I-v curve-based system and method for detecting abnormality of photovoltaic module

Abstract

The present invention relates to an I-V curve-based system for detecting an abnormality of a photovoltaic module. More specifically, the I-V curve-based system for detecting an abnormality of a photovoltaic module may comprise: a collection unit for collecting, from an RTU, I-V curve data for each MPPT for a plurality of photovoltaic modules; a normalization unit for normalizing, for each MPPT, each piece of I-V curve data on the basis of maximum output I-V curve data of a corresponding MPPT; and an abnormality detection unit that determines whether the photovoltaic module is normal or abnormal for each MPPT by inputting the normalized I-V curve data to an image classification model that has learned whether a photovoltaic module is abnormal on the basis of normalized maximum output I-V curve data for each power plant or for each MPPT.

Description

IV-curve-based solar module anomaly detection system and method

[0001] The present invention relates to a system and method for detecting anomalies in solar modules based on an IV curve, and more particularly, to a system and method for detecting anomalies in solar modules based on an IV curve, which can detect anomalies in MPPT solar modules corresponding to abnormal IV curve data by normalizing IV curve data based on the maximum output of each MPPT and classifying the normalized IV curve data as normal or abnormal based on an image classification model.

[0002] Recently, solar power generation technology has been rapidly advancing, and fault detection technologies for solar modules and arrays are gaining attention. These technologies have the advantage of quickly detecting system faults and enabling timely maintenance, thereby enabling efficient management of solar power generation.

[0003] However, prior art primarily utilized a method of monitoring the voltage and current of a solar string comprised of multiple solar modules, monitoring the current and voltage values ​​flowing through the multiple solar modules, and simply comparing these values ​​to reference voltage and current ranges, and issuing an alert if the values ​​exceeded the range. Because this prior art made it very difficult to precisely detect module-specific errors, there is an urgent need for fault detection that can precisely diagnose the location of faulty modules.

[0004] (Patent Document 0001) Republic of Korea Patent Publication No. 10-2017-0028630

[0005] The problem to be solved by the present invention is to provide a system and method for detecting abnormalities in solar modules based on IV curves, which can detect abnormalities in solar modules of MPPTs corresponding to abnormal IV curve data by normalizing IV curve data based on the maximum output of each MPPT and classifying the normalized IV curve data as normal or abnormal based on an image classification model.

[0006] The tasks of the present invention are not limited to the tasks mentioned above, and other tasks not mentioned will be clearly understood by those skilled in the art of the present invention from the description below.

[0007] In order to solve the above-described problem, an IV-curve-based solar module anomaly detection system according to an embodiment of the present invention may include a collection unit that collects IV curve data for each MPPT for a plurality of solar modules from an RTU, a normalization unit that normalizes each IV curve data based on the maximum output IV curve data of each MPPT, and an anomaly detection unit that inputs the normalized IV curve data into an image classification model that has learned whether the solar module is abnormal based on the normalized maximum output IV curve data for each power plant or each MPPT, and determines whether the solar module is normal or abnormal for each MPPT.

[0008] In addition, the normalization unit can perform primary normalization to unitize the IV curve data for each MPPT of each power plant according to the number of solar modules included, and secondary normalization to convert the primary normalized IV curve data into a value of 0 to 1 based on the maximum output of the corresponding MPPT.

[0009] In addition, the normalization unit can perform the first normalization of the IV curve data according to the following mathematical formula.

[0010]

[0011]

[0012] Here, V1 is the first normalized voltage, I1 is the first normalized current, V org is the voltage measurement, I org is the current measurement value, Voc unit Silver open circuit voltage, Isc unit Short-circuit current, n sr is the number of series-connected modules of the corresponding MPPT, n pr can be the number of parallel-connected modules of the corresponding MPPT.

[0013] Additionally, the normalization step can perform secondary normalization on the IV curve data that has been first normalized according to the mathematical formula below.

[0014]

[0015]

[0016]

[0017]

[0018] Here, V2 can be the second normalized voltage, I2 can be the second normalized current, Voc,normal can be the open circuit voltage value of the maximum output IV curve data, and Isc,normal can be the short circuit current value of the maximum output IV curve data.

[0019] Additionally, the IV curve data may include power plant ID, inverter ID, measurement time, voltage measurement value, current measurement value, and number of series and parallel connections.

