Intelligent cable fault location method and device for photovoltaic systems

Abstract

The application relates to the technical field of fault positioning, and discloses an intelligent cable fault positioning method and device for a photovoltaic system. The method comprises the following steps: obtaining a feature data matrix containing voltage difference between components, string current change rate, component power value and adjacent component voltage ratio by performing characteristic parameter calculation and processing on component-level synchronous sampling data; obtaining component abnormal state marking data by performing multi-dimensional threshold analysis and sliding variance analysis processing on the feature data matrix; obtaining cable fault type judgment data by performing decision tree classification processing on the component abnormal state marking data; obtaining cable fault point position data by performing voltage gradient analysis processing on the region where the cable fault type judgment data is located; and performing fault information integration processing on the cable fault point position data and the cable fault type judgment data to obtain cable fault diagnosis data. The application improves the efficiency and accuracy of intelligent cable fault positioning of the photovoltaic system.

Description

Technical Field

[0001] This application relates to the field of fault location, and in particular to a method and apparatus for intelligent cable fault location in a photovoltaic system. Background Technology

[0002] Photovoltaic (PV) systems, as a way to utilize renewable energy, are widely used in residential and industrial sectors. However, in PV power generation systems, cables, as a crucial component connecting various PV modules, often face faults due to external environmental factors or aging. Existing methods for PV cable fault location primarily rely on traditional cable voltage and current monitoring, combined with manual inspections. While these methods can detect faults to some extent, they are susceptible to errors and inaccuracies due to their dependence on manual operation and traditional monitoring techniques, and they cannot provide real-time, efficient fault diagnosis and location.

[0003] Existing technologies have the following shortcomings in fault location: First, existing methods usually rely on traditional voltage and current acquisition methods, which cannot achieve accurate real-time monitoring of cable faults; second, there is a lack of intelligent and automated fault diagnosis methods, which makes it impossible to quickly locate faults after they occur, thus affecting the system's operation and maintenance efficiency; third, existing technologies are difficult to accurately locate fault points, especially in complex photovoltaic systems, where traditional methods usually require manual intervention to complete diagnosis and maintenance, which cannot effectively improve the response speed and accuracy of fault handling. Summary of the Invention

[0004] This application provides a method and apparatus for intelligent cable fault location in photovoltaic systems, which improves the efficiency and accuracy of intelligent cable fault location in photovoltaic systems.

[0005] Firstly, this application provides a method for locating intelligent cable faults in a photovoltaic system, the method comprising:

[0006] Each module in the photovoltaic string undergoes voltage and current synchronous sampling processing through a multi-channel synchronous potential acquisition module to obtain module-level synchronous sampling data.

[0007] The component-level synchronous sampling data is processed by characteristic parameter calculation to obtain a feature data matrix containing inter-component voltage difference, string current change rate, component power value and adjacent component voltage ratio;

[0008] The feature data matrix is ​​processed by multi-dimensional threshold analysis and sliding variance analysis to obtain component abnormal state labeling data;

[0009] The abnormal state marking data of the components are classified using a decision tree to obtain cable fault type determination data;

[0010] Voltage gradient analysis is used to process the area where the cable fault type determination data is located to obtain the cable fault location data;

[0011] The cable fault location data and cable fault type determination data are integrated and processed to obtain cable fault diagnosis data.

[0012] Secondly, this application provides an intelligent cable fault location device for a photovoltaic system, the intelligent cable fault location device for a photovoltaic system comprising:

[0013] The acquisition module is used to perform voltage and current synchronous sampling processing on each component in the photovoltaic string through a multi-channel synchronous potential acquisition module to obtain component-level synchronous sampling data.

[0014] The calculation module is used to process the component-level synchronous sampling data through feature parameter calculation to obtain a feature data matrix containing the voltage difference between components, the string current change rate, the component power value and the voltage ratio of adjacent components;

[0015] The analysis module is used to process the feature data matrix through multi-dimensional threshold analysis and sliding variance analysis to obtain component abnormal state labeling data;

[0016] The classification module is used to classify the abnormal state marker data of the components through a decision tree to obtain cable fault type determination data;

[0017] The processing module is used to process the area where the cable fault type determination data is located through voltage gradient analysis to obtain the cable fault location data.

[0018] The integration module is used to integrate and process the cable fault location data and cable fault type determination data to obtain cable fault diagnosis data.

[0019] The technical solution provided in this application accurately identifies the location of cable faults and determines their distribution characteristics through abrupt change feature analysis and spatial distribution analysis of cable fault type determination data. Abrupt change feature analysis can promptly capture abnormal changes occurring during cable operation, such as drastic fluctuations in voltage or current, thereby quickly pinpointing potential fault points. Spatial distribution analysis, through spatial correlation analysis of the states of various components in the cable, helps to clarify the specific location of the fault, greatly improving the accuracy of fault location. Secondly, the location of cable fault points is further optimized through difference calculation and normalization processing, ensuring the accuracy of voltage difference and voltage gradient data. Especially in complex photovoltaic systems, this effectively reduces misjudgments and improves the accuracy of fault location. Regarding the determination of fault occurrence time and the extraction of fault features, this invention accurately records and correlates the time of fault occurrence through timestamp correlation processing, thus providing a clear timeline for subsequent fault analysis. This time correlation not only helps in the timely detection and response to faults but also provides detailed time information after the fault occurs, allowing maintenance personnel to analyze and locate the cause and development process of the fault. By extracting fault features, the type and severity of faults can be further identified, and their specific manifestations, such as abnormal current and voltage fluctuations, can be clarified, thus providing strong data support for system repair. The extraction of these fault features is crucial for determining the nature of the fault and assessing its impact range, ensuring the system's sensitive response to different types of faults. Furthermore, impact range analysis comprehensively considers the time of fault occurrence and its impact on the system, assessing the fault's expansion area and its impact on surrounding components. This analysis not only helps to quickly confirm the fault's impact range but also provides precise decision support in the operation and maintenance management of photovoltaic systems. By integrating multi-dimensional data such as fault occurrence time, fault feature description, and fault impact assessment, the cable fault diagnosis data ultimately generated by this invention provides maintenance personnel with comprehensive and accurate fault diagnosis information, ensuring rapid fault location and timely repair, minimizing system downtime and maintenance costs. Attached Figure Description

[0020] To more clearly illustrate the technical solutions of the embodiments of the present invention, the drawings used in the description of the embodiments will be briefly introduced below. Obviously, the drawings described below are some embodiments of the present invention. For those skilled in the art, other drawings can be obtained based on these drawings without creative effort.

