KR102459015B1 - Failure detection method for photovoltaic power generating system and analysis apparatus - Google Patents

Abstract

A method of diagnosing a failure of a photovoltaic system using an artificial neural network includes the steps of: an analyzing device receiving environmental information of a target unit of a photovoltaic system, and the analyzing device converting the environmental information to a first artificial neural network ) and outputting generation amount-related prediction information, and outputting, by the analysis device, the generation amount-related prediction information and the measured generation amount-related information of the target unit into a second artificial neural network to output the state information of the target unit includes

Description

FAILURE DETECTION METHOD FOR PHOTOVOLTAIC POWER GENERATING SYSTEM AND ANALYSIS APPARATUS

The technology to be described below relates to a technique for predicting the amount of power generation of a photovoltaic system and diagnosing a failure.

Environmental problems such as climate change and resource depletion have caused changes in the energy market. For example, recently, research on renewable energy has received a lot of attention. Solar energy is a representative renewable energy.

A photovoltaic power generation system includes a photovoltaic module and an inverter for AC conversion. A photovoltaic module is a core component of a photovoltaic system and is a component that converts sunlight into electrical energy. The solar power generation system may separately use a monitoring system for monitoring whether there is a failure. Solar modules exposed to the external environment are the main cause of failure.

Korean Patent No. 10-2148761

The defect detection technique of solar modules mainly uses a method of directly checking the high-temperature panel using a thermal infrared camera. This method requires a separate equipment for taking an image, and there are problems in that the accuracy of detecting a defective panel is not high.

In another aspect, a conventional method of monitoring solar power generation includes a method of comparing the previously measured power generation amount and the current power generation amount according to the season and the weather condition of the day, and determining whether the generation is in a normal power generation state according to the difference. This method has a problem in that it is difficult to accurately determine the various environmental information and its subtle changes.

The technology to be described below is intended to provide a method of predicting the amount of power generation of a solar power generation system using an artificial neural network model. In addition, the technology to be described below is intended to provide a method of diagnosing the state of the solar power generation system using different artificial neural network models based on the predicted generation amount and the current generation amount. The technology to be described below is intended to provide a method for diagnosing a failure in a solar module or panel unit.

The method for detecting a failure of a photovoltaic system includes the steps of: an analyzing device receiving environmental information of a target unit of a photovoltaic power generation system; outputting information and outputting, by the analysis device, the power generation related prediction information and the measured power generation related information of the target unit to a second artificial neural network, and outputting the status information of the target unit.

In another aspect, the method for detecting a failure of a photovoltaic system includes the steps of, by an analysis device, receiving environmental information of a target unit of the photovoltaic power generation system and information related to the amount of power generated by the target unit at a specific point in time, wherein the analysis device receives the environmental information and outputting state information of the target unit by inputting the generation amount-related information into an artificial neural network, and determining, by the analysis device, whether the target unit has a failure based on the status information.

In another aspect, the analysis device for monitoring the operating state of the photovoltaic system includes an input device that receives environmental information of the target unit of the photovoltaic system and the measured power generation amount of the target unit, and prediction information related to the amount of power generation based on the environmental information A storage device for storing a first artificial neural network for outputting a second artificial neural network for outputting state information based on prediction information related to generation amount and measured generation amount related information, and a storage device for storing environmental information input to the input device at a specific point in time 1 to generate generation amount-related prediction information by inputting it into an artificial neural network, and input the generation amount-related prediction information and the measured generation amount-related information of the target unit at the specific point in time into the second artificial neural network to obtain the state information of the target unit It contains an arithmetic unit that generates it.

The technology to be described below can quickly diagnose the state of the photovoltaic system in real time based on an artificial neural network model. The technology to be described below can diagnose the state of the photovoltaic system without using expensive equipment. The technology to be described below can diagnose the state of the solar module or panel based on the inverter capable of measuring the output current, and determine the fault location in the solar module or panel unit. Through this, the technology to be described below effectively manages the solar power generation system and greatly reduces the time and cost required for fault diagnosis.

1 is an example of a system for detecting a failure of a photovoltaic system.
2 is an example of an artificial neural network model for predicting the amount of power generation of a solar power generation system.
3 is an example of a process for detecting a failure of a photovoltaic system.
4 is an example of an artificial neural network model for predicting state information of a solar power generation system.
5 is another example of a process for detecting a failure of a photovoltaic system.
6 is another example of an artificial neural network model for predicting state information of a photovoltaic system.
7 is an example of an analysis device.