[0020] In addition, the above-described abnormality detection unit may include a learning unit that learns a model based on a convolutional neural network using images classified into multiple categories, and generates the image classification model by classifying and learning whether there is an abnormality in the solar module of the MPPT using an image of the maximum output IV curve and an image of an IV curve in which the output of the maximum output IV curve and the output of the preset ratio or more are different as the last layer, and a judgment unit that classifies the normalized IV curve data as normal or abnormal based on the image classification model.

[0021] Additionally, the image classification model can be trained based on MobileNet, ResNet50, or DenseNet121 using images classified into multiple categories.

[0022] In addition, the above maximum output IV curve data may be IV curve data having the maximum output value among the IV curve data accumulated in the collection unit.

[0023] In addition, the IV curve-based solar module abnormality detection method according to one embodiment of the present invention may include a step of collecting IV curve data for each MPPT for a plurality of solar modules, a step of normalizing each IV curve data based on the maximum output IV curve data of each MPPT, and a step of inputting the normalized IV curve data into an image classification model that has learned whether the solar module is abnormal based on the normalized maximum output IV curve data for each power plant or each MPPT, thereby determining whether the solar module is normal or abnormal for each MPPT.

[0024] In addition, the normalizing step may perform primary normalization to unitize the IV curve data for each MPPT of each power plant according to the number of solar modules included, and secondary normalization to convert the primary normalized IV curve data into a value of 0 to 1 based on the maximum output of the corresponding MPPT.

[0025] In addition, the above normalizing step can first normalize the IV curve data according to the following mathematical formula.

[0026]

[0027]

[0028] Here, V1 is the first normalized voltage, I1 is the first normalized current, V or is the voltage measurement, I or is the current measurement value, Voc unit Open circuit voltage, Isc unit is the short-circuit current, n sr is the number of series-connected modules of the corresponding MPPT, n pr can be the number of parallel-connected modules of the corresponding MPPT.

[0029] Additionally, the normalization step can perform secondary normalization on the IV curve data that has been first normalized according to the mathematical formula below.

[0030]

[0031]

[0032]

[0033]

[0034] Here, V2 can be the second normalized voltage, I2 can be the second normalized current, Voc,normal can be the open circuit voltage value of the maximum output IV curve data, and Isc,normal can be the short circuit current value of the maximum output IV curve data.

[0035] Additionally, the IV curve data may include power plant ID, inverter ID, measurement time, voltage measurement value, current measurement value, and number of series and parallel connections.

[0036] In addition, a model is trained based on a convolutional neural network using images classified into multiple categories, and a step of generating the image classification model by classifying and learning whether there is an abnormality in the solar module of the MPPT using an image of the maximum output IV curve and an image of an IV curve in which the output of the maximum output IV curve and the output of the preset ratio are different as the last layer may be further included.

[0037] In addition, the step of determining whether the above is normal or abnormal can classify the MPPT solar module corresponding to the normalized IV curve data as normal or abnormal based on the image classification model.

[0038] In addition, the step of generating the image classification model can train a model based on MobileNet, ResNet50, or DenseNet121 using images classified into multiple categories.

[0039] In addition, the above maximum output IV curve data may be IV curve data having the maximum output value among the IV curve data accumulated in the collection unit.

[0040] Specific details of other embodiments are included in the detailed description and drawings.

[0041] The IV curve-based solar module abnormality detection system and method according to one embodiment of the present invention can diagnose the location of a faulty module in a solar power plant from a remote location by utilizing an IV curve scan, thereby maximizing the efficiency of solar power generation and minimizing the scope of repair.

[0042] In particular, considering the IV curve that varies depending on environmental conditions (irradiance, temperature, etc.) even in the same MPPT, the IV curve can be unitized and maximum-minimum normalized, and solar module anomaly detection can be performed based on the normalized IV curve.

[0043] The effects according to the present invention are not limited to those exemplified above, and other effects can be clearly understood by those skilled in the art from the description of the following specification.

[0044] FIG. 1 is a diagram showing a schematic configuration of a solar module abnormality detection system according to one embodiment of the present invention.

[0045] Figure 2 is a block diagram showing a schematic configuration of the anomaly detection server of Figure 1.