[0021] Figure 1 This is a schematic diagram of an embodiment of the intelligent cable fault location method for a photovoltaic system in this application.

[0022] Figure 2This is a schematic diagram of one embodiment of the intelligent cable fault location device for a photovoltaic system in this application. Detailed Implementation

[0023] This application provides a method and apparatus for intelligent cable fault location in a photovoltaic system. The terms "first," "second," "third," "fourth," etc. (if present) in the specification, claims, and accompanying drawings are used to distinguish similar objects and are not necessarily used to describe a specific order or sequence. It should be understood that such data can be interchanged where appropriate so that the embodiments described herein can be implemented in a sequence other than that illustrated or described herein. Furthermore, the terms "comprising" or "having" and any variations thereof are intended to cover a non-exclusive inclusion; for example, a process, method, system, product, or apparatus that comprises a series of steps or units is not necessarily limited to those steps or units explicitly listed, but may include other steps or units not explicitly listed or inherent to such processes, methods, products, or apparatus.

[0024] For ease of understanding, the specific process of the embodiments of this application is described below. Please refer to [link / reference]. Figure 1 One embodiment of the intelligent cable fault location method for photovoltaic systems in this application includes:

[0025] Step S101: Perform voltage and current synchronous sampling processing on each component in the photovoltaic string through a multi-channel synchronous potential acquisition module to obtain component-level synchronous sampling data;

[0026] Step S102: The component-level synchronous sampling data is processed by characteristic parameter calculation to obtain a feature data matrix containing the voltage difference between components, the string current change rate, the component power value and the voltage ratio of adjacent components;

[0027] Step S103: Process the feature data matrix through multi-dimensional threshold analysis and sliding variance analysis to obtain component abnormal state labeling data;

[0028] Step S104: The abnormal state marking data of the components is processed by decision tree classification to obtain cable fault type determination data;

[0029] Step S105: The area where the cable fault type determination data is located is processed by voltage gradient analysis to obtain the cable fault location data;

[0030] Step S106: Integrate the fault information data of the cable fault location and the cable fault type determination data to obtain cable fault diagnosis data.

[0031] It is understood that the executing entity of this application can be a smart cable fault location device for a photovoltaic system, or it can be a terminal or a server; no specific limitation is made here. This application's embodiment uses a server as an example for illustration.

[0032] Specifically, for each module in the photovoltaic string, the voltage and current of the module are synchronously sampled using a multi-channel synchronous potential acquisition module. Each photovoltaic module's voltage and current data are synchronously measured through a specific voltage and current acquisition module, and the acquired data is transmitted to the control system via the multi-channel synchronous module. This synchronous sampling module ensures that the voltage and current data of each module are accurately acquired at the same time through a time synchronization mechanism, avoiding errors caused by sampling delays between different modules, thus ensuring data accuracy. Taking a module in the photovoltaic string as an example, if the voltage of this module is 12V and the current is 5A, the data acquired by this module will be recorded as (12V, 5A), and this data will be acquired synchronously with the data of other modules. Based on the synchronous sampling data, a feature data matrix is ​​obtained through characteristic parameter calculation, including the voltage difference between modules, the string current change rate, the module power value, and the voltage ratio of adjacent modules. The voltage difference represents the deviation of the voltage between each module, the string current change rate reflects the degree of current fluctuation, the module power value is the product of voltage and current, and the voltage ratio of adjacent modules shows the ratio relationship of voltage between adjacent modules. For example, assuming the voltages of two adjacent modules in a photovoltaic string are 12V and 11.8V respectively, the voltage difference can be calculated as 12V - 11.8V = 0.2V; assuming the string current changes from 4A to 5A, the current change rate is (5A - 4A) / 4A = 0.25. These calculation results are then integrated into the feature data matrix.

[0033] The feature data matrix underwent multi-dimensional threshold analysis and moving variance analysis to obtain component abnormal state labeling data. Multi-dimensional threshold analysis sets a threshold by comparing with historical data; data exceeding this threshold is labeled as abnormal. Moving variance analysis detects abnormal fluctuations in components by calculating the variance of the data within a time window. For example, if the voltage and current differences of a component exceed a set threshold, the component is considered potentially faulty. Assuming that the variance of a component's voltage difference over the past 10 sampling points is 0.05 using moving window processing, if this variance exceeds the preset threshold of 0.03, the component is labeled as abnormal. The abnormal state labeling data is then processed through decision tree classification to obtain cable fault type determination data. Decision tree classification determines the type of cable fault based on a series of features. The decision tree here generates a tree structure through classification training on the abnormal labeling data, progressively considering features such as component abnormal state, power attenuation, and voltage ratio to ultimately determine the specific type of fault. For example, if the current of a component drops sharply and is accompanied by an increase in voltage difference, the decision tree algorithm may determine it as an "open circuit" fault; if the voltage fluctuates greatly but the current changes little, it may determine it as a "poor contact" fault.

[0034] Voltage gradient analysis maps cable fault type determination data to the spatial layout of the photovoltaic system, revealing the location of the cable fault point. Voltage gradient analysis is based on voltage variations within the cable's area. By calculating the voltage difference between modules near the fault point and normalizing this difference with module spacing, voltage gradient data is obtained, thus determining the approximate location of the fault. For example, if a fault occurs in a section of the cable, the voltage difference can be calculated; if the voltage gradient in that area is large, the system identifies that section as the fault area. The cable fault location data and cable fault type determination data are then integrated to generate complete cable fault diagnosis data. This integration process includes correlating the fault occurrence time with timestamps and further assessing the fault's impact range by combining the fault type and location. For example, if a cable fault occurs at the connection point to the main controller, and the current change rate at that location is large, the impact range of the fault may cover the entire system. Through this data integration, the system can provide a comprehensive fault report, including the time of fault occurrence, specific location, and fault type.

