KR102059472B1 - A System and Method for Prediction of Geomagnetic Disturbance Strength based on Solar Coronal Hole Information - Google Patents

Abstract

The present invention relates to a system and method for predicting a geomagnetic disturbance intensity using solar coronal hole information. Specifically, the present invention relates to a system and method for predicting a geomagnetic disturbance intensity using solar coronal hole information wherein a correlation between the information on an occurrence of solar coronal holes accumulated over the past one solar activity cycle (10-11 years) and a change of the earth′s geomagnetic disturbance intensity index is analyzed using a machine learning neural network technique, and the correlation allows the trained neural network to predict a magnitude of the geomagnetic disturbance depending on the occurrence of the solar coronal hole.

Description

System and Method for Prediction of Geomagnetic Disturbance Strength based on Solar Coronal Hole Information}

The present invention relates to a system and method for predicting geomagnetic disturbance intensity using corona hole information of a sun. Specifically, the correlation between the information of the solar corona hole occurrence accumulated over the past one solar activity cycle (10-11 years) and the change of the Earth's geomagnetic disturbance intensity index is analyzed using the machine learning neural network technique. The present invention relates to a system and method for predicting geomagnetic disturbance intensity using solar corona hole information for predicting geomagnetic disturbance intensity according to the occurrence of solar corona holes through a learned neural network.

The solar corona hole refers to the dark and dark part of the sun that looks like a huge hole when the sun is viewed in extreme ultraviolet or X-rays. The corona hole appears dark because the density of the plasma is relatively lower than other parts of the sun's surface, and in this part, the solar magnetic field has an open magnetic field, causing strong solar wind with a speed of about 500 km / s or more. Comes out. The magnetic field structure of the open magnetic field, extending from the corona hole domain into space, bends in a spiral form due to the sun's rotating effect and exerts its influence throughout the solar system. At this time, the high speed solar wind compresses the low speed solar wind emitted from the sun to create a high density co-rotating interaction region, reaching the Earth's magnetic field and affecting the earth's magnetic field.

The solar corona hole often maintains its force for several cycles of solar rotation, so the effect of the earth's magnetic field on the corona hole is repeated over the 27 days of the solar rotation cycle. This repetitive disturbance of the Earth's magnetic field is known to occur frequently, especially when solar activity falls from the extremes to the smallest.

Therefore, in order to predict geomagnetic disturbance at least a little, various attempts have been made in the past to develop a model for predicting geomagnetic disturbance based on solar activity phenomena. However, these prediction models often use solar wind-related physical quantities measured by satellites as the main forecasting factors, mainly around the earth. That is, in most cases, it is a method of predicting the intensity of geomagnetic disturbance based on this result by analyzing the change data of the geomagnetic disturbance intensity index according to the solar wind physical quantity during the past period.

The existing prediction models have the advantage of predicting the intensity of geomagnetic disturbance according to the solar wind physical quantity that changes every time, but it is about 1 ~ 2 even if the time interval between the measurement of solar wind physical quantity and the geomagnetic disturbance is long. Since it is only about time, there is a limit that the prediction leading time is relatively short. Therefore, there is a need for a model that can be predicted with a longer leading time. This requires the development of predictive models based on more fundamental phenomena that cause geomagnetic disturbances.

To that end, the development of a predictive model that uses corona hole occurrence information as a predictor could be a new alternative. Corona Hall is known to be sufficiently associated with geomagnetic disturbances and is a relatively constant phenomenon observed in the sun itself, and thus serves as a predictor for a predictive model with a sufficiently long forecast lead time (up to about 7 days). Can be done. Moreover, it has the full potential as a new direction of geomagnetic disturbance prediction model by utilizing digital techniques together to acquire and process solar extreme ultraviolet image data to detect the occurrence of corona hole.

Korea Patent Registration No. 10-1845762

In the present invention, AI-based neural networks learn relevant data accumulated over a long period of time for various geomagnetic indices (Kp, Dst, AE) representing the location information on the solar surface of the region where corona holes are generated and geomagnetic disturbance intensities. The correlation between the geomagnetic disturbance intensity and the corona hole generation state for a predetermined prediction time is learned by the neural network based on the occurrence time according to the generation area of the corona hole, and the correlation is corona based on the learned neural network. The purpose of the present invention is to provide a reliable prediction result of the strength of geomagnetic disturbances for a predetermined time prediction time based on the occurrence time of the hole.

