A method and system for photovoltaic module hot spot diagnosis and adaptive cleaning

Abstract

The application discloses a photovoltaic module hot spot diagnosis and adaptive cleaning method and system, and the method comprises the following steps: in the cleaning process of a cleaning robot, temperature distribution data of a photovoltaic module surface is acquired in real time through a rapid temperature measurement module; based on the temperature distribution data, a suspected hot spot area with temperature difference anomaly is identified; the cleaning robot is controlled to move to the suspected hot spot area, high-precision temperature scanning is performed through a hot spot diagnosis module, and a hot spot position and range are confirmed; surface state analysis is performed on the confirmed hot spot area, and adaptive cleaning instructions are generated based on the analysis result. Through a cleaning and detection two-stage cooperative mechanism and a dynamic parameter optimization strategy, the application improves the accuracy of photovoltaic module hot spot diagnosis and adaptive cleaning and the operation and maintenance decision efficiency.

Description

Technical Field

[0001] This invention relates to the field of photovoltaic module cleaning technology, and in particular to a method and system for diagnosing hot spots and adaptive cleaning of photovoltaic modules. Background Technology

[0002] Current photovoltaic module cleaning and hot spot detection technologies mainly rely on installing temperature sensors on photovoltaic cleaning robots to determine hot spots by using a fixed temperature difference threshold.

[0003] The existing technology has the following technical defects: low accuracy of judgment, static threshold cannot adapt to the dynamic changes in ambient temperature, and the false alarm rate increases significantly during dawn and dusk; lack of abnormal area re-inspection mechanism, preliminary detection results need to be manually verified again, and there is redundancy in the operation and maintenance process; it cannot automatically distinguish between battery defect hot spots and stain hot spots, resulting in operation and maintenance personnel frequently performing ineffective on-site inspections and serious waste of resources.

[0004] The aforementioned problems have led to low operation and maintenance efficiency and soaring costs for photovoltaic power plants, necessitating a new technical solution to improve the accuracy of hot spot diagnosis and adaptive cleaning of photovoltaic modules, as well as the efficiency of operation and maintenance decision-making. Summary of the Invention

[0005] The main objective of this invention is to propose a method for hot spot diagnosis and adaptive cleaning of photovoltaic modules, aiming to solve the technical problems of low accuracy and low operation and maintenance efficiency in the existing photovoltaic module hot spot diagnosis and adaptive cleaning methods.

[0006] To achieve the above objectives, the first aspect of this invention proposes a method for hot spot diagnosis and adaptive cleaning of photovoltaic modules, comprising:

[0007] Step S100: During the cleaning process of the cleaning robot, the temperature distribution data on the surface of the photovoltaic module is obtained in real time through the rapid temperature measurement module;

[0008] Step S200: Based on the temperature distribution data, identify suspected hot spot areas with abnormal temperature differences;

[0009] Step S300: Control the cleaning robot to move to the suspected hot spot area, and perform a high-precision temperature scan through the hot spot diagnosis module to confirm the location and range of the hot spot;

[0010] Step S400: Perform surface condition analysis on the confirmed hot spot area and generate adaptive cleaning instructions based on the analysis results.

[0011] Preferably, in step S100, the step of acquiring the temperature distribution data of the photovoltaic module surface in real time through the rapid temperature measurement module includes:

[0012] Step S110: When the cleaning robot moves along the surface of the photovoltaic module, it synchronously collects temperature data at multiple locations on the surface of the photovoltaic module at a preset sampling frequency through the multi-temperature sensor array it is equipped with.

[0013] Step S120: Calculate the spatial coordinates of each temperature sampling point on the surface of the photovoltaic module based on the position distribution of the multi-temperature sensor array and the real-time displacement data of the robot;

[0014] Step S130: Integrate the temperature values ​​corresponding to the spatial coordinates to generate a real-time two-dimensional temperature distribution map of the photovoltaic module surface.

[0015] Preferably, in step S200, the step of identifying suspected hot spot areas with abnormal temperature differences based on the temperature distribution data includes:

[0016] Step S210: Based on the real-time two-dimensional temperature distribution map, divide the photovoltaic module cell into grid units according to the physical structure of the cell, with each grid unit containing at least one temperature sampling point;

[0017] Step S220: Calculate the absolute value of the temperature difference between each grid cell and its adjacent grid cells. If the difference exceeds the first threshold, mark the grid cell as a suspected abnormal cell.

[0018] Step S230: Spatial clustering of adjacent suspected abnormal units to form continuous clustered regions; if the clustered region meets any of the following conditions, it is determined to be a suspected hot spot region: the absolute value of the maximum temperature difference in the region exceeds the second threshold, or the absolute value of the highest temperature in the region exceeds the safety threshold.

[0019] Preferably, the first threshold and the second threshold are dynamically set through the following steps:

[0020] Step S231: Real-time acquisition of the operating electrical parameters and environmental parameters of the target photovoltaic module, wherein the operating electrical parameters include operating current and operating voltage, and the environmental parameters include irradiance and ambient temperature;

[0021] Step S232: Generate electrical parameter correction coefficients based on the deviation between the operating electrical parameters and the rated electrical parameters;

[0022] Step S233: Determine the temperature reference value based on the matching result between the ambient temperature and the historical temperature database, and generate an environmental correction coefficient by combining it with the real-time irradiance;

[0023] Step S234: Match a benchmark threshold from a preset database based on the target photovoltaic module model, and dynamically adjust the threshold by overlaying the electrical parameter correction coefficient and the environmental correction coefficient to generate a first threshold.

[0024] Step S235: Based on the first threshold and the preset grade difference ratio, generate a second threshold that is greater than the first threshold.

[0025] Preferably, in step S220, the step of calculating the absolute value of the temperature difference between each grid cell and its adjacent grid cells includes:

[0026] Step S221: Determine the set of adjacent grid cells of the current grid cell based on the physical structure of the target photovoltaic module cell;

[0027] Step S222: Based on the preset temperature feature extraction rules, generate the representative temperature value of the current mesh cell;

[0028] Step S223: Calculate the absolute value of the difference between the temperature value represented by the current grid cell and the temperature value represented by each adjacent grid cell;

[0029] Step S224: Select the maximum value among the absolute values ​​of the differences as the absolute value of the temperature difference between the current grid cell and the adjacent grid cell.

[0030] Preferably, in step S222, the temperature feature extraction rule is set to extract representative temperature values ​​from the grid cell temperature data. The implementation methods include extracting the highest temperature value of all temperature sampling points, or calculating the arithmetic mean temperature value of all temperature sampling points, or calculating the weighted average temperature value based on the location weight.

[0031] Preferably, in step S300, the step of performing a high-precision temperature scan through the hot spot diagnostic module to confirm the location and extent of the hot spot includes:

[0032] Step S310: Control the cleaning robot to move to the suspected hot spot area, wherein the hot spot diagnosis module includes an infrared temperature sensor;

[0033] Step S320: Perform a high-precision temperature scan using the infrared temperature sensor, input the temperature field distribution data obtained from the scan into a pre-trained hot spot recognition neural network model, and generate a hot spot probability distribution map;

[0034] Step S330: Integrate the physical structure information of the photovoltaic module cells, perform boundary optimization processing on the hot spot probability distribution map, and output the geometric center coordinates and physical boundary polygon of the hot spot.

[0035] Preferably, in step S400, the step of performing surface state analysis on the confirmed hot spot area and generating an adaptive cleaning command based on the analysis results includes:

[0036] Step S410: Measure the hot spot area using the surface analysis module to obtain surface ash distribution data;

[0037] Step S420: Calculate a comprehensive score for the degree of dirtiness based on the surface ash distribution data and the historical cleaning records of the hot spot area;

[0038] Step S430: When the overall dirt level score exceeds the dirt threshold, it is determined to be a dirt-induced hot spot, and the deep cleaning process is started; when the overall dirt level score does not exceed the dirt threshold, a defect score is generated by combining the temperature distribution data of the hot spot area; if the defect score exceeds the defect threshold, a component defect alarm command is generated.

