Method and apparatus for diagnosing failures of solar modules

Abstract

A method and apparatus for diagnosing a fault in a solar module according to one embodiment may include the steps of: inputting a plurality of heterogeneous data related to solar energy into a fault diagnosis model to calculate a model estimate value; calculating an upper weight and a lower weight based on a predetermined fuzzy number of the fault diagnosis model; predicting an upper and lower limit value of the power generation amount of the solar module from the model estimate value based on the upper and lower weights; and diagnosing a fault in the solar module based on the solar energy generation amount received from the solar module and the predicted upper and lower limits.

Description

Technology Field

[0001] Below, a technology for diagnosing failures in solar modules using a linguistic model is provided. Background Technology

[0003] A photovoltaic power generation system refers to a system that supplies direct current (DC) electrical energy generated from sunlight to an inverter to convert it into usable power, thereby providing stable power to consumers. A photovoltaic power generation system can be broadly composed of photovoltaic modules and power converters. Photovoltaic modules consist of solar cells that can generate electricity by receiving sunlight. Additionally, power converters can convert the DC power generated by the photovoltaic modules into alternating current (AC). With the emergence of international issues such as the Fukushima nuclear accident and the Paris Agreement on climate change, such photovoltaic power generation systems are already playing a significant role as eco-friendly energy sources capable of replacing nuclear power or fossil fuels. In particular, as Korea has a high dependence on overseas energy sources and is significantly and deeply affected by fluctuations in oil prices, there is a growing need for stable energy supply. Consequently, the necessity for technological development in photovoltaic power generation systems is increasing to reduce reliance on foreign energy and to prepare for energy supply instability.

[0004] The background technology described above is possessed or acquired by the inventor in the process of deriving the content of the disclosure of the present application, and cannot necessarily be considered as prior art disclosed to the general public prior to the filing of this application. Prior art literature

[0006] Korean Published Patent Application No. 10-2020-0114130 (published October 7, 2020) presents a method and system for diagnosing photovoltaic power generation faults. Korean Registered Patent Application No. 10-12238980000 (published January 12, 2013) presents a method for diagnosing and predicting faults in photovoltaic power generation modules. The problem to be solved

[0007] A method and device for diagnosing a fault in a photovoltaic module according to one embodiment may provide a range such as an upper limit and a lower limit for fault diagnosis.

[0008] However, technical challenges are not limited to the technical challenges described above, and other technical challenges may exist. means of solving the problem

[0010] A solar module fault diagnosis method according to one embodiment comprises the steps of: inputting a plurality of heterogeneous data related to solar energy into the fault diagnosis model to calculate a model estimate value (y); and, based on a predetermined fuzzy number of the fault diagnosis model, an upper limit weight (w t+ ) and lower weight (w t- A step of calculating ) and, based on the upper and lower weights, an upper limit value (y) of the power generation amount of the solar module from the model estimate. + ) and lower limit (y - It may include a step of predicting ) and a step of diagnosing a failure of the solar module based on the solar power generation amount received from the solar module and the predicted upper and lower limits.

[0011] The step of calculating the above model estimate comprises a degree of membership (u) indicating the extent to which the plurality of heterogeneous data belong to each of the plurality of partial clusters according to the context. i,k A step of calculating ) as an element of the membership matrix, and based on the degrees of membership of subclusters of the same cluster among a plurality of subclusters, the output value (Z) of the corresponding group t A step of calculating ) and weights (w) on the output values ​​of multiple clusters t It may include a step of calculating the above model estimate (y) based on the sum of the results of applying ).

[0012] The step of predicting the upper and lower limits above comprises the step of calculating a bias value (w0) based on the difference between the model estimate and the target value, and the output value (Z) of the group with the bias value, the upper weight and the lower weight. t It may include a step of predicting the upper and lower limits based on ).

[0013] The above-described solar module fault diagnosis method further includes the step of modeling the fault diagnosis model, and the step of modeling the fault diagnosis model may include the steps of determining the number of contexts and the number of clusters to be formed based on each context, initializing a membership matrix, calculating the cluster center for each of the plurality of clusters based on the result of clustering the plurality of heterogeneous training data, calculating an objective function value based on the cluster centers of the plurality of clusters, and adjusting the clustering of the plurality of clusters based on the objective function value.