[0035] For example, suppose a cable in a photovoltaic system fails during operation. Following the steps outlined above, voltage and current data for each component are first collected, and then the voltage difference and current change rate between components are calculated. After threshold and variance analysis, a component is identified as being in an abnormal state, and a decision tree classification determines the fault to be a poor connection. Voltage gradient analysis then shows the fault is located in the connecting cable section of the photovoltaic string. Finally, the integrated fault information indicates the fault occurred on November 1, 2024, and the fault point is located in the cable connecting component number 5 to the main controller. With this precise data, maintenance personnel can quickly locate the fault and take corrective action, greatly improving the efficiency and accuracy of system maintenance.

[0036] In this embodiment, by analyzing the abrupt change characteristics and spatial distribution of cable fault type determination data, the occurrence point of cable faults can be accurately identified and their distribution characteristics determined. The abrupt change characteristic analysis can promptly capture abnormal changes occurring during cable operation, such as drastic fluctuations in voltage or current, thereby quickly pinpointing potential fault points. Spatial distribution analysis, through spatial correlation analysis of the states of various components in the cable, helps to clarify the specific location of the fault, greatly improving the accuracy of fault location. Secondly, the location of cable fault points is further optimized through difference calculation and normalization processing, ensuring the accuracy of voltage difference and voltage gradient data. Especially in complex photovoltaic systems, this effectively reduces misjudgments and improves the accuracy of fault location. Regarding the determination of fault occurrence time and the extraction of fault features, this invention uses timestamp correlation processing to accurately record and correlate the time of fault occurrence, thus providing a clear timeline for subsequent fault analysis. This time correlation not only helps in the timely detection and response to faults but also provides detailed time information after the fault occurs, allowing maintenance personnel to analyze and locate the cause and development process of the fault. By extracting fault features, the type and severity of faults can be further identified, and their specific manifestations, such as abnormal current and voltage fluctuations, can be clarified, thus providing strong data support for system repair. The extraction of these fault features is crucial for determining the nature of the fault and assessing its impact range, ensuring the system's sensitive response to different types of faults. Furthermore, impact range analysis comprehensively considers the time of fault occurrence and its impact on the system, assessing the fault's expansion area and its impact on surrounding components. This analysis not only helps to quickly confirm the fault's impact range but also provides precise decision support in the operation and maintenance management of photovoltaic systems. By integrating multi-dimensional data such as fault occurrence time, fault feature description, and fault impact assessment, the cable fault diagnosis data ultimately generated by this invention provides maintenance personnel with comprehensive and accurate fault diagnosis information, ensuring rapid fault location and timely repair, minimizing system downtime and maintenance costs.

[0037] In one specific embodiment, the process of performing step S101 may specifically include the following steps:

[0038] (1) Send a synchronous acquisition broadcast command to the POWERBUS fieldbus master station to perform time synchronization processing, obtain the IEEE1588 clock synchronization signal, and perform master-slave clock synchronization processing on the IEEE1588 clock synchronization signal to obtain slave station synchronization clock data.

[0039] (2) The slave station synchronization clock data is processed by clock discipline to obtain a 1MHz sampling frequency signal, and the 1MHz sampling frequency signal is divided to obtain a sampling control signal.

[0040] (3) The output voltage signal of the photovoltaic module is processed by high-precision analog-to-digital conversion to obtain the digital value of the module voltage, and the current signal of the photovoltaic string is processed by the shunt monitor to obtain the digital value of the string current.

[0041] (4) The digital values ​​of component voltage and string current are transmitted and processed through the MODBUS protocol to obtain component-level raw data, and the component-level raw data is processed through data verification to obtain component-level verification data.

[0042] (5) The component-level verification data is processed by data latching to obtain component-level latched data, and the component-level latched data is processed by data packaging to obtain component-level synchronous sampling data.

[0043] Specifically, the synchronous acquisition broadcast command sent by the POWERBUS fieldbus master station undergoes time synchronization processing to obtain the IEEE 1588 clock synchronization signal, and slave station synchronous clock data is obtained through master-slave clock synchronization processing. IEEE 1588 is a precise time synchronization protocol used to ensure time consistency between different devices in a distributed system. In this process, the master station transmits clock signals between various devices in the photovoltaic system by sending synchronization commands. The clock difference between the master and slave stations is adjusted through the master-slave clock synchronization protocol to achieve clock synchronization, ensuring that the timestamps of the acquired voltage and current data are consistent and avoiding data corruption caused by time deviations. In the second step, the slave station synchronous clock data is processed by clock discipline to obtain a 1MHz sampling frequency signal, which is then divided to obtain the sampling control signal. Clock discipline technology adjusts the slave station's clock source to keep it consistent with the master station's clock. Through clock frequency division, the high-frequency clock signal (such as 1MHz) is converted into a frequency suitable for device sampling, ensuring the stability and accuracy of the data acquisition system. For example, suppose the system requires data sampling at a frequency of 1 MHz per second. In this case, clock division technology is used to convert the high-frequency clock into the actual sampling control signal, thereby adjusting the timing of data acquisition from the photovoltaic module.

[0044] The output voltage signal of the photovoltaic module is processed by a high-precision analog-to-digital converter (ADC) to obtain the digital value of the module voltage. An ADC is used to convert analog signals into digital signals; a high-precision ADC is typically used to ensure the accuracy of the converted data. Assuming the voltage of a photovoltaic module is 12V, after ADC conversion, a digital value corresponding to 12V, such as 2000, might be obtained. Simultaneously, the string current signal is processed by a shunt monitor to obtain the digital value of the string current. The shunt monitor calculates the current value by measuring the voltage drop across a known resistor. For example, if the shunt monitor detects a voltage of 0.05V and a resistance of 0.01Ω, the current value is calculated to be 5A using Ohm's law. The digital values ​​of the module voltage and string current are transmitted and processed via the MODBUS protocol to obtain the module-level raw data. The MODBUS protocol is a commonly used communication protocol for communication between different devices. In this system, the module voltage and current data are sent to the control center via the MODBUS protocol. During transmission, data verification ensures the correctness of the transmission. Common verification methods include parity check and CRC check, used to detect whether errors occur during transmission. For example, if a component's voltage is 2000 and its current is 5000 during data transmission, and this data is transmitted via the MODBUS protocol, then if the CRC check result is correct, the data is considered valid.