[0039] Preferably, in step S430, the deep cleaning process includes the following steps:

[0040] Step S431: Determine the initial cleaning intensity level and the maximum allowable number of cleaning cycles based on the comprehensive score of the degree of dirtiness;

[0041] Step S432: Control the cleaning robot to perform a single-wheel cleaning operation;

[0042] Step S433: Immediately after cleaning, detect the current dust concentration and calculate the dust concentration change rate for this round;

[0043] Step S434: When the rate of change of ash concentration is greater than the attenuation threshold, the cleaning intensity level is reduced; when the rate of change of ash concentration is less than the minimum effective threshold, the cleaning intensity level and the number of cleaning cycles are increased.

[0044] Step S435: Determine whether the current dust concentration meets the standard. If it does, terminate the process. Determine whether the cumulative number of rounds exceeds the maximum allowable number of rounds. If it does not exceed the limit, return to step S432 to perform a new round of cleaning. If it exceeds the limit, mark the stubborn dirt and terminate the process.

[0045] A second aspect of this invention provides a photovoltaic module hot spot diagnosis and adaptive cleaning system, comprising:

[0046] Rapid temperature measurement module: used to acquire real-time temperature distribution data on the surface of photovoltaic modules during the cleaning process of the cleaning robot;

[0047] Temperature difference anomaly detection module: Based on the temperature distribution data, identify suspected hot spot areas with temperature difference anomalies;

[0048] Hot spot recognition module: used to control the cleaning robot to move to the suspected hot spot area, and to perform high-precision temperature scanning through the hot spot diagnosis module to confirm the location and range of the hot spot;

[0049] Adaptive cleaning module: Performs surface condition analysis on the identified hot spot areas and generates adaptive cleaning instructions based on the analysis results.

[0050] The photovoltaic module hot spot diagnosis and adaptive cleaning method and system provided by this invention establishes a detection basis by simultaneously acquiring high-density temperature distribution data during the cleaning process; accurately captures suspected hot spot areas based on a dynamic temperature difference recognition algorithm; automatically completes high-precision hot spot positioning and re-inspection through the equipment return path; intelligently identifies the cause of hot spots by integrating surface ash and temperature gradient features; dynamically optimizes the cleaning command execution decision by combining a closed-loop feedback mechanism; realizes integrated cleaning and diagnosis functions for a single device to eliminate manual secondary verification; and ultimately significantly improves the timeliness of operation and maintenance response and reduces the misjudgment rate.

[0051] Furthermore, this invention achieves precise temperature measurement at the cell level by matching the spatial distribution of a temperature sensor array with the structure of the photovoltaic module. A non-contact acquisition mechanism avoids mechanical wear, and an active dwell strategy in the cell area ensures high-quality temperature data acquisition. Adaptive parameter design guarantees overall system compatibility. The invention divides the cell's physical structure into grid cells and calculates adjacent temperature differences based on electrical characteristics. Region growth clustering locates continuous abnormal regions, and a dual dynamic threshold determination mechanism enhances the sensitivity and anti-interference capability of hot spot detection. Thresholds are corrected in real-time using electrical and environmental parameters, and detection sensitivity is dynamically optimized based on historical temperature databases and environmental parameters, improving the robustness of hot spot detection and reducing the risk of omissions. Temperature feature extraction rules are adapted to non-uniform temperature field characteristics, and adjacent relationships are optimized using electrical topology. The system defines and filters maximum values ​​to enhance the ability to capture local abnormal temperature rises and the sensitivity of identifying strong hot spots. A multi-mode temperature feature extraction mechanism improves differentiated analysis capabilities; the maximum value method enhances the sensitivity of high-hot spot capture, the arithmetic mean method ensures the stability of routine detection, and the position-weighted method optimizes the accuracy of local hot spot location. High-precision infrared scanning and neural network recognition work together to improve sub-centimeter-level hot spot location capabilities, while boundary optimization algorithms constrained by the physical structure of the solar cells enhance contour accuracy and structural matching. A dual-mode analysis mechanism combining laser graying and visible light images, combined with historical data fusion, improves the accuracy of dirt scoring and enhances the ability to distinguish the causes of dirt and hot spot defects. A dynamic bidirectional adjustment mechanism for intensity and cycle, combined with cycle upper limit constraints, improves the efficiency and reliability of adaptive cleaning. Attached Figure Description

[0052] To more clearly illustrate the technical solutions in the embodiments of the present invention or the prior art, the drawings used in the description of the embodiments or the prior art will be briefly introduced below. Obviously, the drawings described below are only some embodiments of the present invention. For those skilled in the art, other drawings can be obtained based on the structures shown in these drawings without creative effort.

[0053] Figure 1 This is a flowchart of a photovoltaic module hot spot diagnosis and adaptive cleaning method provided in an embodiment of the present invention;

[0054] Figure 2 This is a flowchart of a real-time cleaning and rapid temperature measurement method provided in an embodiment of the present invention;

[0055] Figure 3 This is a flowchart of a method for identifying suspected hot spot regions according to an embodiment of the present invention;

[0056] Figure 4 This is a flowchart of a method for dynamically setting a threshold according to an embodiment of the present invention;

[0057] Figure 5 This is a flowchart of a method for calculating the absolute value of temperature difference in a grid cell according to an embodiment of the present invention;

[0058] Figure 6 A flowchart of a high-precision scanning method for a hot spot diagnostic module provided in an embodiment of the present invention;

[0059] Figure 7 This is a flowchart of a method for measuring hot spot regions using a surface analysis module according to an embodiment of the present invention;

[0060] Figure 8 A flowchart of a deep cleaning method provided in an embodiment of the present invention;

[0061] Figure 9 This is a schematic diagram of the principle of a photovoltaic module hot spot diagnosis and adaptive cleaning system provided in an embodiment of the present invention;

[0062] Figure 10 This is a schematic diagram of the structure of a photovoltaic module hot spot diagnosis and adaptive cleaning device provided in an embodiment of the present invention.

[0063] The realization of the objective, functional features and advantages of the present invention will be further explained in conjunction with the embodiments and with reference to the accompanying drawings. Detailed Implementation

[0064] 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 a part of the embodiments of the present invention, and not all of the embodiments. Based on the embodiments of the present invention, all other embodiments obtained by those skilled in the art without creative effort are within the scope of protection of the present invention.

[0065] It should be noted that if the embodiments of the present invention involve directional indicators, such as up, down, left, right, front, back, etc., the directional indicators are only used to explain the relative positional relationship and movement of the components in a specific posture. If the specific posture changes, the directional indicators will also change accordingly.

[0066] Furthermore, if the embodiments of this invention involve descriptions such as "first" or "second," these descriptions are for descriptive purposes only and should not be construed as indicating or implying their relative importance or implicitly specifying the number of technical features indicated. Therefore, a feature defined with "first" or "second" may explicitly or implicitly include at least one of those features. Additionally, the meaning of "and / or" throughout the text includes three parallel solutions; for example, "A and / or B" includes solution A, solution B, or a solution where both A and B are satisfied simultaneously. Furthermore, the technical solutions of the various embodiments can be combined with each other, but this must be based on the ability of those skilled in the art to implement them. When the combination of technical solutions is contradictory or impossible to implement, it should be considered that such a combination of technical solutions does not exist and is not within the scope of protection claimed by this invention.

[0067] Definitions:

[0068] Hot Spots: Solar panels are typically installed in open, sunny locations. Over long-term use, birds, dust, fallen leaves, and other obstructions inevitably accumulate, creating shadows on the solar panels. In large solar panel arrays, improper row spacing can also lead to mutual shadows. Due to these localized shadows, the current and voltage of certain individual cells within the solar panel change. This results in an increase in the product of current and voltage in these localized areas, causing localized temperature rises. Defects in some individual cells can also cause localized heating during operation; this phenomenon is called the "hot spot effect."

[0069] Therefore, the main objective of this invention is to propose a method and system for hot spot diagnosis and adaptive cleaning of photovoltaic modules, aiming to solve the technical problems of low accuracy and low operation and maintenance efficiency in the existing technology of hot spot diagnosis and adaptive cleaning of photovoltaic modules.