[0046] Figures 3 and 4 are flowcharts for explaining IV curve data collection.

[0047] Figure 5 is a drawing for explaining the characteristics of a solar module.

[0048] Figure 6 shows an IV curve for each MPPT according to one embodiment of the present invention.

[0049] Fig. 7 is a block diagram showing a schematic configuration of the anomaly detection unit of Fig. 2.

[0050] Figure 8 is maximum output IV curve data (normal pattern) according to one embodiment of the present invention.

[0051] Figure 9 is abnormally classified IV curve data according to one embodiment of the present invention.

[0052] FIG. 10 is a diagram illustrating the accuracy of an anomaly detection unit (image classification model) according to one embodiment of the present invention.

[0053] Figure 11 is a flowchart for explaining a solar module abnormality detection method according to one embodiment of the present invention.

[0054] Figure 12 is a flowchart for explaining the IV curve data normalization (S200) of Figure 11.

[0055] *Explanation of key symbols in the drawing*

[0056] 50 solar module groups (MPPT groups)

[0057] 100: IV Curve Scanner

[0058] 200: Inverter

[0059] 300: RTU

[0060] 400: Network

[0061] 500: Anomaly Detection Server

[0062] 510: Control unit 520: Collection unit

[0063] 530: Normalization section 540: Anomaly detection section

[0064] 541: Learning Department 542: Judgment Department

[0065] The following merely exemplifies the principles of the invention. Therefore, those skilled in the art will be able to implement the principles of the invention and invent various devices within the scope and spirit of the invention, even if not explicitly described or illustrated herein. Furthermore, all conditional terms and embodiments listed herein are expressly intended, in principle, to facilitate understanding of the invention, and should be understood as being in no way limited to the specifically listed embodiments and conditions.

[0066] Additionally, in the following description, ordinal expressions such as first, second, etc. are intended to describe objects that are equal and independent of each other, and should be understood as having no meaning in terms of main / sub or master / slave.

[0067] The above-described purposes, features and advantages will become clearer through the following detailed description with reference to the attached drawings, so that a person having ordinary skill in the art to which the invention pertains can easily practice the technical idea of ​​the invention.

[0068] The individual features of the various embodiments of the present invention can be partially or wholly combined or combined with each other, and as can be fully understood by those skilled in the art, various technical connections and operations are possible, and each embodiment can be implemented independently of each other or can be implemented together in a related relationship.

[0069] Hereinafter, various embodiments of the present invention will be described in detail with reference to the attached drawings.

[0070] For example, a 100kW solar power plant currently in operation typically houses an average of 220 modules with an output of 450W or more. These modules are connected to the inverter via power lines called strings. Each string typically has 13 modules connected in series, resulting in an average of more than 10 strings per plant.

[0071] The average operating warranty period for solar modules is typically 25 years, but the average solar module installed as above is continuously losing efficiency, and in particular, as the warranty period is long, the power generation efficiency is decreasing due to partial shading caused by changes in surrounding structures or damage to the surface of the solar module.

[0072] In order to identify problems with partial shading and module surface damage caused by surrounding structures, the voltage and current of all strings connected to the inverter were measured through offline monthly inspections. Once a string with reduced voltage and current was identified, the module connected to that string was inspected to identify and resolve the problem.

[0073] However, due to the global spread of eco-friendly policies and the increase in solar power plants, it has become increasingly difficult to diagnose module fault locations by checking the voltage of all strings offline. Therefore, the inventors of the present invention aim to remotely diagnose solar module abnormalities and the location of anomaly detection modules using IV curve data.

[0074] Meanwhile, even if the IV curve data is derived from the same MPPT, the shape of the IV curve varies depending on temperature and environmental conditions (solar radiation, temperature, etc.). Therefore, in order to compare IV curve data, the present invention uses normalized IV curve data, and builds an image classification model that classifies normal patterns and abnormal patterns based on the curve image of the normalized IV curve data, so that the location of an error in a solar module corresponding to the abnormal pattern IV curve data can be detected and provided from a remote location.

[0075] FIG. 1 is a schematic diagram illustrating the configuration of a solar module anomaly detection system according to an embodiment of the present invention. FIG. 2 is a block diagram illustrating the schematic configuration of the anomaly detection server of FIG. 1.

