Photovoltaic module subfissure identification method and device

Abstract

The invention relates to the technical field of data processing, and discloses a photovoltaic module subfissure identification method and device. The method comprises the following steps: performing statistical feature extraction on voltage data after time synchronization calibration to obtain a voltage feature vector and a preliminary hidden crack suspicious point mark; performing maximum power point tracking controller working characteristic data acquisition on the photovoltaic module corresponding to the voltage characteristic vector to obtain a power characteristic parameter set; performing feature weight distribution and standardization processing on the voltage feature vector and the power feature parameter set to obtain a standardized feature vector; and performing multi-level threshold discriminant analysis on the standardized feature vector to obtain photovoltaic module subfissure diagnosis data. According to the invention, the efficiency and accuracy of hidden crack identification of the photovoltaic module are improved.

Description

Technical Field

[0001] This application relates to the field of data processing, and particularly to a method and device for identifying hidden cracks in photovoltaic modules. Background Art

[0002] In the field of photovoltaic energy, as the core component of a photovoltaic power generation system, the performance of a photovoltaic module is directly related to the power generation efficiency and stability of the entire system. However, during long-term use, due to environmental factors (such as sand, wind, rain, snow, temperature changes, etc.) and improper operations during installation and maintenance, a photovoltaic module may develop minor damages such as hidden cracks. These hidden cracks are not easily detectable by the naked eye but can seriously affect the power generation efficiency and lifespan of the photovoltaic module. Therefore, it is particularly important to identify and diagnose hidden cracks in photovoltaic modules.

[0003] Currently, there are various methods for identifying hidden cracks in photovoltaic modules on the market, such as infrared thermal imaging, electroluminescence, and ultrasonic detection. However, most of these methods have drawbacks such as complex operation, high cost, and low detection efficiency, making it difficult to promote them on a large scale in practical applications. In addition, with the continuous development of photovoltaic technology, the requirements for the accuracy and real-time performance of hidden crack identification are also increasing, and traditional detection methods are no longer able to meet the current needs. Summary of the Invention

[0004] This application provides a method and device for identifying hidden cracks in photovoltaic modules, which are used to improve the efficiency and accuracy of hidden crack identification in photovoltaic modules.

[0005] In a first aspect, this application provides a method for identifying hidden cracks in photovoltaic modules. The method for identifying hidden cracks in photovoltaic modules includes: performing fieldbus connection and voltage acquisition module deployment on each photovoltaic module in a photovoltaic string, and performing master-slave communication control through a bus communication protocol to obtain an original voltage sampling data stream; performing time synchronization protocol calibration processing on the original voltage sampling data stream, transmitting timestamp information via synchronization messages and follow-up messages, and measuring transmission delay through delay request messages and delay response messages to obtain voltage data after time synchronization calibration; performing statistical feature extraction on the voltage data after time synchronization calibration, and obtaining a voltage feature vector and a preliminary hidden crack suspicious point mark by calculating the voltage difference between adjacent components, voltage mean, standard deviation, and fluctuation characteristics; collecting operating characteristic data of a maximum power point tracking controller for the photovoltaic module corresponding to the voltage feature vector, and obtaining a set of power characteristic parameters by analyzing power tracking efficiency, dynamic response, and boost characteristics; performing feature weight assignment and standardization processing on the voltage feature vector and the set of power characteristic parameters, and obtaining a standardized feature vector by constructing a multi-dimensional feature space mapping; performing multi-level threshold discrimination analysis on the standardized feature vector, and obtaining hidden crack diagnosis data for the photovoltaic module by calculating the feature space distance and historical data trend.

[0006] In a second aspect, the present application provides a device for identifying hidden cracks in a photovoltaic module. The device for identifying hidden cracks in a photovoltaic module includes:

[0007] An acquisition module, configured to perform fieldbus connection and voltage acquisition module deployment on each photovoltaic module in a photovoltaic string, and perform master-slave communication control through a bus communication protocol to obtain an original voltage sampling data stream;

[0008] A calibration module, configured to perform time synchronization protocol calibration processing on the original voltage sampling data stream, transmit timestamp information via synchronization messages and follow-up messages, and measure transmission delay through delay request messages and delay response messages to obtain voltage data after time synchronization calibration;

[0009] An extraction module, configured to perform statistical feature extraction on the voltage data after time synchronization calibration, and obtain a voltage feature vector and a preliminary hidden crack suspicious point mark by calculating the voltage difference between adjacent components, voltage mean, standard deviation, and fluctuation characteristics;

[0010] An acquisition module, configured to collect maximum power point tracking controller operating characteristic data of the photovoltaic module corresponding to the voltage feature vector, and obtain a set of power characteristic parameters by analyzing power tracking efficiency, dynamic response, and boost characteristics;

[0011] A processing module, configured to perform feature weight assignment and normalization processing on the voltage feature vector and the set of power characteristic parameters, and obtain a normalized feature vector by constructing a multi-dimensional feature space mapping;

[0012] A discrimination module, configured to perform multi-level threshold discrimination analysis on the normalized feature vector, and obtain hidden crack diagnosis data of the photovoltaic module by calculating the feature space distance and historical data trend.

[0013] In the technical solution provided by this application, through the deployment of fieldbus connection and voltage acquisition module, this solution realizes the accurate acquisition of voltage data of each photovoltaic module in the photovoltaic string. This not only improves the efficiency and accuracy of data acquisition, but also reduces the cost of manual intervention. The application of master-slave communication control ensures the stability and reliability of the data stream, providing a solid foundation for subsequent data processing. The introduction of time synchronization protocol calibration processing technology effectively solves the problem of data error caused by the clock asynchronization between devices. Through the interaction of synchronization messages, follow-up messages, and delay request and response messages, this solution can accurately measure the transmission delay and perform time calibration on the original voltage sampling data, thus ensuring the timing consistency and accuracy of the data and providing a reliable data source for subsequent statistical analysis. By extracting the statistical features of the voltage data after time synchronization calibration, this solution can comprehensively capture the voltage change characteristics of the photovoltaic module, including the voltage difference between adjacent modules, voltage mean value, standard deviation, and fluctuation characteristics, etc. The extraction of these feature vectors provides a strong basis for marking the preliminary suspected points of hidden cracks. At the same time, combined with the acquisition and analysis of the working characteristic data of the maximum power point tracking controller, this solution further enriches the dimension of diagnostic information and improves the accuracy and reliability of hidden crack identification. The application of feature weight allocation and standardization processing technology reasonably reflects the importance of different features in the diagnostic process, avoiding diagnostic deviation caused by the difference in feature dimensions. By constructing a multi-dimensional feature space mapping, this solution realizes a comprehensive and accurate description of the hidden crack characteristics of the photovoltaic module, providing strong support for subsequent multi-level threshold discrimination analysis. BRIEF DESCRIPTION OF THE DRAWINGS

[0014] In order to more clearly illustrate the technical solutions of the embodiments of the present invention, the following will briefly introduce the drawings required for the description of the embodiments. Obviously, the following drawings are some embodiments of the present invention. For those of ordinary skill in the art, without creative efforts, other drawings can also be obtained based on these drawings.