[0014] The step of adjusting the clustering may include comparing the difference between the objective function of the current period and the objective function of the previous period with a threshold value; terminating the adjustment of the clustering if the difference between the objective function of the current period and the objective function of the previous period is smaller than the threshold value; and calculating the cluster center by calculating a new membership matrix if the difference between the objective function of the current period and the objective function of the previous period is larger than the threshold value.

[0015] The step of adjusting the clustering may further include the step of shifting the cluster center and calculating a new membership matrix when the difference between the objective function of the current period and the objective function of the previous period is greater than the threshold value.

[0016] The step of calculating the objective function value may include the step of calculating the Euclidean distance between the cluster centers of the plurality of clusters and the input elements of the plurality of heterogeneous training data, and the step of calculating the objective function value based on the Euclidean distance and the degree of membership.

[0017] The above-described solar module fault diagnosis method may further include a step of additionally training the fault diagnosis model using the plurality of heterogeneous data and the solar power generation amount.

[0018] The above plurality of heterogeneous data may include horizontal irradiance, slope irradiance, and temperature.

[0019] A solar module fault diagnosis device according to one embodiment may include a communication unit that receives solar power generation amount from a solar module, and a processor that inputs a plurality of heterogeneous data related to solar power into a fault diagnosis model to calculate a model estimate value, calculates an upper weight and a lower weight based on a predetermined fuzzy number of the fault diagnosis model, predicts an upper and lower limit value of the solar module's power generation amount from the model estimate value based on the upper and lower weights, and diagnoses a fault of the solar module based on the solar power generation amount and the predicted upper and lower limits.

[0020] The processor can calculate the degree of membership, which indicates the extent to which the plurality of heterogeneous data belongs to each of the plurality of sub-clusters according to the context, as a component of the membership matrix, calculate the output value of the corresponding group based on the degrees of membership of the sub-clusters of the same cluster among the plurality of sub-clusters, and calculate the model estimate value based on the sum of the results of applying weights to the output values ​​of the plurality of clusters.

[0021] The processor can calculate a bias value based on the difference between the model estimate and the target value, and predict the upper and lower limits based on the bias value, the upper and lower weights, and the output value of the group.

[0022] The processor can determine the number of contexts and the number of clusters to be formed based on each context to model the fault diagnosis model, initialize a membership matrix, calculate cluster centers for each of the multiple clusters based on the result of clustering multiple heterogeneous training data, calculate an objective function value based on the cluster centers of the multiple clusters, and adjust the clustering of the multiple clusters based on the objective function value.

[0023] The processor can terminate the clustering adjustment if the difference between the objective function of the current period and the objective function of the previous period is smaller than a threshold value, and calculate the cluster center by calculating a new membership matrix if the difference between the objective function of the current period and the objective function of the previous period is greater than the threshold value.

[0024] The processor can calculate a new membership matrix by shifting the cluster center when the difference between the objective function of the current period and the objective function of the previous period is greater than the threshold value.

[0025] The processor can calculate the Euclidean distance between each cluster center and the input elements of the plurality of heterogeneous training data, and calculate an objective function value based on the Euclidean distance and the degree of membership of each context.

[0026] The above processor can further train the fault diagnosis model using the plurality of heterogeneous data and the solar power generation amount.

[0027] The above plurality of heterogeneous data may include horizontal solar radiation, inclined solar radiation, and temperature. Effects of the invention

[0029] A method and device for diagnosing a fault in a photovoltaic module according to one embodiment can perform accurate fault diagnosis for a photovoltaic module by providing a range such as an upper limit and a lower limit for fault diagnosis.

[0030] A method and device for diagnosing a fault in a photovoltaic module according to one embodiment can maximize the management of a photovoltaic facility by monitoring the condition of the photovoltaic facility in real time and performing an accurate fault diagnosis. Brief explanation of the drawing

[0032] FIG. 1 shows a solar module fault diagnosis system according to one embodiment. FIG. 2 shows the flow of diagnosing a fault in a solar module in a solar module fault diagnosis system according to one embodiment. FIG. 3 illustrates a flowchart of a solar module fault diagnosis method according to one embodiment. FIG. 4 illustrates the flow of the step of calculating a model estimate among a solar module fault diagnosis method according to one embodiment. Figure 5 shows a weighting range that varies according to the fuzzy number in a solar module fault diagnosis method according to one embodiment. FIG. 6 shows the performance evaluation results of a photovoltaic module fault diagnosis device according to one embodiment. FIG. 7 illustrates a flowchart of a method for modeling a fault diagnosis model among a photovoltaic module fault diagnosis method according to one embodiment. Specific details for implementing the invention