Since the technology to be described below can have various changes and can have various embodiments, specific embodiments are illustrated in the drawings and described in detail. However, this is not intended to limit the technology described below to specific embodiments, and it should be understood to include all modifications, equivalents, and substitutes included in the spirit and scope of the technology described below.

Terms such as first, second, A, and B may be used to describe various components, but the components are not limited by the above terms, and only for the purpose of distinguishing one component from other components. used only as For example, a first component may be named as a second component, and similarly, the second component may also be referred to as a first component without departing from the scope of the technology to be described below. and/or includes a combination of a plurality of related listed items or any of a plurality of related listed items.

In terms of terms used herein, the singular expression should be understood to include the plural expression unless the context clearly dictates otherwise, and terms such as "comprises" include the described feature, number, step, operation, element. , parts or combinations thereof are to be understood, but not to exclude the possibility of the presence or addition of one or more other features or numbers, step operation components, parts or combinations thereof.

Prior to a detailed description of the drawings, it is intended to clarify that the classification of the constituent parts in the present specification is merely a division according to the main function that each constituent unit is responsible for. That is, two or more components to be described below may be combined into one component, or one component may be divided into two or more for each more subdivided function. In addition, each of the constituent units to be described below may additionally perform some or all of the functions of other constituent units in addition to the main function it is responsible for. Of course, it can also be performed by being dedicated to it.

In addition, in performing the method or operation method, each process constituting the method may occur differently from the specified order unless a specific order is clearly described in context. That is, each process may occur in the same order as specified, may be performed substantially simultaneously, or may be performed in the reverse order.

Hereinafter, configurations or terms used in the description will be briefly described.

There are various types of solar power generation systems. A photovoltaic power generation system basically includes a photovoltaic module that converts sunlight into direct current, an inverter that converts direct current into alternating current, and a control device. The following description will focus on the necessary components of the solar power generation system. However, the technology to be described below can of course be applied to various types of photovoltaic power generation systems including a distribution system, a power regulator, and a grid-connected device.

A solar module is composed of a plurality of cells. In addition, in general, a photovoltaic panel means a configuration in which a plurality of photovoltaic modules are aggregated. The technology to be described below diagnoses a failure of a photovoltaic module or a photovoltaic panel in an operating state (failure state or normal state). A photovoltaic module or photovoltaic panel means one unit unit for power generation. The solar power generation system includes one or more unit units.

Hereinafter, the target unit means at least one solar module or at least one solar panel. The target unit is the target of operation state diagnosis. The solar power generation system may be composed of a plurality of target units.

The technology described below predicts the amount of solar power generation using a learning model and diagnoses the operating state of the target unit. The learning model refers to a machine learning model. A learning model is meant to include various types of models. For example, the learning model includes a decision tree, a random forest, a K-nearest neighbor (KNN), a naive Bayes, a support vector machine (SVM), an artificial neural network, and the like. The technology to be described below may use an artificial neural network. The following description will focus on artificial neural networks.

Artificial neural networks are statistical learning algorithms that mimic the neural networks of living things. Various neural network models are being studied. Recently, deep learning networks (DNNs) are attracting attention.

DNN is an artificial neural network model consisting of several hidden layers between an input layer and an output layer. DNN can model complex non-linear relationships like general artificial neural networks. Various types of DNN models have been studied. For example, there are Convolutional Neural Network (CNN), Recurrent Neural Network (RNN), Restricted Boltzmann Machine (RBM), Deep Belief Network (DBN), Generative Adversarial Network (GAN), Relation Networks (RL), and the like.

1 is an example of a system 100 for detecting a failure of a photovoltaic system. The configuration indicated by A in FIG. 1 is a configuration included in the solar power generation system.

The solar power generation system A includes target units 111 , 112 , 113 , inverters 121 , 122 , 123 , and a storage battery 130 . 1 exemplarily shows three target units and three inverters. The target units 111 , 112 , and 113 correspond to at least one solar module or at least one solar panel as described above.

1 illustrates a form in which an inverter is connected to each target unit. Hereinafter, it is assumed that individual inverters are connected to each target unit for convenience of description. The photovoltaic power generation system (A) transmits the generation amount related information to the analysis device 150 in order to predict the generation amount or detect a failure. The power generation-related information is a value measured by each target unit and individual inverter. The generation amount-related information may be transmitted to the analysis device 150 through a control device not shown in FIG. 1 .