[0015] Figure 1 It is a schematic diagram of an embodiment of the method for identifying hidden cracks in photovoltaic modules in an embodiment of this application;

[0016] Figure 2 It is a schematic diagram of an embodiment of the device for identifying hidden cracks in photovoltaic modules in an embodiment of this application. DETAILED DESCRIPTION OF THE EMBODIMENTS

[0017] The embodiments of the present application provide a method and device for identifying hidden cracks in photovoltaic modules. Terms such as "first", "second", "third", "fourth", etc. (if any) in the specification, claims and the above-mentioned drawings of the present application are used to distinguish similar objects, and do not necessarily describe a specific order or sequence. It should be understood that such data can be interchanged under appropriate circumstances, so that the embodiments described here can be implemented in an order other than that illustrated or described here. In addition, the term "comprising" or "having" and any variations thereof are intended to cover non-exclusive inclusion. For example, a process, method, system, product or device including a series of steps or units does not necessarily limit to those steps or units clearly listed, but may include other steps or units not clearly listed or inherent to these processes, methods, products or devices.

[0018] For ease of understanding, the specific process of the embodiments of the present application is described below. Please refer to Figure 1 , an embodiment of the method for identifying hidden cracks in photovoltaic modules in the embodiments of the present application includes:

[0019] Step S101: Perform fieldbus connection and voltage acquisition module deployment on each photovoltaic module in the photovoltaic string, and perform master-slave communication control through the bus communication protocol to obtain the original voltage sampling data stream;

[0020] Step S102: Perform time synchronization protocol calibration processing on the original voltage sampling data stream, transmit timestamp information via synchronization messages and follow-up messages, and measure the transmission delay through delay request messages and delay response messages to obtain the voltage data after time synchronization calibration;

[0021] Step S103: Extract statistical features from the voltage data after time synchronization calibration. By calculating the voltage difference between adjacent components, voltage mean, standard deviation and fluctuation characteristics, obtain the voltage feature vector and the preliminary hidden crack suspicious point mark;

[0022] Step S104: Collect the operating characteristic data of the maximum power point tracking controller for the photovoltaic module corresponding to the voltage feature vector. By analyzing the power tracking efficiency, dynamic response and boost characteristics, obtain the power characteristic parameter set;

[0023] Step S105: Perform feature weight assignment and standardization processing on the voltage feature vector and the power characteristic parameter set. By constructing a multi-dimensional feature space mapping, obtain the standardized feature vector;

[0024] Step S106: Perform multi-level threshold discrimination analysis on the standardized feature vector. By calculating the feature space distance and historical data trend, obtain the hidden crack diagnosis data of the photovoltaic module.

[0025] It can be understood that the execution entity of this application can be a photovoltaic module crack identification device, or it can also be a terminal or a server, and specific details are not limited here. In the embodiments of this application, the server is taken as an example of the execution entity for illustration.

[0026] Specifically, each photovoltaic module in the photovoltaic string is connected through a fieldbus. Combining the arrangement of the voltage acquisition module, centralized data acquisition is realized. "Fieldbus" is a communication method dedicated to industrial control systems, allowing multiple devices to exchange data on the same bus. Through the master-slave communication control strategy, the master device sequentially sends acquisition requests to the slave devices, and each slave device returns a voltage sampling data stream. This voltage data stream records the real-time voltage value of each module, providing basic voltage information for subsequent processing steps. To ensure that the original voltage data collected has strict time consistency, the scheme uses a time synchronization protocol to calibrate the data. The time synchronization protocol is a communication protocol that keeps the time bases of multiple devices consistent. This step transmits timestamps through synchronization messages and follow-up messages, and measures the delay during data transmission using delay request and delay response messages to obtain synchronized voltage data. This calibration process aligns the data of different modules under the same time base, effectively eliminating the impact of data acquisition delay on subsequent analysis.

[0027] The system extracts statistical features from the voltage data after time synchronization calibration. Specifically, the statistical features include the voltage difference between adjacent modules, the voltage mean, the standard deviation, and the voltage fluctuation characteristics. These features can reveal the variation law of voltage among each module. For example, the voltage difference can reflect whether there is an abnormal voltage distribution, and the standard deviation measures the stability of the voltage. These features are summarized to form a voltage feature vector and mark the preliminary suspicious points that may have cracks.

[0028] For the modules with suspicious crack points, the maximum power point tracking (MPPT) controller is used to collect working characteristic data, including power tracking efficiency, dynamic response, and boost characteristics. These power characteristic parameters can reflect the actual output performance of photovoltaic modules under different load and environmental conditions. For example, the higher the power tracking efficiency, the better the module can convert light energy into electrical energy, and the dynamic response and boost characteristics show the adaptability of the module under load changes or light intensity changes. These data form a set of power characteristic parameters for further analyzing the health status of the module.

[0029] The voltage feature vector and the set of power characteristic parameters are assigned different weights and standardized to form a unified standardized feature vector. This standardization process normalizes features in different dimensions through multi-dimensional feature space mapping, facilitating subsequent analysis and comparison.

[0030] Finally, through multi-level threshold discriminant analysis of the standardized eigenvectors, the system calculates the distances in the feature spaces of each component and evaluates them in combination with the historical data trends, ultimately obtaining the hidden crack diagnosis data of the photovoltaic modules. For example, if the voltage difference feature of a group of modules is significantly abnormal and the power characteristics show poor dynamic response, the system will mark this module as having a high hidden crack risk and further determine the possibility of hidden cracks by comparing with the historical data trends. This discriminant method based on multi-dimensional features not only improves the accuracy of hidden crack detection but also reduces the false alarm rate, thereby enhancing the reliability and practicality of the diagnosis.

