Photovoltaic power generation system and photovoltaic power generation method capable of detecting power generation abnormality and diagnosing cause thereof

Abstract

The present invention relates to a photovoltaic power generation system and a photovoltaic power generation method capable of detecting a power generation abnormality and diagnosing a cause thereof, wherein an artificial intelligence model is constructed by collecting MPPT results of MPPT units and performing artificial intelligence learning using MPPT results of the MPPT units excluding an MPPT result of a specific MPPT unit, and by comparing a pattern of an MPPT result estimated by the artificial intelligence model and a pattern of an actual MPPT result, an MPPT unit in which an abnormality in MPPT performance has occurred is detected early and a cause of the abnormality is diagnosed.

Description

A solar power generation system and solar power generation method capable of detecting and diagnosing the cause of power generation abnormalities

[0001] The present invention belongs to the field of solar power generation technology, and more specifically, relates to a solar power generation system and a solar power generation method capable of detecting power generation abnormalities in a solar power generation system and diagnosing their causes.

[0002] Solar power plants generate electricity by converting sunlight into electrical energy through the photovoltaic effect. The power generation system of a solar power plant consists of a number of modules made up of solar cells (solar cells) connected in series and parallel to form a solar panel.

[0003] A string is formed by connecting multiple modules in series or multiple solar panels in series, and when one or more strings are connected, an MPPT unit, which is a unit that performs maximum power point tracking (MPPT), is formed. Two or more MPPT units are connected to an inverter, and the inverter converts the power produced by the module or solar panel from direct current to alternating current and supplies electricity to the grid.

[0004] Each MPPT unit connected to the inverter performs maximum power point tracking individually, and maximum power point tracking controls the impedance so that the inverter operates close to the maximum power point as conditions such as the amount of solar radiation, the temperature around the solar power plant, and the load on the solar power plant change, thereby allowing the solar power generation system to operate at the maximum power point.

[0005] The power output from a solar cell module can be expressed as the product of current and voltage, and the current-voltage characteristic curve of the module according to the solar cell's irradiance and temperature is provided by the manufacturer as the solar cell's specification, so the MPPT unit can measure the irradiance and temperature of the solar cell incident on the solar cell to generate a current-voltage characteristic curve for the given irradiance and temperature, and in the maximum power point tracking process, the inverter adjusts the circuit to achieve the current and voltage at which the maximum output is generated from the MPPT unit.

[0006] In addition, since the maximum power point tracking of a solar power generation system is performed on a string basis, output reduction may occur due to various causes, and if the output reduction that occurred continues, problems may occur in the power generation output of the entire solar power generation system, so the cause of the output reduction must be identified and measures taken.

[0007] As a method to identify the cause of the output drop, a current-voltage scan method can be applied, and a problem occurring in the MPPT unit is detected by extracting the current-voltage characteristic curve at a specific point in time and comparing its shape with the current-voltage characteristic curve provided by the solar cell manufacturer.

[0008] As a method for diagnosing the occurrence of output reduction in such a solar power generation system, Korean Patent Publication No. 10-1729217 (registered on April 17, 2017) proposes an inverter MPPT performance diagnosis device and method for a solar power generation system, which measures the irradiance and temperature of each solar cell array through a sensor, constructs an approximate model of the maximum output voltage based on the measured irradiance and temperature, calculates the maximum output voltage, and compares the measured input voltage with the result of the maximum power point tracking control according to the calculated maximum output voltage, thereby diagnosing whether the maximum power point tracking control is operating normally.

[0009] However, if a sensor is installed for each MPPT unit or string, the cost of building a solar power generation system increases significantly, and since the current-voltage scan is performed on an inverter basis, even if the cause of the decrease in power generation is identified, it is difficult to identify the problem early because the individual string or MPPT unit where the problem occurred must be found by examining the maximum power point tracking control results.

[0010] In particular, when the power generation output of an MPPT unit gradually decreases due to the aging of a specific solar cell module, it is difficult to diagnose whether the solar cell module is aging simply by calculating the maximum output voltage and comparing the measured input voltage. Therefore, there is a need to develop a method for detecting and diagnosing the cause of power generation abnormalities in a solar power generation system that can early identify individual strings or MPPT units that have problems as a result of maximum power point tracking control and easily diagnose whether the solar cell module is aging.

[0011] (Patent Document 1) Republic of Korea Patent Publication No. 10-1729217 (registered on April 17, 2017)

[0012] The purpose of the present invention is to provide a method for detecting and diagnosing the cause of a power generation abnormality that can diagnose the cause of a power generation decrease when an output decrease occurs in an individual string or MPPT unit of a solar power generation system, identify the individual string or MPPT unit in which the output decrease occurred at an early stage, and easily diagnose whether aging of a solar cell module has occurred.

[0013] A solar power generation system capable of detecting power generation abnormalities and diagnosing their causes according to the present invention comprises: an inverter in which at least two MPPT units are connected, each MPPT unit formed by connecting at least one string in which at least two solar cell modules or solar panels are connected in series; an inverter for performing MPPT in each MPPT unit; a server including an artificial intelligence model for detecting MPPT units in which MPPT results of each MPPT unit are continuously measured over time from the inverter and performing artificial intelligence learning using the MPPT results of the MPPT units excluding the MPPT results of a specific MPPT unit; and an MPPT unit in which a pattern of an MPPT result estimated through the learned model is different from an MPPT result actually measured, thereby detecting an abnormality in MPPT performance and diagnosing the cause thereof.

[0014] According to an embodiment of the present invention, the artificial intelligence model uses the first half of the MPPT results measured at each data point, which is a preset measurement time interval for a certain period of time, from MPPT units as learning data, and uses the MPPT results of the remaining MPPT units excluding the MPPT result of the i-th MPPT unit (i is a natural number less than or equal to the number of all MPPT units) among the learning data as independent variables, and performs artificial intelligence learning of the MPPT result of the i-th MPPT unit.

[0015] According to an embodiment of the present invention, after the artificial intelligence learning of the artificial intelligence model is completed, the remaining latter half data excluding the learning data is used as verification data, and the MPPT results of the remaining MPPT units excluding the MPPT result of the i-th MPPT unit among the verification data are input into the artificial intelligence model, thereby predicting and outputting the MPPT estimate value of the i-th MPPT unit.

[0016] According to an embodiment of the present invention, if the error between the MPPT estimate at each data point predicted through verification data and the actual MPPT result is greater than a preset error reference value, the artificial intelligence model determines that an abnormality has occurred during MPPT execution and records it on the server. If the number of abnormal points recorded on the server is greater than or equal to a preset value and the abnormal point occurrence cycle is less than or equal to a preset value, the artificial intelligence model determines that an abnormality has occurred during MPPT execution.