The generation amount-related information is information that can confirm the generation amount generated by the target unit. The generation amount-related information may be single information or a plurality of pieces of information. For example, the amount of power generation-related information includes an output current of at least one of three-phase wiring of the inverter to which the current output by the target unit is applied, the power of the inverter, the current of the solar cell, the voltage of the solar cell, and the output value including the power of the solar cell may include at least one of Taeang battery means the target unit. The three-phase inverter refers to a wiring method that transmits alternating current, and the three phases are also called R-phase, T-phase, and S-phase, respectively.

The sensor device 140 is disposed at a predetermined distance from the target unit to collect environmental information. The sensor device 140 may be configured in various types according to collected information, and a plurality of sensor devices may be used. In some cases, the sensor device 140 may be a component included in the photovoltaic power generation system (A). In this case, the analysis device 150 may receive the environment information through the control device.

The environmental information may include at least one of an information group including external temperature, accumulated precipitation, wind direction, wind speed, humidity, atmospheric pressure, insolation, sunlight, slope insolation, and rear temperature of the solar module.

Meanwhile, at least some of the environment information may be transmitted from a separate object. For example, the analysis device 150 may receive weather information from the database 50 managed by an organization or company that provides weather-related information.

The analysis device 150 may predict the amount of power generation by using the received environmental information and the amount of power generation related information, and may diagnose the operating state of the target unit. The analysis device 150 may input the input data into the previously learned artificial neural network, and may predict the amount of power generation or diagnose the operating state of the target unit. The analysis device 150 may be implemented as a device such as a smart device, a PC, or a server. Furthermore, the analysis device 150 may transmit the predicted result and the diagnosis result to the manager terminal 80 .

The analysis device 150 may be a component (monitoring device) included in the photovoltaic system, or may be a separate object located outside the photovoltaic system.

2 is an example of an artificial neural network model for predicting the amount of power generation of a solar power generation system. The artificial neural network model for predicting the amount of power generation is called the first artificial neural network 200 .

FIG. 2A is an example illustrating the operation of the learned first artificial neural network 200 . The first artificial neural network 200 receives environmental information at a specific point in time and outputs prediction information related to the amount of power generated by the target unit at that point in time.

The environmental information may include at least one of an information group including external temperature, accumulated precipitation, wind direction, wind speed, humidity, atmospheric pressure, insolation, sunlight, slope insolation, and rear temperature of the solar module.

If the environmental information is a single type of information, it may be configured in a one-dimensional form. FIG. 2 is shown in the form of a two-dimensional matrix, assuming that environment information is composed of a plurality of types. Environment information is composed of constant vector values through pre-processing. For example, the analyzer may convert the environment information into a vector value using a technique such as one-hot encoding.

The generation amount-related prediction information is information serving as a criterion for determining the generation amount of the target unit. The generation amount-related prediction information includes the output current of at least one of the three-phase wiring of the inverter to which the current output by the target unit is applied, the power of the inverter, the current of the solar cell, and the voltage of the solar cell. It may include at least one of the output values including the power of the solar cell.

The input data used in the prediction process is preferably the same as the input data used in the learning process.

FIG. 2B is an example illustrating a process of learning the first artificial neural network 200 . The learning process is performed through the aforementioned analysis device or a separate computer device. The first artificial neural network 200 learning process includes forward propagation and backward propagation.

Describe the forward transcription process. The input data is environmental information at a specific point in time. The environment information is the same as described above. The first artificial neural network 200 receives environmental information, which is input data, and outputs a predetermined output value. The output value is information that can determine the amount of power generation of the target unit. The output value is composed of the same type of information as the above-described generation amount-related prediction information.

Describe the reverse transcription process. The label value (label) is information received from the photovoltaic system. The label value is a value actually measured by the photovoltaic system at a specific point in time. The label value is information related to the amount of power generation corresponding to the output value. The generation amount-related prediction information means information predicted by the first artificial neural network 200, and the generation amount-related information means a value actually measured by the solar power generation system. The amount of power generation-related information includes the output current of at least one of the three-phase wiring of the inverter to which the current output by the target unit is applied, the power of the inverter, the current of the solar cell, and the voltage of the solar cell. It may include at least one of the output values including the power of the solar cell. The reverse transcription process corresponds to a process of updating the weight of the artificial neural network layer so that the difference between the output value and the label value is minimized by comparing the output value of the first artificial neural network 200 with the label value. The weight setting process may utilize a technique such as AutoML.

