String fault identification and loss assessment method and system in photovoltaic power generation scene

Abstract

The invention provides a string fault identification and loss assessment method and system in a photovoltaic power generation scene, and relates to the technical field of fault identification, and the method comprises the steps: carrying out the correction through importing the electrical quantity data of a target photovoltaic string, carrying out the low-efficiency degree evaluation based on the corrected electrical quantity data, and screening to obtain an abnormal low-efficiency string; performing performance degradation analysis on the abnormal low-efficiency string to obtain key degradation indexes, and drawing a current curve graph; based on the key degradation index, fault cause diagnosis is carried out on the abnormal low-efficiency string, and a fault diagnosis result is generated; and based on the fault diagnosis result, calculating historical power generation loss, performing future loss trend prediction, and generating a power generation loss evaluation result. The technical problem of low accuracy and reliability of string fault identification and loss assessment in the prior art is solved. The technical effects of effectively improving the efficiency, accuracy and reliability of group cascade fault diagnosis and improving the power generation efficiency, safety and stability of a power station are achieved.

Description

Technical Field

[0001] This invention relates to the field of fault identification technology, specifically to a method and system for identifying and assessing string faults in photovoltaic power generation scenarios. Background Technology

[0002] In a photovoltaic (PV) power generation system, several PV modules are connected in series to form a circuit unit with a certain DC output, referred to as a module string or array. In a string inverter scheme, each PV string is connected to a designated inverter DC input terminal, thus avoiding significant power loss caused by shading of some PV modules. As the core unit of a PV power generation system, the operating status of the PV string directly affects the power generation efficiency and safety stability of the entire power station. With the continuous expansion of PV power station installed capacity, real-time monitoring and fault diagnosis of the string's operating status have become crucial. However, in actual operation, PV strings are susceptible to various factors, such as uneven illumination, shading, module aging, PID effects, and junction box failures. These factors can cause changes in the string's output characteristics, thereby affecting power generation and even causing safety issues.

[0003] Currently, commonly used string fault identification methods mainly rely on manual inspection or simple discrete rate or electrical quantity threshold judgments. Manual inspection is inefficient and difficult to implement for real-time monitoring of large-scale power plants. Threshold-based methods, while simple to implement, are easily affected by environmental factors and noise, leading to a high false alarm rate. Furthermore, these methods often only identify obvious faults that have already occurred, failing to effectively warn of early or potential faults. Existing fault detection methods cannot quickly and accurately locate faulty photovoltaic strings and cannot accurately identify shading, resulting in low detection accuracy. Regarding loss prediction, most existing methods rely on statistical analysis of historical data, making it difficult to accurately predict power generation losses over a future period. Simultaneously, these methods typically ignore the differences between strings and the influence of environmental factors, leading to low accuracy in prediction results.

[0004] In summary, existing technologies for string fault identification and damage assessment suffer from low accuracy and reliability. Summary of the Invention

[0005] The purpose of this application is to provide a method and system for identifying and assessing string faults in photovoltaic power generation scenarios, in order to solve the technical problems of low accuracy and reliability in existing string fault identification and assessment technologies.

[0006] In view of the above problems, this application provides a method and system for string fault identification and damage assessment in photovoltaic power generation scenarios.

[0007] The first aspect of this application provides a method for identifying and assessing string faults in a photovoltaic power generation scenario. The method includes: importing electrical quantity data of a target photovoltaic string, correcting the electrical quantity data based on irradiance data, assessing the degree of inefficiency based on the corrected electrical quantity data, and screening for abnormally inefficient strings; performing performance degradation analysis on the abnormally inefficient strings, obtaining key degradation indicators, and plotting current curves; diagnosing the causes of faults in the abnormally inefficient strings based on the key degradation indicators, and generating fault diagnosis results; calculating historical power generation losses based on the fault diagnosis results, predicting future loss trends, and generating power generation loss assessment results.

[0008] Optionally, the system receives the device identification information and time range of the target photovoltaic string input by the user; obtains the corresponding string electrical quantity data and site irradiance data based on the device identification information and time range; performs linear correction on the string electrical quantity data using the irradiance data, and inputs the linearly corrected string electrical quantity data into a pre-trained MiniRocket binary classification model to determine whether the target photovoltaic string is in an abnormally inefficient state; and selects the abnormally inefficient strings based on the inefficiency state determination result.