[0076] Referring to FIGS. 1 and 2, the solar module anomaly detection system (hereinafter referred to as the system) of the present invention may include a plurality of solar module groups (hereinafter referred to as MPPT groups, 50) connected in series, an IV curve scanner (100), an inverter (200), an RTU (300), and an anomaly detection server (500), and the anomaly detection server (500) may include a control unit (510), a collection unit (520), a normalization unit (530), and an anomaly detection unit (540).

[0077] At this time, each component of the anomaly detection server (500) may be implemented as a separate server connected via a network (400), a software module of a single computing device, a hardware module, or a combination of software and hardware.

[0078] An IV curve scanner (100) is connected between each MPPT group and the inverter, and can scan the IV curve of each MPPT module. The IV curve is a curve representing the characteristics of a solar module, and for example, as shown in Fig. 6, it is a curve representing the change in current according to the change in voltage applied to the solar module. The IV curve scanner (100) can derive an IV curve for each MPPT (each MPPT group) based on the voltage-current applied to each MPPT group.

[0079] An inverter (200) is a device that converts variable direct current (DC) generated from a solar module into power system frequency alternating current (AC) so that it can be used in a commercial power grid or a local non-grid network, and can transmit, for example, design measurement values ​​and IV curve data to an RTU (300).

[0080] Meanwhile, the RTU (300, Remote Terminal Unit) can collect design measurement data collected from the inverter, measurement data from a weather sensor (not shown), and IV curve data and transmit them to the anomaly detection server (500) via a network (400).

[0081] At this time, the RTU (300) can transmit IV curve data including the power plant ID, inverter ID, measurement time, voltage measurement value, current measurement value, and the number of series / parallel connections.

[0082] The collection unit (520) can receive and store data collected from the RTU (300) and can be implemented as a collection server, for example.

[0083] For example, referring to FIG. 3, the RTU (300) can periodically request the inverter (200) for equipment measurement values ​​(e.g., current values ​​and / or voltage values ​​of a real-time solar module MPPT group) every 1 to 10 minutes, preferably every 5 minutes, and the inverter (200) transmits the equipment measurement values ​​to the RTU (300).

[0084] In this case, the RTU (300) sends a UDP transmission request to the communication unit (350), and when the communication unit (350) transmits the equipment measurement value to the collection unit (520) and receives an Ack response, it transmits an Ack response to the RTU (300) again to complete the transmission of the equipment measurement value.

[0085] Meanwhile, the control unit (510) can transmit an IV curve scan diagnosis request to the inverter, and can also collect IV curve data, transmit learning and abnormality detection commands to the abnormality detection unit (540), and can be implemented as, for example, a control server that performs overall control.

[0086] For example, referring to FIG. 4, the RTU (300) can receive equipment measurement values ​​from the inverter and transmit them to the control unit (600), and the control unit (600) can transmit a periodic or aperiodic IV curve scan request to the RTU (300), and at this time, the RTU (300) can transmit an IV curve scan command to the inverter (200). In this case, the inverter (200) can cause the IV curve scanner (100) to perform an IV curve scan, and after the IV curve scan, can transmit IV curve data to the RTU (300). At this time, the RTU (300) can transmit the IV curve data to the control unit (600).

[0087] IV curve data can be formed by the characteristics of a solar module as in Fig. 5. Fig. 12 (a) is an equivalent circuit of a solar module, and an IV curve pattern can be formed as in Fig. 12 (b), and the open circuit voltage (Voc), short circuit current (Isc), maximum voltage (Vmpp), maximum current (Impp), and maximum output (Pmpp) can be confirmed from the IV curve pattern.

[0088] The IV curve data collected as shown in FIGS. 3 and 4 can be illustrated as in FIG. 6, for example, and can be stored separately by MPPT, normalized by the normalization unit (530), and stored as normalized IV curve data in the collection unit (510). In addition, the anomaly detection unit (540) can detect the location of an abnormal solar module using an image classification model learned based on the normalized IV curve data by MPPT.

[0089] The normalization unit (530) can normalize each IV curve data based on the maximum output IV curve data of each MPPT. More specifically, the unit can perform primary normalization to convert the IV curve data of each MPPT of each power plant into units according to the number of solar modules included, and secondary normalization to convert the primary normalized IV curve data into values ​​between 0 and 1 based on the maximum output of the corresponding MPPT.

