Anomaly detection method and system for photovoltaic string, product and medium

Abstract

The invention belongs to the technical field of photovoltaic power generation, and discloses an anomaly detection method and system for a photovoltaic string, a product and a medium. The method comprises the following steps: under a low-illumination stable condition, through an inverter direct control mode, successively applying two current pulses with increasing amplitudes to a photovoltaic group string; a dynamic resistance value corresponding to each pulse is measured and calculated, and a dynamic resistance change factor can be determined based on the difference between the two dynamic resistance values; and when the dynamic resistance change factor exceeds a preset threshold value, judging the corresponding string as an abnormal string with a nonlinear high-resistance fault. According to the invention, the accuracy of performing anomaly detection on the photovoltaic string is effectively improved.

Description

Technical Field

[0001] This invention belongs to the field of photovoltaic power generation technology, and relates to a method, system, product and medium for detecting anomalies in photovoltaic strings. Background Technology

[0002] As the global energy structure accelerates its transformation towards cleaner and lower-carbon energy, photovoltaic (PV) power generation technology has emerged as a core pillar of the new energy system, shouldering the important mission of promoting sustainable energy development. Against this backdrop, the construction of PV power plants is no longer limited to flat areas but is gradually extending to complex terrains such as mountains and hills to fully utilize abundant solar energy resources. However, environmental factors such as foundation subsidence or soil erosion caused by complex terrain have become new challenges to the stable operation of PV power plants. These factors can easily lead to slight displacements of the PV support structure, posing a potential threat to the long-term safety of the system.

[0003] In the operation and maintenance of photovoltaic power plants, to effectively monitor electrical path degradation that may be caused by environmental factors, the industry commonly employs high-precision online impedance spectroscopy analysis or refined IV curve scanning technology. These technologies calculate the total series resistance (Rs) of the photovoltaic strings by accurately measuring their electrical characteristics and then meticulously comparing it with a health fingerprint baseline from the initial commissioning phase. This method can identify abnormal resistance increases caused by poor connector contact or cable damage, providing crucial information for assessing the structural health of the strings.

[0004] However, while existing technologies have achieved some success in monitoring electrical path degradation, they fall short in addressing early, critical-state faults caused by mechanical tension. Specifically, when a connector is subjected to a small tensile force and reaches a critical stretching point, the contact pressure of its internal metal contacts may only decrease slightly, failing to form a stable, high-resistance state that can be identified by static resistance measurement techniques. In this metastable state, the connector can still maintain a seemingly normal conductive path, making the measured series resistance value likely still within the normal error range.

[0005] This significantly reduces the accuracy of existing technologies in anomaly detection, making it impossible to accurately detect the risk of momentary open circuits or arcing that may occur at connection points when encountering sudden current changes or vibrations. The existence of this risk could not only trigger sudden failures in photovoltaic systems, affecting the normal operation of the power station, but also pose a serious threat to the safety of the power station.

[0006] Therefore, improving the operational stability of photovoltaic systems in complex terrain, especially effectively monitoring and providing early warnings of early, critical-state faults caused by mechanical stress, has become an urgent problem to be solved in the current development of photovoltaic power generation technology. In the future, more sensitive and accurate monitoring technologies need to be developed to comprehensively improve the safety and reliability of photovoltaic systems. Summary of the Invention

[0007] To address the problems existing in the prior art, this invention provides a method, system, product, and medium for anomaly detection of photovoltaic strings, thereby solving the technical problem of insufficient accuracy in anomaly detection when monitoring early, critical state faults caused by mechanical tension, due to the inability to identify potential risks caused by minute changes in connectors under metastable conditions.

[0008] This invention is achieved through the following technical solution: A method for detecting anomalies in photovoltaic strings includes the following steps: S1: When the real-time ambient irradiance is in the preset low light intensity range and continues for more than the preset duration, the inverter working mode will be switched from maximum power point tracking mode to direct command control mode. S2: Control the output current of each photovoltaic string to a preset first current value, and record the corresponding first voltage value; synchronously apply a first current pulse command to raise the output current to a second current value, and record the corresponding second voltage value; and obtain the first dynamic resistance value of each photovoltaic string from the first current value to the second current value. S3: Simultaneously apply a second current pulse command to raise the current to the third current value, record the third voltage value, and obtain the second dynamic resistance value from the first current value to the third current value; S4: Based on the first dynamic resistance value and the second dynamic resistance value, obtain the dynamic resistance change factor of each photovoltaic string; S5: When any dynamic resistance change factor is greater than the preset resistance change factor threshold, the corresponding photovoltaic string is an abnormal string.

[0009] Preferably, in step S1, after the real-time ambient irradiance is in a preset low light intensity range and continues for more than a preset duration, the method further includes: Acquire real-time wind speed data and real-time dust concentration data of the photovoltaic power station; calculate the support vibration frequency and dust impact frequency based on the real-time wind speed data and the real-time dust concentration data; When the vibration frequency of the support or the sand and dust impact frequency is greater than the corresponding preset frequency threshold, a waiting time is entered. During the waiting time, the real-time wind speed data and the real-time sand and dust concentration data are periodically acquired again at preset time intervals, and the vibration frequency of the support and the sand and dust impact frequency are calculated again. The waiting time ends when both the real-time calculated vibration frequency of the support and the sand and dust impact frequency are less than the corresponding preset frequency threshold.

[0010] Preferably, during the waiting time, the following further applies: For each photovoltaic string, the open-circuit voltage at both ends is acquired at the preset time interval to obtain a voltage noise sequence; the vibration frequency of the support and the sand and dust impact frequency are used as the characteristic frequencies of the excitation source; the voltage response amplitude associated with the voltage noise sequence and the characteristic frequency of the excitation source is extracted; the ratio of the voltage response amplitude to the excitation source intensity is calculated to obtain the vibration sensitivity factor characterizing the corresponding photovoltaic string; when the vibration sensitivity factor exceeds the preset coupling threshold, the corresponding photovoltaic string is marked as a suspected abnormal string, and a suspected abnormality warning indication is generated.

[0011] Preferably, the dynamic resistance variation factor of each photovoltaic string is determined based on the first dynamic resistance value and the second dynamic resistance value, specifically including: Calculate the actual absolute difference between the second dynamic resistance value and the first dynamic resistance value; subtract the expected change in dynamic resistance from the actual absolute difference to obtain the residual value; calculate the ratio of the absolute value of the residual value to the first dynamic resistance value to obtain the dynamic resistance change factor.

[0012] Preferably, after determining the corresponding photovoltaic string as an abnormal string, the method further includes: For the abnormal string, the second current pulse command is applied again, and the corresponding fourth voltage value is recorded; based on the first voltage value and the fourth voltage value, the third dynamic resistance value is calculated; according to the third dynamic resistance value and the second dynamic resistance value, the resistance value attenuation rate is determined; when the resistance value attenuation rate exceeds the preset healing threshold, the fault type indicated by the abnormality is marked as intermittent self-healing connection abnormality.

[0013] Preferably, a first current pulse command and a second current pulse command are applied based on the correction of the first current value, the correction of the second current value, and the correction of the third current value; The process of obtaining the corrected first current value, the corrected second current value, and the corrected third current value is as follows: Obtain the real-time temperature data of any photovoltaic string; divide the photovoltaic string into multiple temperature groups according to the real-time temperature data, and ensure that the temperature difference between photovoltaic strings in each temperature group is less than a preset temperature difference threshold. The temperature correction coefficient for the corresponding temperature group is obtained by using the temperature difference value; Then, the corrected first current value, corrected second current value, and corrected third current value corresponding to the first current value, the second current value, and the third current value are obtained through the temperature correction coefficient.

[0014] An anomaly detection system for photovoltaic strings includes: The data acquisition module is used to switch the inverter's operating mode from maximum power point tracking mode to direct command control mode when the real-time ambient irradiance is in a preset low light intensity range and continues for a preset duration. The first data processing module is used to control the output current of all photovoltaic strings to a preset first current value and record the corresponding first voltage value; synchronously apply a first current pulse command to raise the output current to a second current value and record the corresponding second voltage value; and obtain the first dynamic resistance value of each string from the first current value to the second current value. The second data processing module is used to synchronously apply a second current pulse command to raise the current to a third current value, record the third voltage value, and obtain the second dynamic resistance value from the first current value to the third current value. The third data processing module is used to obtain the dynamic resistance change factor of each photovoltaic string based on the first dynamic resistance value and the second dynamic resistance value. The fourth data processing module is used to identify photovoltaic strings as abnormal strings when a dynamic resistance change factor is greater than a preset resistance change factor threshold.

[0015] A computer system includes a memory, a processor, and a computer program stored in the memory, wherein the processor executes the computer program to implement the steps of the method described above.

[0016] A computer-readable storage medium storing a computer program, which, when executed by a processor, implements the steps of the above-described method.

[0017] A computer program product includes a computer program / instructions that, when executed by a processor, implement the steps of the above-described method.