[0033] Specific structural or functional descriptions of the embodiments are disclosed for illustrative purposes only and may be modified and implemented in various forms. Accordingly, actual implementations are not limited to the specific embodiments disclosed, and the scope of this specification includes modifications, equivalents, or substitutions included in the technical concept described by the embodiments.

[0034] Terms such as "first" or "second" may be used to describe various components, but these terms should be interpreted solely for the purpose of distinguishing one component from another. For example, the first component may be named the second component, and similarly, the second component may be named the first component.

[0035] When it is stated that a component is "connected" to another component, it should be understood that it may be directly connected to or coupled with that other component, or that there may be other components in between.

[0036] The singular expression includes the plural expression unless the context clearly indicates otherwise. In this specification, terms such as "comprising" or "having" are intended to specify the existence of the described features, numbers, steps, actions, components, parts, or combinations thereof, and should be understood as not precluding the existence or addition of one or more other features, numbers, steps, actions, components, parts, or combinations thereof.

[0037] Unless otherwise defined, all terms used herein, including technical or scientific terms, have the same meaning as generally understood by those skilled in the art. Terms such as those defined in commonly used dictionaries should be interpreted as having a meaning consistent with their meaning in the context of the relevant technology, and should not be interpreted in an ideal or overly formal sense unless explicitly defined in this specification.

[0038] Hereinafter, embodiments will be described in detail with reference to the attached drawings. In the description with reference to the attached drawings, identical components are given the same reference numeral regardless of the drawing number, and redundant descriptions thereof will be omitted.

[0039] Solar modules have a lifespan of over 30 years and can maintain an energy conversion efficiency of approximately 80% even after the performance guarantee period (e.g., 20 years) has passed. Therefore, it is important to maintain solar power plant systems to ensure the sustainability of the solar industry and minimize environmental impact.

[0040] However, government policy focuses only on the new installation of solar power plants and is lacking in aspects related to maintenance that maximize power generation after installation. In fact, regarding solar power generation by public institutions nationwide, the Korea Institute of Energy Research analyzed that as of the end of 2019, 1,084 units, or 34.3% of the total 3,160 solar units installed by local governments and / or public institutions, were poorly managed.

[0041] Accordingly, with the increasing number of solar power generation facilities, technological development in the maintenance of solar modules is crucial for the stable operation of power plants, regular reliability enhancement, and preventing efficiency degradation in the power generation system. In particular, while recently produced products have significantly improved in performance, they have the disadvantage of rapidly dropping in efficiency if dust issues or abnormalities are not addressed early. To prevent such efficiency reduction, solar power generation systems require the accurate diagnosis of failures occurring in solar modules; thus, the importance of developing monitoring systems to diagnose the condition of modules is also emerging.

[0042] Furthermore, although failures in a photovoltaic power generation system can occur in various parts, it may be difficult to determine from the surface whether the system is operating, stopped, running at maximum performance, or if a partial failure has occurred, as the system itself operates silently and lacks moving parts. To solve this difficulty, a method and device for diagnosing a photovoltaic module failure according to an embodiment of the present invention can perform a fault diagnosis of a photovoltaic module using a linguistic model.

[0043] FIG. 1 shows a solar module fault diagnosis system according to one embodiment.

[0044] A solar module fault diagnosis system (100) according to one embodiment may include a solar module (110) and a solar module fault diagnosis device (120).

[0045] A solar module (110) may refer to an arrangement of multiple solar cells that convert light energy within a frame into electrical energy. The solar module fault diagnosis system (100) may monitor a single solar module (110), but is not limited thereto and may also monitor multiple solar modules (110). Additionally, the solar module fault diagnosis device (120) may be connected to an output unit (e.g., a display) to output the diagnosis result directly through the output unit, or connected to a separate management server to transmit the diagnosis result to the management server.