Table 1 below is an example of a training data set used in the process of FIG. 2(B).

hour input data label value t ₁ Environmental information 1 Information about power generation 1 t ₂ Environmental information 2 Information about power generation 2 ... ... ... t _n environmental information n Information about power generation n

The input data is environmental information, and the label value is information related to power generation. The training data set consists of environmental information and power generation related information actually measured at a plurality of time points t ₁ to t _n . The first artificial neural network 200 is learned by repeating the learning process using a plurality of learning data.

When the first artificial neural network 200 of FIG. 2 is used, the amount of power generation of the target unit can be predicted at the time when the environment information is generated.

3 is an example of an artificial neural network model for predicting state information of a photovoltaic system. The first artificial neural network 200 has the same configuration as described with reference to FIG. 2 . The artificial neural network model for predicting the state information of the solar power generation system is called the second artificial neural network 300 .

FIG. 3A is an example showing the operations of the learned first artificial neural network 200 and the second artificial neural network 300 .

The first artificial neural network 200 receives environmental information at a specific point in time and outputs prediction information related to the amount of power generated by the target unit at that point in time.

The environmental information may include at least one of an information group including external temperature, accumulated precipitation, wind direction, wind speed, humidity, atmospheric pressure, insolation, sunlight, slope insolation, and rear temperature of the solar module. 3(A) shows environmental information in the form of a two-dimensional matrix as shown in FIG. 2(A).

The second artificial neural network 300 receives the generation amount-related prediction information and the generation amount-related information output by the first artificial neural network 200 and outputs the state information of the solar power generation system. The status information means information on the operating status (normal or faulty) of the target unit to be diagnosed.

The information related to the amount of power generation is information actually measured in the photovoltaic power generation system as described above. The amount of power generation-related information includes the output current of at least one of the three-phase wiring of the inverter to which the current output by the target unit is applied, the power of the inverter, the current of the solar cell, and the voltage of the solar cell. It may include at least one of the output values including the power of the solar cell.

The input data used in the prediction process is preferably the same as the input data used in the learning process.

FIG. 3B is an example illustrating a process of learning the second artificial neural network 300 . The learning process is performed through the aforementioned analysis device or a separate computer device. It is assumed that the first artificial neural network 200 is in a state where learning is completed. The second artificial neural network 300 learning process includes forward transcription and backward transcription.

Describe the forward transcription process. The input data is generation amount-related prediction information and generation amount-related information at a specific point in time.

The generation amount-related prediction information is information output by the first artificial neural network 200 that has received environmental information at a specific point in time. The information related to the amount of power generation is information actually measured in the photovoltaic power generation system at the same specific point in time.

The items of the generation amount-related prediction information and the generation amount-related information may correspond. Accordingly, the generation amount-related prediction information and the generation amount-related information are the output current of at least one of the three-phase wiring of the inverter to which the current output by the target unit is applied, the power of the inverter, the current of the solar cell, and the voltage of the solar cell, respectively. It may include at least one of the output values including the power of the solar cell.

The second artificial neural network 300 receives input data and outputs an output value. The output value corresponds to the operation state information of the photovoltaic system or the target unit.

Describe the reverse transcription process. The label value (label) is information received from the photovoltaic system. The label value is a value actually measured by the photovoltaic system at a specific point in time, and is information on the operating state (normal state or failure state) of the photovoltaic system or the target unit. The reverse transcription process corresponds to a process of updating the weight of the artificial neural network layer so that the difference between the output value and the label value is minimized by comparing the output value and the label value of the second artificial neural network 300 . The weight setting process may utilize a technique such as AutoML.

Table 2 below is an example of a training data set used in the process of FIG. 3(B). The input data includes input data 1 and input data 2 . Input data 1 is output from the first artificial neural network 200 It is forecast information related to power generation. Input data 2 is information related to power generation. The label value is operation state information of the photovoltaic system or the target unit.