[0017] According to an embodiment of the present invention, the artificial intelligence model repeatedly performs artificial intelligence learning of MPPT results, prediction of MPPT estimation values, and detection of abnormal condition occurrence for each MPPT unit.

[0018] According to an embodiment of the present invention, the error reference value is set within a range of 10 to 20% of the average MPPT result in the learning data for the target MPPT to be predicted.

[0019] According to an embodiment of the present invention, the abnormal point occurrence period setting value is set as a variance reference value, which is the square of the number of all data points during the time when solar irradiance exists during the day, and if the variance value of the time interval between each abnormal state is less than the variance reference value, it is estimated that an abnormality has occurred during the MPPT execution.

[0020] According to an embodiment of the present invention, the MPPT result and the MPPT estimation value are any one selected from among the voltage value, the current value, and the power value of the MPPT unit, and the detection of power generation abnormality and diagnosis of the cause based on the voltage value, the current value, and the power value are performed independently.

[0021] According to an embodiment of the present invention, the artificial intelligence model compares the MPPT estimation value of a specific MPPT unit with the actual MPPT result measured from the inverter, and if the error between the actual MPPT result and the MPPT estimation value continues to exceed a value set in the artificial intelligence model, it is diagnosed as a problem with the MPPT unit itself, or if a difference between the MPPT estimation value and the actual MPPT result, which did not occur during the initial operation of the solar power generation system, is formed only during a specific time period, it is diagnosed as a problem with shading of the solar cells or solar panels connected to the MPPT unit.

[0022] According to an embodiment of the present invention, each MPPT unit is equipped with a sensor that measures irradiance or temperature of a solar cell or solar panel, and an artificial intelligence model performs artificial intelligence learning through the sensor measurement results, and compares the irradiance and temperature measurement results of the sensor with the irradiance and temperature estimation values ​​of the artificial intelligence model, and determines that aging of the solar cell or solar panel has occurred when a deviation exceeding a preset size is formed.

[0023] According to an embodiment of the present invention, before constructing an artificial intelligence model, the server forms a group of MPPT units connected in series or parallel, collects the MPPT results of each MPPT unit through the initial operation of the solar power generation system, forms an MPPT cluster by MPPT group with similar MPPT result patterns, and diagnoses that a problem has occurred when forming multiple clusters from the MPPT results collected during the initial operation process, thereby improving problems in the initial installation and structure of the solar power generation system.

[0024] According to an embodiment of the present invention, in a case where multiple MPPT clusters are formed from MPPT results collected during the initial operation of a solar power generation system, if the MPPT result of a specific MPPT cluster is smaller than the MPPT results of other MPPT clusters, it is determined that there is a problem with the solar cell or solar panel itself or an installation problem, or if multiple MPPT clusters are formed only at a specific time period and a single MPPT cluster is formed at other time periods, it is determined that there is a shading problem, or if there is no problem with the solar cell or solar panel itself or an installation problem, it is determined that there is an inverter defect problem.

[0025] A solar power generation method capable of detecting abnormal power generation and diagnosing the cause thereof according to the present invention comprises: an MPPT result measurement step of measuring and collecting MPPT results of each MPPT unit through an inverter that performs MPPT in each MPPT unit, wherein at least two MPPT units are formed by connecting at least one string in which at least two solar cell modules or solar panels are connected in series; an AI model construction step of performing AI learning through MPPT results of the remaining MPPT units excluding the MPPT result of a specific MPPT unit using the collected MPPT result data; an MPPT result comparison step of comparing the pattern of the MPPT result estimated by the constructed AI model with the collected actual MPPT result; and a detection and diagnosis step of detecting an MPPT unit in which an MPPT estimated value of the AI ​​model and a pattern of an actual MPPT result are formed differently, thereby detecting an abnormality in MPPT performance and diagnosing the cause thereof.

[0026] According to an embodiment of the present invention, in the artificial intelligence model construction step, the first half of the MPPT results measured at each data point, which is a preset measurement interval for a certain period of time, from the MPPT units are used as learning data, and the MPPT results of the remaining MPPT units, excluding the MPPT result of the i-th MPPT unit (i is a natural number less than or equal to the number of all MPPT units) among the learning data, are used as independent variables to conduct artificial intelligence learning of the MPPT result of the i-th MPPT unit.

[0027] According to an embodiment of the present invention, the MPPT result comparison step uses the remaining latter half data excluding the learning data as verification data, inputs the MPPT results of the remaining MPPT units excluding the MPPT result of the i-th MPPT unit among the verification data into the artificial intelligence model, thereby predicting the MPPT estimation value of the i-th MPPT unit.

[0028] According to an embodiment of the present invention, the detection and diagnosis step includes an outlier designation process for designating data points in which an error between the MPPT estimate at each data point predicted by the artificial intelligence model through verification data and the actual MPPT result is greater than a preset error reference value, as outliers, and an outlier recording process for removing data points that are consecutive for less than a preset number of times among the designated outliers and recording the remaining outliers on a server.

[0029] According to an embodiment of the present invention, after the abnormal point recording process is completed in the detection and diagnosis step, an abnormal point count determination process for determining whether the number of abnormal points recorded in the server is greater than or equal to a set value, and an abnormal point occurrence cycle determination process for determining whether the abnormal point occurrence cycle is less than or equal to a set value when the number of abnormal points is greater than or equal to the set value are added, and if the number of abnormal points is less than or equal to the set value in the abnormal point count determination process or the abnormal point occurrence cycle exceeds the set value in the abnormal point occurrence cycle determination process, it is assumed that normal MPPT is performed, and if the abnormal point occurrence cycle is less than or equal to the set value in the abnormal point occurrence cycle determination process, it is assumed that an abnormal state has occurred during MPPT performance.

[0030] According to an embodiment of the present invention, before performing the artificial intelligence model construction step, an initial problem diagnosis and improvement step is added, which comprises a process in which a server collects MPPT results of each MPPT unit through initial operation of a solar power generation system, a process in which MPPT clusters are formed by MPPT groups having similar MPPT result patterns of MPPT units connected in series or in parallel, a diagnosis process in which a problem occurs when forming multiple MPPT clusters from MPPT results collected during the initial operation, and an improvement process in which problems in the initial installation and structure of the diagnosed solar power generation system are resolved.

[0031] According to an embodiment of the present invention, the MPPT result and the MPPT estimation value are any one selected from among the voltage value, the current value, and the power value of the MPPT unit, and the detection of power generation abnormalities and diagnosis of causes based on the voltage value, the current value, and the power value are performed independently.

[0032] According to an embodiment of the present invention, the artificial intelligence model building step and the abnormality detection and cause diagnosis step are repeatedly performed for each MPPT unit.