[0018] Compared with the prior art, the present invention has the following beneficial technical effects: This invention discloses a method for detecting anomalies in photovoltaic (PV) strings. Under stable low-light conditions, the inverter's operating mode is switched from maximum power point tracking (MPPT) to direct command control (DCC). Then, the output current of each PV string is controlled to a preset first current value, and the corresponding first voltage value is recorded. Simultaneously, a first current pulse command is applied to raise the output current to a second current value, and the second voltage value is recorded, obtaining a first dynamic resistance value from the first current value to the second current value. Then, a second current pulse command is applied simultaneously to raise the current to a third current value, and the third voltage value is recorded, obtaining a second dynamic resistance value from the first current value to the third current value. Based on the first and second dynamic resistance values, a dynamic resistance variation factor for each PV string is obtained. Due to early defects such as loose connections or oxidation at connection points, the equivalent resistance exhibits significant nonlinear characteristics due to current stress changes, while healthy connection points exhibit stable linear behavior. This invention calculates and compares the dynamic resistance values ​​under two different current steps to form a dynamic resistance variation factor, which can capture metastable fault characteristics that are difficult to detect with static measurements. This effectively identifies early, critical state faults caused by mechanical tension, improving the accuracy of PV string anomaly detection. Attached Figure Description

[0019] To more clearly illustrate the technical solutions of the embodiments of the present invention, the accompanying drawings used in the embodiments will be briefly introduced below. It should be understood that the following drawings only show some embodiments of the present invention and should not be regarded as a limitation on the scope. For those skilled in the art, other related drawings can be obtained based on these drawings without creative effort.

[0020] Figure 1 This is a flowchart illustrating an anomaly detection method for photovoltaic strings according to the present invention; Figure 2 This is a flowchart illustrating an anomaly detection method for photovoltaic strings in one embodiment of the present invention; Figure 3 This is another flowchart illustrating the anomaly detection method for photovoltaic strings in this embodiment of the invention; Figure 4 This is a schematic diagram of an exemplary hardware structure of the photovoltaic string anomaly detection system in an embodiment of the present invention; Figure 5 This is a schematic diagram of the structure of an anomaly detection system for photovoltaic strings according to the present invention. Detailed Implementation

[0021] To make the objectives, technical solutions, and advantages of the embodiments of the present invention clearer, the technical solutions of the embodiments of the present invention will be clearly and completely described below with reference to the accompanying drawings. Obviously, the described embodiments are only some embodiments of the present invention, and not all embodiments. The components of the embodiments of the present invention described and shown in the accompanying drawings can generally be arranged and designed in various different configurations.

[0022] Therefore, the following detailed description of the embodiments of the invention provided in the accompanying drawings is not intended to limit the scope of the claimed invention, but merely to illustrate selected embodiments of the invention. All other embodiments obtained by those skilled in the art based on the embodiments of the invention without inventive effort are within the scope of protection of the invention.

[0023] It should be noted that similar labels and letters in the following figures indicate similar items. Therefore, once an item is defined in one figure, it does not need to be further defined and explained in subsequent figures.

[0024] In the description of the embodiments of the present invention, it should be noted that if terms such as "upper," "lower," "horizontal," or "inner" indicate the orientation or positional relationship based on the orientation or positional relationship shown in the accompanying drawings, or the orientation or positional relationship commonly used when the product of the invention is in use, they are only for the convenience of describing the present invention and simplifying the description, and do not indicate or imply that the device or element referred to must have a specific orientation, or be constructed and operated in a specific orientation, and therefore should not be construed as a limitation of the present invention. Furthermore, terms such as "first" and "second" are only used to distinguish descriptions and should not be construed as indicating or implying relative importance.

[0025] Furthermore, the use of the term "horizontal" does not imply that the component must be absolutely horizontal, but rather that it can be slightly tilted. For example, "horizontal" simply means that its direction is more horizontal than "vertical," and does not mean that the structure must be completely horizontal, but can be slightly tilted.

[0026] The present invention will now be described in further detail with reference to the accompanying drawings: Example 1 To address the limitations of existing technologies in identifying metastable critical faults, this invention proposes a dynamic disturbance testing method. Under stable low-light conditions, this method applies two current pulses of different intensities to the photovoltaic string via an inverter, recording the voltage change corresponding to each current step. Healthy electrical connections exhibit stable linear (ohmic) characteristics, with their dynamic resistance remaining essentially constant under different current stresses. However, connections with early defects such as loose connections or oxidation show significant nonlinear responses due to their equivalent resistance being sensitive to current changes. Therefore, by calculating and comparing the first and second dynamic resistance values ​​generated under these two different current pulses, and constructing a dynamic resistance change factor to quantify this nonlinearity, this invention can amplify and identify metastable fault characteristics that are imperceptible by static measurements in existing technologies, thereby improving the accuracy of anomaly detection for photovoltaic strings.

[0027] Specifically, such as Figure 1 As shown, the present invention provides an anomaly detection method for photovoltaic strings, comprising the following steps: S1: When the real-time ambient irradiance is in the preset low light intensity range and continues for more than the preset duration, the inverter working mode will be switched from maximum power point tracking mode to direct command control mode. S2: Control the output current of all photovoltaic strings to a preset first current value, and record the corresponding first voltage value; synchronously apply a first current pulse command to raise the output current to a second current value, and record the corresponding second voltage value; and obtain the first dynamic resistance value of each photovoltaic string from the first current value to the second current value. S3: Simultaneously apply a second current pulse command to raise the current to the third current value, record the third voltage value, and obtain the second dynamic resistance value from the first current value to the third current value; S4: Based on the first dynamic resistance value and the second dynamic resistance value, obtain the dynamic resistance change factor of each photovoltaic string; S5: When the dynamic resistance change factor is greater than the preset resistance change factor threshold, the corresponding photovoltaic string is an abnormal string.

[0028] In step S1, after detecting that the real-time ambient irradiance is within a preset low light intensity range, the method further includes: The system acquires real-time wind speed and dust concentration data of the photovoltaic power station; calculates the support vibration frequency and dust impact frequency based on the real-time wind speed and dust concentration data; the support vibration frequency is obtained by substituting the real-time wind speed data into a pre-calibrated wind speed-vibration frequency characteristic curve; the dust impact frequency is equal to a preset proportionality coefficient multiplied by the square of the real-time dust concentration data and the real-time wind speed data; the proportionality coefficient is proportional to the cosine of the tilt angle of the photovoltaic module; when the support vibration frequency or the dust impact frequency is greater than the corresponding preset frequency threshold, a waiting period is entered; during the waiting period, the real-time wind speed and dust concentration data are acquired again periodically at preset time intervals, and the support vibration frequency and the dust impact frequency are calculated again; when the real-time calculated support vibration frequency and dust impact frequency are both less than the corresponding preset frequency threshold, the waiting period ends.

[0029] In precise electrical diagnostics, mechanical disturbances caused by environmental factors such as wind speed and dust are a key source of interference. Therefore, instead of direct measurement, the system first calculates the vibration frequency of the support structure and the impact frequency of dust to quantitatively assess the current level of mechanical interference in real time. Only when both key frequency indicators fall below a preset low threshold, confirming that the external physical disturbance is sufficiently weak, is the subsequent diagnostic process triggered. This proactive approach of selecting quiet measurement windows ensures that electrical measurement data is acquired in a quasi-static environment, avoiding the misinterpretation of transient data jumps caused by environmental disturbances as permanent faults in the photovoltaic string itself, ultimately improving the accuracy of anomaly detection for photovoltaic strings.

[0030] More preferably, during the waiting time, the following further applies: For each photovoltaic string, the open-circuit voltage at both ends is acquired at preset time intervals to obtain a voltage noise sequence; the vibration frequency of the support and the sand and dust impact frequency are used as the characteristic frequencies of the excitation source; the voltage response amplitude associated with the voltage noise sequence and the characteristic frequencies of the excitation source is extracted; the ratio of the voltage response amplitude to the excitation source intensity is calculated to obtain the vibration sensitivity factor characterizing the corresponding photovoltaic string; the excitation source intensity is determined based on the real-time wind speed data and the real-time sand and dust concentration data; when the vibration sensitivity factor exceeds a preset coupling threshold, the corresponding photovoltaic string is marked as a suspected abnormal string, and a suspected abnormality warning indication is generated.

[0031] In the above embodiments, the waiting time caused by environmental disturbances is transformed into an active diagnostic window. Wind speed and dust are no longer considered mere interference, but rather as known mechanical excitation sources. By correlating the voltage noise sequence with the characteristic frequencies of the excitation sources and calculating the core parameter vibration sensitivity factor, this scheme can quantify the sensitivity of each string connection point to external physical disturbances. This allows for the early identification of strings with potential loosening or poor connection risks due to weak physical connections before performing main electrical diagnostics. This diagnostic method adds a completely new physical dimension to traditional electrical testing, thereby improving the accuracy of anomaly detection for photovoltaic strings.

[0032] In some embodiments, the dynamic resistance variation factor of each photovoltaic string is determined based on the first dynamic resistance value and the second dynamic resistance value, specifically including: Calculate the actual absolute difference between the second dynamic resistance value and the first dynamic resistance value; subtract the expected change in dynamic resistance from the actual absolute difference to obtain the residual value; calculate the ratio of the absolute value of the residual value to the first dynamic resistance value to obtain the dynamic resistance change factor.