The training data set includes prediction information related to the amount of generation predicted at a plurality of time points t ₁ to t _n , information related to the actual amount of power generation, and operation state information. The second artificial neural network 300 is learned by repeating the learning process using a plurality of learning data.

hour input data 1 input data 2 label value t ₁ Forecast information related to power generation 1 Information about power generation 1 Status information (Fault/Normal) t ₂ Forecast information related to power generation 2 Information about power generation 2 Status information (Fault/Normal) ... ... ... t _n Forecast information related to power generation n Information about power generation n Status information (Fault/Normal)

4 is an example of a process 400 for detecting a failure of a photovoltaic system. The analysis apparatus learns the first artificial neural network and the second artificial neural network as in the process described above ( 410 ). Alternatively, the analysis device may receive the learned first artificial neural network and the second artificial neural network from another object.

The analysis device collects environmental information and generation amount-related information at a specific time ( 420 ). As described with reference to FIG. 1 , the analysis device may receive environmental information from the sensor device. The analysis device may receive power generation related information from the solar power generation system.

The analysis device inputs the environmental information to the first artificial neural network to generate generation amount-related prediction information ( 430 ).

The analysis device inputs the generation amount-related prediction information and the measured generation amount-related information to the second artificial neural network, and outputs the state information of the solar power generation system (440).

The analysis device may diagnose whether the solar power generation system has failed based on the state information output by the second artificial neural network ( 450 ). For example, (i) the analysis device may diagnose that the photovoltaic system is faulty when the status information is in a faulty state even once. (ii) Alternatively, when a plurality of state information calculated from information measured at different times has a failure state more than a reference number of times, the analysis device may diagnose that the photovoltaic system is malfunctioning. (iii) Alternatively, when a plurality of state information calculated from information measured at different times is in a failure state continuously for a reference number or more, the analysis device may diagnose the photovoltaic power generation system as a failure. (iv) Alternatively, the analysis device may diagnose the photovoltaic system as a failure when the continuously calculated state information for a certain time period is in a failure state.

5 is another example of an artificial neural network model for predicting state information of a photovoltaic system. In FIG. 5 , the artificial neural network model for predicting the state information of the solar power generation system is called a third artificial neural network 500 .

FIG. 5A is an example illustrating the operation of the learned third artificial neural network 500 . The third artificial neural network 500 receives environmental information and generation amount-related information at a specific point in time and outputs status information of the solar power generation system at that point in time. The status information means information on the operating status (normal or faulty) of the target unit to be diagnosed.

The environmental information may include at least one of an information group including external temperature, accumulated precipitation, wind direction, wind speed, humidity, atmospheric pressure, insolation, sunlight, slope insolation, and rear temperature of the solar module.

The information related to the amount of power generation is information actually measured by the photovoltaic power generation system. The output current of at least one of the three-phase wiring of the inverter to which the current output by the target unit is applied, the power of the inverter, the current of the solar cell, the voltage of the solar cell ? ? It may include at least one of the output values including the power of the solar cell.

If the environmental information is a single type of information, it may be configured in a one-dimensional form. FIG. 5 is shown in the form of a two-dimensional matrix, assuming that environment information is composed of a plurality of types. If the generation amount-related information is a single type of information, it may be configured in a one-dimensional form. FIG. 5 is shown in the form of a two-dimensional matrix, assuming that the generation amount-related information is composed of a plurality of types. The input data is composed of constant vector values through preprocessing. For example, the analysis device may convert environmental information and power generation-related information into a vector value using a technique such as one-hot encoding. On the other hand, the analysis device may input the environmental information and the generation amount-related information to different input layers, respectively. Alternatively, the analysis device may merge the environmental information and the generation amount-related information into one matrix, and input the merged information to the input layer.

The input data used in the prediction process is preferably the same as the input data used in the learning process.

FIG. 5B is an example illustrating a process of learning the third artificial neural network 500 . The learning process is performed through the aforementioned analysis device or a separate computer device. The third artificial neural network 500 learning process includes forward propagation and backward propagation.

The third artificial neural network 500 learning process includes forward propagation and backward propagation.

Describe the forward transcription process. The input data is environmental information and power generation related information at a specific point in time. The environmental information and the information related to the amount of power generation are the same as described above, respectively. The third artificial neural network 500 receives input data, such as environmental information and generation amount-related information, and outputs a constant output value. The output value is operation state information (normal state or failure state) of the photovoltaic system or the target unit.

Describe the reverse transcription process. The label value (label) is information received from the photovoltaic system. The label value is a value actually measured by the photovoltaic system at a specific point in time. The label value is operation state information of the photovoltaic power generation system or the target unit at a specific point in time. The reverse transcription process corresponds to a process of updating the weight of the artificial neural network layer so that the difference between the output value and the label value is minimized by comparing the output value of the third artificial neural network 500 with the label value. The weight setting process may utilize a technique such as AutoML.