[0033] The solar power generation system and solar power generation method according to the present invention collect MPPT results of each MPPT unit connected to an inverter, perform artificial intelligence learning using the MPPT results of MPPT units excluding the MPPT result of a specific MPPT unit, and construct an artificial intelligence model, and compare the MPPT result estimated by the artificial intelligence model with the pattern of the actual MPPT result, thereby detecting an MPPT unit in which an abnormality in MPPT performance has occurred at an early stage and diagnosing the cause of the abnormality.

[0034] Figure 1 is a drawing showing an example of a solar cell, which is the most basic unit of a solar power generation system, a solar panel composed of multiple modules made of solar cells connected in series and parallel, and a connection structure between the solar panel and an inverter.

[0035] Figure 2 is a drawing showing an example of a connection structure of a string formed by connecting multiple modules made of solar cells in series or connecting multiple solar panels in series.

[0036] Figure 3 is a drawing showing the connection structure of a string and MPPT unit, the connection structure of an MPPT unit and an inverter, and the connection structure of an inverter and a server.

[0037] Figure 4 is a graph showing an example of MPPT results according to voltage values, current values, and power values ​​measured in the MPPT unit.

[0038] Figures 5 and 6 are diagrams that simplify the process of detecting power generation abnormalities and diagnosing causes based on learning of an artificial intelligence model using learning data among the entire data of MPPT results and predicting MPPT estimation values ​​using verification data.

[0039] FIG. 7 is a diagram that simplifies the MPPT estimation value prediction process of inputting the MPPT results of the remaining sequential MPPT units into an artificial intelligence model to predict the MPPT estimation value of a specific MPPT unit and outputting the MPPT estimation value of the corresponding MPPT unit.

[0040] Figure 8 is a flowchart showing each step of the process of detecting power generation abnormalities and diagnosing the cause of a solar power generation system.

[0041] Figure 9 is an algorithm flowchart showing the overall structure of an artificial intelligence model constructed through artificial intelligence learning and an ideal MPPT estimation algorithm using the constructed artificial intelligence model.

[0042] FIG. 10 is a diagram showing an example of an MPPT group formed by grouping MPPT units connected in series or parallel, and an MPPT cluster formed by grouping MPPT groups with similar MPPT result patterns during initial operation, for the purpose of diagnosing and improving initial problems in a solar power generation system.

[0043] Figure 11 is a drawing showing a configuration of a power plant and an MPPT unit equipped in each power plant according to an embodiment of a solar power generation system.

[0044] Figure 12 shows the result of estimating the ideal MPPT from the MPPT result measured based on the voltage values ​​of power plants 1 to 4 in the solar power generation system embodiment.

[0045] FIGS. 13 to 15 are graphs showing voltage values ​​and occurrence times of abnormal points in the 9th MPPT unit of power plant 1, the 0th MPPT unit of power plant 3, and the 5th MPPT unit of power plant 4, respectively, from which 5 or more abnormal points were extracted from the abnormal MPPT estimation results according to FIG. 12.

[0046] Figure 16 shows the result of estimating the ideal MPPT from the MPPT result measured based on the current values ​​of power plants 1 to 4 in the solar power generation system embodiment.

[0047] FIG. 17 is a graph showing the current value and the occurrence time of the abnormal points in the 5th MPPT unit of power plant 5, from which 5 or more abnormal points were extracted from the abnormal MPPT estimation results according to FIG. 16.

[0048] *Explanation of key symbols in the drawing*

[0049] 10: Inverter

[0050] 11: MPPT group

[0051] 12: MPPT cluster

[0052] 20: Server

[0053] 21: Artificial Intelligence Model

[0054] 100: MPPT unit

[0055] 101: Strings

[0056] 110: Solar panels

[0057] 111: Solar cell

[0058] Hereinafter, embodiments of the present invention will be described in detail with reference to the attached drawings.

[0059] When describing embodiments of the present invention, descriptions of technical contents that are well known in the technical field to which the present invention belongs and are not directly related to the present invention will be omitted.

[0060] This is to convey the gist of the present invention more clearly without obscuring it by omitting unnecessary explanations.

[0061] In addition, when describing the components of the present invention, different reference numerals may be given to components with the same name depending on the drawing, and the same reference numerals may be given to different drawings.

[0062] However, even in such cases, this does not mean that the components have different functions depending on the embodiment, or that they have the same function in different embodiments, and the function of each component should be judged based on the description of each component in the embodiment.

[0063] In addition, the technical terms used in this specification should be interpreted as having a meaning generally understood by a person of ordinary skill in the technical field to which the present invention pertains, unless specifically defined otherwise in this specification, and should not be interpreted in an overly comprehensive or overly narrow sense.

[0064] Additionally, singular expressions used herein include plural expressions unless the context clearly indicates otherwise.

[0065] In this application, terms such as “comprises” or “comprising” should not be construed to necessarily include all of the components or steps described in the specification, and should be construed to mean that some of the components or some of the steps may not be included, or that additional components or steps may be included.

[0066]

[0067] In the solar power generation system capable of detecting abnormalities in power generation and diagnosing their causes according to the present invention, the solar cell (111) is the most basic unit of the solar power generation system, and as shown in Fig. 2, a plurality of modules composed of solar cells (111) are connected in series and parallel to form a solar panel (110).

[0068] And, as shown in FIGS. 3a and 3b, a plurality of solar cell (111) modules are connected in series, or as shown in FIG. 3c, a plurality of solar panels (110) are connected in series to form one string (101), and as shown in FIG. 4, one or more strings (101) are connected to form an MPPT unit (100).

[0069] Two or more MPPT units (100) are connected to the inverter (10), and the inverter (10) individually performs MPPT to track the maximum power point for each connected MPPT unit (100). Generally, about 10 or fewer MPPT units (100) are connected to one inverter (10).

[0070] The solar power generation system of the present invention manages the series / parallel connection structure of the MPPT unit (100) to determine whether there is an installation error in a module composed of solar cells (111), and in particular, in the case of a small-scale solar power generation system having a power production capacity of less than 1 MW, it is a prerequisite that each solar panel (110) constituting the solar power generation system is installed in a location that receives similar solar irradiance.

[0071] The MPPT unit (100) connected to each inverter (10) generates an MPPT result, which is the maximum output value according to the current irradiance at the time of MPPT execution, and the MPPT result of each MPPT unit (100) continuously measured over time from the inverter (10) is collected and stored in a server (20) connected to the inverter (10) via a network.

[0072] The maximum output, which is the MPPT result generated from each MPPT unit (100), is the power produced from the corresponding MPPT unit (100), and the power P, voltage V, and current I are Since the relationship is satisfied, the MPPT result collected and stored in the server (20) as shown in Fig. 5 is composed of at least one selected from among power values, voltage values, or current values.

[0073] The server (20) includes an artificial intelligence model (21) that detects abnormal occurrences in the MPPT unit (100) and diagnoses the cause thereof, and the artificial intelligence model (21) constructs a learning model by performing artificial intelligence learning using the MPPT results of the remaining MPPT units (100) excluding the MPPT results of a specific MPPT unit (100) from among the MPPT results stored in the server (20).