[0009] Optionally, the abnormally inefficient strings can be visually marked on the user interface according to a predefined color mapping specification.

[0010] Optionally, for the abnormally inefficient string, a multilayer perceptron is used to calculate and obtain key degradation indicators, which include at least the degradation rate; based on the key degradation indicators, current curves are plotted in the same coordinate system to generate a first current curve and a second current curve; wherein, the first current curve is the current curve of all strings in the inverter within a preset period, and the second current curve is the current curve of the abnormally inefficient string during the same period.

[0011] Optionally, based on the key degradation indicators, multiple electrical parameters of the abnormally inefficient string are collected within the abnormal duration interval; the multiple electrical parameters are input into a pre-trained multi-class recognition engine to identify the fault defect type and its relative weight; the fault defect type and the relative weight are visualized in a pie chart as the fault diagnosis result.

[0012] Optionally, based on each fault type, corresponding historical normal operation data is extracted, and its theoretical power generation curve under ideal conditions is reconstructed; the theoretical power generation curve is compared point by point with the actual power generation curve of the abnormally inefficient string, and the difference sequence between the two is calculated; the difference sequence is integrated within the abnormal duration interval to generate single-type fault power generation loss, and weighted by the relative weight to calculate the cumulative power generation loss; a linear regression algorithm is used to predict the predicted power generation loss of the abnormally inefficient string in a preset future period; based on the cumulative power generation loss and the predicted power generation loss, a power generation loss assessment result is generated.

[0013] Optionally, the service port of the unmanned patrol device can be linked to retrieve the optical images and thermal radiation maps of the abnormally inefficient cluster within a preset historical period; and the optical images and thermal radiation maps can be displayed in the interactive screen in a predetermined order.

[0014] A second aspect of this application provides a string fault identification and loss assessment system for photovoltaic power generation scenarios. The system includes: a data evaluation module for importing electrical quantity data of a target photovoltaic string, correcting the electrical quantity data based on irradiation data, evaluating the degree of inefficiency based on the corrected electrical quantity data, and screening for abnormally inefficient strings; a performance degradation analysis module for performing performance degradation analysis on the abnormally inefficient strings, obtaining key degradation indicators, and plotting current curves; a fault cause diagnosis module for diagnosing the fault causes of the abnormally inefficient strings based on the key degradation indicators, and generating fault diagnosis results; and a loss trend prediction module for calculating historical power generation losses based on the fault diagnosis results, predicting future loss trends, and generating power generation loss assessment results.

[0015] One or more technical solutions provided in this application have at least the following technical effects or advantages:

[0016] The string fault identification and damage assessment method for photovoltaic power generation scenarios provided in this application imports electrical quantity data of the target photovoltaic string, corrects the electrical quantity data based on irradiation data, assesses the degree of inefficiency based on the corrected electrical quantity data, and screens out abnormally inefficient strings; performs performance degradation analysis on the abnormally inefficient strings, obtains key degradation indicators, and plots current curves; based on the key degradation indicators, diagnoses the fault causes of the abnormally inefficient strings, and generates fault diagnosis results; based on the fault diagnosis results, calculates historical power generation losses, predicts future loss trends, and generates power generation loss assessment results. This achieves intelligent analysis of the entire process of photovoltaic string faults, from inefficiency identification, degradation assessment, root cause diagnosis to loss quantification, improving the accuracy and efficiency of string-level fault diagnosis, providing maintenance personnel with precise maintenance basis, helping to eliminate hidden dangers in a timely manner, minimizing power generation losses, and thus improving the overall operation and management level and economic benefits of the photovoltaic power station.

[0017] The above description is merely an overview of the technical solution of this application. To enable a clearer understanding of the technical means of this application and to facilitate its implementation according to the description, and to make the above and other objects, features, and advantages of this application more apparent, specific embodiments of this application are described below. It should be understood that the content described in this section is not intended to identify key or important features of the embodiments of this application, nor is it intended to limit the scope of this application. Other features of this application will become readily apparent through the following description. Attached Figure Description

[0018] To more clearly illustrate the technical solutions in this application or the prior art, the drawings used in the description of the embodiments or the prior art will be briefly introduced below. Obviously, the drawings described below are merely exemplary. For those skilled in the art, other drawings can be obtained based on the provided drawings without creative effort.