[0090] At this time, the first normalization can be performed using the following mathematical formula 1.

[0091] [Mathematical Formula 1]

[0092]

[0093]

[0094] Here, V1 is the first normalized voltage, I1 is the first normalized current, V org is the voltage measurement, I org is the current measurement value, Voc unit Silver open circuit voltage, Isc unit is the short-circuit current, n sr is the number of series-connected modules of the corresponding MPPT, n pr can be the number of parallel-connected modules of the corresponding MPPT.

[0095] Additionally, secondary normalization can be performed using the following mathematical expression 2.

[0096] [Equation 2]

[0097]

[0098]

[0099]

[0100]

[0101] Here, V2 can be the second normalized voltage, I2 can be the second normalized current, Voc,normal can be the open circuit voltage value of the maximum output IV curve data, and Isc,normal can be the short circuit current value of the maximum output IV curve data.

[0102] Here, the normalized IV curve data can form an IV curve pattern in which current and voltage have values ​​between 0 and 1, as in FIGS. 8 and 9.

[0103] The abnormality detection unit (540) inputs the IV curve data normalized by the normalization unit (530) into an image classification model that has learned whether a solar module is abnormal based on the normalized maximum output IV curve data for each power plant or MPPT, and can classify and determine whether the solar module is normal or abnormal for each MPPT.

[0104] Here, the maximum output IV curve data can be extracted for each MPPT and each power plant, and can be the IV curve data having the maximum output value among the IV curve data accumulated in the collection unit (520).

[0105] Fig. 7 is a block diagram showing a schematic configuration of the anomaly detection unit of Fig. 2.

[0106] Referring to FIG. 7, the anomaly detection unit (540) may include a learning unit (541) and a judgment unit (542).

[0107] The learning unit (541) can learn an image classification model based on transfer learning and fine-tuning techniques. Specifically, a model is learned based on a convolutional neural network using images classified into multiple categories, and the last layer is an image of a maximum output IV curve (maximum output IV curve pattern) and an image of an IV curve in which the output of the maximum output IV curve differs by a preset ratio (e.g., 30%) or more, thereby classifying and learning whether there is an abnormality in the solar module of the MPPT, thereby generating an image classification model.

[0108] At this time, you can train a model based on MobileNet, ResNet50, or DenseNet121 using images classified into multiple categories.

[0109] The judgment unit (542) can classify the IV curve data normalized by the normalization unit (530) as normal or abnormal based on the image classification model, and can determine and provide the error location of the solar module of the MPPT corresponding to the IV curve data classified as abnormal.

[0110] In addition, in one embodiment of the present invention, the image classification model is described as distinguishing between normal and abnormal, but it can also distinguish between types of errors by distinguishing types of IV curve patterns (normal patterns, voltage reduction patterns, current reduction patterns, module aging patterns, and partial shade patterns).

[0111] Fig. 8 is maximum output IV curve data (normal pattern) according to one embodiment of the present invention. Fig. 9 is abnormally classified IV curve data according to one embodiment of the present invention.

[0112] For example, Figure 8 represents the normalized maximum output IV curve patterns of each of the three power plants, which can serve as a reference normal pattern. Furthermore, Figure 9 represents an abnormal pattern in which the output differs by more than 30% compared to the normal pattern.

[0113] FIG. 10 is a diagram illustrating the accuracy of an anomaly detection unit (image classification model) according to one embodiment of the present invention.

[0114] Based on the maximum output IV curve data for each power plant, Gaussian noise was added to the data to multiply the data and learn the normal pattern. In this example, the image classification model was trained using the Pytorch - ResNet50 Pretrained Model.

[0115] As can be seen from Figure 10, 34 out of 34 abnormal patterns were classified as abnormal patterns, and 8 out of 8 normal patterns were classified as normal patterns, indicating a very high level of accuracy.

[0116] Fig. 11 is a flowchart illustrating a solar module anomaly detection method according to one embodiment of the present invention. Fig. 12 is a flowchart illustrating the IV curve data normalization (S200) of Fig. 11.