[0045] Component-level verification data undergoes data latching to obtain component-level latched data, and then is packaged to obtain component-level synchronous sampling data. Data latching technology is used to save the verified data to a buffer to prevent data loss or overwriting. At this point, the data latching module saves the received data to the storage unit according to the latching clock, ensuring data persistence. Next, data packaging process packages all latched data into a single data packet for easy subsequent transmission or storage. The packaged data format typically includes information such as timestamps, component numbers, voltage values, and current values, ensuring data integrity and ease of subsequent analysis.

[0046] For example, suppose a photovoltaic (PV) system string samples a voltage of 12V and a current of 5A. After analog-to-digital conversion, this results in digital values ​​of 2000 and 5000. These data are transmitted via the MODBUS protocol and then checked against a CRC checksum to ensure accuracy. After data latching, the data is stored in a buffer and subsequently packaged into synchronous sampling data, containing the digital values ​​of voltage and current, a timestamp, and component information. Ultimately, this data will be used for subsequent PV module performance analysis and fault diagnosis, helping to locate cable faults in the system and optimize power output.

[0047] In one specific embodiment, the process of performing step S102 may specifically include the following steps:

[0048] (1) The component voltage in the component-level synchronous sampling data is processed by mean calculation to obtain the string average voltage data, and the component voltage and string average voltage data are processed by difference calculation to obtain the voltage difference between components.

[0049] (2) The string current in the component-level synchronous sampling data is processed by time difference to obtain the string current change rate, and the component voltage and string current in the component-level synchronous sampling data are processed by product operation to obtain the component power value.

[0050] (3) The voltage of adjacent components is processed by ratio calculation to obtain the voltage ratio of adjacent components. The voltage difference between components, string current change rate, component power value and voltage ratio of adjacent components are processed by matrix splicing to obtain the feature data matrix.

[0051] Specifically, the component voltage data undergoes averaging to obtain the string's average voltage data. This process aims to derive a representative voltage level by statistically calculating the voltage values ​​of each component within the string. For example, in a string, if each component's voltage is 12.1V, 12.0V, 11.9V, 12.2V, etc., the averaging process involves adding these voltage values ​​and dividing by the number of components to obtain the string's average voltage value. This step reduces fluctuations caused by abnormal voltages in individual components, ensuring a more stable voltage benchmark and providing an accurate reference value for subsequent comparative analysis. The difference between the component voltages and the string's average voltage data is calculated to obtain the inter-component voltage difference. This process measures the deviation between each component's voltage and the string's average voltage. By calculating the voltage difference, components with significant voltage inconsistencies can be identified, which may be due to faults, poor connections, or environmental factors. Large changes in inter-component voltage differences usually indicate potential faults or anomalies, thus guiding the fault localization process.

[0052] The string current signal is processed using time-difference analysis to obtain the string current change rate. Time-difference analysis describes the rate of current change over time by calculating the difference between the current at a given moment and the current at the previous moment. This rate of change reflects current fluctuations. A large current change rate may indicate abnormal system load changes or potential faults. The current change rate is a very important indicator in fault detection because abnormal current fluctuations may be caused by cable breaks, poor contacts, or circuit faults. The component power value is obtained by multiplying the component voltage and string current. The power value calculation can comprehensively reflect the actual operating status of the component by considering both voltage and current changes. Low or unstable power values ​​are often precursors to system faults, possibly caused by component failure, voltage instability, or current problems. The power value change is closely related to voltage and current; therefore, power value analysis helps to discover some fault types that are difficult to detect by voltage or current alone.

[0053] The voltage ratio between adjacent components is obtained by calculating the voltage ratio of adjacent components. This calculation helps in further analyzing the voltage differences between components in the string, especially between adjacent components. A large voltage ratio may indicate abnormal operation of some components or problems with cable connections, thus affecting the performance of adjacent components. Finally, the voltage difference between components, the string current change rate, component power values, and the voltage ratio between adjacent components are processed through matrix concatenation to obtain a comprehensive feature data matrix. This feature data matrix not only integrates various performance indicators but also reflects the overall operating status of the components and the string. Matrix concatenation combines various feature data, which is helpful for subsequent anomaly detection and fault analysis. For example, if the component voltage difference is large, the current change rate is also abnormal, and the power value is low, the combination of these features can often indicate potential cable fault points, thus providing strong data support for fault location.

[0054] For example, suppose the average voltage of a string is calculated to be 12.02V, but the voltage of some components deviates significantly from this average, perhaps by 0.5V or higher. Such discrepancies suggest potential cable connection problems or component failures. Simultaneously, the rate of change of string current indicates large current fluctuations and low power values. Combined with variations in the voltage ratios of adjacent components, these characteristic data are ultimately pieced together into a feature data matrix. If this matrix contains multiple abnormal indicators, the potential location of the fault can be inferred, allowing for further cable fault diagnosis.

[0055] In one specific embodiment, the process of executing step S103 may specifically include the following steps:

[0056] (1) The feature data matrix is ​​statistically processed using historical data to obtain the feature parameter threshold, and the parameters in the feature data matrix are compared with the feature parameter threshold to obtain the threshold exceeding mark;

[0057] (2) The feature data matrix is ​​divided into short-term time window data and long-term time window data by time window division, and the variance ratio data is obtained by variance calculation of the short-term time window data and long-term time window data.

[0058] (3) The variance ratio data and threshold over-limit markers are processed by logical combination to obtain component abnormal state marker data.