[0019] Figure 1 This is a flowchart illustrating the string fault identification and damage assessment method for photovoltaic power generation scenarios provided in this application.

[0020] Figure 2 This is a schematic diagram of the string fault identification and damage assessment system for photovoltaic power generation scenarios provided in this application.

[0021] Figure labeling: Data evaluation module 11, performance degradation analysis module 12, fault cause diagnosis module 13, loss trend prediction module 14. Detailed Implementation

[0022] This application provides a method and system for string fault identification and damage assessment in photovoltaic power generation scenarios, addressing the technical problems of low accuracy and reliability in existing string fault identification and damage assessment technologies. It effectively improves the efficiency, accuracy, and reliability of string-level fault diagnosis, thereby enhancing the power generation efficiency and safety stability of the power plant.

[0023] The technical solutions of the present invention will now be clearly and completely described with reference to the accompanying drawings. Obviously, the described embodiments are only a part of the embodiments of the present invention, and not all of them. It should be understood that the present invention is not limited to the exemplary embodiments described herein. All other embodiments obtained by those skilled in the art based on the embodiments of the present invention without creative effort are within the scope of protection of the present invention. It should also be noted that, for ease of description, only the parts related to the present invention are shown in the accompanying drawings, not all of them.

[0024] Example 1, as Figure 1 As shown, this application provides a method for identifying and assessing string faults in a photovoltaic power generation scenario. The method includes:

[0025] Import the electrical quantity data of the target photovoltaic string, correct the electrical quantity data based on the irradiance data, evaluate the degree of inefficiency based on the corrected electrical quantity data, and screen out abnormally inefficient strings.

[0026] Furthermore, the electrical quantity data of the target photovoltaic string is imported, and the electrical quantity data is corrected based on the irradiance data. An inefficiency assessment is performed based on the corrected electrical quantity data, and abnormally inefficient strings are selected. This includes: receiving the device identification information and time range of the target photovoltaic string input by the user; obtaining the corresponding string electrical quantity data and site irradiance data based on the device identification information and time range; linearly correcting the string electrical quantity data using the irradiance data, and inputting the linearly corrected string electrical quantity data into a pre-trained MiniRocket binary classification model to determine whether the target photovoltaic string is in an abnormally inefficient state; and selecting the abnormally inefficient strings based on the inefficiency state determination result.

[0027] Specifically, the user provides the device identification information and time range for the target photovoltaic string to be analyzed. The device identification information refers to the identification content used to identify a specific photovoltaic string, such as the specific site name, transformer serial number, inverter serial number, and specific string serial number. The time range refers to the target date range, which specifies the start and end time periods for the data to be analyzed. Based on the device identification information and time range of the target photovoltaic string input by the user, the corresponding string electrical quantity data and site irradiance data are obtained from the photovoltaic power plant operation database or related acquisition system. The string electrical quantity data includes parameters reflecting the power generation performance of the photovoltaic string, such as voltage, current, and power. The site irradiance data refers to the solar radiation intensity information received by the photovoltaic site within the corresponding time period.

[0028] Since the irradiance of a photovoltaic (PV) power plant directly affects the power generation performance of PV strings, the actual power generation efficiency reflected by the electrical quantity data of the same string under different irradiance conditions varies. Linear correction of string electrical quantity data can be performed using irradiance data. For example, irradiance data of the PV power plant and the corresponding electrical quantity data of the target PV string are obtained within a time range. The average irradiance intensity of the area where the PV power plant is located under ideal weather conditions is used as the standard irradiance value. The ratio of the actual irradiance value to the standard irradiance value is calculated. The string's electrical quantity data at that moment is then multiplied by this ratio to achieve linear correction of the string's electrical quantity data. This linear correction method eliminates the impact of irradiance fluctuations on the electrical quantity data, thereby more accurately assessing the power generation performance and inefficiency of the PV string.