[0033] Here, by introducing the parameter of expected dynamic resistance change, and based on a pre-calibrated baseline thermal drift coefficient, this normal resistance drift caused purely by thermal effects is quantified. By subtracting this expected change from the actual measured resistance difference to obtain the residual value, this step separates the nonlinear resistance change caused by the fault from the normal physical thermal drift. The final dynamic resistance change factor is based on this purified residual value, thus more accurately reflecting the health status of the connection point, eliminating artifact interference from thermal effects, and improving the accuracy of anomaly detection in photovoltaic strings.

[0034] Furthermore, specifically, in step S4, determining the dynamic resistance variation factor of each photovoltaic string based on the first dynamic resistance value and the second dynamic resistance value specifically includes: Calculate the actual absolute difference between the second dynamic resistance value and the first dynamic resistance value; subtract the expected dynamic resistance change from the actual absolute difference to obtain the residual value; the expected dynamic resistance change is the product of the reference thermal drift coefficient and the additional energy impact; the reference thermal drift coefficient is the expected rate of change of the dynamic resistance value of a healthy photovoltaic string due to the Joule heating effect of the pulsed current injection, obtained based on a pre-calibrated test of the photovoltaic string; the additional energy impact is the additional energy introduced by the pre-set second current pulse command relative to the first current pulse command; calculate the ratio of the absolute value of the residual value to the first dynamic resistance value to obtain the dynamic resistance change factor.

[0035] In some embodiments, after determining the corresponding photovoltaic string as an abnormal string, the method further includes: For the abnormal string, the second current pulse command is applied again, and the corresponding fourth voltage value is recorded; based on the first voltage value and the fourth voltage value, the third dynamic resistance value is calculated; according to the third dynamic resistance value and the second dynamic resistance value, the resistance value attenuation rate is determined; when the resistance value attenuation rate exceeds the preset healing threshold, the fault type indicated by the abnormality is marked as intermittent self-healing connection abnormality.

[0036] In the above embodiments, after initially determining that the string is abnormal, in order to further identify the specific physical nature of the fault, a current pulse of the same intensity is immediately applied again to the abnormal string. The core of this step lies in utilizing the thermal healing characteristics of certain early connection faults (such as slight oxidation or loose connection), that is, the instantaneous high-current Joule heating effect may temporarily break down the thin oxide layer or improve physical contact, causing the contact resistance to decrease in a short period of time. By calculating and comparing the dynamic resistance values ​​under two consecutive high-current pulses, and introducing the resistance decay rate to quantify the degree of this resistance decrease, it is possible to clearly distinguish between this intermittent, critical fault with self-healing phenomenon and stable high-resistance faults, giving the fault a more accurate physical label, avoiding the generalization of faults of different natures into one category, thereby improving the accuracy of anomaly detection of photovoltaic strings.

[0037] In some embodiments, before applying a preset first current pulse command synchronously to all photovoltaic strings, the method further includes: The process involves: acquiring real-time temperature data for each photovoltaic (PV) string; dividing the PV strings into multiple temperature groups based on the real-time temperature data, where the temperature difference between PV strings within each temperature group is less than a preset temperature difference threshold; subtracting the average temperature of the temperature group from the preset standard test temperature to obtain the temperature difference value; the average temperature of the temperature group being the arithmetic mean of each temperature group; multiplying the temperature difference value by a preset temperature coefficient to obtain a temperature compensation value; the temperature coefficient being the absolute value of the open-circuit voltage temperature coefficient of the PV module; adding the temperature compensation value to a reference correction coefficient to obtain the temperature correction coefficient for the corresponding temperature group; the reference correction coefficient being the correction coefficient at the standard test temperature; for each PV string within a temperature group, multiplying the temperature correction coefficient corresponding to the temperature group by the first current value and the second current value respectively to obtain a corrected first current value and a corrected second current value; and using the corrected first current value and the corrected second current value when applying the current pulse command to the PV strings within each temperature group.

[0038] Here, considering that the voltage response of photovoltaic strings is significantly affected by temperature, directly comparing strings at different temperatures would introduce substantial errors. Therefore, the strings are first grouped according to real-time temperature, and a unique temperature correction coefficient is calculated for each temperature group based on the deviation of each group from the standard temperature and the inherent voltage temperature coefficient of the modules. Before applying the current pulse, this coefficient is used to pre-correct the current command value, thus proactively eliminating voltage response differences caused by temperature variations at the measurement source. This ensures that the subsequently calculated dynamic resistance value accurately reflects the electrical health of the string itself, rather than a temperature artifact, improving the comparability of diagnostic results between different strings and ultimately enhancing the accuracy of anomaly detection in photovoltaic strings.

[0039] In addition, after dividing the photovoltaic string into multiple temperature groups according to the real-time temperature data, the method further includes: Obtain the number of temperature groups; when the number of temperature groups is greater than a preset threshold, calculate the average temperature difference between adjacent temperature groups; sort the average temperature differences in ascending order; select the two adjacent temperature groups with the smallest average temperature difference and merge them to obtain a merged temperature group; the average temperature of the merged temperature group is the weighted average temperature of the two temperature groups before merging; the weighted average temperature is calculated based on the number of photovoltaic strings in each temperature group as the weight; repeat the temperature group merging operation until the number of temperature groups is not greater than the preset threshold; recalculate the temperature correction coefficient for each merged temperature group.

[0040] In the above embodiments, to address the issues of excessive grouping, computational redundancy, and unstable correction coefficients caused by initial temperature grouping, the system prioritizes merging adjacent groups with the smallest temperature differences and uses a weighted average based on the number of strings to update the average temperature of the merged groups. This reduces the number of groups and improves computational efficiency while maximizing temperature homogeneity within the new merged groups by selecting the optimal group for merging each time. This results in more stable and representative temperature correction coefficients calculated for each group, thereby reducing the interference of temperature differences on diagnostic results and improving the accuracy of anomaly detection for photovoltaic strings.

[0041] Example 2 To further explain the technical solution of the present invention, the following embodiments are provided: Figure 2 This is a flowchart illustrating the anomaly detection method for photovoltaic strings used in the embodiments of this application, including the following steps: S101. When it is detected that the real-time ambient irradiance is in the preset low light intensity range and the time in the preset low light intensity range exceeds the preset stable low light duration, the working mode of the inverter connected to all photovoltaic strings will be switched from the maximum power point tracking mode to the direct command control mode.

[0042] Among them, real-time ambient irradiance refers to the solar power density currently projected onto the photovoltaic module plane, measured in real time by an irradiance sensor (such as a photodiode or a small reference cell), typically measured in watts per square meter; preset low irradiance range represents a range of low irradiance values ​​set for this detection method, designed to ensure that the background current generated by the photovoltaic effect is sufficiently small to avoid interfering with subsequent electrical characteristic measurements; preset stable low irradiance duration refers to the minimum duration for which the irradiance must remain within the above range, used to avoid false triggering caused by instantaneous changes such as passing clouds; photovoltaic string refers to a basic power generation unit composed of multiple photovoltaic modules (commonly known as solar panels) connected in series; inverter is the core equipment used to convert the DC power generated by the photovoltaic string into AC power and feed it into the grid; Maximum Power Point Tracking (MPPT) mode is the standard operating mode of the inverter during normal power generation, which continuously adjusts the input impedance to ensure that the photovoltaic string always outputs maximum power; direct command control mode is a special engineering or diagnostic mode in which the inverter can abandon the MPPT target and instead execute specific current or voltage commands from the central controller.

[0043] Specifically, the output characteristics of photovoltaic (PV) modules are affected by light intensity and temperature. Under normal or high light conditions, the current generated by the photovoltaic effect is large, which can mask the weak electrical characteristics caused by internal high-resistance faults (such as poor contact). Therefore, choosing to test under low light intensity can minimize the interference of the photovoltaic effect. Furthermore, this period of low light intensity must be continuous and stable to eliminate the influence of unstable factors such as sudden weather changes and cloud cover on the measurement benchmark, ensuring that all strings are compared under a common and stable initial condition. Once this environmental condition is met, the inverter is switched from the MPPT mode, which pursues maximum efficiency, to the direct command control mode, which can execute external commands. This switch is crucial for the subsequent active application of specific current pulses and measurement, transforming the inverter from an adaptive power optimizer into a high-precision programmable power supply.

[0044] S102. Control the output current of all photovoltaic strings to be the preset first current value, and record the first voltage value corresponding to the first current value of each photovoltaic string.

[0045] The first current value refers to a small DC current used to establish the initial operating point. The value must be much smaller than the rated short-circuit current of the photovoltaic module to ensure that all strings are in a uniform and comparable weak positive bias state at the start of the test. The first voltage value refers to the DC voltage across each photovoltaic string, measured by the voltage sampling circuit inside the inverter or by an external high-precision voltmeter when the first current value is applied.

[0046] Specifically, after the mode switch is completed, the output of all photovoltaic strings is theoretically controlled by the inverter. At this time, a weak first current value is forced into or output to each photovoltaic string through direct command control mode, so that all PN junctions (P stands for Positive, representing a semiconductor type dominated by positive charge carriers (holes); N stands for Negative, representing a semiconductor type dominated by negative charge carriers (electrons)) and potential contact defects within the string are in a weak conducting state. Since the photocurrent is extremely small and negligible at this time, the measured first voltage value mainly reflects the total voltage drop of the entire string (including all cells, internal wiring, connectors, bypass diodes, and potential fault points) under this specific bias current. The first voltage values ​​of all strings under the same first current value are recorded simultaneously.