[0059] Specifically, the construction of the feature data matrix includes various key parameters reflecting the operating status of components. The key to this method is how to analyze the system using these parameters and accurately identify potential fault points. The following section explains in detail how fault detection and location are achieved through the feature data matrix. By performing historical data statistical processing on the feature data matrix, threshold values ​​for feature parameters can be obtained. The core of historical data statistical processing lies in setting normal value ranges for various feature parameters through the analysis of a large amount of historical data. For example, based on historical data statistics, the normal fluctuation range for voltage might be ±5%, and the threshold for current fluctuation might be ±3%. Through these statistical data, a threshold can be defined for each feature parameter. During actual operation, each parameter in the feature data matrix is ​​compared with the set threshold. This comparison yields a threshold exceedance marker, indicating whether a particular feature parameter has exceeded its normal range. For example, if the voltage of a component exceeds the normal fluctuation range of ±5%, then the voltage parameter of that component will be marked as exceeding the limit. This exceedance marker is the basis for subsequent fault analysis and location.

[0060] Dividing the feature data matrix into time windows allows for more refined analysis of the data's temporal dimension. By categorizing the feature data into short-term and long-term time windows, the system's performance at different time scales can be observed more accurately. Short-term time window data reflects the system's changing trends over a shorter period, while long-term time window data reflects a more stable system state. The division of time windows is typically determined based on the data collection frequency and actual needs; short-term windows may be minutes or hours, while long-term windows may be days or longer. This approach allows for a comprehensive analysis of changes at different time scales. Then, the short-term and long-term time window data are processed through variance calculation to obtain the variance ratio. Variance is an important indicator of the dispersion of data distribution; a larger variance indicates more drastic data fluctuations. In photovoltaic systems, fluctuations in data such as voltage and current are usually closely related to the system's health. By calculating the variance of the short-term and long-term window data, the difference between the system's short-term volatility and long-term stability can be analyzed. For example, if the variance of short-term data is large while the variance of long-term data is small, this may indicate abnormal fluctuations in the short term, requiring further analysis to determine the cause, which could be due to factors such as poor contact or temperature changes. Conversely, if both short-term and long-term variances are large, it may indicate a persistent system malfunction.

[0061] Finally, the variance ratio data and threshold exceedance markers are logically combined to obtain the abnormal state marker data for the components. Logical combination refers to performing logical operations on multiple conditions to arrive at a comprehensive judgment result. In this scheme, the combination of variance ratio and threshold exceedance markers forms a comprehensive judgment standard. If the variance ratio is too large and some parameters exceed the threshold, then the component can be determined to be in an abnormal state. Through this logical combination, multiple data features can be combined to accurately determine which components may be faulty and which may be in normal working condition. For example, suppose the normal fluctuation range of the component voltage in a photovoltaic system is determined to be ±4% in historical data statistics. In actual operation, the voltage of a component suddenly exceeds this range, reaching ±7%. According to the threshold exceedance marker rule, the voltage of this component is marked as exceeding the limit. Then, the voltage data of this component is divided into time windows. The variance calculation result in the short-term window is 0.5, while the variance in the long-term window is 0.1. This indicates that the component voltage fluctuates more in the short term and less in the long term, which may be due to some temporary faults. Finally, by combining the variance ratio data (0.5 / 0.1 = 5) with the out-of-limit marker, the system determines the abnormal state of the component as a "fault" state. In this way, not only can the abnormality of the component itself be captured, but also the persistence of the fault can be analyzed, further providing a precise basis for fault location.

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

[0063] (1) The abnormal state marking data of the components is processed by mutation feature analysis to obtain mutation feature data, and the abnormal state marking data of the components is processed by spatial distribution analysis to obtain abnormal distribution data;

[0064] (2) The power values ​​of the components are processed by attenuation characteristic analysis to obtain power attenuation data, and the voltage ratios of adjacent components are processed by abnormal characteristic analysis to obtain voltage ratio abnormal data;

[0065] (3) The data on sudden change characteristics, abnormal distribution, power attenuation and voltage ratio abnormality are processed by decision tree nodes to obtain cable fault type determination data.

[0066] Specifically, after processing the component abnormal state marker data through mutation feature analysis, mutation feature data can be obtained. Mutation feature analysis focuses on sudden changes in component data during monitoring, especially drastic fluctuations in key parameters such as voltage and current, which usually indicate an anomaly or fault in the system. For example, if the voltage of a component changes significantly within a short period, exceeding the normal fluctuation range, it may indicate a fault in that component. By identifying such "mutation" phenomena, mutation feature analysis can effectively capture instantaneous fault events, helping to diagnose potential problems in the system in a timely manner. In practice, if the voltage of a photovoltaic module drops sharply from 4V to 0V within one hour, this mutation will be marked as an anomaly, serving as a basis for further fault location. Simultaneously, the component abnormal state marker data also undergoes spatial distribution analysis to obtain anomaly distribution data. Spatial distribution analysis focuses on the spatial distribution patterns of faults or anomalies among different components. If multiple adjacent components exhibit similar anomalies, it may indicate a common fault factor in the cables or connections between these components. By analyzing the geographical location of the abnormal state markers, spatial distribution analysis can help identify the possible range of the fault source. For example, if multiple abnormal markings appear on a certain part of a cable, it can be inferred that there may be a problem with poor contact or insulation damage in that part, thereby narrowing down the scope of troubleshooting and improving the efficiency of fault location.

[0067] Power attenuation data can be obtained by performing attenuation characteristic analysis on component power values. Power attenuation is usually related to factors such as component aging, contamination, poor contact, or internal damage. Attenuation characteristic analysis identifies attenuation patterns by calculating the trend of component power change over time, thereby revealing potential causes of failure. If the power of a component continues to decrease over time, and the rate of decrease is significantly higher than that of other normal components, it can be determined that the component may be faulty. For example, if the power of a component decreases by 8% and 12% respectively over two consecutive days, while the power changes of other components in the same area remain stable, attenuation characteristic analysis can determine that the component has a high risk of failure. Voltage ratio anomaly data is obtained by performing anomaly characteristic analysis on the voltage ratio of adjacent components. The voltage ratio of adjacent components can be used to assess the balance of electrical connections. If the voltage ratio suddenly deviates from the normal range, it usually indicates a problem with the cable connection or contact. For example, in a parallel cable, if the voltage ratio of two adjacent components changes significantly, it indicates that the cable or connection may be damaged or have poor contact. Voltage ratio anomaly analysis can identify abnormal fluctuations in these ratios, enabling timely detection of potential electrical faults in cable systems and providing a more accurate basis for fault location.