[0029] Linearly corrected string electrical quantity data, including current, voltage, and power, are input to a pre-trained MiniRocket binary classification model for normal or abnormal classification. The model outputs a judgment result indicating whether the target photovoltaic string is in an abnormally inefficient state. Based on the inefficiency state judgment result, abnormally inefficient strings are selected. This involves collecting a large number of labeled photovoltaic string electrical quantity data samples, including various situations under normal and abnormally inefficient states, with each sample having a clear label indicating its state category. The photovoltaic string electrical quantity data samples are divided into training, validation, and test sets in an 8:1:1 ratio. The training set is used to train the MiniRocket binary classification model. The MiniRocket binary classification model is an efficient time-series data feature extraction and classification model that can automatically extract key features from time-series data and learn the feature patterns of data under different states. During training, the MiniRocket binary classification model continuously adjusts its internal parameters to minimize loss functions, such as cross-entropy loss, thereby improving the classification accuracy for normal and abnormally inefficient states. The performance of the MiniRocket binary classification model is monitored and evaluated in real time using a validation set. Performance metrics, such as the learning rate, are used to assess the classification performance of the MiniRocket binary classification model, preventing overfitting or underfitting. Once the performance of the MiniRocket binary classification model reaches a stable state, a test set is used to evaluate the final MiniRocket binary classification model, ensuring that it has good generalization ability and can reliably, accurately, and quickly identify abnormally inefficient states of photovoltaic strings.

[0030] By correcting electrical quantity data using irradiation data and using a pre-trained model to assess inefficiency, the efficiency and accuracy of identifying and screening abnormal inefficient strings are improved, reducing false positives and false negatives, thereby improving the efficiency and reliability of string fault identification.

[0031] Furthermore, after obtaining the abnormal inefficient strings based on the inefficiency state discrimination result, the method further includes: visually marking the abnormal inefficient strings on the user interface according to a predefined color mapping specification.

[0032] Specifically, after obtaining abnormally inefficient strings, according to a predefined color mapping specification, the corresponding color codes are assigned to the corresponding string elements on the user interface through programming logic, thus visually indicating the abnormally inefficient strings on the user interface. The predefined color mapping specification clearly defines the colors corresponding to different levels of inefficiency. For example, the evaluation results are marked on the user interface using an intuitive three-color coding method of red, yellow, and green, such as a string layout diagram or list. Red indicates severe inefficiency, yellow indicates moderate inefficiency, and green indicates normal or slight inefficiency.

[0033] Visualized labels enable users to quickly identify problematic strings. In large-scale photovoltaic power plants with numerous strings, users can quickly locate abnormal and inefficient strings and make corresponding decisions by using colors, saving a lot of time and effort and improving the overall operation and maintenance efficiency of photovoltaic power plants.

[0034] The performance degradation analysis of the abnormally inefficient string was performed to obtain key degradation indicators and plot the current curve.

[0035] Furthermore, a performance degradation analysis is performed on the abnormally inefficient string to obtain key degradation indicators and plot current curves. This includes: for the abnormally inefficient string, using a multilayer perceptron to calculate and obtain key degradation indicators, the key degradation indicators including at least the degradation rate; based on the key degradation indicators, plotting current curves in the same coordinate system to generate a first current curve and a second current curve; wherein, the first current curve is the current curve of all strings in the inverter within a preset period, and the second current curve is the current curve of the abnormally inefficient string during the same period.

[0036] Specifically, a multilayer perceptron (MLP) algorithm is used to process and calculate the selected abnormally inefficient strings. The MLP algorithm is a feedforward artificial neural network consisting of an input layer, multiple hidden layers, and an output layer. It learns and calculates by inputting historical electrical quantity data of the abnormally inefficient strings, and outputs key degradation indicators. These key degradation indicators include at least the degradation rate of the abnormally inefficient strings, used to characterize the degree of performance degradation of the abnormally inefficient strings.

[0037] Current curves are plotted based on key degradation indicators. Two current curves are generated on the same coordinate system based on these indicators: a first current curve and a second current curve. The first current curve represents the current changes of all strings under the same inverter belonging to the currently abnormally inefficient string within a preset period. The preset period is set according to actual needs, such as one week. By collecting current data from all strings under the inverter for seven consecutive days and plotting the curves, the current fluctuation range and trend of normal strings under the same environmental conditions can be displayed. The second current curve is the current curve of the abnormally inefficient string within the same time range, reflecting the actual current change trend of the abnormally inefficient string within the same time period. By plotting the first and second current curves on the same coordinate system, the current differences between the abnormally inefficient string and the normal string can be visually compared. For example, if the current curve of the abnormally inefficient string is significantly lower than that of the normal string and fluctuates abnormally, it may indicate that the abnormally inefficient string has degradation phenomena such as connection failure, low current, or shading.