[0070] like Figures 1 to 8 As shown, the first aspect of this invention proposes a method for hot spot diagnosis and adaptive cleaning of photovoltaic modules, comprising:

[0071] Step S100: During the cleaning process of the cleaning robot, the temperature distribution data on the surface of the photovoltaic module is obtained in real time through the rapid temperature measurement module;

[0072] Step S200: Based on the temperature distribution data, identify suspected hot spot areas with abnormal temperature differences;

[0073] Step S300: Control the cleaning robot to move to the suspected hot spot area, and perform a high-precision temperature scan through the hot spot diagnosis module to confirm the location and range of the hot spot;

[0074] Step S400: Perform surface condition analysis on the confirmed hot spot area and generate adaptive cleaning instructions based on the analysis results.

[0075] For details, see Figure 1 In a specific embodiment of the present invention, a flat-sweeping cleaning robot is used to implement a photovoltaic module hot spot diagnosis and adaptive cleaning method. The robot includes a walking mechanism spanning the photovoltaic module array, which integrates a front brush, a rapid temperature measurement module, a rear brush, and a movable hot spot diagnosis module. The front brush is wider than the rear brush to enhance pre-cleaning capabilities and provide clean module surface conditions for the rapid temperature measurement module. The rapid temperature measurement module consists of a linearly arranged close-packed thermocouple array, fixedly mounted on a bracket between the front and rear brushes, with its height slightly lower than the brushes to avoid direct friction with the module surface. The hot spot diagnosis module is linked to a drive motor via a connector and its horizontal movement is controlled by a top guide rail; it integrates a high-precision infrared temperature sensor and a laser dust sensor. After the robot starts, it enters cleaning mode: moving along the photovoltaic module surface to perform cleaning operations, while the rapid temperature measurement module continuously collects module surface temperature data at a preset frequency and transmits it to the processor in real time. The processor generates a temperature distribution map based on the sensor positions and robot displacement data, and identifies and caches the coordinates of suspected hot spot areas with excessive temperature differences based on a dynamic threshold algorithm. After the cleaning task is completed, the robot switches to inspection mode: The robot returns to the center of the first suspected hot spot area, the drive motor moves the hot spot diagnosis module along the guide rail to the target position, the infrared temperature sensor is activated to perform millimeter-level precision scanning and calibrate the hot spot boundary; then the laser dust sensor is activated to analyze the surface dust accumulation state of the hot spot, and commands are triggered according to the detection results: if the dust concentration exceeds the threshold, a pressurized cleaning command is triggered, and the brush pressure and cleaning frequency are automatically adjusted to perform deep cleaning; if the dust accumulation is normal but the temperature is abnormal, a component defect report is generated and uploaded to the monitoring center; when all suspected areas have been inspected and there are no new commands, the robot returns to the initial standby point.

[0076] Understandably, this embodiment employs a dual detection mechanism to significantly improve diagnostic accuracy: rapid temperature scanning in cleaning mode initially identifies abnormal areas; subsequently, high-precision infrared thermography in inspection mode performs secondary verification, effectively overcoming misjudgments caused by ambient temperature fluctuations during dawn and dusk. Simultaneously, this embodiment integrates surface dust distribution characteristics and temperature gradient data to automatically distinguish between hot spots caused by dirt and those caused by battery defects, thereby reducing the need for manual on-site verification and lowering maintenance labor costs. For dirt-type hot spots, this embodiment intelligently adjusts cleaning intensity and frequency based on the rate of change in dust concentration, avoiding resource waste caused by fixed cleaning cycles and improving cleaning efficiency. Furthermore, this embodiment integrates cleaning and diagnostic functions into a single device and automatically completes re-inspection of abnormal areas during the device's return process, completely eliminating the manual secondary verification step in traditional solutions and improving maintenance response time. It should be noted that this embodiment is only a preferred solution; the robot implementing the photovoltaic module hot spot diagnosis and adaptive cleaning method is not limited to a flat sweeping robot, but can also use other types of robots or intelligent maintenance equipment, such as track-mounted, wheeled, or drone inspection platforms. Based on the above technical solutions, those skilled in the art can make corresponding equivalent improvements according to the application scenario. For example, the rapid temperature measurement module can be replaced with a technical solution based on a non-contact infrared temperature sensor array to achieve non-contact rapid scanning; or a flexible thin-film temperature sensor array can be attached to the brush surface for follow-up detection, which is suitable for the surface of wavy components; or the hot spot diagnosis module can be replaced with a multispectral imaging module to synchronously invert the temperature distribution and dust accumulation degree through spectral data from visible light to thermal infrared bands; or a composite diagnostic module based on ultrasonic / electromagnetic non-destructive testing can be deployed to scan and verify internal defects of the battery in combination with the location of the hot spot.

[0077] Preferably, in step S100, the step of acquiring the temperature distribution data of the photovoltaic module surface in real time through the rapid temperature measurement module includes:

[0078] Step S110: When the cleaning robot moves along the surface of the photovoltaic module, it synchronously collects temperature data at multiple locations on the surface of the photovoltaic module at a preset sampling frequency through the multi-temperature sensor array it is equipped with.

[0079] Step S120: Calculate the spatial coordinates of each temperature sampling point on the surface of the photovoltaic module based on the position distribution of the multi-temperature sensor array and the real-time displacement data of the robot;

[0080] Step S130: Integrate the temperature values ​​corresponding to the spatial coordinates to generate a real-time two-dimensional temperature distribution map of the photovoltaic module surface.

[0081] For details, see Figure 2In one specific embodiment of the present invention, the multi-temperature sensor array is adapted to the physical structure of the photovoltaic module. For example, for a standard 60-cell photovoltaic module (6x10 array, size 1.65m×1m), the multi-temperature sensor is configured as 18 closely packed thermocouples arranged in a horizontal single row, fixed at the center line of the support between the front and rear brushes. The array layout satisfies the following: the 18 thermocouples are evenly spaced in a single row along the width of the photovoltaic module; the horizontal width range of each cell corresponds to the monitoring area of ​​3 consecutive thermocouples; the thermocouples are installed at a height slightly lower than the working plane of the brushes to reduce contact, and the surface temperature information is indirectly obtained by measuring the air layer temperature at a small distance above the module surface. During the robot's movement, the displacement sensor records the position information in real time, and each temperature acquisition point is mapped to the global coordinate system of the module through a fixed offset. Furthermore, when the cleaning robot moves horizontally along the length of the component and enters a new battery cell area, it automatically slows down and stops. During this stop, it simultaneously performs temperature sampling through three thermocouples that laterally cover the battery cell. Each battery cell area completes three independent temperature samplings, with each sampling including data from the three thermocouples. At the same time, displacement sensors record position information in real time. Each temperature sampling point is mapped to the component's global coordinate system through a fixed offset, and a two-dimensional temperature distribution matrix of the component surface is constructed using the representative temperature values ​​of 60 battery cells. It should be noted that the temperature sampling frequency, the number of thermocouples, and their placement in this embodiment can all be adjusted according to the actual application scenario. For example, the number of sensors can be increased for large components, the sampling frequency can be increased in high-wind-speed scenarios, the thermocouple suspension height can be optimized to adapt to surface undulations for uneven installation surfaces, and special component layouts can be matched to the battery cell distribution characteristics by adjusting the lateral spacing.

[0082] Understandably, this embodiment ensures that the spatial distribution of temperature sampling points corresponds to the physical location of the battery cells through a matching design that aligns the temperature sensor array layout with the component structure. The non-contact mechanism based on air-layer temperature measurement effectively avoids mechanical wear between the sensors and the component surface, extending the equipment's lifespan. Combined with a displacement synchronization mapping and a data acquisition strategy that allows active dwell time in the battery cell area, accurate and low-interference temperature datasets can be acquired during dynamic movement. Simultaneously, the temperature distribution matrix construction process is closely linked to the battery cell structural units, providing a precise and reliable data foundation for subsequent temperature difference analysis. The flexible design, allowing for adaptive adjustment of the number of thermocouples and sampling frequency, enables the solution to adapt to changing requirements of different component sizes, installation environments, and operating conditions, enhancing the overall system adaptability.