[0031] In the embodiment of the present application, through the deployment of the fieldbus connection and the voltage acquisition module, this solution achieves the accurate acquisition of the voltage data of each photovoltaic module in the photovoltaic string. This not only improves the efficiency and accuracy of data acquisition but also reduces the cost of manual intervention. The application of the master-slave communication control ensures the stability and reliability of the data stream, providing a solid foundation for subsequent data processing. The introduction of the time synchronization protocol calibration processing technology effectively solves the data error problem caused by the clock asynchronization between devices. Through the interaction of synchronization messages, follow-up messages, and delay request and response messages, this solution can accurately measure the transmission delay and perform time calibration on the original voltage sampling data, thereby ensuring the timing consistency and accuracy of the data and providing a reliable data source for subsequent statistical analysis. By extracting the statistical features of the voltage data after time synchronization calibration, this solution can comprehensively capture the voltage change characteristics of the photovoltaic modules, including the voltage difference between adjacent modules, voltage mean, standard deviation, and fluctuation characteristics, etc. The extraction of these feature vectors provides a strong basis for marking the preliminary hidden crack suspicious points. At the same time, combined with the acquisition and analysis of the working characteristic data of the maximum power point tracking controller, this solution further enriches the dimension of the diagnostic information and improves the accuracy and reliability of hidden crack identification. The application of the feature weight assignment and standardization processing technology reasonably reflects the importance of different features in the diagnosis process and avoids diagnostic deviations caused by the difference in feature dimensions. By constructing a multi-dimensional feature space mapping, this solution achieves a comprehensive and accurate description of the hidden crack characteristics of the photovoltaic modules, providing strong support for subsequent multi-level threshold discriminant analysis.

[0032] In a specific embodiment, the process of executing step S101 may specifically include the following steps:

[0033] (1) Install a voltage acquisition module on the positive and negative terminals of each component in the photovoltaic string. The sampling accuracy of the voltage acquisition module is 0.5%, obtaining a voltage acquisition network;

[0034] (2) Perform optoelectronic isolation settings on the voltage acquisition network. The optoelectronic isolation voltage is greater than 2.5 kV, obtaining an acquisition module with electrical isolation;

[0035] (3) Cascade-connect the acquisition modules with electrical isolation through the fieldbus. The maximum communication distance of the fieldbus is 3000 meters to obtain the acquisition module communication network;

[0036] (4) Establish a communication connection between the acquisition modules in the acquisition module communication network and the master station controller at the busbar box, and perform data interaction through the master-slave communication protocol to obtain the master-slave communication link;

[0037] (5) Control data acquisition for the master-slave communication link according to the preset sampling period, and trigger each acquisition module to perform voltage latching through the synchronous acquisition instruction to obtain the original voltage sampling data stream.

[0038] Specifically, high-precision voltage acquisition modules are installed at the positive and negative terminals of each component in the photovoltaic string to form a complete voltage acquisition network. The sampling accuracy of the voltage acquisition module reaches 0.5%, ensuring that the collected data has sufficient details and accuracy to capture the voltage status of each photovoltaic component. This accuracy requirement enables the network to record and reflect the subtle voltage changes of each component in detail, providing the basic raw data for subsequent analysis. To ensure that the collected voltage signal is not interfered by the outside world and remains independent during transmission, an optoelectronic isolation module is introduced into the voltage acquisition network. The optoelectronic isolation voltage is set to 2.5 kV to ensure complete electrical isolation of the signal. Optoelectronic isolation plays the role of electrical isolation in this solution, that is, without affecting data acquisition, it avoids possible electrical interference between the acquisition module and the component. This design enhances the stability of the acquisition module and ensures the signal integrity of each module, ensuring the accuracy of subsequent data processing.

[0039] After completing the electrical isolation, cascade-connect the acquisition modules after electrical isolation through the fieldbus. The fieldbus has a maximum communication distance of 3000 meters, which can realize long-distance centralized management and real-time data sharing of each acquisition module. Through the cascade connection of the fieldbus, a stable acquisition module communication network is established, enabling the data of each acquisition module to be transmitted in a timely manner within the specified distance, meeting the requirements of scenarios where photovoltaic components are widely distributed, and laying a stable communication foundation for large-scale data acquisition. Further, establish a communication connection between each acquisition module in the acquisition module communication network and the master station controller at the busbar box to form a master-slave communication link, and use the master-slave communication protocol to perform data interaction. In this communication link, the master station controller serves as the main management device, controlling and collecting the data of all slave acquisition modules. The master station controller sends acquisition instructions one by one according to the set communication sequence and receives the voltage data of the acquisition module. This data interaction process has high flexibility and efficiency in this solution, and can achieve accurate and timely voltage acquisition in scenarios with a large-scale component distribution.

[0040] After the master-slave communication link is established, the master station controller issues a data acquisition control instruction to the acquisition module according to a preset sampling period. The synchronization acquisition instruction is issued by the master station controller. After receiving the synchronization instruction, each acquisition module performs a voltage latching operation, that is, locks the voltage value at the same time point, ensuring strict time synchronization of data sampling. The resulting original voltage sampling data stream not only contains the voltage values of each component but also ensures the time consistency of the data, facilitating subsequent data processing and analysis.

[0041] For example, assume that there are multiple acquisition modules in a photovoltaic string, and each module records the voltage data of the components it is connected to. For example, the acquisition data stream within one acquisition period shows that the voltage of a certain component suddenly drops, and in the isolated acquisition module, the fluctuation value deviates from the average value of the string. This difference is accurately recorded after optoelectronic isolation and time synchronization locking, and then through data analysis by the busbar box controller, the anomaly can be quickly identified. Such a sampling design ensures the authenticity and usability of the data, enabling voltage characteristic changes to be accurately used for crack analysis and significantly improving the accuracy and reliability of crack identification.

[0042] In a specific embodiment, the process of executing step S102 may specifically include the following steps:

[0043] (1) Set the clock synchronization level for the acquisition module communication network. By setting the master station controller as the master clock source and the voltage acquisition module as the slave clock, a clock level structure is obtained;

[0044] (2) Send synchronization messages and follow-up messages to the master clock in the clock level structure. By carrying the timestamp information of the sending moment in the messages, the master clock time reference is obtained;

[0045] (3) Measure the delay of the slave clock. By sending delay request messages and receiving delay response messages, the transmission path delay value is obtained;

[0046] (4) Calculate the deviation of the local time of the slave clock. By using the master clock time reference and the transmission delay value, a clock correction parameter is obtained;

[0047] (5) Perform time calibration on the original voltage sampling data stream. By correcting the sampling moment with the clock correction parameter, the voltage data after time synchronization calibration is obtained.