[0074] The artificial intelligence model (21) estimates the MPPT result of the MPPT unit (100) connected to each inverter (10) through the learned model, and detects an MPPT unit (100) in which the pattern of the MPPT result according to the estimated MPPT result and the data point actually measured through the inverter (10) is formed differently, thereby detecting an abnormality in the MPPT unit (100) and diagnosing the cause of the abnormality.

[0075] The artificial intelligence model (21) performs artificial intelligence learning using the MPPT results measured for each data point, and the data points are defined by preset measurement time intervals over a certain period of time.

[0076] In order to perform artificial intelligence learning of a specific MPPT unit (100), the MPPT result measured at each data point and stored in the server (20) is used as learning data, and the MPPT results of the MPPT units (100) in the remaining sequential numbers excluding the learning data MPPT result of the ith MPPT unit (100) among all MPPT units (100) connected to the inverter (10) (i is a natural number less than or equal to the number of all MPPT units) are used as independent variables to perform artificial intelligence learning, thereby performing artificial intelligence learning of the MPPT result of the ith MPPT unit (100), and performing artificial intelligence learning of the MPPT result as many times as the number of all MPPT units (100) connected to the inverter (10) for each data point.

[0077] At this time, the learning data of the artificial intelligence model (21) is composed of data of data points belonging to the first half including the starting point in the order of time arrangement among the MPPT results measured for each data point and stored in the server (20) as shown in FIGS. 6 and 7.

[0078] When the artificial intelligence learning of the artificial intelligence model (21) is completed through the above algorithm, the MPPT estimation value of a specific MPPT unit (100) is predicted through the artificial intelligence model (21) in order to detect power generation abnormalities and diagnose the cause of the solar power generation system.

[0079] In addition, the artificial intelligence model (21) predicts the MPPT estimation value through verification data, and the verification data is composed of the remaining data excluding the learning data used for artificial intelligence learning of the artificial intelligence model (21), that is, the data of the remaining data points belonging to the latter half including the end point in the order of time arrangement among the MPPT results measured for each data point and stored in the server (20).

[0080] Accordingly, as shown in FIG. 8, the MPPT results of the MPPT units (100) of the remaining sequential numbers, excluding the MPPT results according to the verification data (i is a natural number less than or equal to the number of all MPPT units) of the ith MPPT unit (100) among all MPPT units (100) connected to the inverter (10), are input into the artificial intelligence model (21), thereby predicting and outputting the MPPT estimation value of the ith MPPT unit (100), and the prediction model of the artificial intelligence model (21) uses the LightGBM regression model.

[0081] The artificial intelligence model (21) compares the MPPT estimate of the i-th MPPT unit (100) predicted for each data point with the actual MPPT result, and if the error between them is greater than the error reference value preset in the artificial intelligence model (21), it determines that an abnormal point has occurred at the corresponding data point, and records data related to the occurrence of the abnormal point in the server (20).

[0082] The artificial intelligence model (21) estimates that an abnormal state has occurred during the MPPT execution when the number of abnormal points recorded in the server (20) is greater than or equal to a set value and, at the same time, the abnormal point occurrence cycle is less than or equal to the set value. The artificial intelligence model (21) performs artificial intelligence learning of the MPPT results, predicts the MPPT estimation value, and detects the occurrence of an abnormal state repeatedly for each MPPT unit (100).

[0083] At this time, the error reference value of the i-th MPPT unit (100) for detecting the occurrence of an abnormal state during MPPT execution is set within a range of 10 to 20% of the average MPPT result in the learning data for the target MPPT to be predicted, and more preferably, it is set to about 15% of the average MPPT result in the learning data for the target MPPT to be predicted.

[0084] In addition, the first condition for estimating that an abnormal state has occurred during MPPT execution is satisfied only when an error between the MPPT estimated value and the actual MPPT result that is greater than the error reference value set in advance in the artificial intelligence model (21) is detected in the data points according to the MPPT execution at different time periods of the i-th MPPT unit (100).

[0085] Even if the error between the MPPT estimated value and the actual MPPT result is greater than the error reference value, if the number of detected error occurrences is less than the set number, it is likely to be a temporary problem such as a change in the output of the MPPT unit (100) due to a change in weather conditions, so by setting the first condition for detecting the occurrence of an abnormal state during MPPT execution, the accuracy of the estimation of the occurrence of an abnormal state is improved.

[0086] And, in a state where the first condition for occurrence of an abnormal state is satisfied, only when the abnormal point occurrence cycle recorded in the server (20) is less than or equal to the time interval size preset in the artificial intelligence model (21), the second condition for estimating occurrence of an abnormal state during the MPPT execution is satisfied, and the artificial intelligence model (21) estimates that an abnormal state has occurred during the MPPT execution only when the first condition and the second condition are satisfied at the same time.

[0087] To this end, the artificial intelligence model (21) filters only the outlier data that satisfies the first condition for estimating the occurrence of an abnormal state during MPPT execution, lists the time intervals between the outliers of the filtered outlier data, and records them as data for determining whether the second condition for detecting the occurrence of an abnormal state is satisfied, and if the period of the recorded list of outlier time intervals is less than the set value, it is estimated that an abnormal state has occurred during the MPPT execution.

[0088] At this time, the abnormal point occurrence period setting value set in the artificial intelligence model (21) becomes the variance reference value, and the variance reference value is defined as the square of the number of all data points during the time when solar irradiance exists during the day.

[0089] Accordingly, in the second condition for estimating the occurrence of an abnormal state, it is estimated that an abnormal state has occurred during the MPPT execution only when the variance of the time interval between each abnormal point is less than or equal to the variance reference value.

[0090] In particular, even if the first condition for occurrence of an abnormal state is satisfied, if the variance of the time interval between each abnormal point exceeds the variance reference value, there is a high possibility that it is a cause that occurs repeatedly, such as the occurrence of shadows in the MPPT unit (100) where the MPPT is performed. Therefore, by setting the second condition for detecting occurrence of an abnormal state, the accuracy of estimating occurrence of an abnormal state during MPPT performance can be further improved.

[0091] Through this, it is possible to identify an MPPT unit (100) among a plurality of MPPT units (100) connected to an inverter (10) in which an abnormality in MPPT performance has occurred at an early stage, and follow-up measures such as replacing a solar cell (111) module or solar panel (110) in which a problem has occurred can be quickly taken depending on the cause of the diagnosed problem.

[0092] In addition, when constructing an artificial intelligence model (21) through artificial intelligence learning or estimating the occurrence of an abnormality during MPPT performance through an artificial intelligence model (21), the MPPT result or MPPT estimation value is composed of one of the voltage value, current value, or power value output from the MPPT unit (100), and the artificial intelligence learning of the artificial intelligence model (21) based on the voltage value, current value, or power value and the detection of power generation abnormalities and diagnosis of causes are performed independently of each other.