[0038] By analyzing performance degradation and identifying key degradation indicators, the degree of performance degradation in abnormally inefficient strings can be quantitatively assessed. Furthermore, by plotting current curves under these key degradation indicators and comparing their shape and amplitude with those of other strings, the current changes in abnormally inefficient strings can be visually displayed, improving the accuracy and efficiency of string fault identification and diagnosis.

[0039] Based on the key degradation indicators, the fault causes of the abnormally inefficient series are diagnosed, and fault diagnosis results are generated.

[0040] Furthermore, based on the key degradation indicators, fault cause diagnosis of the abnormally inefficient series is performed, and fault diagnosis results are generated, including: based on the key degradation indicators, collecting multiple electrical parameters of the abnormally inefficient series within the abnormal duration interval; inputting the multiple electrical parameters into a pre-trained multi-class recognition engine to identify fault defect types and their relative weights; and visualizing the fault defect types and the relative weights in a pie chart as the fault diagnosis results.

[0041] Specifically, based on key degradation indicators, data acquisition devices, such as smart meters and data collectors, are used to collect multiple electrical parameters of abnormally inefficient strings within the abnormal duration interval. The abnormal duration interval refers to the time period from the onset of the string's abnormal inefficiency to the current moment. These multiple electrical parameters include various data reflecting the string's electrical performance, such as voltage, current, and power, reflecting electrical characteristics from different dimensions. These multiple electrical parameters are input into a pre-trained multi-class recognition engine. This multi-class recognition engine is a model built based on machine learning algorithms, such as a neural network model. Through training with a large amount of known fault types, including but not limited to photovoltaic string faults such as shading, surface contamination, diode failure, hot spot effect, and open circuit, and corresponding electrical parameters, it can learn the complex mapping relationship between different fault types and electrical parameters. During training, the multi-class recognition engine continuously adjusts its internal parameters to minimize prediction errors and improve the accuracy of identifying different fault types.

[0042] After inputting multiple electrical parameters of an abnormally inefficient string, the multi-class recognition engine identifies the fault defect type corresponding to the abnormally inefficient string based on the learned pattern, and calculates the relative weight of each fault type relative to other types. The relative weight reflects the proportion of different fault types in causing the string to be abnormally inefficient. For example, if the relative weight of the occlusion fault type is high, it means that the fault type has a greater impact on the performance of the current abnormally inefficient string.

[0043] The identified fault types and their relative weights are visualized using a pie chart, which serves as the fault diagnosis result, enabling a visual representation of the diagnosis. The pie chart displays the proportion of different anomaly types in a ring shape. Based on the identified fault type and its relative weight, the ring is divided into different sector regions, each representing a fault type. The size of the sector region is proportional to the relative weight of that fault type; that is, the area of ​​each sector corresponds to the specific proportion of that anomaly type in the overall diagnostic result.

[0044] By collecting multiple electrical parameters and inputting them into a multi-category identification engine, the algorithm model accurately identifies fault types and their relative weights, improving the accuracy and comprehensiveness of fault diagnosis. Visualizing the diagnostic results using a pie chart allows users to clearly understand the main causes of string inefficiency and their contribution, improving the targeting and reliability of maintenance decisions, and ultimately effectively improving photovoltaic power generation efficiency and operational stability.

[0045] Based on the fault diagnosis results, historical power generation losses are calculated, future loss trends are predicted, and power generation loss assessment results are generated.

[0046] Furthermore, based on the fault diagnosis results, historical power generation losses are calculated, and future loss trends are predicted to generate power generation loss assessment results. This includes: extracting corresponding historical normal operation data for each fault type and reconstructing its theoretical power generation curve under ideal conditions; comparing the theoretical power generation curve with the actual power generation curve of the abnormally inefficient string at each point and calculating the difference sequence between them; performing an integral operation on the difference sequence within the abnormal duration interval to generate single-type fault power generation loss, and weighting it with the relative weights to calculate the cumulative power generation loss; using a linear regression algorithm to predict the predicted power generation loss of the abnormally inefficient string within a preset future period; and generating power generation loss assessment results based on the cumulative power generation loss and the predicted power generation loss.