[0048] Specifically, by establishing a clock synchronization hierarchy and achieving master-slave clock coordination, voltage data after time synchronization calibration is finally obtained, providing an accurate timing basis for subsequent hidden crack identification. In the clock synchronization hierarchy setting of the communication network of the acquisition module, the master station controller is designated as the master clock source, and all voltage acquisition modules act as slave clocks. In this way, a hierarchical time synchronization architecture is formed between the master clock and the slave clocks, enabling the master clock to provide a unified time reference for the entire acquisition network. This clock hierarchy structure lays the foundation for the time synchronization mechanism, ensuring the consistency of the sampling times of each acquisition module and avoiding sampling errors caused by local time differences between different modules.

[0049] Secondly, in order to transmit an accurate time reference between the master clock and the slave clocks, the master station controller (acting as the master clock) periodically sends synchronization messages and follow-up messages to the slave clocks. In the synchronization message, the timestamp information of the master clock's sending moment is carried, and this timestamp is a specific time data that records the accurate time of the master clock. The follow-up message further confirms and conveys the time information, enabling the slave clock to obtain the accurate master clock time reference. Through the transmission of these two types of messages and the sharing of timestamp information, the current time of the master clock is synchronized to the slave clocks, providing an effective reference point for subsequent time correction. Next, in order to compensate for the transmission delay between the master clock and the slave clocks, delay measurement is performed on the slave clock side. Delay measurement is achieved by sending a delay request message and receiving a delay response message. During this process, the slave clock sends a delay request message to the master clock, and the master clock immediately returns a delay response message after receiving the request. In this way, the slave clock calculates the transmission path delay value between the master clock and the slave clock based on the time difference between the request sent and the response received. This delay value can be used as part of the time deviation correction to make up for the time consumption during data transmission and ensure precise time synchronization between the slave clock and the master clock.

[0050] After completing the delay measurement, the slave clock uses the time reference of the master clock and the transmission delay value to calculate the deviation, thereby obtaining the clock correction parameter. Specifically, the slave clock corrects the local time through the master clock time and the delay value, adjusting the local time to be consistent with the master clock. The calculation of this clock correction parameter is based on the time difference between the master clock and the slave clock and the transmission delay value, thereby obtaining an accurate correction amount, making the local times of each acquisition module consistent with the master clock and ensuring the alignment of the acquired data in the time dimension. Finally, after the time synchronization of all acquisition modules is completed, the scheme uses this clock correction parameter to perform time calibration on the original voltage sampling data stream. Specifically, the calibration process adjusts the sampling moment through the clock correction parameter to ensure that all voltage data is aligned with the same time reference. This voltage data after time calibration has high timing accuracy and eliminates the interference of acquisition delay on data analysis.

[0051] For example, assume that within a certain sampling period, the time reference of the master station controller is 12:00:00.000, and the slave clock shows 12:00:00.002 due to delay. After delay measurement and clock deviation calculation, the correction parameter of the slave clock is -0.002 seconds. Then, in the actual voltage data acquisition, the sampling moments of all data will be synchronized to the reference moment 12:00:00.000 of the master clock, so that the collected voltage data has strict time consistency during analysis. This clock synchronization and calibration process ensures time accuracy, enabling voltage fluctuations and crack characteristics in subsequent analysis to be compared and identified under a unified time reference, greatly improving the accuracy and reliability of photovoltaic module crack identification.

[0052] In a specific embodiment, the process of performing step S103 may specifically include the following steps:

[0053] (1) Perform filtering on the voltage data after time synchronization and calibration. Remove power frequency interference through a band-stop filter and impulse noise through median filtering to obtain filtered voltage data;

[0054] (2) Compare the voltages of adjacent components within a string for the filtered voltage data. Calculate the voltage difference and difference ratio between adjacent components to obtain a voltage matching degree parameter;

[0055] (3) Calculate the 24-hour statistics for the filtered voltage data. Calculate the voltage mean, standard deviation, skewness, and kurtosis to obtain voltage statistical characteristics;

[0056] (4) Conduct a rate-of-change analysis on the filtered voltage data. Calculate the voltage change rate and change acceleration to obtain voltage dynamic characteristics;

[0057] (5) Combine the voltage matching degree parameter, voltage statistical characteristics, and voltage dynamic characteristics to obtain a voltage feature vector and a preliminary crack suspect point mark.

[0058] Specifically, filter the voltage data after time synchronization and calibration through a band-stop filter and a median filter to remove various unnecessary interference signals. The band-stop filter is specifically used to remove power frequency interference, that is, the 50Hz or 60Hz power frequency noise generated by the power system, which will significantly affect the stability of the voltage signal. By setting the center frequency of the filter at the power frequency, the interference signals in this frequency band are effectively suppressed. In addition, to remove impulse noise in the data (such as isolated peak signals generated by instantaneous interference), the median filter is used in the solution. The median filter has a strong suppression effect on short-term impulse interference and does not affect the trend of the signal, making the filtered voltage data smooth and close to the actual voltage change situation, facilitating further analysis.

[0059] After obtaining the filtered voltage data, the solution compares and analyzes the voltages of adjacent components within the string. This process calculates the voltage difference and difference ratio for each pair of adjacent components to obtain the voltage matching parameter. The voltage difference reflects the voltage balance between components, while the difference ratio further quantifies the relative magnitude of this difference. By comparing the matching degrees of each pair of adjacent component voltages, pairs of components with abnormal voltages can be quickly located. The larger the voltage difference and the more the difference ratio deviates from the normal value, the greater the likelihood of potential hidden cracks or other faults within the component group. Subsequently, 24-hour statistical calculations are performed on the filtered voltage data to extract the statistical characteristics of the voltage. The data within 24 hours can reflect the operating conditions of the photovoltaic components under all-weather conditions. The voltage mean represents the average voltage level within a day, the standard deviation reflects the voltage fluctuation range, the skewness depicts the symmetry of the voltage data distribution, and the kurtosis measures the spiky characteristics of the data. Through these statistics, the basic characteristics of the component voltage can be understood. If there are significant deviations from the normal range in the statistical characteristics (such as a high standard deviation or abnormal kurtosis), it may indicate potential risks of voltage instability or hidden cracks in the components.