[0047] In some embodiments, current control and data recording in this step can be implemented in a variety of ways: Optionally, a closed-loop control method based on inverters is used: the central controller issues a command containing a target first current value to all inverters. Each inverter's digital signal processor (DSP) initiates a current closed-loop control algorithm, such as a proportional-integral (PI) controller, to monitor the output current of its corresponding string in real time. The DSP continuously adjusts the switching duty cycle of its internal power devices, such as insulated-gate bipolar transistors (IGBTs), until the actual output current of the string stabilizes within the error range (e.g., ±1%) of the target "first current value." After stabilization, the DSP records the voltage across the string at this point as the first voltage value and uploads it to the central controller.

[0048] It is understandable that other methods can be used to control and record this step, such as combining more complex adaptive control algorithms, etc., which are not limited here.

[0049] S103. Apply a preset first current pulse command to all photovoltaic strings synchronously, so that the output current of each photovoltaic string increases from the first current value to the second current value corresponding to the first current pulse command, and record the second voltage value corresponding to the second current value of each photovoltaic string.

[0050] The first current pulse command is a transient command generated and issued by the central controller, which requires the inverter to change the output current in a very short time. It defines the target value of the current and the rate of change. The second current value is the target current of the first current pulse command. The second current value is greater than the first current value but is still at a low level, representing the first detection step. The second voltage value is the corresponding voltage value recorded when the string current stably reaches the second current value.

[0051] Specifically, after establishing the static operating point, a dynamic disturbance is introduced to observe the string's response. By synchronously sending a first current pulse command to all inverters, the output current of all strings is forced to jump from a first current value to a second current value within a short period (e.g., milliseconds). This rapid current step change can detect the transient voltage response of the strings under different current densities. For a healthy string with linear resistive characteristics, its voltage change should be proportional to the current change. However, for strings with nonlinear faults such as poor contact, the voltage drop at the fault point may change nonlinearly with increasing current. Applying the pulse synchronously ensures that all strings are disturbed at the same time, eliminating measurement errors caused by small changes in environmental parameters (such as temperature) due to time differences. The second voltage value is recorded when the current stabilizes at the second current value.

[0052] In some embodiments, before applying a preset current pulse command to all photovoltaic strings synchronously, current command correction based on dynamic temperature grouping and iterative merging can be used to compensate for measurement errors caused by uneven operating temperatures of each string in a large-scale photovoltaic power plant, thereby improving the accuracy of anomaly diagnosis.

[0053] Specifically, this temperature compensation process aims to reduce the interference of temperature variables on diagnostic results at the source by feedforward correction of the input excitation, so that all photovoltaic strings can be tested under equivalent and standardized electrical conditions.

[0054] First, real-time temperature data for each photovoltaic (PV) string within the PV power plant is acquired. This data typically comes from high-precision temperature sensors installed on the backsheet of the PV modules. Due to differences in microenvironmental factors such as shading, air convection, and installation location, the actual operating temperatures of different strings in a large PV power plant may vary. PV modules exhibit a distinct negative temperature coefficient in their voltage output characteristics; that is, the higher the temperature, the lower the terminal voltage at the same current. Ignoring this difference and directly using a uniform current command for testing may cause healthy strings with higher temperatures to exhibit different voltage responses than healthy strings with lower temperatures. This temperature-induced difference may be misinterpreted as nonlinearity caused by a fault, thus interfering with the diagnosis of the actual fault.

[0055] Secondly, all photovoltaic (PV) strings are initially grouped according to their real-time temperature. The grouping principle is to ensure that the temperature difference between any two strings within each group is less than a preset temperature difference threshold. This preset threshold is a key parameter pre-set based on the voltage-temperature characteristics of the PV modules and the measurement accuracy required by the diagnostic system. It provides a quantitative basis for temperature grouping, ensuring that all PV strings within the same group can share a unified temperature correction coefficient, while controlling the diagnostic error introduced by residual temperature differences within the group within an acceptable range. However, in extreme cases of extremely uneven temperature distribution, this initial grouping may generate a large number of sparsely populated temperature groups. This not only increases the subsequent computational burden but may also lack statistical representativeness due to insufficient sample size. Therefore, it is checked whether the total number of current temperature groups exceeds a preset group number threshold. If it does, an iterative merging algorithm is initiated to calculate the average temperature difference between all adjacent temperature groups after sorting by temperature, and then the two most similar temperature groups with the smallest difference are merged. It is worth noting that the average temperature of the newly generated merged temperature group is not a simple arithmetic average but is calculated based on the weighted average of the number of PV strings contained in the two original groups. This weighted processing ensures that groups with more members have greater influence in determining the new average temperature, making the newly generated average temperature more accurately represent the thermodynamic state of the merged large group. This merging operation is performed repeatedly, merging the closest pair each time, until the total number of temperature groups does not exceed a preset group number threshold.

[0056] After obtaining the final, reasonably sized temperature groups, a temperature correction factor is calculated independently for each temperature group. The calculation logic for this temperature correction factor follows the physical characteristics of semiconductor devices: First, using the internationally recognized standard test conditions (STC) for photovoltaic modules (typically 25°C) as a benchmark, the average operating temperature of the current temperature group is subtracted to obtain a temperature difference (ΔT). Next, this temperature difference is multiplied by the absolute value of the open-circuit voltage temperature coefficient provided by the photovoltaic module manufacturer or experimentally calibrated (e.g., the absolute value of -0.3% / °C, i.e., 0.003 / °C) to obtain an initial voltage compensation percentage. Finally, this compensation value is algebraically added to a benchmark correction factor (which is 1 at the standard test temperature) to obtain the final temperature correction factor for that temperature group. The physical meaning of this temperature correction factor is the amount of pre-adjustment required to the input excitation (i.e., the test current) to compensate for the voltage change caused by ΔT. When applying current pulses subsequently, a uniform nominal current value is no longer used. Instead, the first and second nominal current values ​​are multiplied by a specific correction factor for the temperature group of the corresponding string, thus obtaining the corrected current command for that string. The inverter will then use this corrected current value to perform subsequent pulse injection and measurement tasks.

[0057] By applying a temperature-corrected current command, the physical influence of temperature on the string voltage response is actively counteracted, ensuring that the testing process of all strings is equivalent to being carried out under uniform standard test conditions. This removes the interference variable of temperature from the measurement results, allowing the subsequently calculated dynamic resistance change factor to more purely reflect the nonlinear characteristics caused by the fault itself, thus improving the accuracy of diagnosis.

[0058] S104. Calculate the first dynamic resistance value of each photovoltaic string from the first current value to the second current value.

[0059] The first dynamic resistance value is a quantitative indicator that characterizes the equivalent resistance of the photovoltaic string within the first detection interval (i.e., when the current changes from the first current value to the second current value). It is a differential resistance or incremental resistance (ΔV / ΔI), where ΔV is the difference between the second voltage value and the first voltage value, and ΔI is the difference between the second current value and the first current value.

[0060] Specifically, dynamic resistance is defined as the ratio of a small change in voltage to a small change in current that causes that change. In this application, the voltage difference between two detection points is divided by the current difference, specifically calculated as: First dynamic resistance value = (Second voltage value - First voltage value) / (Second current value - First current value). This calculation result reflects the average electrical impedance characteristics of the string within the current range of [first current value, second current value]. For an ideal, healthy string, its internal resistance should remain essentially constant under low current conditions. Therefore, the first dynamic resistance value mainly reflects the sum of the inherent series resistances of the cells, solder ribbons, connectors, etc.

[0061] S105. A preset second current pulse command is synchronously applied to all photovoltaic strings, so that the output current of each photovoltaic string is increased to the third current value corresponding to the second current pulse command, and the third voltage value corresponding to the third current value of each photovoltaic string is recorded.

[0062] The second current pulse command is the second transient command, and its application method is similar to that of the first current pulse command. The third current value is the target current of the second pulse, and its value is designed to be significantly greater than the second current value, in order to push the string to a detection area with a higher current density. The third voltage value is the corresponding voltage reading recorded when the current stabilizes at the third current value.

[0063] Specifically, this step involves a second, more intense dynamic detection to reveal and amplify potential nonlinear effects. After completing the first dynamic resistance calculation, a second, larger current pulse is applied, causing the string current to jump from the first (or second, depending on the implementation) current value to a third. The second, larger pulse is necessary because some early or minor contact defects may not exhibit significant nonlinear resistance characteristics (e.g., resistance increases with current, or vice versa) under small current disturbances. Applying a larger current (the third current value) can more significantly elicit these nonlinear effects. For example, a poor contact point might heat up rapidly with increased current, causing a significant change in resistivity. This change might be masked by noise under a small current pulse, but under a large current pulse, it will be clearly reflected in a disproportionate increase in voltage. Record the third set of key data points (the third current value and the third voltage value).