[0047] Specifically, monitoring equipment records the actual power generation loss of the abnormally inefficient string within a historical timeframe, such as the past month, forming an actual power generation curve. Based on the fault diagnosis results, for each fault type, historical normal operation data of the abnormally inefficient string or similar normal strings under similar weather conditions during the same historical period are extracted from the photovoltaic power generation operation database. This historical normal operation data includes parameters such as current, voltage, power, and irradiance of the abnormally inefficient string or similar normal strings under normal operating conditions. Using a multilayer sensor model, based on the historical normal operation data, a theoretical power generation curve of the abnormally inefficient string under ideal conditions is fitted and reconstructed. This theoretical power generation curve is the curve showing the change in power generation of the abnormally inefficient string over time under ideal conditions, i.e., without any abnormalities. The theoretical power generation curve and the actual power generation curve are compared point-by-point to calculate the difference sequence between them. The difference sequence is then integrated within the abnormal duration interval. The sum of the differences within the abnormal duration interval yields the power generation loss for a single type of fault. Similarly, the power generation loss for each fault type is obtained. Based on the power generation losses of multiple single-type faults, and combined with the relative weights in the fault diagnosis results, the cumulative power generation loss is generated by weighted summation.

[0048] A linear regression algorithm is used to predict the power generation loss of abnormally inefficient strings within a preset future period, such as one week. The algorithm establishes a linear relationship model between the dependent variable (power generation loss) and independent variables (time, environmental factors, etc.) to predict future power generation losses based on historical data and current fault conditions. A power generation loss assessment result is generated based on the cumulative and predicted power generation losses. This assessment result includes not only the historical cumulative power generation loss due to faults over a past period but also the predicted power generation loss. The assessment results are visualized using tools such as charts, such as bar charts or line charts.

[0049] By calculating historical power generation losses and predicting future loss trends, the impact of faults on the economic benefits of photovoltaic power plants can be comprehensively assessed. This provides data support for the economic benefit assessment and maintenance prioritization of photovoltaic power plants, enabling the rational allocation of resources and improving the overall operating efficiency and power generation benefits of photovoltaic power generation.

[0050] Furthermore, the method also includes: linking the service port of the unmanned patrol device to retrieve the optical images and thermal radiation spectra of the abnormally inefficient cluster within a preset historical period; and displaying the optical images and thermal radiation spectra in an interactive screen in a predetermined order.

[0051] Specifically, unmanned inspection equipment, such as inspection systems mounted on drones, is equipped with various sensors capable of collecting optical images and thermal radiation information of photovoltaic strings. During operation, the collected data is transmitted in real time to the drone inspection platform for storage and management. The service port of the unmanned inspection equipment is linked, and by calling the preset API interface of the drone inspection platform, optical images and thermal radiation maps of the currently abnormally inefficient strings, captured by the drone within a preset historical period (e.g., the most recent quarter), are obtained. The optical images are visual images of the string appearance collected by visible light sensors, clearly showing whether there are obvious stains, bird droppings, obstructions, or physical damage on the surface of abnormally inefficient strings. The thermal radiation map is an image of the string temperature distribution obtained using infrared thermal imaging technology; different temperature regions are imaged in different colors on the map, allowing for the visual identification of hot spots or other abnormal temperature areas within the strings. After acquiring the optical images and thermal radiation maps, they are displayed in a predetermined order on the interactive screen. For example, the interactive screen can be divided into two areas, with the left area displaying optical images and the right area displaying thermal radiation images, or a top-to-bottom layout can be used, or the images can be arranged according to time sequence or different parts of a series.

[0052] By linking with unmanned inspection equipment to retrieve optical images and thermal radiation spectra, a comprehensive analysis of abnormal and inefficient strings can be performed from two dimensions: physical appearance and thermal characteristics. Combined with the diagnostic results of electrical parameters, this further improves the accuracy and reliability of string fault diagnosis.

[0053] Example 2, based on the same inventive concept as the string fault identification and damage assessment method in the photovoltaic power generation scenario in the previous examples, such as... Figure 2 As shown, this application provides a string fault identification and damage assessment system for photovoltaic power generation scenarios, wherein the string fault identification and damage assessment system for photovoltaic power generation scenarios includes:

[0054] The data evaluation module 11 is used to import the electrical quantity data of the target photovoltaic string, correct the electrical quantity data based on the irradiance data, evaluate the degree of inefficiency based on the corrected electrical quantity data, and screen out abnormally inefficient strings; the performance degradation analysis module 12 is used to perform performance degradation analysis on the abnormally inefficient strings, obtain key degradation indicators, and draw current curves; the fault cause diagnosis module 13 is used to diagnose the fault causes of the abnormally inefficient strings based on the key degradation indicators and generate fault diagnosis results; the loss trend prediction module 14 is used to calculate historical power generation loss based on the fault diagnosis results, predict future loss trends, and generate power generation loss assessment results.