[0083] Preferably, in step S200, the step of identifying suspected hot spot areas with abnormal temperature differences based on the temperature distribution data includes:

[0084] Step S210: Based on the real-time two-dimensional temperature distribution map, divide the photovoltaic module cell into grid units according to the physical structure of the cell, with each grid unit containing at least one temperature sampling point;

[0085] Step S220: Calculate the absolute value of the temperature difference between each grid cell and its adjacent grid cells. If the difference exceeds the first threshold, mark the grid cell as a suspected abnormal cell.

[0086] Step S230: Spatial clustering of adjacent suspected abnormal units to form continuous clustered regions; if the clustered region meets any of the following conditions, it is determined to be a suspected hot spot region: the absolute value of the maximum temperature difference in the region exceeds the second threshold, or the absolute value of the highest temperature in the region exceeds the safety threshold.

[0087] For details, see Figure 3 In a specific embodiment of the present invention, the grid cells are divided according to the physical cell boundaries of the photovoltaic module. Each grid cell corresponds to a cell region and its average temperature is used as a representative value. When calculating the temperature difference, the temperature values ​​of adjacent cells in the same row or column within the same cell string are compared first. If the temperature difference with any adjacent cell exceeds a first threshold, it is marked as a suspected abnormal cell. When performing spatial clustering on the marked cells, a region growing algorithm is used to automatically merge directly adjacent abnormal cells starting from the first abnormal cell to form a temperature abnormality clustering region. When the absolute difference between the highest and lowest temperatures within the region exceeds a second threshold or the highest temperature in the region exceeds a preset safety threshold, it is determined to be a suspected hot spot region and its spatial boundary coordinates are recorded. It should be noted that the typical values ​​for the first threshold are 10℃, the second threshold is 20℃, and the safety threshold is 100℃, all of which can be dynamically adjusted based on real-time environmental parameters. Mesh partitioning can also employ an equal division method, except for cell boundaries. For example, each cell region can be subdivided into 2×2 or 3×3 sub-mesh units to improve the accuracy of small-sized hot spots, or a non-uniform mesh can be customized based on the statistical characteristics of the actual failure distribution of the components. In addition to being based on direct adjacency, the region clustering strategy can also optimize the definition range of adjacent units by combining the series and parallel topology of the components. All algorithms can be executed in real-time on the embedded processor during implementation. For example, by collecting the current irradiance, ambient temperature, and component operating current through the embedded processor, when the irradiance is below 500W / m², the first threshold is automatically lowered to 8℃ and the second threshold to 15℃. When the ambient temperature exceeds 40℃, the threshold temperature rise compensation mechanism is activated. The region clustering strategy optimizes the definition of adjacency relationships based on the component topology. For example, for series-connected cell strings, adjacent units within the same string are compared first, while for parallel structures, both horizontal and vertical adjacent units are compared simultaneously.

[0088] Understandably, this embodiment uses the physical structure alignment design of the cell-level grid units to closely link temperature difference analysis with the electrical transmission characteristics of the photovoltaic module; it employs a region growing clustering algorithm to efficiently locate continuous abnormal regions, avoiding misjudgment of discrete points; a dual judgment mechanism combining absolute temperature difference and safe temperature rise threshold significantly improves the sensitivity and reliability of hot spot identification; the dynamic threshold adjustment function adapts to the needs of complex environmental changes, ensuring detection robustness under different operating conditions; and the embedded real-time processing capability meets the online analysis requirements of the equipment, providing effective target positioning for subsequent accurate re-inspection.

[0089] Based on the above technical solutions, those skilled in the art can make corresponding equivalent improvements according to the application scenario. For example, the grid cell division method can be adjusted to non-uniform partitioning, and the differentiated grid density can be dynamically generated according to the historical hot spot distribution probability of the component, and the grid points can be automatically densified in high-risk areas; or the temperature field spatial clustering algorithm can be extended, and the maximum connected component identification algorithm based on graph theory can be used to replace the region growing method, and cross-regional anomaly correlation analysis can be achieved by constructing a topological map of cell temperature differences; or a lightweight convolutional neural network model can be introduced to perform end-to-end hot spot region semantic segmentation on the real-time two-dimensional temperature distribution map, and directly output the boundary of the abnormal region and the confidence score; for large-scale photovoltaic power station scenarios, the initial screening clustering algorithm can be run on the edge computing unit to identify suspected hot spot regions, and the temperature distribution data can be uploaded to the cloud simultaneously to perform complex model verification; in micro-component or mobile detection platform scenarios, the grid cell definition can be simplified, and the adjacent temperature difference comparison can be performed directly with a single cell as the smallest analysis unit to improve the processing efficiency of the embedded system.

[0090] Preferably, the first threshold and the second threshold are dynamically set through the following steps:

[0091] Step S231: Real-time acquisition of the operating electrical parameters and environmental parameters of the target photovoltaic module, wherein the operating electrical parameters include operating current and operating voltage, and the environmental parameters include irradiance and ambient temperature;

[0092] Step S232: Generate electrical parameter correction coefficients based on the deviation between the operating electrical parameters and the rated electrical parameters;

[0093] Step S233: Determine the temperature reference value based on the matching result between the ambient temperature and the historical temperature database, and generate an environmental correction coefficient by combining it with the real-time irradiance;

[0094] Step S234: Match a benchmark threshold from a preset database based on the target photovoltaic module model, and dynamically adjust the threshold by overlaying the electrical parameter correction coefficient and the environmental correction coefficient to generate a first threshold.

[0095] Step S235: Based on the first threshold and the preset grade difference ratio, generate a second threshold that is greater than the first threshold.

[0096] For details, see Figure 4 In a specific embodiment of the present invention, the system monitors the output current and voltage data of the photovoltaic module in real time, and simultaneously reads the values ​​of the environmental irradiance sensor and the atmospheric temperature sensor. When the operating current is lower than 15% of the rated value, an electrical parameter correction coefficient K1=1.2 is generated; when the voltage fluctuation exceeds ±5%, K1=1.1 is generated. Based on the current ambient temperature, the system queries the historical database of the same season to match the temperature benchmark value of similar operating conditions, and generates an environmental correction coefficient K2=0.8 when the real-time irradiance is lower than 800W / m². The first threshold for matching the module model is 10℃, and after correction calculation, the first threshold is adjusted to 10×K1×K2=8.4℃. The second threshold is adjusted to 8.4×1.8≈15℃ according to the preset 1.8-fold difference ratio. The safety threshold is fixed at 100℃. It should be noted that the dynamic threshold setting step is automatically executed by the embedded processor throughout the process, the database supports real-time expansion and updates, the difference ratio can be adaptively adjusted according to the degree of module attenuation, and the benchmark threshold can be optimized and iterated with the actual operating data of the power station.

[0097] Understandably, this embodiment overcomes the detection blind spots of low-irradiation and aging components by correcting the threshold through multi-dimensional parameter fusion; the environmental temperature benchmark matching based on the historical database can improve the detection robustness during dawn and dusk; the dynamic correction mechanism of electrical parameters can effectively identify abnormal operating conditions of components and avoid misjudgment of temperature difference caused by sudden current drop; the grade difference ratio can be adapted according to the component's lifespan decay characteristics, and global protection is provided through safety thresholds to ensure zero omission of high-risk hot spots.

[0098] Based on the above technical solutions, those skilled in the art can make corresponding equivalent improvements according to the application scenario. For example, the linear superposition calculation of electrical parameter correction coefficient and environmental correction coefficient can be changed to a nonlinear combination algorithm, and a comprehensive correction factor can be generated by fusing current decay trend and irradiance change rate through a polynomial fitting model; or an end-to-end deep neural network model can be used to replace the traditional calculation steps, inputting real-time current voltage, irradiance, ambient temperature and historical database features, and directly outputting dynamically adjusted first and second thresholds; in newly built power plants lacking historical databases, the standard deviation of adjacent component temperatures can be used as a substitute calculation basis for temperature benchmark values; in scenarios where the component temperature gradient distribution pattern is obvious, the grade difference ratio can be set as a function of the regional temperature difference median to achieve adaptive adjustment.

