absolute value of the difference between the mean relative error and 1—are mapped to... Within the specified range, the three standardized sub-indicators are summed to obtain the final correlation compliance index.

[0094] S15. Based on the comprehensive power generation efficiency deviation and correlation compliance index, diagnose the cause of the abnormal photovoltaic string efficiency, and output the diagnosis result.

[0095] In a preferred embodiment of the present invention, the process of diagnosing the cause of the abnormal low efficiency of the photovoltaic string includes: establishing an abnormal cause feature library, which pre-stores efficiency deviation features corresponding to various abnormal causes and typical abnormal patterns of each parameter's correlation compliance index.

[0096] Calculate the degree of deviation in power generation efficiency of the abnormal photovoltaic string, the combination of correlation conformity indicators of each parameter, and the matching degree with each abnormal cause pattern in the feature library.

[0097] It should be noted that the above matching degree is calculated by combining the deviation of the power generation efficiency of the abnormal photovoltaic string with the correlation conformity index of each parameter to form a feature vector. The cosine similarity or the reciprocal of the Euclidean distance between the feature vector of the current abnormal string and the feature vector of each abnormal cause template in the feature library is then calculated. The higher the similarity or the closer the distance, the higher the matching degree.

[0098] Based on the matching degree calculation results, a weighted decision algorithm is used to determine the type of abnormal cause, where the weights are assigned according to the importance of each parameter in the abnormal diagnosis.

[0099] It should be noted that the weighted decision algorithm specifically refers to the linear weighted fusion calculation method. The weights are pre-allocated based on the diagnostic specificity of each parameter pair for different abnormal causes. For a given abnormal cause, the parameter pair that can uniquely indicate the abnormality will be given a higher weight.

[0100] When multiple parameters simultaneously indicate an anomaly in the same equipment component according to the correlation compliance index, the anomaly score for that component is superimposed to enhance it.

[0101] It should also be noted that the superposition enhancement refers to the weighted summation of the matching degree of all associated parameter pairs under the device link, and the summation result is multiplied by an enhancement coefficient greater than 1.

[0102] The output includes a diagnostic report containing the type of cause of the anomaly, the location of the abnormal equipment or component, and an anomaly confirmation score.

[0103] The above-mentioned anomaly confirmation score is the final score calculated for each candidate anomaly cause in the weighted decision algorithm. Its value is the sum of all relevant parameters weighted according to their matching degree and after superposition enhancement adjustment. The candidate anomaly cause with the highest score is selected as the final diagnosis result.

[0104] This invention establishes a rule base for association between key operating parameters, deeply integrating the physical principles of photovoltaic power generation with the historical operating parameter set of the power station. This constructs a dynamic and adaptive equipment health behavior reference system. By calculating the correlation conformity index between the actual operating parameters of abnormal photovoltaic strings and the rule base, it achieves in-depth diagnosis of the root causes behind inefficiency, elevating the analysis capability from phenomenon identification to causal localization, and providing clear guidance for subsequent operation and maintenance.

[0105] In a preferred embodiment of the present invention, the following closed-loop optimization steps are also included: automatically generating maintenance work orders based on the diagnosed causes of inefficiency, and scheduling operation and maintenance resources to perform on-site maintenance.

[0106] The diagnostic accuracy is evaluated based on the on-site maintenance verification results, and the parameter thresholds in the association rule base are dynamically adjusted based on the diagnostic accuracy.

[0107] It should be noted that the specific strategy for the above dynamic adjustment is as follows: a target range for diagnostic accuracy is preset. When the actual diagnostic accuracy is consistently higher than the upper limit of the target range, the confidence threshold in the association rule base is increased by a first preset step size to improve rule quality and reduce false positives. When the accuracy is consistently lower than the lower limit of the target range, the confidence threshold is decreased by a second preset step size to uncover more potential rules and reduce false negatives.

[0108] Verify the efficiency recovery of the photovoltaic strings after maintenance, and update the operation data and verification results during the maintenance process to the historical database.

[0109] The association rule base is periodically retrained using updated historical data containing maintenance verification results to continuously adapt to changes in device status.