[0046] The solar module fault diagnosis device (120) may include a communication unit (121) and a processor (122). Additionally, the solar module fault diagnosis device (120) may further include a fault diagnosis model.

[0047] The communication unit (121) can be connected to the solar module (110) via wired or wireless connection and receive the amount of solar power generated from the solar module. Additionally, the communication unit (121) can receive data regarding the weather conditions at the location where the solar module (110) is placed through a separately provided sensor. The data regarding the weather conditions may include horizontal irradiance, slope irradiance, and temperature, but is not limited thereto. Additionally, the data regarding the weather conditions may include humidity and wind speed. Furthermore, the communication unit (121) can receive learning data regarding the weather conditions through a separate storage or external server for the training of a fault diagnosis model.

[0048] The processor (122) can diagnose a fault in the solar module based on data regarding the weather conditions received through the communication unit (121).

[0049] FIG. 2 shows the flow of diagnosing a fault in a solar module in a solar module fault diagnosis system according to one embodiment.

[0050] The processor (122) can receive data (210) regarding horizontal solar radiation, inclined solar radiation, and temperature from the sensor through the communication unit (121). The data (210) regarding horizontal solar radiation, inclined solar radiation, and temperature may be average values ​​over a certain period of time (e.g., 1 hour). Accordingly, the processor (122) can perform fault diagnosis on the solar module for each period, with the certain period of time as one period. Additionally, the processor (122) can perform fault diagnosis on the solar module when the value of horizontal solar radiation or inclined solar radiation is greater than or equal to a predetermined solar radiation value (e.g., during the time when there is solar radiation).

[0051] The processor (122) can calculate the predicted solar power generation amount (220) when data (210) (e.g., multiple heterogeneous data) regarding horizontal solar irradiance, inclined solar irradiance, and temperature is input into the fault diagnosis model. For example, the data input into the fault diagnosis model may be in the form of a 3x8 matrix if the input data (210) regarding horizontal solar irradiance, inclined solar irradiance, and temperature is input in the form of 3-dimensional data for 8 hours a day. Since the processor (122) generates one output per cycle using the fault diagnosis model, if 8 3-dimensional vectors are input into the fault diagnosis model, 8 1-dimensional values ​​can be output. The predicted solar power generation amount (220) is output in the form of a range with upper and lower limits, and the fault of the solar module can be diagnosed based on whether the actual solar power generation amount (230) is within the range (240).

[0052] FIG. 3 illustrates a flowchart of a solar module fault diagnosis method according to one embodiment.

[0053] In step (310), the processor can input multiple heterogeneous data related to sunlight (e.g., horizontal solar radiation, inclined solar radiation, and temperature data) into the fault diagnosis model to calculate a model estimate (y). Prior to step (310), the processor can model the fault diagnosis model. The modeling method is described in detail later in FIG. 7.

[0054] In step (320), the processor determines the upper limit weight (w) based on the predetermined fuzzy number (α) of the fault diagnosis model. t+ ) and lower weight (w t- ) can be calculated. The fuzzy number (α) is a value greater than or equal to 0 and less than 1, and the upper weight (w) using this can be calculated. t+ ) and lower weight (w t- The calculation is described in detail later in Fig. 5.

[0055] In step (330), the processor has an upper weight (w t+ ) and lower weight (w t-Based on ), the upper limit of the solar module's power generation (y) from the model estimate + ) and lower limit (y - It can predict ). The processor can calculate a bias value (w0) based on the difference between the model estimate and the target value. The processor calculates the bias value, upper weight (w t+ ) and lower weight (w t- ) and the output value of each group (Z t Based on ), the upper limit (y + ) and lower limit (y - ) can be predicted. Using the bias value (w0), the upper limit (y + ) and lower limit (y - The method for calculating ) will be described in detail later in mathematical formula 5.

[0056] In step (340), the processor receives the solar power generation amount and the predicted upper limit (y) from the solar module. + ) and lower limit (y - Solar module failures can be diagnosed based on ). The received solar power generation is the predicted upper limit (y + ) and lower limit (y - If it is not within the ) range, the processor can diagnose that the corresponding solar module has failed. In addition, the processor upper limit value (y + ) and lower limit (y - It can additionally perform an operation to determine whether the solar module is experiencing a simple failure, efficiency degradation due to dust, or other abnormal conditions based on calculating the degree to which it deviates from the range or the solar power generation patterns of previous cycles. Meanwhile, the received solar power generation is the predicted upper limit value (y + ) and lower limit (y - If it is within the range, the processor can diagnose that the solar module is operating normally.