[0099] Preferably, in step S220, the step of calculating the absolute value of the temperature difference between each grid cell and its adjacent grid cells includes:

[0100] Step S221: Determine the set of adjacent grid cells of the current grid cell based on the physical structure of the target photovoltaic module cell;

[0101] Step S222: Based on the preset temperature feature extraction rules, generate the representative temperature value of the current mesh cell;

[0102] Step S223: Calculate the absolute value of the difference between the temperature value represented by the current grid cell and the temperature value represented by each adjacent grid cell;

[0103] Step S224: Select the maximum value among the absolute values ​​of the differences as the absolute value of the temperature difference between the current grid cell and the adjacent grid cell.

[0104] For details, see Figure 5 In a specific embodiment of the present invention, the temperature difference calculation process is illustrated using a standard photovoltaic module with 60 solar cells and a single row of 18 thermocouples as an example: The grid cells are divided based on individual solar cells. The representative temperature value of the current solar cell is the arithmetic mean of nine temperature points obtained from three independent sampling data points completed by the three thermocouples covering the cell during their dwell time. The set of adjacent grid cells is defined according to the module's electrical topology as cells directly electrically connected to the current solar cell, including the previous and next solar cells in the same column of the same cell string, and the previous and next solar cells in adjacent cell strings in the same row. During calculation, the representative temperature values ​​of all adjacent cells of the current solar cell are first determined, and then the absolute value of the difference between the current cell temperature and the temperature of each adjacent cell is calculated. The maximum value is selected as the final absolute temperature difference. For example, the adjacent cells of the solar cell in the 3rd row, 2nd column are defined as the solar cells in the 2nd row, 2nd column, 4th row, 1st row, and 3rd row, 3rd column. The temperature difference calculation takes the maximum value of the temperature difference between the current cell temperature and the temperatures of these four adjacent cells. It should be noted that the definition of adjacency can be extended to diagonal solar cells, as well as other units that have a direct electrical connection or spatial thermal conduction relationship with the current solar cell.

[0105] Understandably, this embodiment can effectively capture local abnormal temperature rise characteristics through focused analysis of representative temperature values ​​at the cell level; avoid the influence of weak interference from the neighborhood through the maximum temperature difference screening mechanism, and improve the sensitivity of strong hot spot identification; adapt to non-uniform temperature field scenarios through representative value extraction rules, and improve scalability; and meet the needs of multi-scale spatial correlation analysis through the definition of adjacent relationships. Based on the above technical solutions, those skilled in the art can make corresponding equivalent improvements according to the application scenario. For example, the definition of adjacent units can be adjusted to be based on the composite relationship of electrical path and heat conduction path, and a dynamic adjacent weight coefficient matrix can be generated through a finite element analysis model; or a transfer learning mechanism can be introduced to train a temperature propagation feature model based on historical hot spot data to generate a customized set of adjacent units for cells in different locations; in the micro-module scenario, it can be simplified to the principle of spatial distance priority, and only spatially adjacent units in the same row and column can be calculated; for photovoltaic building integrated modules with severe backsheet heat accumulation, it can be extended to diagonal units and given higher weights; for special scenarios with abnormally uniform heat field distribution, the calculation of adjacent temperature difference can be skipped and the absolute temperature exceeding the threshold area can be directly detected; or a graph neural network can be used to model the temperature correlation of cells, and the optimal adjacent relationship judgment rule can be automatically learned through node embedding features.

[0106] Preferably, in step S222, the temperature feature extraction rule is set to extract representative temperature values ​​from the grid cell temperature data. The implementation methods include extracting the highest temperature value of all temperature sampling points, or calculating the arithmetic mean temperature value of all temperature sampling points, or calculating the weighted average temperature value based on the location weight.

[0107] Specifically, in one embodiment of the present invention, taking the aforementioned standard photovoltaic module with 60 solar cells and a single row of 18 thermocouples as an example, three methods for calculating representative temperature values ​​are explained. For the solar cell unit in the 3rd row and 2nd column, 9 temperature sampling points are obtained: 32.1℃, 31.8℃, 32.3℃, 33.5℃, 33.9℃, 33.2℃, 32.7℃, 34.1℃, and 33.6℃. When using the highest temperature value rule, the maximum value of 34.1℃ among the sampling points is directly extracted as the representative value; when using the arithmetic mean rule, the sum of the 9 temperature values ​​(287.2℃) is divided by 9 to obtain the average value (31.91℃), which is rounded to 31.9℃ as the representative temperature value; under the weighted average temperature value calculation method based on position weight, the system assigns a weight of 1.2 to the three sampling points in the central area (33.5℃, 33.9℃, and 34.1℃), and the remaining edge points... The six sampling points in the edge region are assigned a weight of 0.8. The calculation process is (33.5×1.2+33.9×1.2+34.1×1.2)+(32.1×0.8+31.8×0.8+32.3×0.8+33.2×0.8+32.7×0.8+33.6×0.8)=278.36. Dividing this by the total weight (3×1.2+6×0.8=8.4) yields 33.14℃ as the final representative temperature value. It should be noted that the system can automatically switch or combine the above temperature extraction rules according to the component status. The highest temperature value extraction method is suitable for the detection of old components with local strong hot spots on the surface. The arithmetic mean temperature value extraction method is suitable for the routine detection of new components with uniform temperature distribution. The position weighted average temperature extraction method is suitable for the detection of specific components with uneven temperature distribution.

[0108] Understandably, this embodiment provides differentiated analysis strategies for different types of components through a multi-mode temperature feature extraction mechanism; maximum value extraction ensures sensitive capture of strong hot spots; arithmetic mean ensures the stability of conventional detection; position weighting optimizes the accuracy of local hot spot identification; adaptive switching mechanism significantly improves the system's compatibility with different operating conditions; and weight allocation rules can be dynamically adjusted according to component attenuation characteristics to continuously optimize detection effectiveness.

[0109] Based on the above technical solutions, those skilled in the art can make corresponding equivalent improvements according to the application scenario. For example, they can add a backsheet temperature weighting algorithm to bifacial power generation modules and optimize the weight allocation rules; or introduce a Gaussian mixture model to automatically identify the cluster center point of temperature distribution features; or define the position weight as an inverse proportional function of the distance from the center of the battery cell; or use a convolutional neural network to directly extract feature vectors from the temperature distribution map to replace manual rules.

[0110] Preferably, in step S300, the step of performing a high-precision temperature scan through the hot spot diagnostic module to confirm the location and extent of the hot spot includes:

[0111] Step S310: Control the cleaning robot to move to the suspected hot spot area, wherein the hot spot diagnosis module includes an infrared temperature sensor;

[0112] Step S320: Perform a high-precision temperature scan using the infrared temperature sensor, input the temperature field distribution data obtained from the scan into a pre-trained hot spot recognition neural network model, and generate a hot spot probability distribution map;

[0113] Step S330: Integrate the physical structure information of the photovoltaic module cells, perform boundary optimization processing on the hot spot probability distribution map, and output the geometric center coordinates and physical boundary polygon of the hot spot.

[0114] For details, see Figure 6 In a specific embodiment of the present invention, after the cleaning robot returns to the center point of the suspected hot spot area, the drive motor controls the hot spot diagnosis module to move along the guide rail to the target position; a high-precision infrared temperature sensor is activated to scan the target area at millimeter-level resolution to obtain temperature field distribution data; during the scanning process, high-density dot matrix temperature measurement is performed along the horizontal and vertical sides of the component surface, and the collected temperature field distribution data is input into the pre-loaded hot spot recognition neural network model in the embedded processor; the model adopts a lightweight convolutional neural network architecture, with the original temperature matrix as input and the hot spot probability distribution map of the same resolution as output; subsequently, the processor loads the physical structure parameters of the component's battery cells, and performs morphological boundary optimization on the hot spot probability distribution map according to the boundary position of the battery cells: first, an erosion algorithm is applied to eliminate discrete noise points, then an expansion algorithm is used to connect adjacent hot spot areas, and finally, the hot spot boundary polygon is corrected along the edge of the battery cells; the output result is the coordinates of the geometric center point of the hot spot area and the minimum bounding polygon composed of a multi-vertex coordinate chain; this polygon is strictly aligned with the physical boundary of the battery cells to ensure that subsequent diagnostic operations accurately locate the target area. It should be noted that the neural network model is trained based on historical hot spot temperature field data and supports receiving cloud model updates through the system message center; the boundary optimization process can match different sized cell structure templates according to the component model.