[0064] In some embodiments, the pulse application in this step can be implemented in a variety of ways: Optionally, the current jumps directly from the first current value: After recording the second voltage value, the command current first drops back to the first current value and stabilizes; then, the central controller issues a second current pulse command with a target value of the third current value; the inverter control current jumps directly from the first current value to the third current value, and after stabilizing, the third voltage value is recorded. Optionally, the current jumps continuously from the second current value: After recording the second voltage value, the current remains at the second current value; a second current pulse command is issued with a target value of the third current value; the inverter control current continues to increase from the second current value to the third current value, and after stabilizing, the third voltage value is recorded.

[0065] It is understandable that other methods can be used to apply the pulse in this step, such as inserting a specified waiting time between two pulses to observe the thermal relaxation effect, etc., which are not limited here.

[0066] S106. Calculate the second dynamic resistance value of each photovoltaic string from the first current value to the third current value.

[0067] The second dynamic resistance value is a quantitative indicator that characterizes the equivalent resistance of the photovoltaic string in the second, larger detection range (i.e., when the current changes from the first current value to the third current value). It reflects the overall impedance performance of the string under higher current stress.

[0068] Specifically, the calculation principle in this step is similar to that in S104, but different data points are used. The starting point of the first detection (first current value, first voltage value) and the ending point of the second detection (third current value, third voltage value) are used to calculate a macroscopic dynamic resistance covering the entire detection range. The calculation formula is: Second dynamic resistance value = (Third voltage value - First voltage value) / (Third current value - First current value). By calculating the second dynamic resistance value, a value reflecting the average impedance characteristics of the string throughout the entire current change process from low to high is obtained. Comparing this value with the first dynamic resistance value calculated only in the low current range effectively determines whether the string's resistance characteristics change with increasing current, i.e., whether nonlinearity exists.

[0069] S107. Based on the first dynamic resistance value and the second dynamic resistance value, determine the dynamic resistance variation factor of each photovoltaic string.

[0070] The dynamic resistance variation factor is a dimensionless, normalized key performance indicator used to quantify and amplify the difference between two dynamic resistance values, thereby characterizing the degree of resistance nonlinearity of the photovoltaic string. For an ideal linear string, the dynamic resistance variation factor should be close to zero; conversely, the more severe the nonlinear fault, the larger the absolute value of its dynamic resistance variation factor.

[0071] Specifically, after obtaining the first and second dynamic resistance values, representing the low-current and full-current (or high-current) ranges respectively, a unified metric is needed to assess whether the difference between them is significant. Directly comparing the difference (second dynamic resistance value - first dynamic resistance value) may be affected by the total resistance of the string itself (the difference for longer strings may naturally be greater than that for shorter strings). Therefore, using a normalized "variance factor" is more scientific. A typical calculation method is to calculate its relative rate of change, with the formula: Dynamic resistance variation factor = |second dynamic resistance value - first dynamic resistance value| / first dynamic resistance value. This factor indicates how much the dynamic resistance deviates from the dynamic resistance under a larger current excitation compared to the dynamic resistance under the initial small current excitation. Through this normalization process, the factor can effectively eliminate the influence of inherent attributes such as string length and component model, making comparisons between different strings, as well as comparisons between the same string and its own health benchmark, more fair and sensitive, thereby accurately capturing the nonlinear characteristics caused by faults.

[0072] In some embodiments, the factor determination in this step can be achieved in a variety of ways.

[0073] Optionally, a calculation method based on relative rate of change is as follows: Obtain the first and second dynamic resistance values ​​of the specified string; calculate the absolute difference between the two resistance values: ΔR = |second dynamic resistance value - first dynamic resistance value|; calculate the change factor: Factor = ΔR / first dynamic resistance value. If the first dynamic resistance value may be zero or extremely small, to avoid the denominator being zero, the method Factor = ΔR / ((first dynamic resistance value + second dynamic resistance value) / 2) can be used, using the average value as the denominator.

[0074] Optionally, a statistical Z-score-based calculation method is used: Maintain a historical database consisting of dynamic resistance change values ​​(second dynamic resistance value - first dynamic resistance value) measured from a large number of healthy strings, and calculate the mean μ (theoretically close to 0) and standard deviation σ of the change values ​​for this healthy group; for the current string to be tested, calculate its dynamic resistance change value ΔR_test = second dynamic resistance value - first dynamic resistance value; determine its dynamic resistance change factor as the Z-score value: Factor = (ΔR_test - μ) / σ; this factor directly represents the degree of nonlinearity of the current string deviating from the standard deviation of the healthy group, with clear statistical significance.

[0075] It is understandable that other methods can be used to determine the factors in this step, such as introducing a more complex machine learning model, using the first dynamic resistance value and the second dynamic resistance value as feature inputs, and having the model directly output a fault probability score as a change factor, etc., which are not limited here.

[0076] S108. When a dynamic resistance change factor exceeds a preset resistance change factor threshold, the corresponding photovoltaic string is identified as an abnormal string, and an abnormal indication is generated.

[0077] Among them, the preset resistance change factor threshold is a pre-set critical value used to distinguish between healthy and abnormal states. This threshold can be obtained through experimental testing and statistical analysis of a large number of samples; abnormal string refers to a photovoltaic string that is judged to have a high probability of nonlinear high resistance fault; abnormal indication is an alarm or work order generated by the system that contains diagnostic conclusions and necessary information, used to notify operation and maintenance personnel, including the physical location of the abnormal string.

[0078] Specifically, the dynamic resistance variation factor of each photovoltaic string calculated in the previous step is compared with a pre-set resistance variation factor threshold. If a string's factor is less than or equal to the threshold, its dynamic resistance variation is considered to be within the normal error or measurement noise range, and it is judged as a healthy string. Conversely, if its factor exceeds the threshold, it is considered that the string has significant nonlinear resistance characteristics, which is usually directly related to high-resistance faults such as poor connector contact, cold solder joints, and microcracks in the cells, and therefore it is judged as an abnormal string. Once an abnormality is determined, an abnormality indication is generated. To facilitate quick location and handling by maintenance personnel, this indication must include the physical location information of the abnormal string, such as its combiner box number, branch number, and row / column number in the array.

[0079] In some embodiments, after the photovoltaic string is identified as an abnormal string, a verification secondary current pulse injection test can be performed on the abnormal string to deeply characterize the physical characteristics of the abnormality, thereby identifying and marking intermittent connection faults with thermal healing effects.

[0080] Specifically, after identifying a photovoltaic string as abnormal in S108, a verification procedure is initiated to investigate the inherent physical properties of the abnormality. The inverter is controlled to apply a second current pulse command with identical parameters to that in S105 to the diagnosed abnormal string, raising the output current of the abnormal string back to the third current value, and the corresponding voltage, the fourth voltage value, is recorded. A repeated electrical stress impact of the same intensity is then applied to the fault point; the core purpose is to observe whether the electrical characteristics of the fault point have changed after the first impact. Subsequently, based on the starting voltage of the first detection (first voltage value) and the ending voltage of this verification detection (fourth voltage value), a new dynamic resistance value is calculated, called the third dynamic resistance value. The calculation formula is: Third dynamic resistance value = (fourth voltage value - first voltage value) / (third current value - first current value). The third dynamic resistance value and the second dynamic resistance value calculated in S106 were measured within the exact same current variation range (from the first current value to the third current value), therefore they are directly comparable. The second dynamic resistance value represents the resistance characteristic of the fault point when it is first subjected to electrical stress, while the third dynamic resistance value represents its resistance characteristic after being subjected to one electrical stress. By comparing these two values, a resistance attenuation rate can be determined, calculated as: Resistance attenuation rate = (Second dynamic resistance value - Third dynamic resistance value) / Second dynamic resistance value. This attenuation rate quantifies the degree of change in dynamic resistance between two consecutive high-current tests. For certain specific connection failures, such as those with a slight oxide film on the connector surface, the Joule heat and electric field force generated by the first high-current pulse are sufficient to break down this thin insulating layer, producing a microscopic sintering or arc cleaning effect, thereby significantly reducing the contact resistance instantaneously. This phenomenon is intermittent self-healing or thermal healing. When the calculated resistance attenuation rate exceeds a preset healing threshold calibrated through a large amount of experimental data, the fault is determined to have this characteristic, and the fault type in the initially generated anomaly indication will be marked as intermittent self-healing connection anomaly.

[0081] By performing secondary pulse verification on identified abnormal strings and calculating the resistance attenuation rate, the degree of healing of fault points after being subjected to electrical stress can be quantified. This allows for the identification of intermittent connection faults with unique physical characteristics from general high-resistance anomalies. This avoids missed detections or misjudgments by maintenance personnel during on-site troubleshooting when the fault temporarily disappears and cannot be reproduced. Ultimately, by providing deeper qualitative information about the fault, it improves the overall accuracy of photovoltaic string anomaly detection and the value of maintenance guidance.

[0082] In the above embodiments, static resistance measurement is insufficient to identify critical faults caused by reduced contact pressure due to mechanical tension, before a stable high-resistance state has been formed. This is addressed by applying two progressively increasing current pulses to the string under stable low-light conditions. The core principle is that a connection point with early defects such as poor connection or oxidation exhibits significantly nonlinear equivalent resistance due to changes in current stress, while a healthy connection point displays stable linear (ohmic) behavior. This solution calculates and compares the dynamic resistance values ​​under two different current steps, ultimately forming a dynamic resistance variation factor. This captures metastable fault characteristics that are difficult to detect with static measurements, thereby improving the ability to identify early-stage problems and increasing the accuracy of anomaly detection in photovoltaic strings.