Table 3 below is an example of a training data set used in the process of FIG. 5(B). The input data includes input data 1 and input data 2 . Input data 1 is environment information. Input data 2 is information related to power generation. The label value is operation state information of the photovoltaic system or the target unit.

hour input data 1 input data 2 label value t ₁ Environmental information 1 Information about power generation 1 Status information (Fault/Normal) t ₂ Environmental information 2 Information about power generation 2 Status information (Fault/Normal) ... ... ... t _n environmental information n Information about power generation n Status information (Fault/Normal)

The learning data set is composed of environmental information at a plurality of time points t ₁ to t _n , information related to the measured power generation amount, and operation state information. The third artificial neural network 500 is learned by repeating the learning process using a plurality of learning data.

6 is another example of a process 600 of detecting a failure of a photovoltaic system. FIG. 6 corresponds to a process of diagnosing the operating state of the photovoltaic system using the third artificial neural network 500 of FIG. 5 .

The analysis apparatus learns and prepares the third artificial neural network as in the above-described process (610). Alternatively, the analysis device may receive the learned third artificial neural network from another object.

The analysis device collects environmental information and generation amount-related information at a specific time ( 620 ). As described with reference to FIG. 1 , the analysis device may receive environmental information from the sensor device. The analysis device may receive power generation related information from the solar power generation system.

The analysis device inputs the environmental information and the measured power generation amount-related information to the third artificial neural network and outputs status information of the solar power generation system (63).

The analysis device may diagnose whether the solar power generation system has failed based on the state information output by the third artificial neural network ( 640 ). For example, (i) the analysis device may diagnose that the photovoltaic system is faulty when the status information is in a faulty state even once. (ii) Alternatively, when a plurality of state information calculated from information measured at different times has a failure state more than a reference number of times, the analysis device may diagnose that the photovoltaic system is malfunctioning. (iii) Alternatively, when a plurality of state information calculated from information measured at different times is in a failure state continuously for a reference number or more, the analysis device may diagnose the photovoltaic power generation system as a failure. (iv) Alternatively, the analysis device may diagnose the photovoltaic system as a failure when the continuously calculated state information for a certain time period is in a failure state.

7 is an example of the analysis device 700 . The analysis device 700 of FIG. 7 corresponds to the analysis device 150 of FIG. 1 . The analysis device 700 may predict the amount of solar power generation using an artificial neural network based on the input data or diagnose the operating state of the solar power generation system. The analysis device 700 may be any one of various types, such as a smart device, a PC, or a server on a network.

The analysis device 700 may include a storage device 710 , a memory 720 , an arithmetic device 730 , an interface device 740 , a communication device 750 , and an output device 760 .

The storage device 710 stores the learning model. The storage device 710 may store at least one of the aforementioned (i) first artificial neural network, (ii) first and second artificial neural networks, and (iii) third artificial neural network.

The storage device 710 may store learning data for artificial neural network learning. Also, the storage device 710 may store source codes or programs for controlling input data preprocessing, artificial neural network control, and output data processing operations.

The storage device 710 may store environmental information received from an external object and information related to the amount of power generation. The storage device 710 may store power generation related prediction information output by the first artificial neural network. The storage device 710 may store operation state information of the solar power generation system output by the second artificial neural network or the third artificial neural network.

The memory 720 may store temporary data generated by the analysis device 700 in the process of constructing an artificial neural network, predicting the amount of power generation, or diagnosing the operating state of the solar power generation system.

The interface device 740 is a device that receives a predetermined command or information from a user. The interface device 740 may receive a predetermined command or data from an external input device. The interface device 740 may receive learning data, an artificial neural network model, a control program, and input data used for predicting the amount of power generation or the operation state of the solar power generation system, etc. from a physically connected external storage device.

The communication device 750 refers to a configuration for receiving and transmitting predetermined information through a network. The communication device 750 may receive learning data, an artificial neural network model, a control program, and input data used for predicting the amount of power generation or the operation state of the photovoltaic system from an external object. Also, the communication device 750 may transmit the analysis result to the external object.

The communication device 750 and the interface device 740 are devices that receive predetermined data or commands from the outside. The communication device 750 or the interface device 740 may be referred to as input devices.