[0115] Understandably, this embodiment achieves high-precision hot spot localization through the collaboration of high-resolution infrared scanning and lightweight neural networks; it eliminates sensor noise interference through a boundary optimization algorithm based on the cell structure; it ensures that the hot spot contour conforms to the electrical structure characteristics of the component through the output of boundary polygons with strong physical constraints, improving the accuracy of subsequent operations; and it meets the real-time analysis needs on-site through an embedded processing architecture, providing a reliable spatial benchmark for subsequent dirt detection and defect judgment. Based on the above technical solutions, those skilled in the art can make corresponding equivalent improvements according to the application scenario, such as: replacing the neural network model with a support vector machine decision model to improve adaptability to small sample scenarios; or using a random forest algorithm to fuse visible light images and temperature features to generate a composite hot spot probability map; or adding a curved surface projection coordinate transformation module for curved surface components; or simplifying it to fixed threshold segmentation instead of neural network inference in microprocessor scenarios; or optimizing the constraints of the hot spot boundary diffusion path based on the serial-parallel relationship of the components; or adapting the hot spot pattern recognition of new components through transfer learning technology; or introducing an attention mechanism to enhance the neural network's ability to capture small hot spots; or integrating a hot spot diffusion trend prediction model into the boundary optimization.

[0116] Preferably, in step S400, the step of performing surface state analysis on the confirmed hot spot area and generating an adaptive cleaning command based on the analysis results includes:

[0117] Step S410: Measure the hot spot area using the surface analysis module to obtain surface ash distribution data;

[0118] Step S420: Calculate a comprehensive score for the degree of dirtiness based on the surface ash distribution data and the historical cleaning records of the hot spot area;

[0119] Step S430: When the overall dirt level score exceeds the dirt threshold, it is determined to be a dirt-induced hot spot, and the deep cleaning process is started; when the overall dirt level score does not exceed the dirt threshold, a defect score is generated by combining the temperature distribution data of the hot spot area; if the defect score exceeds the defect threshold, a component defect alarm command is generated.

[0120] For details, see Figure 7In a specific embodiment of the present invention, the surface analysis module employs two implementation methods: a laser dust sensor or a visible light image analysis system. The first method uses a laser dust sensor to emit a near-infrared beam towards the hot spot area, calculating the micron-level dust coverage density based on reflection intensity and scattering characteristics, and generating a surface dust concentration distribution cloud map. The second method uses a regular visible light camera to capture high-definition images of the hot spot area, analyzing the image's grayscale distribution and texture features based on a pre-trained dust accumulation recognition model, and outputting the percentage of dirty coverage area and dust thickness level. The system integrates real-time measured dust accumulation data and historical cleaning records of the area's residual dirt trend parameters, calculating a comprehensive dirt level score using a weighted formula. When the score exceeds a dynamic dirt threshold, a pressurized cleaning command is triggered, automatically increasing brush pressure and cleaning frequency to perform deep cleaning. If the score does not exceed the threshold but the temperature in the area remains abnormal, a defect score is generated based on the temperature difference gradient between adjacent grids, and a component defect alarm is uploaded for areas exceeding the defect threshold. It should be noted that the dirt threshold is set to a default value based on the component installation environment and can be adaptively optimized. The selection of the surface analysis module type does not affect the subsequent cleaning decision process, and the visible light image analysis solution can achieve real-time processing by deploying a lightweight model through the edge computing unit.

[0121] Understandably, this embodiment adapts to different cost and accuracy requirements through a dual-mode surface state analysis mechanism. The laser graying solution provides micron-level precision detection, while the visible light imaging solution utilizes existing hardware to reduce system costs. A dirt scoring model based on historical data fusion can dynamically optimize cleaning decision thresholds, improving system adaptability. A defect scoring mechanism effectively isolates the causes of dirt-type and defect-type hot spots, improving accuracy. Adaptive cleaning parameter adjustment enables precise resource allocation, avoiding the resource waste of traditional fixed-cycle cleaning. Modular design supports seamless switching between multiple detection technologies, enhancing system deployment flexibility. Based on the above technical solutions, those skilled in the art can make corresponding equivalent improvements according to application scenarios. For example: using multispectral imaging technology to simultaneously acquire reflectivity and thermal radiation data to generate a composite dirt score; or deploying the image analysis module on a cloud server for asynchronous processing in UAV inspection scenarios; or combining acoustic resonance detection technology to quantify the degree of surface adhesion-type dirt; or introducing a reinforcement learning-based cleaning decision model to replace fixed scoring rules; or adapting differentiated cleaning strategies through dirt type identification; or adding a humidity compensation coefficient to correct the dirt score calculation model for high-humidity environments.

[0122] Preferably, in step S430, the deep cleaning process includes the following steps:

[0123] Step S431: Determine the initial cleaning intensity level and the maximum allowable number of cleaning cycles based on the comprehensive score of the degree of dirtiness;

[0124] Step S432: Control the cleaning robot to perform a single-wheel cleaning operation;

[0125] Step S433: Immediately after cleaning, detect the current dust concentration and calculate the dust concentration change rate for this round;

[0126] Step S434: When the rate of change of ash concentration is greater than the attenuation threshold, the cleaning intensity level is reduced; when the rate of change of ash concentration is less than the minimum effective threshold, the cleaning intensity level and the number of cleaning cycles are increased.

[0127] Step S435: Determine whether the current dust concentration meets the standard. If it does, terminate the process. Determine whether the cumulative number of rounds exceeds the maximum allowable number of rounds. If it does not exceed the limit, return to step S432 to perform a new round of cleaning. If it exceeds the limit, mark the stubborn dirt and terminate the process.

[0128] For details, see Figure 8 In a specific embodiment of the present invention, the system sets initial cleaning parameters based on a comprehensive score of the degree of dirtiness. The following is an example of such a setting: When the score is 75 points, the initial cleaning intensity is set to 130% of the standard pressure, and the maximum allowed number of cleaning cycles is set to 4. After the first round of cleaning, the detected change rate of the accumulated ash concentration is 40%, which is greater than the preset attenuation threshold of 30%, so the cleaning intensity level is reduced to 120%. After the second round of cleaning, the change rate is 18%, which is lower than the minimum effective threshold of 25%. At this time, the cleaning intensity is simultaneously increased to 125%, and the maximum number of cleaning cycles is increased by one to 5. After the third round of cleaning, the concentration still does not meet the standard, but the cumulative number of cycles is 3, which is less than the maximum of 5, so the system returns to execute the fourth round of cleaning. After the fourth round, the concentration is retested and meets the standard, so the cleaning process is terminated. If the concentration still does not meet the standard after the fifth round, since the cumulative number of cycles has reached the maximum number of cleaning cycles (5), the system marks the current hot spot as stubborn dirt and uploads alarm information. It should be noted that the number of cleaning cycles is not unlimited. Generally, an upper limit is set based on the actual operating conditions of the photovoltaic modules. In this embodiment, after comprehensively considering the mechanical fatigue of the equipment and the efficiency of operation and maintenance, the upper limit is set to 5 cycles.

[0129] Understandably, this embodiment significantly improves deep cleaning efficiency while ensuring component safety through a dual adjustment mechanism that reduces intensity to prevent damage and enhances strength to break stubborn stains; it meets the needs of tackling different types of stains through a flexible cycle expansion design; it effectively prevents equipment overload through a hard cycle upper limit constraint; and it achieves a dynamic balance between energy consumption, timeliness, and cleanliness by optimizing cleaning parameters in real time through feedback, ultimately minimizing operation and maintenance costs. Those skilled in the art can make corresponding equivalent improvements based on the above technical solution according to the application scenario. For example, the cycle upper limit can be set as a function of the working years of the battery cells, dynamically relaxing the limit range according to the aging degree of the components; or a deep reinforcement learning model can be used to dynamically generate cleaning parameter combinations to replace empirical formula settings; or a motor temperature rise compensation algorithm can be added in high-altitude, low-oxygen environments to adjust the cycle upper limit in real time.