[0057] After step (340), the processor can further train the fault diagnosis model using multiple heterogeneous data and solar power generation amounts used in the cycle. The operation of further training while performing fault diagnosis can be optimized so that the fault diagnosis model can perform more accurate fault diagnosis. In addition, the operation of further training can reduce the burden of training even when modeling the fault diagnosis model.

[0058] Figure 4 shows the flow of the step of calculating a model estimate among a solar module fault diagnosis method according to one embodiment.

[0059] In step (410), the processor uses a fault diagnosis model to apply a context-based fuzzy clustering algorithm to the membership degree (u) of a three-dimensional input value (x) (e.g., multiple heterogeneous data) in one cycle. i,k ) can be calculated. The fault diagnosis model can form p clusters (412) by arbitrarily selecting cluster centers (411). This may mean that each cluster (412) contains c cluster centers (411). The fault diagnosis model [calculates] each degree of membership (u) for the center i,k Can calculate ). Degree of membership (u i,k ) can represent a fuzzy degree of membership, which is the degree to which it belongs to each of multiple partial clusters (e.g., cluster centers (411)) depending on the context. For example, the degree of membership (u i,k ) can represent the degree of membership in which the k-th data value belongs to the i-th cluster. In addition, the degree of membership (u i,k ) can be an element of the membership matrix (U). Accordingly, the degree of membership (u i,k ) is the degree of membership (u) of the location closest to the location of the input value (x) among each center. i,k ) can have the largest value. Degree of membership (u i,k ) can be calculated using mathematical formula 1.

[0060]

[0061] In the aforementioned mathematical formula 1, m can represent the fuzzification coefficient. f k W(d) is a value greater than or equal to 0 and less than 1 that represents the inclusion degree of the k-th data in the fuzzy set (W). k It can be calculated using ) and can represent the membership degree in a specific context. In addition, f k is a condition, degree of membership (u i,k Can represent the sum of ). x k is a data value (e.g., a 3D vector) at an arbitrary time point (k-th), and c i is the center of the i-th cluster, and c j can represent the centers of clusters 1 through c as the center of the j-th cluster.

[0062] In step (420), the processor can use a fault diagnosis model to calculate the output value of the corresponding group based on the degrees of membership of the sub-clusters of the same cluster (412) among a plurality of sub-clusters. For example, the processor can calculate u, which are degree values ​​of membership belonging to the first cluster. 11 , u 1i , u 1c Z1 can be calculated by summing them. Additionally, the processor uses u, which are the degree of membership values ​​belonging to cluster t. t1 , u ti , u tc Add Z t Calculate and u, which are the degree of membership values ​​belonging to the p-th cluster. p1 , u pi , u pc Add Z p It can calculate.

[0063] In step (430), the processor uses a fault diagnosis model to determine the output values ​​(Z) of multiple clusters. t ) weights(w tThe model estimate (y) can be calculated based on the sum of the results of applying ). Weights (w t ) can be iteratively trained by steepest descent to minimize the error between the target value and the estimate. The model estimate (y) can be calculated using Equation 2.

[0064]

[0065] In the aforementioned Equation 2, w0 represents a numerical bias term and is connected to a node in the output layer, which can reduce the error by adjusting the biased output value of the model. Additionally, w0 can be calculated using Equation 3.

[0066]

[0067] In the aforementioned mathematical formula 3, the target k represents the target value of the k-th data, and y k is input data(x k The output of the linguistic model for ) (e.g., the model prediction value for the k-th data) can be represented. Accordingly, the bias value (w0) is not limited to being calculated in step (430) but can be calculated and used after modeling the fault diagnosis model.

[0068] In addition, the processor has a bias value, an upper weight (w t+ ) and lower weight (w t- Upper and lower limits can be predicted based on ) and the output values ​​of each group. Upper weight (w t+ ) and lower weight (w t- ) can be calculated based on the fuzzy number (α), which is represented as in Equation 4.

[0069]

[0070] When the fuzzy number (α) is 0, the fuzzy confidence interval has the output boundary of the existing linguistic model, and as the fuzzy number (α) approaches 1, it may mean that the confidence interval of the model decreases.