[0083] In other embodiments of this application, when performing electrical characteristic diagnosis on photovoltaic strings, severe weather conditions such as strong winds and sandstorms may cause distortion of measurement signals due to external physical disturbances, misinterpreting normal mechanical vibration responses as electrical faults. The photovoltaic string anomaly detection method provided in this application ensures that diagnosis is performed only within a window of sufficiently low environmental interference by setting a waiting time, thereby guaranteeing the prerequisites for fault detection and improving diagnostic accuracy.

[0084] Example 3 Furthermore, such as Figure 3 The diagram shown is another flowchart illustrating the anomaly detection method for photovoltaic strings provided in this application embodiment, including the following steps: S201. When it is detected that the real-time ambient irradiance is in the preset low light intensity range and the time in the preset low light intensity range exceeds the preset stable low light duration, the working mode of the inverter connected to all photovoltaic strings will be switched from the maximum power point tracking mode to the direct command control mode.

[0085] Step S201 and Figure 1 Step S101 in the illustrated embodiment is similar and can be found in the description of step S101, which will not be repeated here.

[0086] S202. Obtain real-time wind speed data and real-time dust concentration data of the photovoltaic power station.

[0087] Among them, real-time wind speed data refers to the air velocity measured in real time by an anemometer installed at the photovoltaic power station site (usually at a representative height and location), which is used to characterize the wind conditions of the current environment; real-time dust concentration data refers to the mass or quantity of suspended dust particles in a unit volume of air measured in real time by equipment such as dust monitors, particulate counters or visibility meters, which is used to quantify the degree of air pollution.

[0088] Specifically, this step forms the data foundation for the environmental suitability assessment conducted before initiating electrical testing in the entire diagnostic process. In large-scale photovoltaic power plants, photovoltaic modules are mounted on massive metal supports, and strong winds can cause mechanical vibrations throughout the array. Simultaneously, in windy and dusty areas, high-speed airflow carrying dust particles continuously impacts the modules and connectors. Both of these physical disturbances—the macroscopic vibration of the supports and the microscopic impact of dust—can cause instantaneous, micrometer-level displacement or vibration at connection points with early contact defects (such as the springs inside connectors), resulting in drastic fluctuations in contact resistance. If precise dynamic resistance measurements are performed in this noisy mechanical environment, these resistance fluctuations caused by external physical disturbances can interfere with or even obscure the nonlinear electrical characteristics of the fault itself revealed by current changes, leading to distorted diagnostic results. Therefore, obtaining real-time wind speed and dust concentration data is a prerequisite for ensuring diagnostic accuracy.

[0089] S203. Calculate the vibration frequency of the support and the impact frequency of sand and dust based on real-time wind speed data and real-time dust concentration data.

[0090] Among them, the support vibration frequency refers to the dominant frequency of mechanical resonance or forced vibration of the photovoltaic support system (including columns, beams, and the modules themselves) caused by wind force, which is obtained by substituting real-time wind speed data into a pre-calibrated wind speed-vibration frequency characteristic curve; the dust impact frequency is a comprehensive physical quantity used to equivalently characterize the intensity of the impact disturbance of dust particles on the surface of photovoltaic modules, reflecting the flux of impact energy, and is equal to a preset proportionality coefficient multiplied by the product of the real-time dust concentration data and the square of the real-time wind speed data; the wind speed-vibration frequency characteristic curve is a curve pre-established through theoretical calculation. The proportionality coefficient is a comprehensive empirical constant that links measurable macroscopic quantities (wind speed, dust concentration) with microscopic impact effects. This coefficient is proportional to the cosine of the tilt angle because when the tilt angle of the component is close to 0 degrees (flat), its effective cross-sectional area facing the wind and dust is the largest, cos(0°)=1, and it withstands the strongest impact. When the tilt angle is close to 90 degrees (vertical), the effective cross-sectional area is the smallest, cos(90°)=0, and it withstands the weakest impact.

[0091] Specifically, by using a pre-calibrated wind speed-vibration frequency characteristic curve (which can be a lookup table or a polynomial function), real-time wind speed is directly mapped to the most likely vibration frequency of the support structure, enabling a more accurate assessment of vibration risk. For dust impacts, the disturbance energy caused to the connector is related to the kinetic energy of the dust particles (proportional to the square of the velocity) and the particle flux (proportional to the product of concentration and velocity). Therefore, using dust concentration × wind speed... 2This combined term provides a more scientific characterization of the total impact energy intensity than considering either factor individually. Introducing a proportionality coefficient proportional to the cosine of the tilt angle is a refined correction to the model, making the calculations more universally applicable to power plants with different installation tilt angles. Ultimately, these two calculations yielded two key, quantified mechanical noise indicators.

[0092] In some embodiments, the frequency calculation in this step can be implemented in a variety of ways: Optionally, one method is based on an embedded model in the embedded firmware: In the firmware program of the inverter or central controller, the "wind speed-vibration frequency characteristic curve" is fixed in the form of a piecewise linear function or a look-up table; the proportional coefficient calculated based on the tilt angle of the power station components is written into the system as a configurable parameter; after the controller obtains real-time data, it directly calls the internal function and quickly calculates the two frequency values ​​by looking up the table or formula.

[0093] It is understandable that other methods can be used to perform the calculation in this step, such as using a machine learning model to predict these two frequencies by training on historical data, etc., which are not limited here.

[0094] S204. When the vibration frequency of the support or the sand and dust impact frequency is greater than the corresponding preset frequency threshold, a waiting time is entered. During the waiting time, real-time wind speed data and real-time sand and dust concentration data are periodically acquired again at preset time intervals, and the vibration frequency of the support and the sand and dust impact frequency are calculated again.

[0095] Among them, the preset frequency threshold refers to the critical value set for support vibration and sandstorm impact respectively, which distinguishes between acceptable and excessively harsh environments. These thresholds are usually determined experimentally, that is, by conducting dynamic resistance tests under different interference intensities to find the highest interference level that does not affect the signal-to-noise ratio of the test results. The waiting time is not a fixed duration, but refers to a working state in which the system is paused and monitored. The preset time interval refers to the time step for periodic environmental re-examination in the waiting state, which is used to balance the timeliness of response and the consumption of computing resources.

[0096] Specifically, after calculating the two frequency indicators characterizing mechanical interference, they are compared with their respective preset thresholds. The logical judgment here is an OR relationship; that is, if either the support vibration frequency or the sandstorm impact frequency exceeds the safety limit, it is determined that it is not suitable to perform precise electrical diagnostics. Once it is determined to be unsuitable, a waiting time state is entered. In this state, all subsequent electrical testing steps will be suspended or prohibited from starting, and an internal periodic timer will be started. Whenever the timer reaches the preset time interval, operations S202 and S203 are actively and repeatedly executed—that is, the latest environmental data is reacquired, and the two frequencies are recalculated. This cyclical waiting-re-checking process can dynamically and continuously monitor environmental changes, patiently waiting for severe weather (such as strong winds and sandstorms) to pass until environmental conditions become suitable again.

[0097] In some embodiments, the waiting and periodic re-examination logic of this step can be implemented in a variety of ways: Optionally, a state machine and timer interrupt-based approach is used: the controller's software architecture is designed as a state machine, including states such as "environmental monitoring," "waiting," and "test execution." When a frequency exceedance is detected in the "environmental monitoring" state, the program switches the state variable to "waiting." This state transition simultaneously starts a hardware timer and sets its overflow time to a preset time interval. Each time the timer overflows and generates an interrupt, the interrupt service routine calls the environment re-check function and re-evaluates whether the conditions for exiting the waiting state are met.

[0098] It is understandable that other methods can be used to implement the control logic of this step, such as using a simpler loop delay structure, etc., which are not limited here.

[0099] In some embodiments, when a waiting period is entered due to harsh environment, the environmental disturbance can also be used as a natural excitation source to passively monitor the photovoltaic strings online, thereby screening out suspected faulty strings that are highly sensitive to mechanical vibration in advance before formal electrical testing.

[0100] Specifically, for each photovoltaic string, its open-circuit voltage (Voc) is continuously acquired at a preset time interval much higher than the rate of environmental change (e.g., 10 samples per second), thus forming a voltage noise sequence that varies over time. Under ideal static conditions, the open-circuit voltage of a healthy string should be very stable, and its noise sequence should be close to a straight line. However, when there are poor contacts (such as loose connectors or improper crimping) or microcracks in the cells, external mechanical vibrations can cause instantaneous and drastic fluctuations in the contact resistance at these defective points. This microscopic resistance change is immediately reflected in macroscopic open-circuit voltage fluctuations, thus generating significant peaks and troughs in the voltage noise sequence.