[0060] After completing the extraction of statistical characteristics, it is also necessary to perform a rate-of-change analysis on the filtered voltage data to identify the dynamic characteristics of the voltage. The voltage change rate represents the amount of voltage change per unit time, while the change acceleration describes the trend of the voltage change rate. If there are abnormal fluctuations in the voltage change rate and acceleration, it may mean that there are abnormalities in the voltage response of the components under changing environmental conditions. For example, if the component voltage shows sharp fluctuations or rapid drops within a short period of time, it may indicate the possibility of hidden cracks or other electrical faults in the component. Finally, by combining the voltage matching parameter, voltage statistical characteristics, and voltage dynamic characteristics, a comprehensive voltage feature vector is formed, and preliminary hidden crack suspicious points are marked. The voltage feature vector contains voltage information at multiple levels, providing a reliable feature combination for identifying hidden cracks. The marking of preliminary hidden crack suspicious points helps to quickly screen out the components most likely to have hidden cracks, facilitating further detailed detection and confirmation.

[0061] For example, assume that during a day of data collection, the 24-hour voltage mean of a certain component is significantly lower than that of other components. At the same time, its voltage change rate shows abnormal fluctuations during multiple daytime periods, and the voltage matching parameter also indicates a large voltage difference between its adjacent components. In such a case, the voltage feature vector of this component may show obvious abnormal signals and thus be marked as a preliminary hidden crack suspicious point. This analysis process combines multiple features, ensuring the comprehensiveness and accuracy of hidden crack identification and helping to detect and handle hidden cracks in photovoltaic components at an early stage.

[0062] In a specific embodiment, the process of performing step S104 may specifically include the following steps:

[0063] (1) Perform a scanning control on the working voltage of the photovoltaic module. By changing the duty cycle of the MPPT controller, obtain the voltage-power curve data;

[0064] (2) Locate the maximum power point for the voltage-power curve data. Calculate the power change rate by the conductance increment method to obtain the working parameters at the maximum power point;

[0065] (3) Measure the dynamic response of the MPPT controller. Through the output characteristics under a step change in light intensity, obtain the dynamic tracking parameters;

[0066] (4) Calculate the conversion efficiency of the MPPT controller. Through the input-output power ratio and the boost ratio characteristics, obtain the efficiency characteristic parameters;

[0067] (5) Integrate the working parameters at the maximum power point, the dynamic tracking parameters, and the efficiency characteristic parameters to obtain the power characteristic parameter set.

[0068] Specifically, perform a scanning control on the working voltage of the photovoltaic module to obtain complete voltage-power curve data. The scanning control is completed by an MPPT (Maximum Power Point Tracking) controller. The method is to adjust the duty cycle of the MPPT controller, thereby changing the output voltage of the photovoltaic module. The change in the duty cycle will cause changes in the output voltage and current of the photovoltaic module. Record these changes and calculate the corresponding power to obtain a series of data points of voltage-power pairs, and finally plot the voltage-power curve. The voltage-power curve can intuitively reflect the power output characteristics of the module at different working voltages, especially identify the trend of power change with voltage and the peak position. Locate the maximum power point for the obtained voltage-power curve data to determine the optimal output state of the photovoltaic module under the current conditions. The maximum power point refers to the voltage-current combination point at which the module can output the highest power. This location process uses the conductance increment method, that is, by calculating the change rates of voltage and power to locate the maximum power point. The conductance increment method analyzes the influence of voltage change on power, continuously adjusts the voltage in the region where power increases with the increase of voltage until reaching the point where power no longer increases, thereby locking the maximum power point. The working parameters at the maximum power point, including the voltage and current values at this point, provide key data support for subsequent component performance analysis.

[0069] After the maximum power point location is completed, the solution measures the dynamic response of the MPPT controller to evaluate the controller's ability to respond to sudden changes in light intensity. The dynamic response measurement is achieved by applying a step change in light intensity and observing the response characteristics of the controller's output power. When the light intensity suddenly rises or falls, the output voltage and current of the controller will change correspondingly, and the system records these changes to evaluate the controller's dynamic tracking ability. The dynamic tracking parameters reflect the adaptability and response speed of the controller to environmental changes, and are of great significance for evaluating the stability of the component under dynamic conditions. Further, the solution calculates the conversion efficiency of the MPPT controller by comparing the ratio of the input power and the output power and combining the boost ratio characteristics to obtain the efficiency characteristic parameters. The conversion efficiency calculation focuses on the ratio of the actual output power of the photovoltaic component to the input light energy power. Usually, the input power is obtained by recording the input voltage and current and calculating their product, and the product of the output voltage and current is used as the output power, and the two are compared to obtain the efficiency parameter. In addition, the boost ratio reflects the stability and performance upper limit of the controller at high power output, and further reflects the conversion ability and efficiency performance of the component.

[0070] Finally, the maximum power point operating parameters, dynamic tracking parameters, and efficiency characteristic parameters are integrated to obtain a comprehensive set of power characteristic parameters. This parameter set contains the performance of the photovoltaic component under different operating conditions, providing richer characteristic information for crack detection. For example, if the dynamic response of a certain component is slow and the efficiency characteristic parameters are significantly lower than those of other components, it may indicate that there are structural problems or crack risks. Based on the integrated set of power characteristic parameters, the system can further perform feature matching and trend analysis to quickly locate the components that may have cracks and conduct in-depth detection.

[0071] In a specific embodiment, the process of executing step S105 may specifically include the following steps:

[0072] (1) Perform dimensional normalization on the voltage feature vector. Through the maximum-minimum normalization method, obtain the normalized voltage feature;

[0073] (2) Perform dimensional normalization on the set of power characteristic parameters. Through the Z-score normalization method, obtain the normalized power feature;

[0074] (3) Evaluate the importance of the normalized voltage feature. Through information gain ratio calculation, obtain the voltage feature weight coefficient;

[0075] (4) Evaluate the importance of the normalized power feature. Through Gini coefficient calculation, obtain the power feature weight coefficient;

[0076] (5) Combine the standardized voltage feature and the standardized power feature with weights, and obtain a standardized feature vector through the feature weight coefficient.