[0130] See Figure 9 The second aspect of this invention provides a photovoltaic module hot spot diagnosis and adaptive cleaning system, comprising:

[0131] Rapid temperature measurement module: used to acquire real-time temperature distribution data on the surface of photovoltaic modules during the cleaning process of the cleaning robot;

[0132] Temperature difference anomaly detection module: Based on the temperature distribution data, identify suspected hot spot areas with temperature difference anomalies;

[0133] Hot spot recognition module: used to control the cleaning robot to move to the suspected hot spot area, and to perform high-precision temperature scanning through the hot spot diagnosis module to confirm the location and range of the hot spot;

[0134] Adaptive cleaning module: Performs surface condition analysis on the identified hot spot areas and generates adaptive cleaning instructions based on the analysis results.

[0135] A third aspect of the present invention also provides a storage medium storing a photovoltaic module hot spot diagnosis and adaptive cleaning program, wherein when the photovoltaic module hot spot diagnosis and adaptive cleaning program is executed by a processor, the photovoltaic module hot spot diagnosis and adaptive cleaning steps as described in any of the above embodiments are implemented.

[0136] See Figure 10 The fourth aspect of the present invention provides a computer device, including a memory, a processor, and a computer program stored in the memory and executable on the processor, characterized in that, when the processor executes the computer program, it implements the photovoltaic module hot spot diagnosis and adaptive cleaning method as described in any embodiment of the first aspect.

[0137] The processor can 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. The general-purpose processor can be a microprocessor or any conventional processor. This processor is the control center for the photovoltaic module hot spot diagnosis and adaptive cleaning system, connecting various parts of the operational device through various interfaces and lines.

[0138] The memory can be used to store the computer programs and / or modules. The processor, by running or executing the computer programs and / or modules stored in the memory and calling the data stored in the memory, realizes various functions of the photovoltaic module hot spot diagnosis and adaptive cleaning system. The memory may mainly include a program storage area and a data storage area. The program storage area may store the operating system, at least one application program required for a function (such as sound playback function, image playback function, etc.), etc.; the data storage area may store data created based on the use of the mobile phone (such as audio data, phonebook, etc.). In addition, the memory may include high-speed random access memory, and may also include non-volatile memory, such as hard disk, memory, plug-in hard disk, smart media card (SMC), secure digital (SD) card, flash card, at least one disk storage device, flash memory device, or other volatile solid-state storage device.

[0139] Compared with the prior art, the beneficial effects of the present invention include at least the following:

[0140] The photovoltaic module hot spot diagnosis and adaptive cleaning method and system provided by this invention establishes a detection basis by simultaneously acquiring high-density temperature distribution data during the cleaning process; accurately captures suspected hot spot areas based on a dynamic temperature difference recognition algorithm; automatically completes high-precision hot spot positioning and re-inspection through the equipment return path; intelligently identifies the cause of hot spots by integrating surface ash and temperature gradient features; dynamically optimizes the cleaning command execution decision by combining a closed-loop feedback mechanism; realizes integrated cleaning and diagnosis functions for a single device to eliminate manual secondary verification; and ultimately significantly improves the timeliness of operation and maintenance response and reduces the misjudgment rate.

[0141] Furthermore, this invention achieves precise temperature measurement at the cell level by matching the spatial distribution of a temperature sensor array with the structure of the photovoltaic module. A non-contact acquisition mechanism avoids mechanical wear, and an active dwell strategy in the cell area ensures high-quality temperature data acquisition. Adaptive parameter design guarantees overall system compatibility. The invention divides the cell's physical structure into grid cells and calculates adjacent temperature differences based on electrical characteristics. Region growth clustering locates continuous abnormal regions, and a dual dynamic threshold determination mechanism enhances the sensitivity and anti-interference capability of hot spot detection. Thresholds are corrected in real-time using electrical and environmental parameters, and detection sensitivity is dynamically optimized based on historical temperature databases and environmental parameters, improving the robustness of hot spot detection and reducing the risk of omissions. Temperature feature extraction rules are adapted to non-uniform temperature field characteristics, and adjacent relationships are optimized using electrical topology. The system defines and filters maximum values ​​to enhance the ability to capture local abnormal temperature rises and the sensitivity of identifying strong hot spots. A multi-mode temperature feature extraction mechanism improves differentiated analysis capabilities; the maximum value method enhances the sensitivity of high-hot spot capture, the arithmetic mean method ensures the stability of routine detection, and the position-weighted method optimizes the accuracy of local hot spot location. High-precision infrared scanning and neural network recognition work together to improve sub-centimeter-level hot spot location capabilities, while boundary optimization algorithms constrained by the physical structure of the solar cells enhance contour accuracy and structural matching. A dual-mode analysis mechanism combining laser graying and visible light images, combined with historical data fusion, improves the accuracy of dirt scoring and enhances the ability to distinguish the causes of dirt and hot spot defects. A dynamic bidirectional adjustment mechanism for intensity and cycle, combined with cycle upper limit constraints, improves the efficiency and reliability of adaptive cleaning.

[0142] The embodiments described above are merely illustrative of several implementations of the present invention, and while the descriptions are specific and detailed, they should not be construed as limiting the scope of the present invention. It should be noted that those skilled in the art can make various modifications and improvements without departing from the concept of the present invention. Equivalent structural transformations made using the description and drawings of the present invention, or direct / indirect applications in other related technical fields, are all included within the scope of patent protection of the present invention. Therefore, the scope of protection of this patent should be determined by the appended claims.

[0143] It should be noted that the embodiments implemented on the photovoltaic module hot spot diagnosis and adaptive cleaning system side in this invention can be referenced in conjunction with the embodiments implemented on the photovoltaic module hot spot diagnosis and adaptive cleaning method side, and will not be described in detail in this invention.

[0144] The above are merely preferred embodiments of the present invention and do not limit the scope of the patent. Any equivalent structural or procedural transformations made based on the description and drawings of the present invention, or direct or indirect applications in other related technical fields, are similarly included within the scope of patent protection of the present invention.