[0101] The support vibration frequency and dust impact frequency calculated in S203 are used as two known, dominant excitation source characteristic frequencies. Then, signal processing techniques, such as Fast Fourier Transform (FFT), lock-in amplifier algorithms, or correlation analysis, are employed to process the voltage noise sequence of each string. The aim is to analyze the spectrum of this voltage noise sequence and specifically extract the voltage response amplitude at the two excitation source characteristic frequencies. If a string's voltage noise spectrum shows significant peaks at these two specific frequency points, it strongly suggests a direct causal relationship between its voltage fluctuations and external physical vibrations. For fair cross-sectional comparisons, the response also needs to be normalized. This calculates a vibration sensitivity factor, which is the extracted voltage response amplitude divided by the excitation source strength characterizing the current total disturbance intensity. This excitation source strength is a comprehensive index, weighted based on real-time wind speed and dust concentration data; for example, strength ∝ aV_wind. 2 +bC_dust*V_wind 2 Among them, intensity represents the excitation source intensity of the current total disturbance intensity, which is a comprehensive indicator; a and b : is a weighting coefficient used to adjust the contribution of wind speed and dust concentration to the total disturbance intensity; V_wind: represents the real-time wind speed; C_dust: represents the real-time dust concentration. This formula shows that the excitation source intensity is proportional to the square of the wind speed (V_wind...). 2 ) and the product of dust concentration and the square of wind speed (C_dust×V_wind) 2 The weighting coefficients a and b are used to determine the relative importance of these two factors in the total disturbance intensity; ∝ in the formula represents proportional to. This factor represents how much voltage response a unit intensity of mechanical disturbance can excite on the string, thus removing the influence of weather change itself on the response magnitude and reflecting the string's own mechanical sensitivity. Finally, when the vibration sensitivity factor of a string exceeds a preset coupling threshold calibrated experimentally and representing the upper limit of the healthy string response, the string is marked as a suspected abnormal string, and an early warning indication is generated to remind maintenance personnel that the string has a potential mechanical connection defect.

[0102] This method transforms adverse weather conditions into a natural vibration table, directly quantifying the mechanical sensitivity of the string by analyzing the coupling strength between voltage noise and environmental vibration frequencies. A healthy string has a robust electrical connection and exhibits extremely low voltage response to mechanical vibration, while a string with potential faults such as poor contact will show strong voltage fluctuations. Therefore, by using a vibration sensitivity factor, high-risk strings with mechanical connection defects can be identified in advance, thus opening up a new dimension of fault detection beyond the main diagnostic process. This improves the detection rate of intermittent and latent faults, ultimately enhancing the overall accuracy of anomaly detection.

[0103] S205. When the real-time calculated vibration frequency of the support and the sand and dust impact frequency are both less than the corresponding preset frequency threshold, the waiting time ends.

[0104] Specifically, in the periodic re-inspection loop described in S204, this step's judgment is performed after each recalculation of the support vibration frequency and sand impact frequency. This judgment is a rigorous composite logic judgment: it requires that the real-time calculated support vibration frequency must be less than the corresponding preset threshold, and simultaneously, the sand impact frequency must also be less than the corresponding preset threshold. Only when both conditions are simultaneously met is it considered that the external mechanical interference has been comprehensively reduced to a safe level. Once this condition is met, the waiting time state immediately ends, and the periodic environmental re-inspection loop stops, thus ensuring that all subsequent precision measurements are performed in a truly quiet physical environment.

[0105] It is understandable that other methods can be used to implement the logical judgment of this step, such as introducing a requirement for the number of consecutive times the condition is met (e.g., exiting the waiting state only after three consecutive qualified tests) to increase the robustness of the judgment. No limitation is made here.

[0106] S206. Control the output current of all photovoltaic strings to be the preset first current value, and record the first voltage value corresponding to the first current value of each photovoltaic string.

[0107] S207. A preset first current pulse command is synchronously applied to all photovoltaic strings, so that the output current of each photovoltaic string increases from the first current value to the second current value corresponding to the first current pulse command, and the second voltage value corresponding to the second current value of each photovoltaic string is recorded.

[0108] S208. Calculate the first dynamic resistance value of each photovoltaic string from the first current value to the second current value.

[0109] S209. Apply a preset second current pulse command to all photovoltaic strings synchronously, so that the output current of each photovoltaic string is increased to the third current value corresponding to the second current pulse command, and record the third voltage value corresponding to the third current value of each photovoltaic string.

[0110] S210. Calculate the second dynamic resistance value of each photovoltaic string from the first current value to the third current value.

[0111] S211. Based on the first dynamic resistance value and the second dynamic resistance value, determine the dynamic resistance variation factor of each photovoltaic string.

[0112] Steps S206-S211 and Figure 1 Steps S102-S107 in the illustrated embodiment are similar and can be found in the descriptions of steps S102-S107, which will not be repeated here.

[0113] In some embodiments, when determining the dynamic resistance variation factor of each photovoltaic string, a physical model-based residual analysis method can also be introduced. By accurately quantifying and eliminating the normal thermal drift introduced by the testing process itself, the technical effect of more realistically reflecting the inherent nonlinear characteristics of the fault point can be achieved.

[0114] Specifically, the core of this step lies in distinguishing between normal, predictable resistance changes and abnormal, fault-induced resistance changes. First, the actual absolute difference between the second and first measured dynamic resistance values ​​(|second dynamic resistance value - first dynamic resistance value|) is calculated; this difference encompasses all variations. However, even a perfectly healthy photovoltaic string will exhibit a slight change in dynamic resistance under two consecutive high-current pulse injections. This is because the current pulse generates Joule heat as it passes through the string, i.e., Q=I. 2 Rt, where Q represents the heat generated (Joule heat), measured in Joules (J); I represents the current, measured in Ampersters (A); R represents the resistance, measured in Ohms (Ω); and t represents the time, measured in seconds (s). This formula shows that the heat generated when current flows through a resistor is proportional to the square of the current, the resistance, and the time. Joule heat causes a momentary increase in temperature in the string (especially its internal resistive portion). Due to the physical properties of semiconductors and metals, resistance values ​​change with temperature; this change caused by the test itself is normal and physically expected. If indiscriminately applied, this normal thermal drift may be misinterpreted as a fault or mask weak, genuine fault signals.

[0115] To address this issue, a projected dynamic resistance change is introduced, calculated using a pre-established physical model to predict the expected resistance change of a healthy photovoltaic string during testing. This is calculated as the product of a baseline thermal drift coefficient and an additional energy surge. The baseline thermal drift coefficient is an empirical constant derived from calibration tests on a large number of healthy photovoltaic strings, characterizing the projected rate of change (in Ω / Joule) of dynamic resistance due to the injection of a unit of Joule of thermal energy. The additional energy surge refers to the energy injected by the second current pulse (the third current value) relative to the first pulse (the second current value), causing a temperature rise, and can be calculated based on the pulse current magnitude and duration. Multiplying these two quantities yields a theoretical resistance change due to the Joule heating effect. Subtracting this calculated projected change from the actual measured absolute resistance difference yields a residual value. This residual value represents the pure resistance change remaining after eliminating normal thermal drift, which cannot be explained by normal physical processes. For a healthy string, its residual value should theoretically be close to zero. Finally, to facilitate comparison and set a uniform threshold, the absolute value of this residual value is divided by the initial first dynamic resistance value to obtain a standardized, dimensionless dynamic resistance change factor.

[0116] By introducing compensation for anticipated changes based on a physical model, the Joule heating effect interference source of the test current can be removed from the measurement results. The dynamically calculated resistance change factor is no longer affected by normal temperature drift, but rather reflects more purely and directly the nonlinear resistance abrupt changes unique to fault points such as connector oxidation and poor solder joints under electrical stress. This improves the signal-to-noise ratio, enabling more reliable identification of weak early faults, while avoiding misjudging normal thermal effects as anomalies, thereby improving the accuracy of photovoltaic string anomaly detection.

[0117] S212. When a dynamic resistance change factor exceeds a preset resistance change factor threshold, the corresponding photovoltaic string is identified as an abnormal string, and an abnormal indication is generated.

[0118] Step S212 and Figure 1 Step S108 in the illustrated embodiment is similar and can be found in the description of step S108, which will not be repeated here.

[0119] The following describes an exemplary photovoltaic string anomaly detection system 300 provided in an embodiment of this application. Figure 3 This is an exemplary hardware structure diagram of the photovoltaic string anomaly detection system 300 provided in the embodiments of this application.

[0120] In some embodiments, the anomaly detection system 300 for the photovoltaic string is a computer device or includes a computer device. The computer device includes a processor, memory, and a network interface connected via a system bus. The processor of the computer device provides computing and control capabilities. The memory of the computer device includes a non-volatile storage medium and internal memory. The non-volatile storage medium stores an operating system, computer programs, and a database. The internal memory provides an environment for the operation of the operating system and computer programs in the non-volatile storage medium. The database of the computer device stores data. The network interface of the computer device is used to communicate with other external terminals or servers via a network connection. In some embodiments, the network interface can be a wired network interface; in some embodiments, the network interface can also be a wireless network interface. When the computer program is executed by the processor, it implements the methods in the embodiments of this application.

[0121] Those skilled in the art will understand that Figure 4 The structure shown is merely a block diagram of a portion of the structure related to the present application and does not constitute a limitation on the computer device to which the present application is applied. Specific computer devices may include more or fewer components than those shown in the figure, or combine certain components, or have different component arrangements.