The computing device 730 may learn and construct the aforementioned artificial neural networks (the first artificial neural network, the second artificial neural network, and/or the third artificial neural network) by using the learning data. The learning data and the learning process are the same as described above.

The computing device 730 may predict the amount of solar power generation based on the input data using the first artificial neural network stored in the storage device 710 . The input data and the prediction process are the same as described above.

The computing device 730 may predict the operating state of the solar power generation system or the target unit based on input data using the first artificial neural network and the second artificial neural network stored in the storage device 710 . The input data and the prediction process are the same as described above.

The computing device 730 may predict the operating state of the solar power generation system or the target unit based on input data using the third artificial neural network stored in the storage device 710 . The input data and the prediction process are the same as described above.

The arithmetic unit 730 may preprocess the input data and convert it into a vector column or matrix form.

The computing device 730 may be a device such as a processor, an AP, or a chip embedded with a program that processes data and processes a predetermined operation.

The output device 760 may output an interface screen required for artificial neural network learning or an analysis process using the learned artificial neural network. Also, the output device 760 may output a prediction result (power generation amount prediction information or operation state information) by analyzing the input data on the screen.

In addition, the artificial neural network learning method as described above, the method of predicting the amount of power generation of the solar power generation system, and/or the method of diagnosing the failure of the solar power generation system is a program (or application) including an executable algorithm that can be executed on a computer. can be implemented. The program may be provided by being stored in a temporary or non-transitory computer readable medium.

The non-transitory readable medium refers to a medium that stores data semi-permanently, rather than a medium that stores data for a short moment, such as a register, cache, memory, and the like, and can be read by a device. Specifically, the various applications or programs described above are CD, DVD, hard disk, Blu-ray disk, USB, memory card, ROM (read-only memory), PROM (programmable read only memory), EPROM (Erasable PROM, EPROM) Alternatively, it may be provided while being stored in a non-transitory readable medium such as an EEPROM (Electrically EPROM) or flash memory.

Temporarily readable media include: Static RAM (SRAM), Dynamic RAM (DRAM), Synchronous DRAM (SDRAM), Double Data Rate SDRAM (DDR SDRAM), Enhanced SDRAM (Enhanced) SDRAM, ESDRAM), Synchronous DRAM (Synclink DRAM, SLDRAM) and Direct Rambus RAM (Direct Rambus RAM, DRRAM) refers to a variety of RAM.

This embodiment and the drawings attached to this specification merely clearly show a part of the technical idea included in the above-described technology, and within the scope of the technical idea included in the specification and drawings of the above-described technology, those skilled in the art can easily It will be apparent that all inferred modified examples and specific embodiments are included in the scope of the above-described technology.

Claims (13)