Claims

1. A method for diagnosing and adaptively cleaning hot spots in photovoltaic modules, characterized in that, include: Step S100: During the cleaning process of the cleaning robot, the temperature distribution data of the photovoltaic module surface is acquired in real time through the rapid temperature measurement module, including: Step S110: When the cleaning robot moves along the photovoltaic module surface, the temperature data of multiple positions on the photovoltaic module surface is collected synchronously at a preset sampling frequency through the multi-temperature sensor array; Step S120: Based on the position distribution of the multi-temperature sensor array and the robot's real-time displacement data, the spatial coordinates of each temperature sampling point on the photovoltaic module surface are calculated; Step S130: The temperature values ​​corresponding to the spatial coordinates are integrated to generate a real-time two-dimensional temperature distribution map of the photovoltaic module surface; Step S200: Based on the temperature distribution data, identify suspected hot spot areas with abnormal temperature differences, including: Step S210: Based on the real-time two-dimensional temperature distribution map, divide the photovoltaic module cells into grid units according to their physical structure, with each grid unit containing at least one temperature sampling point; Step S220: Calculate the absolute value of the temperature difference between each grid unit and its adjacent grid units. If the difference exceeds a first threshold, mark the grid unit as a suspected abnormal unit; Step S230: Spatial clustering of adjacent suspected abnormal units to form continuous clustered regions; if the clustered region meets any of the following conditions, it is determined to be a suspected hot spot area: the absolute value of the maximum temperature difference in the region exceeds a second threshold, or the absolute value of the highest temperature in the region exceeds a safety threshold; the first and second thresholds are dynamically adjusted through the following steps. Setting the Status: Step S231: Real-time acquisition of the operating electrical parameters and environmental parameters of the target photovoltaic module, wherein the operating electrical parameters include operating current and operating voltage, and the environmental parameters include irradiance and ambient temperature; Step S232: Generation of electrical parameter correction coefficients based on the deviation between the operating electrical parameters and the rated electrical parameters; Step S233: Determination of a temperature benchmark value based on the matching result between the ambient temperature and the historical temperature database, and generation of an environmental correction coefficient in combination with the real-time irradiance; Step S234: Matching a benchmark threshold from a preset database according to the target photovoltaic module model, and dynamically adjusting by superimposing the electrical parameter correction coefficient and the environmental correction coefficient to generate a first threshold; Step S235: Generation of a second threshold greater than the first threshold based on the first threshold and a preset differential ratio relationship; Step S300: Control the cleaning robot to move to the suspected hot spot area, and perform a high-precision temperature scan through the hot spot diagnosis module to confirm the location and range of the hot spot; Step S300, the step of performing a high-precision temperature scan through the hot spot diagnosis module to confirm the location and range of the hot spot includes: Step S310: Control the cleaning robot to move to the suspected hot spot area, wherein the hot spot diagnosis module includes an infrared temperature sensor; Step S320: Perform a high-precision temperature scan through the infrared temperature sensor, input the temperature field distribution data obtained by the scan into a pre-trained hot spot recognition neural network model, and generate a hot spot probability distribution map; Step S330: Integrate the physical structure information of the photovoltaic module cells, perform boundary optimization processing on the hot spot probability distribution map, and output the geometric center coordinates and physical boundary polygon of the hot spot; Step S400: Perform surface condition analysis on the confirmed hot spot area and generate an adaptive cleaning command based on the analysis results, including: Step S410: Measure the hot spot area using the surface analysis module to obtain surface dust distribution data; Step S420: Calculate a comprehensive dirt level score based on the surface dust distribution data and the historical cleaning records of the hot spot area; Step S430: When the comprehensive dirt level score exceeds a dirt threshold, it is determined to be a dirt-induced hot spot, and a deep cleaning process is initiated; when the comprehensive dirt level score does not exceed the dirt threshold, a defect score is generated by combining the temperature distribution data of the hot spot area; if the defect score exceeds a defect threshold, a component defect alarm command is generated; In Step S430, the deep cleaning process includes... The following steps are as follows: Step S431: Determine the initial cleaning intensity level and the maximum allowable number of rounds based on the comprehensive score of the degree of dirt; Step S432: Control the cleaning robot to perform a single round of cleaning operation; Step S433: Immediately after cleaning, detect the current dust concentration and calculate the dust concentration change rate of this round; Step S434: When the dust concentration change rate is greater than the attenuation threshold, reduce the cleaning intensity level; when the dust concentration change rate is less than the minimum effective threshold, increase the cleaning intensity level and the number of cleaning rounds; Step S435: Determine whether the current dust concentration meets the standard. If it does, terminate the process; determine whether the cumulative number of rounds exceeds the maximum allowable number of rounds. If it does not exceed the limit, return to step S432 to perform a new round of cleaning. If it exceeds the limit, mark stubborn dirt and terminate the process.

2. The photovoltaic module hot spot diagnosis and adaptive cleaning method as described in claim 1, characterized in that, In step S220, the step of calculating the absolute value of the temperature difference between each grid cell and its adjacent grid cells includes: Step S221: Determine the set of adjacent grid cells of the current grid cell based on the physical structure of the target photovoltaic module cell; Step S222: Based on the preset temperature feature extraction rules, generate the representative temperature value of the current mesh cell; Step S223: Calculate the absolute value of the difference between the temperature value represented by the current grid cell and the temperature value represented by each adjacent grid cell; Step S224: Select the maximum value among the absolute values ​​of the differences as the absolute value of the temperature difference between the current grid cell and the adjacent grid cell.

3. The photovoltaic module hot spot diagnosis and adaptive cleaning method as described in claim 2, characterized in that, In step S222, the temperature feature extraction rule is set to extract representative temperature values ​​from the grid cell temperature data. The implementation methods include extracting the highest temperature value of all temperature sampling points, or calculating the arithmetic mean temperature value of all temperature sampling points, or calculating the weighted average temperature value based on the location weight.

4. A photovoltaic module hot spot diagnosis and adaptive cleaning system, characterized in that, include: Rapid temperature measurement module: used to acquire real-time temperature distribution data of the photovoltaic module surface during the cleaning process of the cleaning robot, including: Step S110: when the cleaning robot moves along the photovoltaic module surface, the temperature data of multiple positions on the photovoltaic module surface are collected synchronously at a preset sampling frequency through the on-board multi-temperature sensor array; Step S120: according to the position distribution of the multi-temperature sensor array and the real-time displacement data of the robot, the spatial coordinates of each temperature sampling point on the photovoltaic module surface are calculated; Step S130: the temperature values ​​corresponding to the spatial coordinates are integrated to generate a real-time two-dimensional temperature distribution map of the photovoltaic module surface; Temperature difference anomaly detection module: Based on the temperature distribution data, it identifies suspected hot spot areas with temperature difference anomalies, including: Step S210: Based on the real-time two-dimensional temperature distribution map, it divides the photovoltaic module cells into grid units according to their physical structure, with each grid unit containing at least one temperature sampling point; Step S220: It calculates the absolute value of the temperature difference between each grid unit and its adjacent grid units. If the difference exceeds a first threshold, the grid unit is marked as a suspected anomaly unit; Step S230: It performs spatial clustering on adjacent suspected anomaly units to form continuous clustered regions; if the clustered region meets any of the following conditions, it is determined to be a suspected hot spot region: the absolute value of the maximum temperature difference in the region exceeds a second threshold, or the absolute value of the highest temperature in the region exceeds a safety threshold; the first and second thresholds are determined through the following steps. Dynamic Setting: Step S231: Real-time acquisition of operating electrical parameters and environmental parameters of the target photovoltaic module, including operating current and operating voltage, and environmental parameters including irradiance and ambient temperature; Step S232: Generation of electrical parameter correction coefficients based on the deviation between the operating electrical parameters and the rated electrical parameters; Step S233: Determination of a temperature benchmark value based on the matching result between the ambient temperature and the historical temperature database, and generation of an environmental correction coefficient based on the real-time irradiance; Step S234: Matching a benchmark threshold from a preset database based on the target photovoltaic module model, and dynamically adjusting by superimposing the electrical parameter correction coefficient and the environmental correction coefficient to generate a first threshold; Step S235: Generation of a second threshold greater than the first threshold based on the first threshold and a preset differential ratio relationship; Hot spot recognition module: Used to control the cleaning robot to move to the suspected hot spot area, and to perform a high-precision temperature scan through the hot spot diagnosis module to confirm the location and range of the hot spot, including: Step S310: Control the cleaning robot to move to the suspected hot spot area, wherein the hot spot diagnosis module includes an infrared temperature sensor; Step S320: Perform a high-precision temperature scan through the infrared temperature sensor, input the temperature field distribution data obtained by the scan into a pre-trained hot spot recognition neural network model, and generate a hot spot probability distribution map; Step S330: Integrate the physical structure information of the photovoltaic module cells, perform boundary optimization processing on the hot spot probability distribution map, and output the geometric center coordinates and physical boundary polygon of the hot spot; Adaptive Cleaning Module: Performs surface condition analysis on the confirmed hot spot area and generates an adaptive cleaning command based on the analysis results, including: Step S410: Measure the hot spot area using the surface analysis module to obtain surface dust distribution data; Step S420: Calculate a comprehensive dirt level score based on the surface dust distribution data and the historical cleaning records of the hot spot area; Step S430: When the comprehensive dirt level score exceeds a dirt threshold, it is determined to be a dirt-induced hot spot, and a deep cleaning process is initiated; when the comprehensive dirt level score does not exceed the dirt threshold, a defect score is generated based on the temperature distribution data of the hot spot area; if the defect score exceeds a defect threshold, a component defect alarm command is generated; In step S430, the deep cleaning process includes... The following steps are as follows: Step S431: Determine the initial cleaning intensity level and the maximum allowable number of rounds based on the comprehensive score of the degree of dirt; Step S432: Control the cleaning robot to perform a single round of cleaning operation; Step S433: Immediately after cleaning, detect the current dust concentration and calculate the dust concentration change rate of this round; Step S434: When the dust concentration change rate is greater than the attenuation threshold, reduce the cleaning intensity level; when the dust concentration change rate is less than the minimum effective threshold, increase the cleaning intensity level and the number of cleaning rounds; Step S435: Determine whether the current dust concentration meets the standard. If it does, terminate the process; determine whether the cumulative number of rounds exceeds the maximum allowable number of rounds. If it does not exceed the limit, return to step S432 to perform a new round of cleaning. If it exceeds the limit, mark stubborn dirt and terminate the process.

Images