The geomagnetic disturbance intensity prediction system using solar corona hole information according to an embodiment of the present invention includes occurrence information including the occurrence date and time and occurrence region of each corona hole occurrence in the sun and Kp, AE, and Dst which are indexes related to geomagnetic disturbance intensity. The data collection unit collects and stores a plurality of different geomagnetic disturbance intensities for each exponent, and a predetermined solar surface grating system in which a surface of the sun-related solar model is divided into a plurality of gratings having a predetermined size. Corona hole generation state related information including corona hole generation by grid is generated by determining whether corona hole is generated according to whether or not the area occupied by the corona hole for each lattice is equal to or greater than a preset reference value. For a preset period of time based on a specific date and time of occurrence A pre-processing unit that generates matching information for each of the exponents and the plurality of exponents of the geomagnetic disturbance intensity and the corona hole state information corresponding to the specific occurrence date and time; And a correlation between the geomagnetic disturbance intensity and the corona hole generation state for a predetermined period of time based on the generation point of the corona hole in the neural network by applying the matching information received from the preprocessor to a corresponding neural network each time the corona hole is generated. Learning and correlating to the corona hole state information of the specific corona hole through the plurality of neural networks when receiving the corona hole state information corresponding to the specific corona hole through the pre-processing unit according to the generation of a specific corona hole. Predicted by index over the forecast period. May include an analysis unit for generating and providing prediction information about the geomagnetic disturbance intensity.

As an example related to the present invention, the prediction period, which is preset based on the occurrence time, based on the prediction information received from the analysis unit, is divided into prediction periods of a predetermined interval for each prediction period belonging to the prediction period. The method may further include an information processor configured to generate and output forecast information for a geomagnetic disturbance prediction including a category value for each index, which is categorized based on whether the geomagnetic disturbance intensity is exceeded by a reference value for each index.

As an example related to the present invention, the information processing unit may include corona hole state information corresponding to the forecast information and include the output in the forecast information.

As an example related to the present invention, the generation information further includes polarity-related polarity information of the magnetic field of the corona hole, and the preprocessing unit determines the polarity for each lattice based on the polarity information included in the generation information to determine the unit grid. Generating the corona hole state information reflecting polarity, and the analysis unit is further configured to correlate a correlation between the corona hole generation state according to the corona hole state information reflecting the polarity and the geomagnetic disturbance intensity for each exponent during the predetermined period of time; It can be characterized by learning by index.

As an example related to the present invention, the analysis unit detects index information for each index collected and stored by the data collection unit for a preset time based on generation information and corona hole generation time according to the generation information whenever corona hole is generated. A correlation obtained by applying information to a correlation neural network preset by the information to correlate the corona hole generation state to the correlation neural network, and learning the correlation between a plurality of exponential corona hole state obtained by applying the corona hole state information of the specific corona hole to the plurality of neural networks. The pattern corresponding to the correlation between the geomagnetic disturbance intensity for each exponent included in the prediction information by applying the information and the corona hole state information of the specific corona hole to the correlation neural network, and the correlation corresponding to the corona hole occurrence state of the specific corona hole. Based on learned information on relationship neural network Calculated by number of the prediction related information pattern correction information for correcting the difference between the pattern of the relationship between the geomagnetic disturbances strength, and on the basis of the pattern correction information may be characterized in that to correct the prediction information.

As an example related to the present invention, the preprocessor determines the sequence number according to a preset number for each unit grid, and sorts the binary type value for whether or not a corona hole is generated for each unit grid according to the sequence number for each unit grid. Corona hole state information including binary values for each grid may be generated.

As an example related to the present invention, when the prediction information is received from the analysis unit, an index for each index collected and stored by the data collection unit for the preset prediction period based on the generation time point of the corona hole included in the prediction information. Collect information as actual information, compare geomagnetic disturbance intensity by index for the prediction period according to the prediction information, and geomagnetic disturbance intensity by index for the period according to the measurement information, and generate error information for each index, The error information of which error is greater than a preset reference value among a plurality of different error information corresponding to the exponent of the index is provided to the analysis unit so that the error of any one of the plurality of neural networks is corrected based on the error information. It may be characterized by further comprising a verification unit.