[0093] As a result of estimating the occurrence of an abnormal condition during MPPT execution through an artificial intelligence model (21), the cause of the abnormal condition detected in the power generation of a solar power generation system can be diagnosed as follows based on the results of comparing the MPPT estimation value in a specific MPPT execution with the actual MPPT result measured from the inverter (10).

[0094] If the error between the actual MPPT result in the power generation process of the solar power generation system and the MPPT estimated value through the artificial intelligence model (21) continues to exceed the value set in the artificial intelligence model (21), a problem in the MPPT unit (100) itself is diagnosed as the cause of the abnormality, and measures are taken accordingly.

[0095] If a difference between the MPPT estimated value and the actual MPPT result, which did not occur during the initial operation of the solar power generation system, occurs repeatedly only during a specific time period, the cause of the abnormality is diagnosed to be a problem of shading occurring in the solar cell (111) or solar panel (110) connected to the MPPT unit (100) where the MPPT is performed, and measures are taken accordingly.

[0096] In addition, even if no problems occur during the initial installation and operation of the solar power generation system, the output of the MPPT unit (100) may decrease over time due to aging of the solar cells (111) or solar panels (110) that constitute the MPPT unit (100).

[0097] In particular, the learning method characteristics of the artificial intelligence model (21) that performs artificial intelligence learning through the MPPT results of the learning data of the remaining MPPT units (100) excluding the MPPT results of the learning data of a specific MPPT unit (100), and the abnormality detection algorithm characteristics of the artificial intelligence model (21) that predicts the MPPT estimated value through the MPPT results of the verification data of the remaining MPPT units (100) excluding the MPPT results of the verification data of a specific MPPT unit (100) and compares the error with the actual MPPT result to determine whether an abnormality has occurred during MPPT performance, the accuracy of detecting the occurrence of aging of the MPPT unit (100) may be lowered when the artificial intelligence model (21) is constructed based on the learning data and verification data according to the actual MPPT result in which the effect of aging of the MPPT unit (100) is reflected.

[0098] Therefore, in order to prevent the accuracy of detecting aging of the MPPT unit (100) from decreasing, the MPPT results formed in the past are stored in the server (20), and the artificial intelligence model (21) determines whether aging of the MPPT unit (100) has occurred by comparing the current MPPT results with the past MPPT results.

[0099] In addition, in order to further improve the accuracy of detecting aging of a specific MPPT unit (100) in a solar power generation system, each MPPT unit (100) may be equipped with a sensor, and the sensor measures the amount of solar radiation or temperature of the solar cells (111) or solar panels (110) forming the MPPT unit (100) and stores the measured data in a server (20).

[0100] When each MPPT unit (100) is equipped with a sensor, the artificial intelligence model (21) uses the sensor measurement result values ​​stored in the server (20) as learning data, and performs artificial intelligence learning of the irradiance and temperature measurement values ​​of the MPPT unit (100) from the irradiance and temperature measurement values ​​of the remaining MPPT units (100) excluding the irradiance and temperature measurement values ​​of a specific MPPT unit (100) from the learning data.

[0101] When the artificial intelligence learning of the artificial intelligence model (21) is completed, the irradiance and temperature measurement values ​​of the remaining MPPT units (100) excluding the irradiance and temperature measurement values ​​of a specific MPPT unit (100) among the verification data are input into the artificial intelligence model (21), thereby predicting and outputting the irradiance and temperature measurement values ​​of the corresponding MPPT unit (100), and when the error between the actual irradiance and temperature measurement results and the irradiance and temperature estimation values ​​of the artificial intelligence model (21) exceeds a preset deviation size, it is determined that aging has occurred in the solar cell (111) or solar panel (110) of the corresponding MPPT unit (100).

[0102] In addition, if an installation defect, such as a series-parallel connection error of the MPPT unit (100), occurs during the installation process of the solar power generation system, or an initial defect occurs in the solar cell (111) or module of the MPPT unit (100), the problem arises that the initial problem of the solar power generation system cannot be detected through the artificial intelligence model (21) because the construction of the artificial intelligence model (21) based on artificial intelligence learning has not been completed.

[0103] Accordingly, in order to diagnose initial problems that occurred during the installation process of a solar power generation system and then improve the diagnosed problems, the server (20) groups MPPT units (100) connected in series or parallel before the artificial intelligence model (21) is built to form multiple MPPT groups (11), and collects MPPT results of each MPPT group (11) by performing initial operation of the solar power generation system, and then groups MPPT groups (11) with similar MPPT result patterns together to form one or more MPPT clusters (12).

[0104] At this time, when MPPT groups (11) of multiple MPPT units (100) using the same solar cell (111) module and having a similar series-parallel connection structure are arranged adjacent to each other, theoretically, the MPPT result patterns of the MPPT groups (11) are formed similarly, so one MPPT cluster (12) should be formed.

[0105] However, due to initial problems occurring during the installation process of a solar power generation system, multiple MPPT clusters (12) may be formed from the MPPT results of MPPT groups (11) collected during the initial operation process of an actual solar power generation system, and when multiple MPPT clusters (12) are formed, it is diagnosed that a problem has occurred in the MPPT group (11) that constitutes the MPPT cluster (12) having a lower MPPT result than other MPPT clusters (12), thereby diagnosing problems in the initial installation and structure of the solar power generation system.

[0106] When multiple MPPT clusters (12) are formed from MPPT results collected during the initial operation of a solar power generation system, initial problems are diagnosed as follows based on the MPPT result patterns of the MPPT clusters (12), and the problems are improved according to the diagnosed problems.

[0107]

[0108] The solar power generation method of a solar power generation system capable of detecting and diagnosing causes of abnormal power generation, which is configured in this manner, is composed of an initial problem diagnosis and improvement step (S10) to a detection and diagnosis step (S50), as shown in Fig. 9.

[0109] The initial problem diagnosis and improvement step (S10) consists of a process of collecting MPPT results (S11), a process of forming an MPPT cluster (12) (S12), a diagnosis process (S13), and an improvement process (S14).

[0110] In the process of collecting MPPT results (S11), after the installation of the solar power generation system is completed, the MPPT results of each MPPT unit (100) are collected through the initial operation of the installed solar power generation system and stored in the server (21), and as shown in FIG. 11, MPPT units (100) of similar series-parallel connection structures arranged adjacent to each other in the solar power generation system form one MPPT group (11), thereby forming multiple MPPT groups (11) in the solar power generation system.