[0117] The solar module anomaly detection method of FIGS. 11 and 12 can be performed based on the solar module anomaly detection system of FIGS. 1 to 10. Accordingly, it can be described with reference to FIGS. 1 to 10, and a detailed description thereof will be omitted as it has been described above.

[0118] Referring to Fig. 11, IV curve data for each MPPT for each of multiple solar modules can be collected (S100).

[0119] Next, for each MPPT, each IV curve data can be normalized based on the maximum output IV curve data of the corresponding MPPT (S200). Specifically, as shown in FIG. 12, primary normalization can be performed to unitize the IV curve data for each MPPT per power plant according to the number of solar modules included (S210). At this time, primary normalization can be performed using the mathematical expression 1 described above.

[0120] In step S210, secondary normalization can be performed to convert the first-normalized IV curve data into a value between 0 and 1 based on the maximum output of the corresponding MPPT (S220). At this time, the secondary normalization can be performed using the mathematical expression 2 described above.

[0121] In addition, an image classification model that learns whether a solar module is abnormal can be created based on the normalized maximum output IV curve data for each power plant or MPPT among the normalized IV curve data.

[0122] In the present invention, an image classification model can be constructed based on transfer learning and fine-tuning techniques. Specifically, a model is trained based on a convolutional neural network using images classified into multiple categories, and the final layer is an image of a maximum output IV curve and an image of an IV curve whose output differs by a preset ratio or more from the maximum output IV curve, thereby classifying and learning whether or not an MPPT solar module is abnormal, thereby generating the image classification model.

[0123] Next, by inputting the normalized IV curve data from step S220 into the learned image classification model, the normality or abnormality of the solar module for each MPPT can be determined (S230). That is, based on the image classification model, the normalized IV curve data from step S200 can be classified as normal or abnormal, and the error location of the solar module of the MPPT corresponding to the abnormally classified IV curve data can be determined and provided.

[0124] In this way, the system of the present invention can precisely diagnose the location of faulty modules in a solar power plant, maximizing solar power generation efficiency and minimizing repair costs. Furthermore, by utilizing IV curve scanning in string inverters to which solar modules are directly connected, faults in solar panels can be quickly detected remotely.

[0125] In addition, considering the IV curve that varies depending on environmental conditions (irradiance, temperature, etc.) even in the same MPPT, the IV curve can be unitized and maximum-minimum normalized, and solar module abnormality detection can be performed based on the normalized IV curve.

[0126] Although the embodiments of the present invention have been described in more detail with reference to the attached drawings, the present invention is not necessarily limited to these embodiments, and various modifications may be implemented without departing from the technical spirit of the present invention. Therefore, the embodiments disclosed in the present invention are not intended to limit the technical spirit of the present invention, but to explain it, and the scope of the technical spirit of the present invention is not limited by these embodiments. Therefore, it should be understood that the embodiments described above are illustrative in all aspects and not restrictive. The protection scope of the present invention should be interpreted by the following claims, and all technical ideas within a scope equivalent thereto should be interpreted as being included in the scope of the rights of the present invention.

Claims

1. A collection unit that collects IV curve data for each MPPT for multiple solar modules from the RTU; For each MPPT, a normalization unit that normalizes each IV curve data based on the maximum output IV curve data of the corresponding MPPT; and An IV curve-based solar module abnormality detection system, comprising an abnormality detection unit that inputs the normalized IV curve data into an image classification model that has learned whether the solar module is abnormal based on the normalized maximum output IV curve data for each power plant or MPPT, and determines whether the solar module is normal or abnormal for each MPPT.

2. In paragraph 1, The above normalization part is, An IV curve-based solar module anomaly detection system that performs primary normalization to unitize the IV curve data of each power plant and each MPPT according to the number of solar modules included, and secondary normalization to convert the primary normalized IV curve data into a value of 0 to 1 based on the maximum output of the corresponding MPPT.

3. In paragraph 2, The above normalization part is, An anomaly detection system for a solar module based on an IV curve, which performs the first normalization of the IV curve data according to the mathematical formula below. Here, V1 is the first normalized voltage, I1 is the first normalized current, V org is the voltage measurement, I org is the current measurement value, Voc unit Silver open circuit voltage, Isc unit is the short-circuit current, n sr is the number of series-connected modules of the corresponding MPPT, n pr is the number of parallel-connected modules of the corresponding MPPT.