According to an embodiment of the present invention, a method of predicting geomagnetic disturbance intensity using solar corona hole information of a service providing apparatus for predicting geomagnetic disturbance intensity using solar corona hole information includes a date and time of occurrence of each occurrence of corona hole in the sun. Collecting and storing a plurality of different exponential geomagnetic disturbance intensities including Kp, AE, and Dst, which are related to occurrence information and indexes related to geomagnetic disturbance intensity, and a plurality of grids having a predetermined size And matching the occurrence information to a predetermined solar surface grid system divided into a plurality of grids, and determining whether corona holes are generated according to whether or not the area occupied by the corona holes for each of the grids is greater than or equal to a predetermined reference value. Generating corona hole state information related to a corona hole generation state; Generating, by the index, matching information in which a plurality of index information of the geomagnetic disturbance intensity and the corona hole state information corresponding to the specific occurrence date and time are matched with each other based on a specific occurrence date and time according to occurrence information. And applying the matching information generated each time the corona hole is generated to a neural network corresponding to the matching information among a plurality of preset neural networks corresponding to the plurality of indices, respectively, for the geomagnetic field during a preset period based on the generation time point of the corona hole. Learning the correlation between the disturbance intensity and the corona hole generation state and applying the corona hole state information generated according to the specific corona hole generation to the plurality of neural networks, and through the plurality of neural networks, corona hole state information of the specific corona hole. Based on the occurrence time corresponding to With respect to the specified prediction time period may include the step of providing to generate a prediction about the geomagnetic disturbances strength is predicted by each index.

The geomagnetic disturbance intensity prediction system using solar corona hole information according to an embodiment of the present invention is a unit grid of the solar surface lattice system that generates a generation region according to the corona hole generation in the solar surface lattice system in which the solar model is divided into a plurality of lattice In addition, it is possible to standardize the state of occurrence of corona holes and to train the neural network by matching the exponential value of the exponential strength of geomagnetic disturbances with the normalized state of corona hole generation for a preset period of time based on the occurrence of corona holes. By supporting the correlation between the corona hole occurrence state and the geomagnetic disturbance intensity by index, this correlation is learned through the neural network learned, and when the corona hole occurs, the specific corona hole corresponds to the region generated by the sun. Is a predetermined predictor from the time of occurrence of the corona hole To ensure that prediction with respect to a coronal hole generating state and the effect greatly improve the prediction accuracy and reliability of the disturbed terrestrial magnetism according to the strength of geomagnetism disturbance factor corona holes generated on the basis of the relationship between the geomagnetic disturbances strength.

In addition, the present invention can support to accurately predict the change of the geomagnetic disturbance intensity according to the polar distribution of the corona hole as well as the region of the corona hole on the sun surface can support to increase the reliability of the geomagnetic disturbance intensity.

1 is a block diagram of a geomagnetic disturbance intensity prediction system using solar corona hole information according to an embodiment of the present invention.
2 is an illustration of the distribution of solar corona holes.
3 and 4 are exemplary views illustrating a standardization process using a solar model for solar corona holes of a geomagnetic disturbance intensity prediction system using solar corona hole information according to an embodiment of the present invention.
5 is an exemplary diagram of solar corona hole information calculated by a geomagnetic disturbance intensity prediction system using solar corona hole information according to an embodiment of the present invention.
6 is an exemplary operation diagram for providing prediction information and forecast information of a geomagnetic disturbance intensity prediction system using solar corona hole information according to an embodiment of the present invention.
7 is an exemplary diagram illustrating a prediction information correction process of a geomagnetic disturbance intensity prediction system using solar corona hole information according to an embodiment of the present invention.
8 is a flowchart illustrating a method for predicting geomagnetic disturbance intensity using solar corona hole information according to an embodiment of the present invention.

Hereinafter, exemplary embodiments of the present invention will be described with reference to the accompanying drawings.

1 is a block diagram of a geomagnetic disturbance intensity prediction system using solar corona hole information according to an embodiment of the present invention, as shown in the data collection unit 110, the preprocessor 120, and the analysis unit ( 130, the information processing unit 140 and the verification unit 150 may be configured.

At this time, the geomagnetic disturbance intensity prediction system using the solar corona hole information may be configured as a single service providing apparatus 100, the data collection unit 110, pre-processing unit 120, analysis unit in the service providing apparatus 100 130, the information processor 140 and the verification unit 150 may be included.

In addition, the service providing apparatus 100 may be configured as a server, in which the analysis unit 130 performs the overall control function of the service providing apparatus 100 and each component of the service providing apparatus 100. It can be configured as a control unit for controlling. In this case, the controller may include a RAM, a ROM, a CPU, a GPU, a bus, and the RAM, a ROM, a CPU, a GPU, and the like may be connected to each other through a bus.