[0110] In a preferred embodiment of the present invention, the following abnormal handling steps are also included: when no abnormal photovoltaic strings deviating from the average level are identified in the comparable analysis group, the statistical average value of the operating power generation efficiency of the comparable analysis group is compared with the historical group efficiency benchmark.

[0111] When the average operating power generation efficiency is consistently lower than the group efficiency benchmark and the deviation exceeds the set threshold, the system-level anomaly diagnosis process is initiated: by verifying the consistency of regional readings of the irradiance sensor, analyzing the historical efficiency curve of the inverter, and checking the regional environmental images, macroscopic anomalies affecting the entire region are identified.

[0112] It should be noted that the verification of the regional reading consistency of the irradiance sensor refers to the coefficient of variation of the readings of all sensors in the calculated area at the same time. If the coefficient is consistently higher than a preset threshold, it is determined that there is sensor inaccuracy.

[0113] The analysis of the historical efficiency curve of the inverter refers to the trend fitting of the historical efficiency data sequence of the inverter in the region. If the slope is continuously negative and the absolute value exceeds the threshold, it is determined that there is efficiency degradation.

[0114] The environmental images of the inspection area refer to panoramic images of the site collected by drones or high-definition cameras, and the use of image recognition technology to quantitatively assess the pollution and erosion rate of the component surface.

[0115] Output a system-level anomaly diagnostic report, indicating macroscopic anomaly types, including sensor misalignment, inverter efficiency degradation, or component contamination, and corresponding maintenance recommendations.

[0116] This invention effectively combines data analysis with operation and maintenance management processes. Through automated cause diagnosis, work order generation, and maintenance effect verification, a closed-loop intelligent operation and maintenance management system is formed. Furthermore, by introducing a machine learning mechanism to continuously optimize and update the association rule base, the analysis method is ensured to maintain long-term effectiveness and adaptability throughout the entire lifecycle of the photovoltaic power station, thereby fundamentally improving the overall operational efficiency of the photovoltaic power station.

[0117] The above formulas are all dimensionless calculations. The formulas are derived from software simulations based on a large amount of collected data to obtain the most recent real-world results. The preset parameters in the formulas are set by those skilled in the art according to the actual situation.

[0118] The above embodiments can be implemented, in whole or in part, by software, hardware, firmware, or any other combination thereof. When implemented using software, the above embodiments can be implemented, in whole or in part, in the form of a computer program product.

[0119] Those skilled in the art will recognize that the modules and algorithm steps of the various examples described in conjunction with the embodiments disclosed herein can be implemented in electronic hardware, or a combination of computer software and electronic hardware. Whether these functions are implemented in hardware or software depends on the specific application and design constraints of the technical solution. Those skilled in the art can use different methods to implement the described functions for each specific application, but such implementation should not be considered beyond the scope of this application.

[0120] In addition, the functional modules in the various embodiments of this application can be integrated into one processing module, or each module can exist physically separately, or two or more modules can be integrated into one module.

[0121] The above description is merely a specific embodiment of this application, but the scope of protection of this application is not limited thereto. Any variations or substitutions that can be easily conceived by those skilled in the art within the scope of the technology disclosed in this application should be included within the scope of protection of this application. Therefore, the scope of protection of this application should be determined by the scope of the claims.

[0122] Finally, the above description is only a preferred embodiment of the present invention and is not intended to limit the present invention. Any modifications, equivalent substitutions, improvements, etc., made within the spirit and principles of the present invention should be included within the protection scope of the present invention.