4. In paragraph 2, The above normalization part is, An anomaly detection system for solar modules based on the IV curve, which secondarily normalizes the first-normalized IV curve data according to the mathematical formula below. Here, V2 is the second normalized voltage, I2 is the second normalized current, Voc,normal is the open circuit voltage value of the maximum output IV curve data, and Isc,normal is the short circuit current value of the maximum output IV curve data.

5. In paragraph 1, The above IV curve data is an IV curve-based solar module anomaly detection system including power plant ID, inverter ID, measurement time, voltage measurement value, current measurement value, and number of series and parallel connections.

6. In paragraph 1, The above anomaly detection unit, A learning unit that learns a model based on a convolutional neural network using images classified into multiple categories, and creates an image classification model by classifying and learning whether the solar module of the MPPT is abnormal using an image of the maximum output IV curve and an image of an IV curve whose output is different from a preset ratio or more from the output of the maximum output IV curve as the last layer; and An anomaly detection system for a solar module based on an IV curve, comprising a judgment unit that classifies the normalized IV curve data as normal or abnormal based on the image classification model.

7. In paragraph 6, The above image classification model is an IV curve-based solar module anomaly detection system that is trained based on MobileNet, ResNet50, or DenseNet121 using images classified into multiple categories.

8. In paragraph 1, The above maximum output IV curve data is an IV curve data having the maximum output value among the IV curve data accumulated in the above collection unit, and is an anomaly detection system for an IV curve-based solar module.

9. Step of collecting IV curve data for each MPPT for multiple solar modules; For each MPPT, a step of normalizing each IV curve data based on the maximum output IV curve data of the corresponding MPPT; and An IV curve-based solar module abnormality detection method, comprising a step of inputting the normalized IV curve data into an image classification model that has learned whether a solar module is abnormal based on the normalized maximum output IV curve data for each power plant or MPPT, thereby determining whether the solar module is normal or abnormal for each MPPT.

10. In paragraph 9, The above normalizing step is, An anomaly detection method for a solar module based on an IV curve, which performs primary normalization to unitize the IV curve data by MPPT of each power plant according to the number of solar modules included, and secondary normalization to convert the primary normalized IV curve data into a value of 0 to 1 based on the maximum output of the corresponding MPPT.

11. In paragraph 10, The above normalizing step is, An anomaly detection method for a solar module based on an IV curve, which performs the first normalization of the IV curve data according to the mathematical formula below. Here, V1 is the first normalized voltage, I1 is the first normalized current, V org is the voltage measurement, I org is the current measurement value, Voc unit Silver open circuit voltage, Isc unit is the short-circuit current, n sr is the number of series-connected modules of the corresponding MPPT, n pr is the number of parallel-connected modules of the corresponding MPPT.

12. In paragraph 10, The above normalizing step is, An anomaly detection method for a solar module based on an IV curve, which secondarily normalizes the first-normalized IV curve data according to the mathematical formula below. Here, V2 is the second normalized voltage, I2 is the second normalized current, Voc,normal is the open circuit voltage value of the maximum output IV curve data, and Isc,normal is the short circuit current value of the maximum output IV curve data.

13. In paragraph 9, The above IV curve data includes a power plant ID, an inverter ID, a measurement time, a voltage measurement value, a current measurement value, and the number of series and parallel connections, and is an IV curve-based solar module anomaly detection method.

14. In paragraph 9, A model is trained based on a convolutional neural network using images classified into multiple categories, and the last layer is an image of the maximum output IV curve and an image of the IV curve having a difference in output from the maximum output IV curve and a preset ratio or more, and a step of classifying and learning whether the solar module of the MPPT is abnormal is further included to generate the image classification model. The steps for determining whether the above is normal or abnormal are: An abnormality detection method for a solar module based on an IV curve, which classifies a solar module of an MPPT corresponding to the normalized IV curve data as normal or abnormal based on the image classification model.

15. In paragraph 14, The step of generating the above image classification model is an IV curve-based solar module anomaly detection method that trains a model based on MobileNet, ResNet50, or DenseNet121 using images classified into multiple categories.

16. In paragraph 9, The above maximum output IV curve data is an IV curve data having the maximum output value among the accumulated IV curve data, and is an anomaly detection method for a solar module based on an IV curve.

Images