receiving, by the analysis device, environmental information of the target unit of the photovoltaic system;
outputting, by the analysis device, the environmental information to a first artificial neural network and prediction information related to the amount of power generation; and
Comprising the step of the analysis device outputting the state information of the target unit by inputting the power generation related prediction information and the measured power generation related information of the target unit to a second artificial neural network,
The second artificial neural network is learned in advance using the learning data measured at a specific time in the solar power generation system,
The learning data is composed of input data and label values,
The input data is an output of at least one of (i) an output value output by the first artificial neural network receiving the environmental information and (ii) a three-phase wiring of an inverter to which a current output from the target unit is applied at the specific point in time At least one of an output value including a current, the power of the inverter, the current of the solar cell, the voltage of the solar cell, and the power of the solar cell,
The label value is an operation state information of the target unit or the photovoltaic system at the specific point in time. According to claim 1,
The environment information is a value measured at a certain distance from the target unit,
The environmental information includes at least one of an information group including external temperature, accumulated precipitation, wind direction, wind speed, humidity, atmospheric pressure, insolation, sunlight, slope insolation and the rear temperature of the solar module. Solar power generation system using an artificial neural network of fault detection methods. The method of claim 1,
The power generation-related information is
The output current of at least one of the three-phase wiring of the inverter to which the current output from the target unit is applied, the power of the inverter, the current of the solar cell, the voltage of the solar cell ?? A failure detection method of a solar power generation system using an artificial neural network including at least one of output values including power of a solar cell. The method of claim 1,
The first artificial neural network is learned in advance using the learning data measured at a specific time in the solar power generation system,
The learning data is composed of input data and label values,
The input data is a value measured at a certain distance from the target unit, and at least one of the information group including external temperature, accumulated precipitation, wind direction, wind speed, humidity, atmospheric pressure, insolation, sunlight, slope insolation, and rear temperature of the solar module including one,
The label value is the output current of at least one of the three-phase wiring of the inverter to which the current output from the target unit is applied, the power of the inverter, the current of the solar cell, the voltage of the solar cell ?? A method for detecting a failure of a solar power generation system using an artificial neural network including at least one of an output value including power of a solar cell. delete According to claim 1,
The target unit is at least one photovoltaic module or at least one photovoltaic panel. A method for detecting a failure of a photovoltaic system using an artificial neural network. According to claim 1,
The state information indicates a normal state or a failure state,
The analysis device is an artificial neural network that determines whether the target unit is in failure based on the number of times the state information is in a failure state or the length of a time interval indicating a failure state while acquiring the state information continuously or at regular time intervals. A method for detecting a failure of a solar power system using an input device for receiving environmental information of the target unit of the photovoltaic system and the measured power generation amount of the target unit;
a storage device for storing a first artificial neural network outputting generation amount-related prediction information based on environmental information and a second artificial neural network outputting state information based on generation amount-related prediction information and measured generation amount-related information; and
The environment information input to the input device at a specific point in time is input to the first artificial neural network to generate power generation related prediction information, and the power generation related prediction information and the measured power generation related information of the target unit at the specific point in time are generated. Comprising a computing device for generating state information of the target unit by input to the second artificial neural network,
The second artificial neural network is learned in advance using the learning data measured at a specific time in the solar power generation system,
The learning data is composed of input data and label values,
The input data is an output of at least one of (i) an output value output by the first artificial neural network receiving the environmental information and (ii) a three-phase wiring of an inverter to which a current output from the target unit is applied at the specific point in time At least one of an output value including a current, the power of the inverter, the current of the solar cell, the voltage of the solar cell, and the power of the solar cell,
The label value is an analysis device for monitoring the operation state of the photovoltaic system, which is the operational state information of the target unit or the photovoltaic system at the specific time point. 9. The method of claim 8,
The first artificial neural network is learned in advance using the learning data measured at a specific time in the solar power generation system,
The learning data is composed of input data and label values,
The input data includes at least one of an information group including external temperature, accumulated precipitation, wind direction, wind speed, humidity, atmospheric pressure, insolation, sunlight, slope insolation, and rear temperature of the solar module,
The label value is the output current of at least one of the three-phase wiring of the inverter to which the current output from the target unit is applied, the power of the inverter, the current of the solar cell, the voltage of the solar cell ?? An analysis device for monitoring an operating state of a photovoltaic system including at least one of an output value including power of a solar cell. delete receiving, by the analysis device, environmental information of a target unit of the photovoltaic power generation system and information related to the amount of power generated by the target unit at a specific point in time;
outputting state information of the target unit by inputting, by the analysis device, the environment information and the generation amount-related information into an artificial neural network; and
Comprising the step of the analysis device determining whether the failure of the target unit based on the status information,
The artificial neural network is learned in advance using the learning data measured at a specific time in the solar power generation system,
The learning data is composed of first input data, second input data, and a label value,
The first input data is the input data is a value measured from a certain distance from the target unit, external temperature, accumulated precipitation, wind direction, wind speed, humidity, atmospheric pressure, insolation, sunlight, inclined surface insolation and the rear temperature of the solar module At least one of the information groups comprising
The second input data includes an output current of at least one of the three-phase wirings of the inverter to which the current output from the target unit is applied, the power of the inverter, the current of the solar cell, and the voltage of the solar cell. At least one of the output values including the power of the solar cell,
The label value is the operation state information of the target unit or the photovoltaic system at the specific point in time. 12. The method of claim 11,
The environmental information is a value measured at a certain distance from the target unit, and the environmental information includes external temperature, accumulated precipitation, wind direction, wind speed, humidity, atmospheric pressure, insolation, sunlight, slope insolation, and the rear temperature of the solar module. comprising at least one of the information groups;
The generation amount-related information may include an output current of at least one of three-phase wiring of the inverter to which the current output from the target unit is applied, the power of the inverter, the current of the solar cell, and the voltage of the solar cell. A method for diagnosing a failure of a photovoltaic system including at least one of an output value including power of a solar cell. delete

Images