[0068] After analyzing the mutation feature data, abnormal distribution data, power attenuation data, and voltage ratio anomaly data, this data is input into a decision tree for final fault type determination. Decision trees are a commonly used machine learning algorithm capable of making accurate fault judgments based on known feature data. In this scheme, the nodes of the decision tree make judgments based on different feature data. For example, if a component exhibits significant mutation characteristics, large power attenuation, and an abnormal voltage ratio, the decision tree will mark this component as "potentially faulty." Based on this judgment, further fault localization can determine the location and type of the fault. For example, suppose in a photovoltaic system, the voltage of a component experiences a sudden drop from 4V to 0V, and other components in the same area also exhibit similar voltage anomalies. Spatial distribution analysis shows that these anomalies are concentrated in the same area, indicating a potential problem with the cables in that area. Simultaneously, power attenuation feature analysis reveals that multiple components in this area experience rapid power attenuation. Combined with the voltage ratio analysis of adjacent components, the voltage ratio is significantly low, indicating a possible poor contact in the cable connections. Finally, all these abnormal data points are fed into a decision tree for processing. The decision tree ultimately concludes that a cable fault exists in the area. Based on this conclusion, the root cause of the fault can be quickly located and repaired, ensuring the stable operation of the photovoltaic system.

[0069] In one specific embodiment, the process of executing step S105 may specifically include the following steps:

[0070] (1) The cable fault type determination data is processed by spatial mapping to obtain preliminary fault area data, and the voltage of adjacent components in the preliminary fault area data is processed by difference calculation to obtain voltage difference data.

[0071] (2) The voltage difference data is normalized by component spacing to obtain voltage gradient data, and the voltage gradient data is processed by extreme value detection to obtain cable fault location data.

[0072] Specifically, spatial mapping processing is performed on cable fault type determination data to identify preliminary fault area data. Spatial mapping processing refers to analyzing and mapping the operating data of cable components, such as voltage and current, according to their relative spatial distribution patterns to identify potential fault areas. For example, if the voltage values ​​of some components are significantly low or fluctuate abnormally, and these components are physically close to each other, it may indicate that the fault is concentrated in that area. The preliminary fault area is determined by performing spatial clustering analysis on these signals to identify abnormal components and areas, forming preliminary fault area data. The voltage difference between adjacent components within the preliminary fault area data is calculated to further confirm the location and range of the fault. Processing voltage difference data helps determine the specific range of the fault because in an electrical system, a sudden drop or fluctuation in voltage in a certain part usually indicates a fault in that part. By calculating the difference between the voltages of adjacent components, the fault area can be located more accurately, especially when the voltage difference exceeds the normal range, indicating a problem with the cable in that area. This process can determine whether there is a continuous fault signal in the area based on the specific voltage difference, thereby eliminating interference from normal fluctuations and helping to accurately locate the fault point.

[0073] For voltage difference data, further normalization processing of the module spacing is performed to obtain voltage gradient data. Module spacing normalization processing refers to combining the voltage difference with the physical distance between modules to eliminate the influence of the physical distance between different modules on the voltage difference. In this way, changes in voltage gradient can be reflected more accurately without being affected by module spacing. For example, if the voltage difference between two adjacent modules is large and the two modules are physically close, then the voltage gradient is large, which may indicate a serious electrical fault in the area. After extreme value detection, the location data of the cable fault point can be obtained from the voltage gradient data. Extreme value detection uses algorithms to filter out the maximum or minimum values ​​in the voltage gradient to identify the location of the fault point. In electrical systems, faults usually cause drastic voltage changes, so by detecting the extreme values ​​of the voltage gradient, the location of the cable fault can be accurately determined. In this way, accurate cable fault location can be achieved, ensuring that the location of the faulty cable in the photovoltaic system can be quickly located, and then repaired or replaced.

[0074] For example, suppose the voltage of a group of modules in a photovoltaic system gradually decreases. Preliminary analysis shows that the voltage fluctuation in this area is significant. Initial fault area data obtained through spatial mapping processing shows that the fault area includes several adjacent modules. Further calculation of the voltage difference between these modules reveals that the voltage difference exceeds the normal fluctuation range. Therefore, the difference calculation confirms that a fault does exist in this area. Next, after normalizing the voltage difference data by the module spacing, voltage gradient data is obtained. If two modules in this area are close together and have a large voltage gradient, it indicates that the cable in this area may have a serious electrical problem. Finally, through extreme value detection, the fault point is successfully located, determining the cable fault location. Maintenance personnel can then quickly take measures to repair it, avoiding system losses due to delayed repairs.

[0075] In one specific embodiment, the process of executing step S106 may specifically include the following steps:

[0076] (1) The location data of cable fault points is processed by timestamp association to obtain the fault occurrence time data, and the cable fault type determination data is processed by fault feature extraction to obtain fault feature description data.

[0077] (2) The fault occurrence time data and fault characteristic description data are processed by the influence range analysis to obtain fault impact assessment data. The fault occurrence time data, cable fault location data, fault characteristic description data and fault impact assessment data are processed by data integration to obtain cable fault diagnosis data.

[0078] Specifically, after obtaining the cable fault location data, it is correlated with the fault occurrence timestamp. The timestamp records the exact time the cable fault occurred, serving as fundamental data in cable fault analysis. By combining the fault location with time, the specific moment of fault occurrence and its temporal characteristics can be determined, thus clarifying the fault's start time and evolution process. For example, if the voltage in a certain area drops at a certain moment and persists for a certain period, a fault may have occurred at that moment. This timestamp correlation helps clarify the timing of fault occurrence, which is crucial for analyzing the fault's expansion and impact range. Through fault feature extraction, the system can extract features from the fault occurrence time and other relevant data of the fault location, forming fault feature description data. Fault feature description data typically includes the fault type, severity, and specific environmental characteristics (such as temperature, voltage, current, etc.). For example, if the current fluctuates abnormally during a fault occurrence within a specific time period, this current anomaly can be extracted as one of the fault features. The core of fault feature extraction lies in identifying the specific manifestations of the fault, such as voltage drops, current fluctuations, or temperature changes, thereby providing a basis for subsequent fault assessment.