[0077] Specifically, perform dimension normalization on the voltage feature vector using the maximum-minimum normalization method. Maximum-minimum normalization is a linear transformation that normalizes the feature values to a specific range (e.g., between 0 and 1) through a formula, thereby eliminating the dimension differences between different voltage features and ensuring the consistency of voltage features within the numerical range. For example, for a certain voltage feature vector, its original minimum value is 0.5V and the maximum value is 5V. After normalization, the feature data will be adjusted and distributed within the range of 0 to 1. The standardized voltage feature after such processing can be directly compared and calculated in subsequent steps. For the power characteristic parameter set, the Z-score normalization method is used for dimension normalization. Z-score normalization is based on the mean and standard deviation of the feature data, and converts each feature value into a relative deviation from the mean, making it have a zero mean and unit variance. This method is suitable for the normalization of data under a normal distribution and helps to eliminate the influence of extreme values and outliers in the power characteristics on the analysis. For example, if the original mean of a certain power feature is 50W and the standard deviation is 5W, then under Z-score normalization, the feature value of 60W will be converted into a standardized value of 2 after processing. In this way, the standardized power feature can maintain dimensional consistency with the voltage feature in subsequent analysis.

[0078] After obtaining the standardized voltage feature, perform importance evaluation through the information gain ratio. The information gain ratio is a commonly used index for feature selection, which reflects the influence degree of a certain feature on the target variable (crack diagnosis result). By calculating the information gain ratio of each voltage feature on the diagnosis target, the importance ranking of different voltage features can be obtained, and corresponding weight coefficients can be assigned to each feature. The higher the information gain ratio of a feature, the greater its contribution to crack diagnosis, and the larger the weight coefficient. For example, if the gain ratio of a certain voltage feature is significantly higher than other features, then this feature will be assigned a higher weight coefficient in the diagnosis model so that it can have a greater impact in subsequent weighted calculations. For the standardized power feature, the Gini coefficient is used to calculate the importance to obtain the weight coefficient of each power feature. The Gini coefficient is an index used to measure the imbalance of data distribution, and the importance of a certain power feature in target discrimination can be evaluated through the Gini coefficient. The lower the Gini coefficient of a feature, the stronger its discrimination effect on the diagnosis result, so a higher weight coefficient is assigned. Specifically, the calculated Gini coefficient of each power feature reflects its contribution degree to the classification effect of power characteristics in the diagnosis process, and the feature with a larger weight will have a more significant impact on the diagnosis result.

[0079] Finally, the standardized voltage feature and the standardized power feature are combined with weights. Using the feature weight coefficients obtained in the foregoing steps, the voltage feature and the power feature are weighted respectively, so that they have corresponding influence weights in the combined standardized feature vector. In this way, the generated standardized feature vector gathers important information from multiple aspects, providing a comprehensive feature basis for the subsequent hidden crack identification process.

[0080] In a specific embodiment, the process of executing step S106 may specifically include the following steps:

[0081] (1) Perform clustering analysis on the standardized feature vector, and obtain the feature distribution density by calculating the Euclidean distance in the feature space;

[0082] (2) Set multi-level discrimination thresholds for the feature distribution density, and obtain the hidden crack degree classification standard through the statistical distribution of normal samples;

[0083] (3) Perform trend analysis on the standardized feature vector, and obtain the development trend parameters by calculating the feature change rate and acceleration;

[0084] (4) Make a comprehensive judgment on the hidden crack degree classification standard and the development trend parameters, and obtain the hidden crack level through multi-level threshold comparison;

[0085] (5) Integrate the hidden crack level and the development trend to obtain the hidden crack diagnosis data of the photovoltaic module.

[0086] Specifically, perform clustering analysis on the standardized feature vector, calculate the Euclidean distance in the feature space to obtain the feature distribution density. The Euclidean distance is a commonly used measurement method for measuring the similarity between different samples in a multi-dimensional feature space. In clustering analysis, by calculating the Euclidean distance between the component feature vectors, the feature clustering degree of different components is identified. The level of the feature distribution density directly reflects the concentration degree of the photovoltaic module features - when an abnormality appears in a certain feature dense area, it may indicate that the components in this area have similar hidden crack features. After obtaining the feature distribution density, the scheme sets multi-level discrimination thresholds for the density data, and based on the statistical distribution of normal samples, establishes a classification standard for the hidden crack degree. The feature distribution of normal samples is regarded as a benchmark, and the hidden crack risk level is divided by observing the degree of deviation of the feature density from the benchmark. The multi-level discrimination thresholds are used to divide the hidden crack degree into different levels, such as mild, moderate, and severe hidden cracks, etc. A higher density deviation threshold usually corresponds to a more severe hidden crack degree, enabling the system to automatically classify and determine different hidden crack situations.

[0087] Perform trend analysis on the standardized eigenvector. By calculating the change rate and change acceleration of the features, the feature development trend of the component is obtained. The change rate represents the increase or decrease rate of the feature over time, while the change acceleration reflects the dynamic change of the change rate. By observing the feature change rate and acceleration, the change trend of the component performance can be captured. For example, a certain component feature is in a state of gradual deterioration or gradual stabilization. These trend parameters provide a basis for predicting the future state of the component, enabling the hidden crack detection system to not only reflect the current hidden crack situation but also provide early warning information for the subsequent potential development of hidden cracks in the component. Subsequently, a comprehensive judgment is made on the hidden crack degree grading standard and the development trend parameters, and the hidden crack level is determined by using multi-level threshold comparison. By comparing the current feature state with the set discrimination threshold, the system can judge the hidden crack degree of the component. The hidden crack level is divided into multiple levels to intuitively reflect the severity of the hidden crack. Through this multi-level discrimination process, each photovoltaic component is assigned a clear hidden crack level, facilitating rapid identification and classification processing.

[0088] Finally, integrate the data of the hidden crack level and the trend parameters to generate the final hidden crack diagnosis data of the photovoltaic component. This diagnosis data includes both the level information of the current hidden crack degree and the quantitative analysis of the development trend, providing reliable data support for operation and maintenance and risk management. For example, if a certain component is determined to be at a moderate hidden crack level and the trend analysis shows a positive acceleration, then this component may deteriorate further in the future and needs to be repaired or replaced as soon as possible. Through this integrated diagnosis process, the health status of the photovoltaic component can be accurately quantified, providing precise information support for subsequent decision-making.