[0071] Figure 5 shows a weighting range that varies according to the fuzzy number in a solar module fault diagnosis method according to one embodiment.

[0072] In FIG. 5, the fuzzy number (α) can form a wide range of weights (510) as it approaches 0, and a narrow range of weights (520) as it approaches 1.

[0073] Upper limit (y + ) and lower limit (y - ) can be predicted using mathematical equation 5, which is similar to the model estimate (y).

[0074]

[0075]

[0076] FIG. 6 shows the performance evaluation results of a photovoltaic module fault diagnosis device according to one embodiment.

[0077] The performance of the solar module fault diagnosis device was evaluated using solar public data. The solar module fault diagnosis device can predict upper limit values ​​(610) and lower limit values ​​(640) using a fault diagnosis model. The graph shows that while the actual value (630) and the model estimate value (620) show somewhat similar values ​​at some points in time, there are more points with deviations of a certain magnitude than points showing completely identical results. However, as the actual value (630) is generally included within the range of the upper limit value (610) and lower limit value (640), it can be confirmed that diagnosing faults using range values ​​provides higher diagnostic performance than diagnosing faults using a single value.

[0078] Meanwhile, the processor can model a fault diagnosis model to calculate the model estimate (y). The fault diagnosis model can be modeled as a linguistic model (LM), but is not limited thereto.

[0079] FIG. 7 illustrates a flowchart of a method for modeling a fault diagnosis model among a photovoltaic module fault diagnosis method according to one embodiment.

[0080] In step (710), the processor can determine the number of linguistic contexts and the number of clusters to be formed based on each context, and initialize a membership matrix (U). The membership matrix (U) may have multiple degrees of membership clustered to form multiple clusters. As contexts are one of the data processing elements, increasing the number of contexts can improve data processing capabilities, but at the same time, computational costs and processing load increase, so an appropriate value can be set for the number of contexts. Additionally, increasing the number of clusters can improve model performance, but exceeding an appropriate number of clusters may cause the model to become excessively complex, so the number of clusters can be set to an appropriate value. The processor can generate linguistic contexts by a triangular membership function (e.g., a Gaussian function) that is evenly distributed in the output space.

[0081] In step (720), the processor clusters a cluster center (c) for each of the multiple clusters based on the result of clustering multiple heterogeneous learning data. i Can calculate ). Cluster center (c i ) can be calculated using mathematical formula 6.

[0082]

[0083] In step (730), the processor can calculate the objective function value based on the cluster centers of multiple clusters. The objective function value can be calculated using Equation 7.

[0084]

[0085] d 2 ikIt can represent the Euclidean distance between the i-th cluster center among multiple clusters and the k-th data point among the input elements of multiple heterogeneous training data.

[0086] In step (740), the processor can adjust the clustering of the plurality of clusters based on the objective function value. The processor can determine the difference between the objective function of the current period (p) and the objective function of the previous period (p-1) as expressed in Equation 8 below using a threshold value ( It can be compared with ).

[0087]

[0088] Based on the comparison results, the processor may terminate the clustering adjustment or calculate a new membership matrix. If the difference between the objective function of the current period and the objective function of the previous period is smaller than the aforementioned threshold, the processor may terminate the clustering adjustment. This may mean stopping optimization if the degree of improvement compared to the previous period is small. Additionally, if the difference between the objective function of the current period and the objective function of the previous period is greater than the aforementioned threshold, the processor may calculate a new membership matrix by shifting the cluster centers to enable optimal clustering. Accordingly, the processor may calculate new cluster centers based on the new membership matrix. Furthermore, the processor may adjust the variance to enable optimal clustering. Accordingly, the processor may perform either shifting the cluster centers or adjusting the variance, or perform both. For example, when using a Gaussian function defined for three dimensions, the processor can modify the mean_x, mean_y, mean_z, sigma_x, sigma_y, and sigma_z values ​​when shifting cluster centers and adjusting variance. This means that when calculating nine cluster centers, there may be nine mean_x, mean_y, mean_z, sigma_x, sigma_y, and sigma_z values.