[0079] Fault occurrence time data and fault characteristic description data are processed through impact range analysis to obtain fault impact assessment data. Impact range analysis assesses the scope of the fault's impact on the system by combining the fault occurrence time with the status of other relevant areas in the system (such as voltage, current, temperature, etc.). Specifically, this process analyzes signal changes related to the fault occurrence time to determine which areas' performance is affected, thereby estimating the affected area. For example, if a voltage drop occurs in a cable area, impact range analysis considers voltage changes in other areas during the fault time, thus assessing which areas the fault might extend to and which components or cables might be affected. This impact range analysis helps determine the severity of the fault and its impact on the overall operation of the photovoltaic system. The fault impact assessment data is then integrated with fault occurrence time data, fault location data, and fault characteristic description data to ultimately obtain cable fault diagnosis data. This data integrates the fault's occurrence time, location, characteristics, and impact assessment results, providing a comprehensive basis for accurate cable fault diagnosis. Through data integration and processing, the specific circumstances of the fault can be determined more comprehensively and accurately, providing detailed guidance for subsequent repair work. Data integration and processing ensures the accuracy and efficiency of fault diagnosis results by merging various types of data.

[0080] For example, suppose a cable in a photovoltaic system fails. Through timestamp association processing, the specific time of the fault is first determined, such as a sharp drop in cable voltage at a certain moment. Next, through fault feature extraction, the system extracts features such as voltage drops and current fluctuations as fault feature description data. Subsequently, through impact range analysis, the system discovers that the fault may have spread from the affected cable area to adjacent areas, affecting the voltage of several other components. After data integration and processing, the system generates cable fault diagnosis data, clarifying the time, location, and type of the fault, and assessing the scope of its impact, ultimately providing accurate diagnostic information for fault repair.

[0081] The above describes the intelligent cable fault location method for photovoltaic systems in the embodiments of this application. The following describes the intelligent cable fault location device for photovoltaic systems in the embodiments of this application. Please refer to [link / reference]. Figure 2 One embodiment of the intelligent cable fault location device for photovoltaic systems in this application includes:

[0082] The acquisition module 201 is used to perform voltage and current synchronous sampling processing on each component in the photovoltaic string through a multi-channel synchronous potential acquisition module to obtain component-level synchronous sampling data.

[0083] The calculation module 202 is used to process the component-level synchronous sampling data through feature parameter calculation to obtain a feature data matrix containing inter-component voltage difference, string current change rate, component power value and adjacent component voltage ratio.

[0084] Analysis module 203 is used to process the feature data matrix through multi-dimensional threshold analysis and sliding variance analysis to obtain component abnormal state labeling data;

[0085] The classification module 204 is used to classify the abnormal state marking data of the components through a decision tree to obtain cable fault type determination data;

[0086] Processing module 205 is used to process the area where the cable fault type determination data is located through voltage gradient analysis to obtain cable fault location data;

[0087] The integration module 206 is used to integrate and process the cable fault location data and cable fault type determination data to obtain cable fault diagnosis data.

[0088] Through the collaborative efforts of the aforementioned components, and by analyzing the abrupt change characteristics and spatial distribution of cable fault type determination data, the occurrence point of cable faults can be accurately identified, and their distribution characteristics determined. Abrupt change characteristic analysis can promptly capture abnormal changes occurring during cable operation, such as drastic fluctuations in voltage or current, thereby quickly pinpointing potential fault points. Spatial distribution analysis, through spatial correlation analysis of the states of various components within the cable, helps to clarify the specific location of the fault, greatly improving the accuracy of fault location. Furthermore, the location of cable fault points is further optimized through difference calculation and normalization processing, ensuring the accuracy of voltage difference and voltage gradient data. Especially in complex photovoltaic systems, this effectively reduces misjudgments and improves the precision of fault location. Regarding the determination of fault occurrence time and the extraction of fault characteristics, this invention uses timestamp correlation processing to accurately record and correlate the time of fault occurrence, thus providing a clear timeline for subsequent fault analysis. This time correlation not only helps in the timely detection and response to faults but also provides detailed time information after the fault occurs, allowing maintenance personnel to analyze and locate the cause and development process of the fault. By extracting fault features, the type and severity of faults can be further identified, and their specific manifestations, such as abnormal current and voltage fluctuations, can be clarified, thus providing strong data support for system repair. The extraction of these fault features is crucial for determining the nature of the fault and assessing its impact range, ensuring the system's sensitive response to different types of faults. Furthermore, impact range analysis comprehensively considers the time of fault occurrence and its impact on the system, assessing the fault's expansion area and its impact on surrounding components. This analysis not only helps to quickly confirm the fault's impact range but also provides precise decision support in the operation and maintenance management of photovoltaic systems. By integrating multi-dimensional data such as fault occurrence time, fault feature description, and fault impact assessment, the cable fault diagnosis data ultimately generated by this invention provides maintenance personnel with comprehensive and accurate fault diagnosis information, ensuring rapid fault location and timely repair, minimizing system downtime and maintenance costs.

[0089] The above-described embodiments are only used to illustrate the technical solutions of this application, and are not intended to limit them. Although this application has been described in detail with reference to the foregoing embodiments, those skilled in the art should understand that modifications can still be made to the technical solutions described in the foregoing embodiments, or equivalent substitutions can be made to some of the technical features. Such modifications or substitutions 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 this application.

Claims

1. A method for intelligent cable fault location in a photovoltaic system, characterized in that, The intelligent cable fault location method for the photovoltaic system includes: Each module in the photovoltaic string undergoes voltage and current synchronous sampling processing through a multi-channel synchronous potential acquisition module to obtain module-level synchronous sampling data. The component-level synchronous sampling data is processed by characteristic parameter calculation to obtain a feature data matrix containing inter-component voltage difference, string current change rate, component power value and adjacent component voltage ratio; The feature data matrix is ​​processed by multi-dimensional threshold analysis and sliding variance analysis to obtain component abnormal state labeling data; The abnormal state marking data of the components are classified using a decision tree to obtain cable fault type determination data; Voltage gradient analysis is used to process the area where the cable fault type determination data is located to obtain the cable fault location data; The cable fault location data and cable fault type determination data are integrated and processed to obtain cable fault diagnosis data.