[0122] Example 4 In one specific embodiment, such as Figure 5 As shown, the present invention also provides an anomaly detection system for photovoltaic strings, comprising: The data acquisition module is used to switch the inverter's operating mode from maximum power point tracking mode to direct command control mode when the real-time ambient irradiance is in a preset low light intensity range and continues for a preset duration. The first data processing module is used to control the output current of all photovoltaic strings to a preset first current value and record the corresponding first voltage value; synchronously apply a first current pulse command to raise the output current to a second current value and record the corresponding second voltage value; and obtain the first dynamic resistance value of each string from the first current value to the second current value. The second data processing module is used to synchronously apply a second current pulse command to raise the current to a third current value, record the third voltage value, and obtain the second dynamic resistance value from the first current value to the third current value. The third data processing module is used to obtain the dynamic resistance change factor of each photovoltaic string based on the first dynamic resistance value and the second dynamic resistance value. The fourth data processing module is used to identify photovoltaic strings as abnormal strings when a dynamic resistance change factor is greater than a preset resistance change factor threshold.

[0123] Additionally, a schematic diagram of a terminal device according to an embodiment of the present invention is provided. This terminal device includes: a processor, a memory, and a computer program stored in the memory and executable on the processor. When the processor executes the computer program, it implements the steps in the various method embodiments described above. Alternatively, when the processor executes the computer program, it implements the functions of each module / unit in the various device embodiments described above.

[0124] The computer program can be divided into one or more modules / units, which are stored in the memory and executed by the processor to complete the present invention.

[0125] The terminal device may be a desktop computer, laptop, handheld computer, or cloud server, etc. The terminal device may include, but is not limited to, a processor and a memory.

[0126] The processor may be a central processing unit (CPU), or other general-purpose processors, digital signal processors (DSPs), application-specific integrated circuits (ASICs), field-programmable gate arrays (FPGAs), or other programmable logic devices, discrete gate or transistor logic devices, discrete hardware components, etc.

[0127] The memory can be used to store the computer program and / or module. The processor implements various functions of the terminal device by running or executing the computer program and / or module stored in the memory and calling the data stored in the memory.

[0128] If the modules / units integrated into the terminal device are implemented as software functional units and sold or used as independent products, they can be stored in a computer-readable storage medium. Based on this understanding, all or part of the processes in the methods of the above embodiments can also be implemented by a computer program instructing related hardware. The computer program can be stored in a computer-readable storage medium, and when executed by a processor, it can implement the steps of the various method embodiments described above. The computer program includes computer program code, which can be in the form of source code, object code, executable files, or certain intermediate forms. The computer-readable medium can include: any entity or device capable of carrying the computer program code, recording media, USB flash drives, portable hard drives, magnetic disks, optical disks, computer memory, read-only memory (ROM), random access memory (RAM), electrical carrier signals, telecommunication signals, and software distribution media, etc. It should be noted that the content included in the computer-readable medium can be appropriately added or removed according to the requirements of legislation and patent practice in the jurisdiction. For example, in some jurisdictions, according to legislation and patent practice, computer-readable media do not include electrical carrier signals and telecommunication signals. The above are merely preferred embodiments of the present invention and are not intended to limit the present invention. For those skilled in the art, the present invention can have various modifications and variations.

[0129] The above are merely preferred embodiments of the present invention and are not intended to limit the present invention. Various modifications and variations can be made to the present invention by those skilled in the art. Any modifications, equivalent substitutions, improvements, etc., made within the spirit and principles of the present invention should be included within the scope of protection of the present invention.

Claims

1. A method for detecting anomalies in photovoltaic strings, characterized in that, Includes the following steps: S1: When the real-time ambient irradiance is in the preset low light intensity range and continues for more than the preset duration, the inverter working mode will be switched from maximum power point tracking mode to direct command control mode. S2: Control the output current of each photovoltaic string to a preset first current value and record the corresponding first voltage value; synchronously apply a first current pulse command to raise the output current to a second current value and record the corresponding second voltage value. And obtain the first dynamic resistance value of each photovoltaic string from the first current value to the second current value; S3: Apply a second current pulse command synchronously to raise the current to the third current value and record the third voltage value; And obtain the second dynamic resistance value from the first current value to the third current value; S4: Based on the first dynamic resistance value and the second dynamic resistance value, obtain the dynamic resistance change factor of each photovoltaic string; S5: When any dynamic resistance change factor is greater than the preset resistance change factor threshold, the corresponding photovoltaic string is an abnormal string.

2. The method for anomaly detection of photovoltaic strings according to claim 1, characterized in that, In step S1, after the real-time ambient irradiance is within a preset low light intensity range and continues for more than a preset duration, the following steps are also included: Acquire real-time wind speed data and real-time dust concentration data of the photovoltaic power station; calculate the support vibration frequency and dust impact frequency based on the real-time wind speed data and the real-time dust concentration data; When the vibration frequency of the support or the sand and dust impact frequency is greater than the corresponding preset frequency threshold, a waiting time is entered. During the waiting time, the real-time wind speed data and the real-time sand and dust concentration data are periodically acquired again at preset time intervals, and the vibration frequency of the support and the sand and dust impact frequency are calculated again. The waiting time ends when both the real-time calculated vibration frequency of the support and the sand and dust impact frequency are less than the corresponding preset frequency threshold.

3. The method for anomaly detection of photovoltaic strings according to claim 2, characterized in that, The waiting time also includes: For each photovoltaic string, the open-circuit voltage at both ends is acquired at the preset time interval to obtain a voltage noise sequence; the vibration frequency of the support and the sand and dust impact frequency are used as the characteristic frequencies of the excitation source; the voltage response amplitude associated with the voltage noise sequence and the characteristic frequency of the excitation source is extracted; the ratio of the voltage response amplitude to the excitation source intensity is calculated to obtain the vibration sensitivity factor characterizing the corresponding photovoltaic string; when the vibration sensitivity factor exceeds the preset coupling threshold, the corresponding photovoltaic string is marked as a suspected abnormal string, and a suspected abnormality warning indication is generated.

4. The method for anomaly detection of photovoltaic strings according to claim 1, characterized in that, Based on the first and second dynamic resistance values, the dynamic resistance variation factor of each photovoltaic string is determined, specifically including: Calculate the actual absolute difference between the second dynamic resistance value and the first dynamic resistance value; subtract the expected change in dynamic resistance from the actual absolute difference to obtain the residual value; calculate the ratio of the absolute value of the residual value to the first dynamic resistance value to obtain the dynamic resistance change factor.

5. The method for anomaly detection of photovoltaic strings according to claim 1, characterized in that, After identifying the corresponding photovoltaic string as an abnormal string, the process also includes: For the abnormal string, the second current pulse command is applied again, and the corresponding fourth voltage value is recorded; based on the first voltage value and the fourth voltage value, the third dynamic resistance value is calculated; according to the third dynamic resistance value and the second dynamic resistance value, the resistance value attenuation rate is determined; when the resistance value attenuation rate exceeds the preset healing threshold, the fault type indicated by the abnormality is marked as intermittent self-healing connection abnormality.

6. The method for anomaly detection of photovoltaic strings according to claim 1, characterized in that, Based on the correction of the first current value, the correction of the second current value, and the correction of the third current value, a first current pulse command and a second current pulse command are applied. The process of obtaining the corrected first current value, the corrected second current value, and the corrected third current value is as follows: Obtain the real-time temperature data of any photovoltaic string; divide the photovoltaic string into multiple temperature groups according to the real-time temperature data, and ensure that the temperature difference between photovoltaic strings in each temperature group is less than a preset temperature difference threshold. The temperature correction coefficient for the corresponding temperature group is obtained by using the temperature difference value; Then, the corrected first current value, corrected second current value, and corrected third current value corresponding to the first current value, the second current value, and the third current value are obtained through the temperature correction coefficient.

7. An anomaly detection system for photovoltaic strings, characterized in that, include: The data acquisition module is used to switch the inverter's operating mode from maximum power point tracking mode to direct command control mode when the real-time ambient irradiance is in a preset low light intensity range and continues for a preset duration. The first data processing module is used to control the output current of all photovoltaic strings to a preset first current value and record the corresponding first voltage value; and to synchronously apply a first current pulse command to raise the output current to a second current value and record the corresponding second voltage value. And obtain the first dynamic resistance value of each string from the first current value to the second current value; The second data processing module is used to synchronously apply the second current pulse command to raise the current to the third current value and record the third voltage value. And obtain the second dynamic resistance value from the first current value to the third current value; The third data processing module is used to obtain the dynamic resistance change factor of each photovoltaic string based on the first dynamic resistance value and the second dynamic resistance value. The fourth data processing module is used to identify photovoltaic strings as abnormal strings when a dynamic resistance change factor is greater than a preset resistance change factor threshold.

8. A computer system comprising a memory, a processor, and a computer program stored in the memory, characterized in that, The processor executes the computer program to implement the steps of the method as described in any one of claims 1 to 6.

9. A computer-readable storage medium storing a computer program, characterized in that, When the computer program / instructions are executed by the processor, they implement the steps of the method as described in any one of claims 1 to 6.

10. A computer program product comprising a computer program / instructions, characterized in that, When the computer program / instructions are executed by the processor, they implement the steps of the method as described in any one of claims 1 to 6.

Images