[0055] Furthermore, the data evaluation module 11 is also used to: receive the device identification information and time range of the target photovoltaic string input by the user; obtain the corresponding string electrical quantity data and site irradiance data according to the device identification information and time range; perform linear correction on the string electrical quantity data using the irradiance data, and input the linearly corrected string electrical quantity data into a pre-trained MiniRocket binary classification model to determine whether the target photovoltaic string is in an abnormally inefficient state; and filter out the abnormally inefficient strings according to the inefficiency state determination result.

[0056] Furthermore, the data evaluation module 11 is also used to: visually mark the abnormally inefficient strings on the user interface according to a predefined color mapping specification.

[0057] Furthermore, the performance degradation analysis module 12 is also used to: calculate and obtain key degradation indicators for the abnormally inefficient string using a multilayer perceptron, wherein the key degradation indicators include at least the degradation rate; and draw current curves in the same coordinate system based on the key degradation indicators to generate a first current curve and a second current curve; wherein the first current curve is the current curve of all strings in the inverter within a preset period, and the second current curve is the current curve of the abnormally inefficient string during the same period.

[0058] Furthermore, the fault cause diagnosis module 13 is also used to: collect multiple electrical parameters of the abnormal inefficient string within the abnormal duration interval based on the key degradation index; input the multiple electrical parameters into a pre-trained multi-class recognition engine to identify the fault defect type and its relative weight; and visualize the fault defect type and the relative weight in a pie chart as the fault diagnosis result.

[0059] Furthermore, the loss trend prediction module 14 is also used to: extract corresponding historical normal operation data based on each fault defect type, and reconstruct its theoretical power generation curve under ideal conditions; compare the theoretical power generation curve with the actual power generation curve of the abnormal inefficient string point by point, and calculate the difference sequence between the two; perform integral operation on the difference sequence within the abnormal duration interval to generate single-type fault power generation loss, and calculate the cumulative power generation loss by combining the relative weight; use a linear regression algorithm to predict the predicted power generation loss of the abnormal inefficient string in a preset future period; and generate a power generation loss assessment result based on the cumulative power generation loss and the predicted power generation loss.

[0060] Furthermore, the system is also used to: link the service port of the unmanned patrol device to retrieve the optical images and thermal radiation spectra of the abnormally inefficient cluster within a preset historical period; and display the optical images and thermal radiation spectra in an interactive screen in a predetermined order.

[0061] The various embodiments in this specification are described in a progressive manner, with each embodiment focusing on the differences from other embodiments. The string fault identification and damage assessment method and specific examples in the photovoltaic power generation scenario described in the foregoing embodiment one are also applicable to the string fault identification and damage assessment system in the photovoltaic power generation scenario of this embodiment. Through the foregoing detailed description of the string fault identification and damage assessment method in the photovoltaic power generation scenario, those skilled in the art can clearly understand the string fault identification and damage assessment system in the photovoltaic power generation scenario of this embodiment. Therefore, for the sake of brevity, it will not be described in detail here.

[0062] The above description of the disclosed embodiments enables those skilled in the art to make or use this application. Various modifications to these embodiments will be readily apparent to those skilled in the art, and the general principles defined herein may be implemented in other embodiments without departing from the spirit or scope of this application. Therefore, this application is not to be limited to the embodiments shown herein, but is to be accorded the widest scope consistent with the principles and novel features disclosed herein.

[0063] Obviously, those skilled in the art can make several improvements and modifications to this application without departing from the principles of this application, and these improvements and modifications also fall within the protection scope of this application.

Claims

1. A method for identifying and assessing string faults in photovoltaic power generation scenarios, characterized in that, The method includes: Import the electrical quantity data of the target photovoltaic string, and correct the electrical quantity data based on the irradiance data. Based on the corrected electrical quantity data, evaluate the degree of inefficiency and screen out abnormally inefficient strings. The performance degradation analysis of the abnormally inefficient string was performed to obtain key degradation indicators and plot the current curve. Based on the aforementioned key degradation indicators, fault cause diagnosis is performed on the abnormally inefficient series, and fault diagnosis results are generated. Based on the fault diagnosis results, historical power generation losses are calculated, future loss trends are predicted, and power generation loss assessment results are generated.