[0089] The method for identifying hidden cracks in photovoltaic components in the embodiments of the present application is described above. Next, the device for identifying hidden cracks in photovoltaic components in the embodiments of the present application will be described. Please refer to Figure 2 , an embodiment of the device for identifying hidden cracks in photovoltaic components in the embodiments of the present application includes:

[0090] The control module 201 is used to perform field bus connection and voltage acquisition module deployment on each photovoltaic component in the photovoltaic string, and perform master-slave communication control through the bus communication protocol to obtain the original voltage sampling data stream;

[0091] The calibration module 202 is used to perform time synchronization protocol calibration processing on the original voltage sampling data stream, transmit timestamp information via synchronization messages and follow-up messages, and measure the transmission delay through delay request messages and delay response messages to obtain the voltage data after time synchronization calibration;

[0092] The extraction module 203 is used to extract statistical features from the voltage data after time synchronization calibration. By calculating the voltage difference between adjacent components, voltage mean, standard deviation, and fluctuation features, a voltage feature vector and a preliminary hidden crack suspicious point mark are obtained;

[0093] The acquisition module 204 is used to collect the operating characteristic data of the maximum power point tracking controller for the photovoltaic module corresponding to the voltage feature vector, and obtain a set of power characteristic parameters by analyzing the power tracking efficiency, dynamic response, and boost characteristic.

[0094] The processing module 205 is used to perform feature weight assignment and normalization processing on the voltage feature vector and the set of power characteristic parameters, and obtain a normalized feature vector by constructing a multi-dimensional feature space mapping.

[0095] The discrimination module 206 is used to perform multi-level threshold discrimination analysis on the normalized feature vector, and obtain the hidden crack diagnosis data of the photovoltaic module by calculating the feature space distance and the historical data trend.

[0096] Through the collaborative cooperation of the above-mentioned various components, and through the deployment of the field bus connection and the voltage acquisition module, this solution realizes the accurate acquisition of the voltage data of each photovoltaic module in the photovoltaic string. This not only improves the efficiency and accuracy of data acquisition, but also reduces the cost of manual intervention. The application of the master-slave communication control ensures the stability and reliability of the data stream, providing a solid foundation for subsequent data processing. The introduction of the time synchronization protocol calibration processing technology effectively solves the data error problem caused by the clock non-synchronization between devices. Through the interaction of synchronization messages, follow-up messages, and delay request and response messages, this solution can accurately measure the transmission delay and perform time calibration on the original voltage sampling data, thus ensuring the timing consistency and accuracy of the data, and providing a reliable data source for subsequent statistical analysis. By extracting the statistical features of the voltage data after time synchronization calibration, this solution can comprehensively capture the voltage change characteristics of the photovoltaic module, including the voltage difference between adjacent modules, voltage mean, standard deviation, and fluctuation characteristics, etc. The extraction of these feature vectors provides a strong basis for marking the preliminary hidden crack suspicious points. At the same time, combined with the collection and analysis of the operating characteristic data of the maximum power point tracking controller, this solution further enriches the dimension of diagnostic information and improves the accuracy and reliability of hidden crack identification. The application of the feature weight assignment and normalization processing technology reasonably reflects the importance of different features in the diagnostic process, avoiding diagnostic deviations caused by differences in feature dimensions. By constructing a multi-dimensional feature space mapping, this solution realizes a comprehensive and accurate description of the hidden crack characteristics of the photovoltaic module, providing strong support for subsequent multi-level threshold discrimination analysis.

[0097] As described above, the above embodiments are only used to illustrate the technical solutions of the present application, rather than to limit it; although the present application has been described in detail with reference to the foregoing embodiments, those of ordinary skill in the art should understand that: they can still modify the technical solutions recorded in the foregoing embodiments, or perform equivalent replacements on some of the technical features; and these modifications or replacements do not cause the essence of the corresponding technical solutions to deviate from the spirit and scope of the technical solutions of the embodiments of the present application.

Claims

1. A method for identifying hidden cracks in a photovoltaic module, characterized in that, The method for identifying hidden cracks in photovoltaic modules includes: Conduct field bus connection and voltage acquisition module deployment for each photovoltaic module in the photovoltaic string, perform master-slave communication control through the bus communication protocol, and obtain the original voltage sampling data stream; Perform time synchronization protocol calibration processing on the original voltage sampling data stream, transmit timestamp information via synchronization messages and follow-up messages, and measure the transmission delay through delay request messages and delay response messages to obtain the voltage data after time synchronization calibration; Extract statistical features from the voltage data after time synchronization calibration, calculate the voltage difference between adjacent components, voltage mean, standard deviation, and fluctuation characteristics to obtain the voltage feature vector and preliminary hidden crack suspicious point marks; Collect the operating characteristic data of the maximum power point tracking controller for the photovoltaic module corresponding to the voltage feature vector, and obtain the power characteristic parameter set by analyzing the power tracking efficiency, dynamic response, and boost characteristics; Perform feature weight assignment and standardization processing on the voltage feature vector and power characteristic parameter set, and obtain the standardized feature vector by constructing a multi-dimensional feature space mapping; Perform multi-level threshold discriminant analysis on the standardized feature vector, calculate the feature space distance and historical data trend to obtain the hidden crack diagnosis data of the photovoltaic module.

2. The method for identifying hidden cracks in a photovoltaic module according to claim 1, characterized in that, The conduct of field bus connection and voltage acquisition module deployment for each photovoltaic module in the photovoltaic string, perform master-slave communication control through the bus communication protocol, and obtain the original voltage sampling data stream includes: Install voltage acquisition modules on the positive and negative terminals of each component in the photovoltaic string, and the sampling accuracy of the voltage acquisition module is 0.5% to obtain the voltage acquisition network; Set optoelectronic isolation for the voltage acquisition network, and the optoelectronic isolation voltage is greater than 2.5 kV to obtain the acquisition module with electrical isolation; Cascade connect the acquisition modules with electrical isolation through the field bus, and the maximum communication distance of the field bus is 3000 meters to obtain the acquisition module communication network; Establish a communication connection between the acquisition module in the acquisition module communication network and the master controller at the busbar box, and perform data interaction through the master-slave communication protocol to obtain the master-slave communication link; Control data acquisition for the master-slave communication link according to a preset sampling period, trigger each acquisition module to latch the voltage through a synchronous acquisition instruction, and obtain the original voltage sampling data stream.