Claims

1. A method for analyzing the power generation efficiency of a distributed photovoltaic power station, characterized in that, include: Based on the consistency of effective light-receiving conditions and the consistency of component configuration parameters, the photovoltaic strings in the site are divided into multiple comparable analysis groups; The consistency of effective light-receiving conditions is determined based on the comprehensive difference between the real-time effective light-receiving area and the real-time effective light-receiving intensity. For each of the comparable analysis groups, based on real-time environmental data, the theoretical power generation of each photovoltaic string in the group is calculated according to the photovoltaic power generation physical model, and the actual power generation is obtained to calculate the operating power generation efficiency. By comparing the power generation efficiency of each photovoltaic string within the same group, abnormal photovoltaic strings whose power generation efficiency deviates from the average level within the group are identified. Extract the operating parameter sequence of the abnormal photovoltaic string and calculate its correlation conformity index relative to the pre-established association rule base. The association rule base is constructed based on the physical principle of photovoltaic power generation and the historical operating parameter set, and is used to define the expected correlation relationship between key operating parameters based on primary and secondary priorities. By combining the deviation of power generation efficiency with the correlation conformity index, the cause of the abnormal low efficiency of the photovoltaic string is diagnosed, and the diagnosis result is output. It also includes the following exception handling steps: When no abnormal photovoltaic strings deviating from the average level are identified within the comparable analysis group, the statistical average of the operating power generation efficiency of the comparable analysis group is compared with the historical group efficiency benchmark for the same period. When the average operating power generation efficiency remains below the group efficiency benchmark and the deviation exceeds a set threshold, the system-level anomaly diagnosis process is initiated: By verifying the consistency of regional readings from irradiance sensors, analyzing historical efficiency curves of inverters, and examining regional environmental images, macroscopic anomalies affecting the entire region can be identified. Output a system-level anomaly diagnostic report, indicating macroscopic anomaly types, including sensor misalignment, inverter efficiency degradation, or component contamination, and corresponding maintenance recommendations.

2. The method for analyzing the power generation efficiency of a distributed photovoltaic power station according to claim 1, characterized in that, The process of dividing the comparable analysis groups includes: The surrounding environment of the photovoltaic string is acquired by an image acquisition device. The shadow distribution on the surface of the photovoltaic module is analyzed based on image recognition to determine the real-time effective light-receiving area of ​​each photovoltaic string. The real-time effective light intensity of each photovoltaic string is collected synchronously through a distributed irradiance sensor network deployed within the site. Based on the real-time effective light-receiving area and real-time effective light-receiving intensity, the matching degree of light-receiving conditions between each photovoltaic string is calculated. The specific calculation process for the matching degree of light reception conditions between photovoltaic strings is as follows: For any two photovoltaic strings to be compared, their real-time effective light-receiving area and real-time effective light-receiving intensity are obtained respectively, which together constitute their real-time effective light-receiving parameter set; Calculate the absolute difference between two photovoltaic strings on the same type of light-receiving parameters; Obtain the arithmetic mean of all photovoltaic strings in the power station at the current moment for this type of light-receiving parameter, as a reference benchmark; The absolute difference is compared with the arithmetic mean to obtain a dimensionless parameter of difference that characterizes the relative difference. The difference between the effective light-receiving area and the effective light-receiving intensity parameters are obtained separately, and the reciprocal of the sum of the difference between the two parameters is used as the matching degree of the light-receiving conditions. Obtain the component configuration parameters for each photovoltaic string, including cell technology type, module series topology, and inverter electrical characteristics. Photovoltaic strings with a matching degree of light reception conditions higher than the preset matching threshold and consistent component configuration parameters are classified into the same comparable analysis group.

3. The method for analyzing the power generation efficiency of a distributed photovoltaic power station according to claim 1, characterized in that, The theoretical power generation calculation process for each photovoltaic string within the group includes: Obtain real-time environmental parameters, including real-time irradiance and ambient temperature, as well as the rated parameters of the photovoltaic string equipment; The real-time environmental parameters and equipment rated parameters are input into the photovoltaic power generation physical model, which calculates the theoretical power generation by coupling the effects of irradiance and temperature on power generation. The model outputs the theoretical power generation value under the current environmental conditions, and the calculation process includes an estimation step of the component operating temperature based on the ambient temperature and real-time irradiance.

4. The method for analyzing the power generation efficiency of a distributed photovoltaic power station according to claim 1, characterized in that, The operating power generation efficiency is defined as the ratio of actual power generation to theoretical power generation.