2. The string fault identification and damage assessment method in a photovoltaic power generation scenario as described in claim 1, characterized in that, Import the electrical quantity data of the target photovoltaic string, and correct the electrical quantity data based on the irradiance data. Then, assess the inefficiency based on the corrected electrical quantity data and screen out abnormally inefficient strings, including: Receive the device identification information and time range of the target photovoltaic string input by the user; Based on the equipment identification information and time range, obtain the corresponding string electrical quantity data and site irradiance data; The irradiation data is used to linearly correct the string electrical quantity data, and the linearly corrected string electrical quantity data is input into a pre-trained MiniRocket binary classification model to determine whether the target photovoltaic string is in an abnormally inefficient state. Based on the inefficiency state discrimination results, the abnormal inefficient strings are obtained by filtering.

3. The method according to claim 2, wherein the method further comprises: After obtaining the abnormally inefficient strings based on the inefficiency state discrimination result, the process further includes: According to a predefined color mapping specification, the abnormally inefficient strings are visually marked on the user interface. 4.The string fault identification and loss assessment method in photovoltaic power generation scene according to claim 1, characterized in that, The performance degradation analysis of the abnormally inefficient strings was performed to obtain key degradation indicators and to plot current curves, including: For the aforementioned abnormally inefficient strings, a multilayer perceptron is used to calculate and obtain key degradation indicators, which include at least the degradation rate. Based on the key degradation indicators, current curves are plotted in the same coordinate system to generate a first current curve and a second current curve. The first current curve is the current curve of all strings in the inverter within a preset period, and the second current curve is the current curve of the abnormally inefficient string during the same period.

5. The method of claim 1, wherein the method further comprises: Based on the aforementioned key degradation indicators, fault cause diagnosis is performed on the abnormally inefficient series, generating fault diagnosis results, including: Based on the key degradation indicators, multiple electrical parameters of the abnormally inefficient string are collected during the abnormal duration interval. The aforementioned multiple electrical parameters are input into a pre-trained multi-class recognition engine to identify fault and defect types and their relative weights. The fault defect type and the relative weight are visualized in a pie chart as the fault diagnosis result.

6. The method of claim 5, wherein the method further comprises: Based on the fault diagnosis results, historical power generation losses are calculated, future loss trends are predicted, and power generation loss assessment results are generated, including: Based on each fault / defect type, extract the corresponding historical normal operation data and reconstruct its theoretical power generation curve under ideal conditions; The theoretical power generation curve is compared point by point with the actual power generation curve of the abnormally inefficient string, and the difference sequence between the two is calculated. The difference sequence is integrated over the period of abnormality to generate a single type of fault power generation loss, and then weighted by the relative weights to calculate the cumulative power generation loss. Using a linear regression algorithm, the predicted power generation loss of the abnormally inefficient string within a preset future period is calculated. Based on the cumulative power generation loss and the predicted power generation loss, a power generation loss assessment result is generated.

7. The method of claim 1, wherein the method further comprises: The method further includes: The service port of the linked unmanned patrol device is used to retrieve the optical images and thermal radiation spectra of the abnormally inefficient string within a preset historical period. The optical images and thermal radiation maps are displayed in an interactive screen in a predetermined order.

8. A string fault identification and loss assessment system in a photovoltaic power generation scenario, characterized in that, The steps for implementing the string fault identification and damage assessment method in a photovoltaic power generation scenario according to any one of claims 1 to 7 include: The data evaluation module is used to import the electrical quantity data of the target photovoltaic string, correct the electrical quantity data based on the irradiation data, evaluate the degree of inefficiency based on the corrected electrical quantity data, and screen out abnormally inefficient strings. The performance degradation analysis module is used to perform performance degradation analysis on the abnormally inefficient string, obtain key degradation indicators, and draw current curves. The fault cause diagnosis module is used to diagnose the fault causes of the abnormal and inefficient series based on the key degradation indicators and generate fault diagnosis results. The loss trend prediction module is used to calculate historical power generation loss based on the fault diagnosis results, predict future loss trends, and generate power generation loss assessment results.

Images