[0111] In the process of forming an MPPT cluster (S12), MPPT groups (11) having similar MPPT result patterns according to data points collected in the server (21) are grouped together to form an MPPT cluster (12). If no initial problem occurs in the solar power generation system, the MPPT result patterns of each MPPT group (11) are formed similarly, and the MPPT groups (11) form one MPPT cluster (12). If an initial problem occurs in the solar power generation system, two or more types of MPPT result patterns are generated in each MPPT group (11), and multiple MPPT clusters (12) are formed.

[0112] In the diagnosis process (S13), when multiple MPPT clusters (12) are formed in the process of forming an MPPT cluster (S12), it is diagnosed whether an initial problem has occurred in the solar power generation system, and in the improvement process (S14), depending on the cause of the problem diagnosed in the diagnosis process (S13), measures are taken to improve the problem, such as replacing the MPPT unit (100) or inverter (10). If multiple MPPT clusters (12) are not formed in the diagnosis process (S13), the improvement process (S14) is omitted.

[0113] As a result of the diagnostic process (S13),

[0114] 1) If the MPPT result of a specific MPPT cluster (12) is lower than the MPPT result of another MPPT cluster (12), it is judged to be a problem with the solar cell (111) or solar panel (110) itself or an installation problem.

[0115] 2) If multiple MPPT clusters (12) are formed only at a specific time zone and a single MPPT cluster (12) is formed at other time zones, it is judged that this is a problem due to shading occurring in the MPPT unit (100) of the MPPT cluster (12).

[0116] 3) If it is not a problem with the solar cell (111) or solar panel (110) itself or an installation problem, and at the same time, it is not a problem caused by shading of the MPPT unit (100), it is judged to be a problem with the inverter (10).

[0117] Diagnose problems in the solar power generation system, and perform initial defect improvement of the solar power generation system according to the diagnosed problems in the improvement process (S14).

[0118] When the initial problem diagnosis and improvement step (S10) is completed, in order to build an artificial intelligence model (21) as shown in FIG. 10, setting values ​​such as the measurement period of the MPPT units (100), data point interval, error reference value, consecutive number of abnormal points, number of abnormal points, and abnormal point cycle are input into the artificial intelligence model (21).

[0119] In the MPPT result measurement step (S20), the MPPT results are collected and stored in the server (20) by performing MPPT continuously in each MPPT unit (100) over time through the inverter (10), and the MPPT result data is prepared by sequentially selecting all MPPT units (100) to perform MPPT and collecting the measured MPPT results and storing them in the server (20).

[0120] Once the MPPT result data is prepared, the MPPT results for the entire period of data points measured from each MPPT unit (100) are separated into learning data and verification data.

[0121] In the artificial intelligence model construction step (S30), based on the MPPT result data stored in the server (20), artificial intelligence learning of the artificial intelligence model (21) is performed using the MPPT results of the remaining MPPT units (100) excluding the MPPT result of a specific MPPT unit (100), and the artificial intelligence model (21) is constructed by performing artificial intelligence learning using data corresponding to the first half of 80% including the measurement start point among the MPPT results for the data points of the entire period measured in each MPPT unit (100) as learning data.

[0122] In the MPPT result comparison step (S40), the MPPT result pattern according to the data points of each MPPT unit (100) estimated through the constructed artificial intelligence model (21) is compared with the MPPT result pattern according to the data points actually measured in the MPPT result collection step (S20).

[0123] The estimated value of the MPPT result for comparison with the actual measured value of the MPPT result can be predicted and output by inputting the MPPT result of the verification data into the artificial intelligence model (21) by using the data corresponding to the latter 20% including the measurement end point, excluding the data selected as learning data, among the MPPT results for the data points of the entire period measured in each MPPT unit (100), as verification data.

[0124] In the detection and diagnosis step (S50), an MPPT unit (100) in which the pattern of the MPPT estimated value and the pattern according to the data point of the actually measured MPPT result are formed differently is detected, and the MPPT unit (100) in which the pattern between the MPPT estimated value and the actual MPPT result is formed differently is specified, thereby detecting the occurrence of an MPPT abnormality and diagnosing the cause thereof.

[0125] In order to detect the occurrence of an MPPT abnormality in the detection and diagnosis step (S50), in the abnormal point designation process (S51), it is determined whether the error between the MPPT estimate at each data point predicted from the verification data and the actual MPPT result is larger than a preset error reference value through an artificial intelligence model (21), and a data point where an error larger than the error reference value is formed is designated as an abnormal point.

[0126] In the outlier recording process (S52), data points of the outliers specified in the outlier designation process (S51) that are consecutive less than a preset number of times are removed, and only the outliers that are consecutive more than a preset number of times are recorded in the server (20).

[0127] In the process of determining the number of abnormal points (S53), it is determined whether the number of abnormal points recorded in the server (21) is greater than or equal to a preset value. If the number of abnormal points is less than the preset value, it is assumed that normal MPPT has been performed, and the detection and diagnosis step (S50) is terminated. Only when the number of abnormal points is greater than or equal to the preset value, the process of determining the abnormal point occurrence cycle (S54) is performed.

[0128] In the process of determining the abnormal point occurrence period (S54), it is determined whether the abnormal point occurrence period is less than or equal to the pre-set variance reference value, and only when the abnormal point occurrence period is less than or equal to the variance reference value, it is assumed that an abnormal state due to MPPT abnormality has occurred.

[0129] Therefore, in the detection and diagnosis step (S50), when the error between the MPPT estimated value and the actual MPPT result is formed to be larger than the preset error reference value and the prerequisite conditions are satisfied for a set number of consecutive times or more, it is recorded as an abnormality in the server (21), and only when the first condition that the number of abnormal points among the abnormal points recorded in the server (21) is greater than the preset reference value and the second condition that the abnormal point occurrence cycle is less than the variance reference value are sequentially and simultaneously satisfied, it is estimated that an abnormality has occurred due to an abnormality in the MPPT unit (100).

[0130] [Example]

[0131] In an embodiment of a solar power generation system and solar power generation method capable of detecting abnormal power generation and diagnosing the cause of the present invention, as shown in FIG. 12, an artificial intelligence model (21) was constructed by utilizing individual measurement values ​​of MPPT results according to voltage values, current values, or power values ​​of MPPT units (100) in four power plants (pv1, pv2, pv3, pv4) as data, and 10 MPPT units (100) are provided for each power plant.

[0132] The data point measurement period according to the embodiment is set to 7 months (from 2024.01.01. to 2024.07.31.), the data point measurement time interval is set to 5 minutes, and since MPPT must be performed during a time zone where there is solar radiation during the day, in the embodiment, 180 data points are formed for 15 hours excluding the sunset time zone every day during the data point measurement period.

[0133] Among the entire data, the learning data for artificial intelligence learning of the artificial intelligence model (21) consists of MPPT results measured from 2024.01.01 to 2024.06.19, and the verification data for predicting the MPPT estimation value through the constructed artificial intelligence model (21) consists of MPPT results measured from 2024.06.20 to 2024.07.31.