[0089] The embodiments described above may be implemented as hardware components, software components, and / or combinations of hardware and software components. For example, the devices, methods, and components described in the embodiments may be implemented using a general-purpose computer or a special-purpose computer, such as, for example, a processor, a controller, an arithmetic logic unit (ALU), a digital signal processor, a microcomputer, a field programmable gate array (FPGA), a programmable logic unit (PLU), a microprocessor, or any other device capable of executing and responding to instructions. The processing unit may execute an operating system (OS) and software applications executed on said operating system. Additionally, the processing unit may access, store, manipulate, process, and generate data in response to the execution of the software. For ease of understanding, the processing unit may be described as being used as a single unit, but those skilled in the art will understand that the processing unit may include multiple processing elements and / or multiple types of processing elements. For example, the processing unit may include multiple processors or one processor and one controller. In addition, other processing configurations, such as parallel processors, are also possible.

[0090] Software may include computer programs, code, instructions, or a combination of one or more of these, and may configure a processing unit to operate as desired or instruct the processing unit independently or collectively. Software and / or data may be stored on any type of machine, component, physical device, virtual equipment, computer storage medium, or device so as to be interpreted by the processing unit or to provide instructions or data to the processing unit. Software may be distributed over networked computer systems and stored or executed in a distributed manner. Software and data may be stored on computer-readable recording media.

[0091] The method according to the embodiment may be implemented in the form of program instructions that can be executed through various computer means and recorded on a computer-readable medium. The computer-readable medium may include program instructions, data files, data structures, etc., either alone or in combination, and the program instructions recorded on the medium may be those specifically designed and configured for the embodiment or those known and available to those skilled in the art of computer software. Examples of computer-readable recording media include magnetic media such as hard disks, floppy disks, and magnetic tapes; optical recording media such as CD-ROMs and DVDs; magneto-optical media such as floptical disks; and hardware devices specifically configured to store and execute program instructions, such as ROM, RAM, and flash memory. Examples of program instructions include machine code, such as that generated by a compiler, as well as high-level language code that can be executed by a computer using an interpreter, etc.

[0092] The hardware device described above may be configured to operate as one or more software modules to perform the operation of the embodiment, and vice versa.

[0093] Although the embodiments have been described above with reference to the limited drawings, those skilled in the art can apply various technical modifications and variations based thereon. For example, suitable results may be achieved even if the described techniques are performed in a different order than described, and / or if the components of the described system, structure, device, circuit, etc. are combined or assembled in a form different from described, or replaced or substituted by other components or equivalents.

[0094] Therefore, other implementations, other embodiments, and equivalents to the claims also fall within the scope of the claims set forth below.