2. The intelligent cable fault location method for a photovoltaic system according to claim 1, characterized in that, The process involves synchronously sampling the voltage and current of each component in the photovoltaic string using a multi-channel synchronous potential acquisition module to obtain component-level synchronous sampling data, including: The POWERBUS fieldbus master station sends a synchronous acquisition broadcast command to perform time synchronization processing to obtain the IEEE1588 clock synchronization signal. The IEEE1588 clock synchronization signal is then processed by master-slave clock synchronization to obtain slave station synchronous clock data. The slave station synchronization clock data is processed by clock discipline to obtain a 1MHz sampling frequency signal, and the 1MHz sampling frequency signal is divided to obtain a sampling control signal. The output voltage signal of the photovoltaic module is processed by a high-precision analog-to-digital converter to obtain the digital value of the module voltage, and the current signal of the photovoltaic string is processed by a shunt monitor to obtain the digital value of the string current. The digital values ​​of the component voltage and the digital values ​​of the string current are transmitted and processed through the MODBUS protocol to obtain component-level raw data, and the component-level raw data is processed through data verification to obtain component-level verification data. The component-level verification data is processed by data latching to obtain component-level latched data, and the component-level latched data is processed by data packaging to obtain component-level synchronous sampling data.

3. The intelligent cable fault location method for a photovoltaic system according to claim 1, characterized in that, The component-level synchronous sampling data is processed through feature parameter calculation to obtain a feature data matrix containing inter-component voltage differences, string current change rates, component power values, and adjacent component voltage ratios, including: The component voltage in the component-level synchronous sampling data is processed by mean calculation to obtain string average voltage data, and the component voltage and string average voltage data are processed by difference calculation to obtain the voltage difference between components. The string current in the component-level synchronous sampling data is processed by time difference to obtain the string current change rate, and the component voltage and string current in the component-level synchronous sampling data are processed by product operation to obtain the component power value. The voltages of adjacent components are processed by ratio calculation to obtain the voltage ratio of adjacent components. The voltage difference between the components, the string current change rate, the component power value and the voltage ratio of adjacent components are processed by matrix concatenation to obtain the feature data matrix.

4. The intelligent cable fault location method for a photovoltaic system according to claim 1, characterized in that, The feature data matrix is ​​processed through multi-dimensional threshold analysis and moving variance analysis to obtain component abnormal state labeling data, including: The feature data matrix is ​​statistically processed using historical data to obtain feature parameter thresholds, and the parameters in the feature data matrix are compared with the feature parameter thresholds to obtain threshold exceedance markers. The feature data matrix is ​​divided into short-term time window data and long-term time window data by time window division, and the variance ratio data is obtained by variance calculation of the short-term time window data and the long-term time window data. The variance ratio data and the threshold exceedance marker are logically combined to obtain the component abnormal state marker data.

5. The intelligent cable fault location method for a photovoltaic system according to claim 1, characterized in that, The abnormal state marking data of the components is processed through decision tree classification to obtain cable fault type determination data, including: The abnormal state marker data of the components is processed by mutation feature analysis to obtain mutation feature data, and the abnormal state marker data of the components is processed by spatial distribution analysis to obtain abnormal distribution data; The power values ​​of the components are processed by attenuation characteristic analysis to obtain power attenuation data, and the voltage ratios of adjacent components are processed by anomaly characteristic analysis to obtain voltage ratio anomaly data; The cable fault type determination data is obtained by processing the mutation feature data, the abnormal distribution data, the power attenuation data, and the voltage ratio abnormal data through decision tree nodes.

6. The intelligent cable fault location method for a photovoltaic system according to claim 1, characterized in that, The region where the cable fault type determination data is located is processed through voltage gradient analysis to obtain cable fault location data, including: The cable fault type determination data is processed by spatial mapping to obtain preliminary fault area data, and the voltage of adjacent components within the preliminary fault area data is processed by difference calculation to obtain voltage difference data. The voltage difference data is normalized by component spacing to obtain voltage gradient data, and the voltage gradient data is processed by extreme value detection to obtain the cable fault location data.

7. The intelligent cable fault location method for a photovoltaic system according to claim 1, characterized in that, The process of integrating the cable fault location data and cable fault type determination data to obtain cable fault diagnosis data includes: The cable fault location data is processed by timestamp association to obtain fault occurrence time data, and the cable fault type determination data is processed by fault feature extraction to obtain fault feature description data. The fault occurrence time data and the fault characteristic description data are processed through influence range analysis to obtain fault impact assessment data. The fault occurrence time data, the cable fault location data, the fault characteristic description data, and the fault impact assessment data are then processed through data integration to obtain cable fault diagnosis data.

8. A smart cable fault location device for a photovoltaic system, used to implement the smart cable fault location method for a photovoltaic system as described in any one of claims 1-7, characterized in that, The intelligent cable fault location device for the photovoltaic system includes: The acquisition module is used to perform voltage and current synchronous sampling processing on each component in the photovoltaic string through a multi-channel synchronous potential acquisition module to obtain component-level synchronous sampling data. The calculation module is used to process the component-level synchronous sampling data through feature parameter calculation to obtain a feature data matrix containing inter-component voltage difference, string current change rate, component power value and adjacent component voltage ratio; The analysis module is used to process the feature data matrix through multi-dimensional threshold analysis and sliding variance analysis to obtain component abnormal state labeling data; The classification module is used to classify the abnormal state marker data of the components through a decision tree to obtain cable fault type determination data; The processing module is used to process the area where the cable fault type determination data is located through voltage gradient analysis to obtain the cable fault location data. The integration module is used to integrate and process the cable fault location data and cable fault type determination data to obtain cable fault diagnosis data.

Images