5. The method for analyzing the power generation efficiency of a distributed photovoltaic power station according to claim 1, characterized in that, The process for identifying abnormal photovoltaic strings whose power generation efficiency deviates from the group's average level includes: For each comparable analysis group, calculate the statistical mean and standard deviation of the photovoltaic string operating power generation efficiency within the group; Based on the statistical mean and standard deviation, and combined with the efficiency distribution characteristics of similar groups in historical operational efficiency data, the confidence level is analyzed to determine the dynamic efficiency deviation threshold for each comparable analysis group. The operating power generation efficiency of each photovoltaic string in the group is compared with the statistical average value. When the efficiency deviation exceeds the corresponding dynamic efficiency deviation threshold, it is identified as an abnormal photovoltaic string.

6. The method for analyzing the power generation efficiency of a distributed photovoltaic power station according to claim 1, characterized in that, The process of establishing the association rule base includes: Obtain a set of historical operating parameters, which includes synchronous timing data of irradiance, component temperature, DC current, DC voltage, inverter input and output power in multiple consecutive operating cycles; Based on the physical principles of photovoltaic power generation, the theoretical relationships between key operating parameters are predefined, and the primary and secondary priorities of parameter pairs are established to form physical rule constraints. The historical operating parameter set is analyzed by association rule mining algorithm to identify frequently co-occurring parameter combinations. The correlation strength between parameters is calculated and verified based on conditional probability distribution. The association patterns obtained from data mining are compared with the physical rule constraints to retain the association rules that conform to physical principles and are statistically significant, thus generating the association rule library.

7. The method for analyzing the power generation efficiency of a distributed photovoltaic power station according to claim 6, characterized in that, The calculation process for the correlation compliance index includes: Obtain the actual operating parameter sequence of abnormal photovoltaic strings within a preset time window; Based on the expected association relationships between parameters defined in the association rule base, identify each pair of associated parameters and their primary and secondary priorities; Based on the actual change sequence of the primary parameter in the parameter pair, the expected change sequence of the secondary parameter is derived through the expected correlation. Based on the comparison between the expected and actual change sequences of secondary parameters, a multi-dimensional quantitative analysis was conducted on the synchronicity of the change direction, the coordination of the change magnitude, and the statistical correlation of the sequences of primary and secondary parameters. The results of the multi-dimensional quantitative analysis are standardized to generate correlation compliance indices for each pair of related parameters.

8. The method for analyzing the power generation efficiency of a distributed photovoltaic power station according to claim 1, characterized in that, The process of diagnosing the cause of the abnormal low efficiency of the photovoltaic string includes: Establish an anomaly cause feature library, which pre-stores efficiency deviation features corresponding to various anomalies and typical anomaly patterns of each parameter's correlation conformity index. Calculate the degree of deviation in power generation efficiency of the abnormal photovoltaic string, the combination of correlation conformity indicators of each parameter, and the matching degree with each abnormal cause pattern in the feature library; Based on the matching degree calculation results, a weighted decision algorithm is used to determine the type of abnormal cause, wherein the weights are assigned according to the importance of each parameter in the abnormal diagnosis. When multiple parameters simultaneously indicate an anomaly in the same equipment component according to the correlation compliance index, the anomaly score for that component is superimposed to enhance it. The output includes a diagnostic report containing the type of cause of the anomaly, the location of the abnormal equipment or component, and an anomaly confirmation score.

9. The method for analyzing the power generation efficiency of a distributed photovoltaic power station according to claim 1, characterized in that, It also includes the following closed-loop optimization steps: Maintenance work orders are automatically generated based on the diagnosed causes of inefficiency, and maintenance resources are scheduled to perform on-site maintenance. The diagnostic accuracy is evaluated based on the on-site maintenance verification results, and the parameter thresholds in the association rule base are dynamically adjusted based on the diagnostic accuracy. Verify the efficiency recovery of the photovoltaic strings after maintenance, and update the operation data and verification results during the maintenance process to the historical database; The association rule base is periodically retrained using updated historical data containing maintenance verification results to continuously adapt to changes in device status.

Images