Claims

Claim 1 A method for diagnosing a fault in a solar module using a fault diagnosis model, comprising: a step of inputting a plurality of heterogeneous data related to solar energy into the fault diagnosis model to calculate a model estimate value; a step of calculating an upper weight and a lower weight based on a predetermined fuzzy number of the fault diagnosis model; a step of predicting an upper and lower limit value of the power generation amount of the solar module from the model estimate value based on the upper and lower weights; and a step of diagnosing a fault in the solar module based on the solar energy generation amount received from the solar module and the predicted upper and lower limits. The method further comprises a step of modeling the fault diagnosis model, wherein the step of modeling the fault diagnosis model includes: a step of determining the number of contexts and the number of clusters to be formed based on each context and initializing a membership matrix; a step of calculating a cluster center for each of the plurality of clusters based on the result of clustering a plurality of heterogeneous learning data; a step of calculating an objective function value based on the cluster centers of the plurality of clusters; and a step of adjusting the clustering of the plurality of clusters based on the objective function value. Claim 2 A method for diagnosing a photovoltaic module fault according to claim 1, wherein the step of calculating the model estimate comprises: a step of calculating a degree of membership as a component of a membership matrix that indicates the degree to which the plurality of heterogeneous data belong to each of the plurality of partial clusters according to context; a step of calculating an output value of a group based on the degrees of membership of the partial clusters of the same cluster among the plurality of partial clusters; and a step of calculating the model estimate based on the sum of the results of applying weights to the output values ​​of the plurality of clusters. Claim 3 A method for diagnosing a photovoltaic module fault according to claim 1, wherein the step of predicting the upper and lower limits comprises: a step of calculating a bias value based on the difference between the model estimate value and the target value; and a step of predicting the upper and lower limits based on the bias value, the upper limit weight and the lower limit weight and the output value of the group. Claim 4 delete Claim 5 A method for diagnosing a solar module fault according to claim 1, wherein the step of adjusting the clustering comprises: a step of comparing the difference between the objective function of the current period and the objective function of the previous period with a threshold value; a step of terminating the adjustment of the clustering if the difference between the objective function of the current period and the objective function of the previous period is smaller than the threshold value; and a step of calculating a new membership matrix and calculating the cluster center if the difference between the objective function of the current period and the objective function of the previous period is larger than the threshold value. Claim 6 A method for diagnosing solar module failures according to claim 5, wherein the step of adjusting the clustering further includes the step of calculating a new membership matrix by shifting the cluster center when the difference between the objective function of the current cycle and the objective function of the previous cycle is greater than the threshold value. Claim 7 A method for diagnosing a photovoltaic module fault according to claim 1, wherein the step of calculating the objective function value comprises: a step of calculating the Euclidean distance between the cluster centers of the plurality of clusters and the input elements of the plurality of heterogeneous learning data; and a step of calculating the objective function value based on the Euclidean distance and the degree of membership. Claim 8 A solar module fault diagnosis method according to claim 1, further comprising the step of additionally training the fault diagnosis model using the plurality of heterogeneous data and the solar power generation amount. Claim 9 A method for diagnosing a photovoltaic module fault according to claim 1, wherein the plurality of heterogeneous data include horizontal irradiance, slope irradiance, and temperature. Claim 10 A solar module fault diagnosis device comprising: a communication unit that receives solar power generation amount from a solar module; and a processor that inputs a plurality of heterogeneous data related to solar power into a fault diagnosis model to calculate a model estimate value, calculates an upper weight and a lower weight based on a predetermined fuzzy number of the fault diagnosis model, predicts an upper and lower limit value of the solar module's power generation amount from the model estimate value based on the upper and lower weights, and diagnoses a fault of the solar module based on the solar power generation amount and the predicted upper and lower limits, wherein the processor determines the number of contexts and the number of clusters to be formed based on each context to model the fault diagnosis model, initializes a membership matrix, calculates a cluster center for each of the plurality of clusters based on the result of clustering a plurality of heterogeneous learning data, calculates an objective function value based on the cluster centers of the plurality of clusters, and adjusts the clustering of the plurality of clusters based on the objective function value. Claim 11 A solar module fault diagnosis device according to claim 10, wherein the processor calculates a degree of membership indicating the degree to which the plurality of heterogeneous data belongs to each of the plurality of partial clusters according to context as a component of a membership matrix, calculates an output value of the corresponding group based on the degrees of membership of the partial clusters of the same cluster among the plurality of partial clusters, and calculates the model estimate value based on the sum of the results of applying weights to the output values ​​of the plurality of clusters. Claim 12 A solar module fault diagnosis device according to claim 10, wherein the processor calculates a bias value based on the difference between the model estimate value and the target value, and predicts the upper limit value and the lower limit value based on the bias value, the upper limit weight and the lower limit weight and the output value of the group. Claim 13 delete Claim 14 A solar module fault diagnosis device according to claim 10, wherein the processor terminates the adjustment of clustering if the difference between the objective function of the current cycle and the objective function of the previous cycle is smaller than a threshold value, and calculates the cluster center by calculating a new membership matrix if the difference between the objective function of the current cycle and the objective function of the previous cycle is larger than the threshold value. Claim 15 In claim 14, the above processor calculates a new membership matrix by shifting the cluster center when the difference between the objective function of the current cycle and the objective function of the previous cycle is greater than the threshold value, a solar module fault diagnosis device. Claim 16 A solar module fault diagnosis device according to claim 10, wherein the processor calculates the Euclidean distance between each cluster center and the input elements of the plurality of heterogeneous learning data, and calculates an objective function value based on the Euclidean distance and the degree of membership of each context. Claim 17 In claim 10, the above processor further trains the fault diagnosis model using the above plurality of heterogeneous data and the above solar power generation amount, a solar module fault diagnosis device. Claim 18 In item 10, the above-mentioned plurality of heterogeneous data includes horizontal solar radiation, inclined solar radiation, and temperature, a photovoltaic module fault diagnosis device.

Images