[0134] In the abnormal state designation process (S51) according to the embodiment, a value corresponding to 15% of the average size of the MPPT results of the verification data measured from the 0th to 9th MPPT units (100) of each power plant (pv1, pv2, pv3, pv4) was set as the error reference value, and it was estimated that an abnormal point occurred in the MPPT results in the 9th MPPT unit (100) of power plant 1 (pv1), the 0th MPPT unit (100) of power plant 3 (pv3), and the 5th MPPT unit (100) of power plant 4 (pv4), where the error between the MPPT estimation value according to the voltage value predicted by the artificial intelligence model (21) and the actual MPPT result exceeded the error reference value, as shown in FIG. 13.

[0135] In the outlier recording process (S52), data points of estimated outliers that are consecutive less than 3 times are removed, and only outliers that are consecutive 3 or more times are recorded in the server (20). Only when the number of outliers recorded in the server (21) is 5 or more, it is judged as an abnormal MPPT.

[0136] Accordingly, the graphs of occurrence of abnormal points in the MPPT results according to voltage values ​​in the 9th MPPT unit (100) of power plant 1 (pv1), the 0th MPPT unit (100) of power plant 3 (pv3), and the 5th MPPT unit (100) of power plant 4 (pv4) were output as shown in FIGS. 14 to 16.

[0137] In addition, it was estimated that an abnormality occurred in the MPPT result in the 5th MPPT unit (100) of power plant 4 (pv4) where the error between the MPPT estimation value according to the current value predicted by the artificial intelligence model (21) as shown in Fig. 17 and the actual MPPT result exceeded the error standard value.

[0138] Likewise, in the outlier recording process (S52), data points of estimated outliers that are consecutively less than three times are removed, and only outliers that are consecutively more than three times are recorded in the server (20). Only when the number of outliers recorded in the server (21) is five or more, it is judged as an abnormal MPPT.

[0139] Accordingly, the graph of occurrence of abnormal points in the MPPT results according to the current value in the 5th MPPT unit (100) of power plant 4 (pv4) was output as in Fig. 18.

[0140] In the process of determining the abnormal point occurrence period (S54), it is determined whether the abnormal point occurrence period is 32,400 or less, and the numerical value of the variance reference value is determined from the square of 180, which is the number of data points for 15 hours during which sunlight exists.

[0141] If the occurrence period of the abnormal point is less than or equal to 32,400, which is the variance reference value, it is assumed that an abnormal condition has occurred due to an MPPT abnormality, thereby completing the detection of an abnormality in the power generation system according to the embodiment.

[0142]

[0143] Although the embodiments of the present invention have been described with reference to the above, those skilled in the art will understand that the present invention can be implemented in other specific forms without changing the technical idea or essential features thereof.

[0144] Therefore, the embodiments described above should be understood as being exemplary and not restrictive in all respects, and the scope of the present invention described in the detailed description above is indicated by the claims described below, and all changes or modifications derived from the meaning and scope of the claims and their equivalent concepts should be interpreted as being included in the scope of the present invention.

Claims

1. 2 or more solar cell (111) modules or solar panels (110) are connected in series, and at least two MPPT units (100) are formed by connecting one or more strings (101), and an inverter (10) that performs MPPT in each MPPT unit (100); A server (20) including an artificial intelligence model (21) that detects an MPPT unit (100) in which a pattern of an MPPT result estimated through a learned model and an MPPT result actually measured are formed differently by collecting and storing MPPT results of each MPPT unit (100) continuously measured over time from an inverter (10), performing artificial intelligence learning using the MPPT results of the MPPT units (100) excluding the MPPT result of a specific MPPT unit (100), and diagnosing an abnormality in MPPT performance and the cause thereof; A solar power generation system capable of detecting and diagnosing abnormalities in power generation, characterized by comprising:

2. In paragraph 1, The above artificial intelligence model (21) uses the first half of the MPPT results measured at each data point, which is a preset measurement time interval for a certain period of time, from MPPT units (100) as learning data, and uses the MPPT results of the remaining MPPT units (100) excluding the MPPT result of the ith MPPT unit (100) (i is a natural number less than or equal to the number of all MPPT units) among the learning data as independent variables, and is a solar power generation system capable of detecting power generation abnormalities and diagnosing their causes.

3. In paragraph 2, A solar power generation system capable of detecting power generation abnormalities and diagnosing causes thereof, characterized in that after the artificial intelligence learning of the artificial intelligence model (21) is completed, the remaining latter half data excluding the learning data is used as verification data, and the MPPT results of the remaining MPPT units (100) excluding the MPPT result of the i-th MPPT unit (100) among the verification data are input into the artificial intelligence model (21), thereby predicting and outputting the MPPT estimation value of the i-th MPPT unit (100).

4. In paragraph 3, The above artificial intelligence model (21) determines that an abnormality has occurred during MPPT execution if the error between the MPPT estimate at each data point predicted through verification data and the actual MPPT result is greater than a preset error reference value, and records it in the server (20), and if the number of abnormal points recorded in the server (20) is greater than a set value and the abnormal point occurrence cycle is less than a set value, it is estimated that an abnormality has occurred during the MPPT execution, wherein the solar power generation system is capable of detecting abnormalities and diagnosing their causes.

5. In paragraph 4, A solar power generation system capable of detecting power generation abnormalities and diagnosing their causes, characterized in that the above artificial intelligence model (21) repeatedly performs artificial intelligence learning of MPPT results, prediction of MPPT estimation values, and detection of abnormal conditions for each MPPT unit (100).

6. In paragraph 4, A solar power generation system capable of detecting power generation abnormalities and diagnosing their causes, characterized in that the above error reference value is set within a range of 10 to 20% of the average MPPT result in the learning data for the target MPPT to be predicted.

7. In paragraph 4, A solar power generation system capable of detecting and diagnosing causes of power generation abnormalities, characterized in that the above abnormal point occurrence period setting value is the square of the number of all data points during the time when solar irradiance exists during the day, and if the variance value of the time interval between each abnormal state is less than the variance reference value, it is estimated that an abnormality has occurred during the MPPT execution.

8. In any one of the clauses selected from clauses 3 to 7, A solar power generation system capable of detecting and diagnosing power generation abnormalities, characterized in that the MPPT result and MPPT estimated value are any one of the voltage value, current value, or power value of the MPPT unit (100), and the detection of power generation abnormalities and diagnosis of causes based on the voltage value, current value, and power value are independently performed.