3. The method for identifying hidden cracks of a photovoltaic module according to claim 2, wherein The perform time synchronization protocol calibration processing on the original voltage sampling data stream, transmit timestamp information via synchronization messages and follow-up messages, and measure the transmission delay through delay request messages and delay response messages to obtain the voltage data after time synchronization calibration includes: Set the clock synchronization level for the acquisition module communication network, and obtain the clock level structure by setting the master controller as the master clock source and the voltage acquisition module as the slave clock; Send synchronization messages and follow-up messages to the master clock in the clock level structure, and obtain the master clock time reference by carrying the timestamp information of the sending moment in the message; Measure the delay of the slave clock, and obtain the transmission path delay value by sending delay request messages and receiving delay response messages; Calculate the deviation of the local time of the slave clock, and obtain the clock correction parameter through the master clock time reference and the transmission delay value; Perform time calibration on the original voltage sampling data stream, and correct the sampling moment through the clock correction parameter to obtain the voltage data after time synchronization calibration.

4. The method for identifying hidden cracks in a photovoltaic module according to claim 1, wherein Extract the statistical features of the voltage data after time synchronization calibration. By calculating the voltage difference between adjacent components, voltage mean, standard deviation, and fluctuation characteristics, obtain the voltage feature vector and the preliminary hidden crack suspicious point mark, including: Perform filtering processing on the voltage data after time synchronization calibration. Remove power frequency interference through a band-stop filter and remove pulse noise through median filtering to obtain the filtered voltage data; Compare the voltages of adjacent components within the string for the filtered voltage data. By calculating the voltage difference and difference ratio between adjacent components, obtain the voltage matching degree parameter; Calculate the 24-hour statistics for the filtered voltage data. By calculating the voltage mean, standard deviation, skewness, and kurtosis, obtain the voltage statistical features; Perform rate of change analysis on the filtered voltage data. By calculating the voltage change rate and change acceleration, obtain the voltage dynamic features; Combine the voltage matching degree parameter, voltage statistical features, and voltage dynamic features to obtain the voltage feature vector and the preliminary hidden crack suspicious point mark.

5. The method for identifying hidden cracks in a photovoltaic module according to claim 1, wherein Collect the working characteristic data of the maximum power point tracking controller for the photovoltaic module corresponding to the voltage feature vector. By analyzing the power tracking efficiency, dynamic response, and boost characteristic, obtain the power characteristic parameter set, including: Perform scanning control on the working voltage of the photovoltaic module. By changing the duty cycle of the MPPT controller, obtain the voltage-power curve data; Locate the maximum power point for the voltage-power curve data. By calculating the power change rate through the conductance increment method, obtain the working parameters of the maximum power point; Measure the dynamic response of the MPPT controller. By the output characteristics under a step change in light intensity, obtain the dynamic tracking parameters; Calculate the conversion efficiency of the MPPT controller. By the ratio of input and output power and the boost ratio characteristic, obtain the efficiency characteristic parameters; Integrate the working parameters of the maximum power point, dynamic tracking parameters, and efficiency characteristic parameters to obtain the power characteristic parameter set.

6. The method for identifying hidden cracks of a photovoltaic module according to claim 1, wherein Perform feature weight assignment and normalization processing on the voltage feature vector and the power characteristic parameter set. By constructing a multi-dimensional feature space mapping, obtain the standardized feature vector, including: Perform dimensional normalization processing on the voltage feature vector. Through the maximum-minimum normalization method, obtain the standardized voltage feature; Perform dimensional normalization processing on the power characteristic parameter set. Through the Z-score normalization method, obtain the standardized power feature; Evaluate the importance of the standardized voltage feature. Through information gain ratio calculation, obtain the voltage feature weight coefficient; Evaluate the importance of the standardized power feature. Through Gini coefficient calculation, obtain the power feature weight coefficient; Perform weighted combination on the standardized voltage feature and the standardized power feature. Through the feature weight coefficient, obtain the standardized feature vector.

7. The method for identifying hidden cracks in a photovoltaic module according to claim 1, wherein Performing multi-level threshold discrimination analysis on the standardized feature vector, and obtaining the hidden crack diagnosis data of the photovoltaic module by calculating the feature space distance and the historical data trend, including: Performing clustering analysis on the standardized feature vector, and obtaining the feature distribution density by calculating the Euclidean distance in the feature space; Setting multi-level discrimination thresholds for the feature distribution density, and obtaining the hidden crack degree grading standard through the statistical distribution of normal samples; Performing trend analysis on the standardized feature vector, and obtaining the development trend parameters by calculating the feature change rate and acceleration; Performing comprehensive judgment on the hidden crack degree grading standard and the development trend parameters, and obtaining the hidden crack level through multi-level threshold comparison; Integrating the hidden crack level and the development trend to obtain the hidden crack diagnosis data of the photovoltaic module.

8. A photovoltaic module crack identification device for implementing the photovoltaic module crack identification method according to any one of claims 1-7, characterized in that, The photovoltaic module hidden crack identification device includes: An acquisition module, configured to perform field bus connection and voltage acquisition module deployment on each photovoltaic module in the photovoltaic string, perform master-slave communication control through the bus communication protocol, and obtain the original voltage sampling data stream; A calibration module, configured to perform time synchronization protocol calibration processing on the original voltage sampling data stream, transmit timestamp information via synchronization messages and follow-up messages, and measure the transmission delay through delay request messages and delay response messages to obtain the voltage data after time synchronization calibration; An extraction module, configured to perform statistical feature extraction on the voltage data after time synchronization calibration, and obtain the voltage feature vector and the preliminary hidden crack suspicious point mark by calculating the voltage difference between adjacent modules, the voltage mean, the standard deviation, and the fluctuation characteristics; An acquisition module, configured to collect the working characteristic data of the maximum power point tracking controller for the photovoltaic module corresponding to the voltage feature vector, and obtain the power characteristic parameter set by analyzing the power tracking efficiency, the dynamic response, and the boost characteristic; A processing module, configured to perform feature weight assignment and standardization processing on the voltage feature vector and the power characteristic parameter set, and obtain the standardized feature vector by constructing a multi-dimensional feature space mapping; A discrimination module, configured to perform multi-level threshold discrimination analysis on the standardized feature vector, and obtain the hidden crack diagnosis data of the photovoltaic module by calculating the feature space distance and the historical data trend.

Images