9. In paragraph 1, The above artificial intelligence model (21) compares the MPPT estimated value of a specific MPPT unit (100) with the actual MPPT result measured from the inverter (10). If the error between the actual MPPT result and the MPPT estimated value continues to exceed the value set in the artificial intelligence model (21), it is diagnosed as a problem with the MPPT unit (100) itself, or If a difference between the MPPT estimated value and the actual MPPT result, which did not occur during the initial operation of the solar power generation system, is formed only at a specific time, it is diagnosed as a problem of shading of the solar cell (111) or solar panel (110) connected to the MPPT unit (100). A solar power generation system capable of detecting and diagnosing abnormalities in power generation, characterized by:

10. In paragraph 1, A solar power generation system capable of detecting and diagnosing the cause of power generation abnormalities, characterized in that each MPPT unit (100) is equipped with a sensor that measures irradiance or the temperature of a solar cell (111) or a solar panel (110), and an artificial intelligence model (21) performs artificial intelligence learning through the sensor measurement results, and compares the irradiance and temperature measurement results of the sensor with the irradiance and temperature estimation values ​​of the artificial intelligence model (21), and determines that aging of the solar cell (111) or solar panel (110) has occurred when a deviation exceeding a preset size is formed.

11. In paragraph 1, A solar power generation system capable of detecting and diagnosing causes of power generation abnormalities, characterized in that before constructing an artificial intelligence model (21), the server (20) forms a group of MPPT units (100) connected in series or parallel, collects the MPPT results of each MPPT unit (100) through the initial operation of the solar power generation system, forms an MPPT cluster by MPPT group with similar MPPT result patterns, and diagnoses that a problem has occurred when forming multiple clusters from the MPPT results collected during the initial operation process, thereby improving problems in the initial installation and structure of the solar power generation system.

12. In paragraph 11, In the case where multiple MPPT clusters (12) are formed from MPPT results collected during the initial operation of the solar power generation system, If the MPPT result of a specific MPPT cluster (12) is lower than the MPPT result of another MPPT cluster (12), it is judged to be a problem with the solar cell (111) or solar panel (110) itself or an installation problem, or If multiple MPPT clusters (12) occur only at certain times and a single MPPT cluster (12) is formed at other times, it is judged to be a shading problem, or If there is no problem with the solar cell (111) or solar panel (110) itself, or an installation problem, or a shading problem, it is judged to be a defective inverter (10) problem. A solar power generation system capable of detecting and diagnosing abnormalities in power generation, characterized by: 13.2 or more solar cell (111) modules or solar panels (110) are connected in series to form one or more MPPT units (100), and two or more MPPT units (100) are connected, and an MPPT result measurement step (S20) of measuring and collecting the MPPT result of each MPPT unit (100) through an inverter (10) that performs MPPT in each MPPT unit (100); An artificial intelligence model construction step (S30) that performs artificial intelligence learning through the MPPT results of the remaining MPPT units (100) excluding the MPPT result of a specific MPPT unit (100) using the collected MPPT result data; An MPPT result comparison step (S40) for comparing the pattern of the MPPT result estimated by the constructed artificial intelligence model (21) and the collected actual MPPT result; and A detection and diagnosis step (S50) for detecting an MPPT unit (100) in which the MPPT estimated value of the artificial intelligence model (21) and the pattern of the actual MPPT result are formed differently, thereby detecting an abnormality in MPPT performance and diagnosing the cause thereof; A solar power generation method capable of detecting and diagnosing the cause of an abnormality in power generation, characterized by including:

14. In paragraph 13, A solar power generation method capable of detecting power generation abnormalities and diagnosing causes thereof, characterized in that in the above artificial intelligence model construction step (S30), the first half of the MPPT results measured at each data point, which is a preset measurement interval for a certain period of time, from MPPT units (100) are used as learning data, and the MPPT results of the remaining MPPT units (100) excluding the MPPT result of the ith (i is a natural number less than or equal to the number of all MPPT units) MPPT unit (100) among the learning data are used as independent variables, thereby performing artificial intelligence learning of the MPPT result of the ith MPPT unit (100).

15. In paragraph 14, A solar power generation method capable of detecting power generation abnormalities and diagnosing causes thereof, characterized in that the MPPT result comparison step (S40) uses the remaining latter half data excluding the learning data as verification data, inputs the MPPT results of the remaining MPPT units (100) excluding the MPPT result of the ith MPPT unit (100) among the verification data into the artificial intelligence model (21), thereby predicting the MPPT estimation value of the ith MPPT unit (100).

16. In paragraph 13, The above detection and diagnosis step (S50) is characterized by including an outlier designation process (S51) for designating data points in which the error between the MPPT estimate at each data point predicted by the artificial intelligence model (21) through verification data and the actual MPPT result is greater than a preset error reference value as outliers, and an outlier recording process (S52) for removing data points among the designated outliers that are consecutive for less than a preset number of times and recording the remaining outliers on a server. A solar power generation method capable of detecting power generation outliers and diagnosing their causes.

17. In paragraph 16, In the above detection and diagnosis step (S50), after the abnormal point recording process (S52) is completed, an abnormal point count determination process (S53) is added to determine whether the number of abnormal points recorded in the server (21) is greater than or equal to a set value, and an abnormal point occurrence cycle determination process (S54) is added to determine whether the abnormal point occurrence cycle is less than or equal to a set value when the number of abnormal points is greater than or equal to a set value. A solar power generation method capable of detecting power generation abnormalities and diagnosing their causes, characterized in that if the number of abnormal points is less than a set value in the process of determining the number of abnormal points (S53) or the abnormal point occurrence cycle exceeds a set value in the process of determining the abnormal point occurrence cycle (S54), it is assumed that normal MPPT performance has occurred, and if the abnormal point occurrence cycle is less than a set value in the process of determining the abnormal point occurrence cycle (S54), it is assumed that an abnormal state has occurred during MPPT performance.

18. In paragraph 13, A solar power generation method capable of detecting power generation abnormalities and diagnosing causes thereof, characterized in that an initial problem diagnosis and improvement step (S10) is added, which comprises a process (S11) in which a server (21) collects MPPT results of each MPPT unit (100) through initial operation of a solar power generation system before performing the artificial intelligence model construction step (S20), a process (S12) in which MPPT groups (11) having similar MPPT result patterns are formed by groups of MPPT units (100) connected in series or parallel, a diagnosis process (S13) in which a problem is diagnosed when forming multiple MPPT groups (12) from MPPT results collected during the initial operation, and an improvement process (S14) in which problems in the initial installation and structure of the diagnosed solar power generation system are resolved.

19. In any one of the clauses 15 to 17 selected, A solar power generation method capable of detecting and diagnosing power generation abnormalities, characterized in that the MPPT result and MPPT estimated value are any one of the voltage value, current value, or power value of the MPPT unit (100), and the detection of power generation abnormalities and diagnosis of causes based on the voltage value, current value, and power value are independently performed.

20. In any one of the clauses selected from clauses 13 to 18, A solar power generation method capable of detecting and diagnosing power generation abnormalities, characterized in that the artificial intelligence model building step (S20) and the abnormality detection and cause diagnosis step (S50) are repeatedly performed for each MPPT unit (100).

Images