In addition, the data collector 110, the preprocessor 120, the analyzer 130, the information processor 140 and the verification unit 150 may be configured as individual servers, respectively, the data collector 110 The preprocessing unit 120, the analysis unit 130, the information processing unit 140, and the verification unit 150 may be connected to each other to form a geomagnetic disturbance strength prediction system using the solar corona hole information. At this time, the communication network may be applied to a variety of wired and wireless communication methods well known.

Hereinafter, a description will be given based on the service providing apparatus 100, but is not limited thereto. As described above, the data collection unit 110, the preprocessor 120, the analysis unit 130, the information processing unit 140, and the verification unit ( Of course, each of the 150 may be configured as individual servers communicating with each other through a communication network to implement various embodiments implemented by the service providing apparatus 100.

Based on the above-described configuration, the present invention is to learn the area of the corona hole by learning the AI-based neural network by the time-dependent change value of the past accumulated by various index types representing the geomagnetic disturbance intensity generated in the earth according to the region where the corona hole occurs in the sun The correlation between the geomagnetic disturbance intensity and the corona hole occurrence state for a predetermined time based on the occurrence time point according to the occurrence time is learned by the neural network, and the correlation is corona hole which is a generation factor of the solar wind that generates geomagnetic disturbance based on the learned neural network. The purpose of the present invention is to reliably predict the prediction result of the geomagnetic disturbance intensity for a predetermined future period (prediction period) based on the occurrence time.

First, as illustrated in FIG. 1, the data collecting unit 110 receives and collects generation information including a generation date and time (occurring time of occurrence) and a generation region from the external server every time a corona hole is generated on the surface of the sun and then collects the corona hole. Can be stored in DB.

In this case, the generation region may mean an area occupied by corona holes on the surface of the sun.

In addition, the data collection unit 110 includes a plurality of different indices (geomagnetic index) Kp index, AE index, Dst index which is preset in relation to the geomagnetic disturbance intensity according to the generation of corona holes of the sun from an external server Index information for each of the geomagnetic disturbance strengths of a plurality of different exponents may be received and stored in a plurality of different exponent DBs respectively corresponding to the plurality of different exponents.

At this time, the Kp index (Kp index) is a representative index indicating the intensity of disturbance of the geomagnetic field (earth magnetic field, Geomagnetic Field). The index is a comprehensive global geomagnetic disturbance index that is collected from the U.S. Space Environment Forecast Center (SWPC), a local geomagnetic disturbance index from 13 geomagnetic stations scattered across the globe, and is calculated in units of three hours. Each of these 13 stations is located at latitude 44 degrees to 60 degrees north latitude or south latitude and uses a standardized method to calculate the K index, a geomagnetic disturbance index for each station. Kp is called the Planetary K Index and is meant to mean the Global Geomagnetic Disturbance Index. The value of the Kp index may have a value from minimum 0 to maximum 9, and typically exceeds 4 to 5 to correspond to a strong geomagnetic disturbance. Due to the nature of the calculation process, it is used as a measure of the intensity of the mean geomagnetic disturbances in the mid-latitude regions of the Earth. The concept of the Kp index was first conceived in 1938 and has been officially calculated since 1949, and the data has been continuously available worldwide.

In addition, the AE index (AE index) is an Auroral Electrojet Index is an index indicating the intensity of geomagnetic disturbance caused by the current present in the ionosphere in the Aurora region. This index is based on the disturbance intensity data of geomagnetic fields produced by 10 to 13 stations located in the northern high latitudes. The value of the AE index is in the range of approximately 0 to 1000, and the greater the intensity of geomagnetic disturbance, the larger the value. The AE Index is an hourly unit and has been officially calculated since 1957 and has been continuously available.

In addition, the Dst index (Dst index) is an index indicating the intensity of geomagnetic disturbance caused by the ring current (Ring Current) present in the ionosphere over the Earth equator region. Therefore, it is mainly based on geomagnetic disturbance intensity data from four geomagnetic stations located at low latitudes near the equator. When a strong geomagnetic disturbance called Geomagnetic Storm occurs, a phenomenon called Main Phase Depression is caused by the annular current of the low latitude ionospheric layer, which causes the disturbance of the equatorial geomagnetic field to be observed in the low latitude region. Therefore, it is Dst index which shows the intensity | strength of this disturbance phenomenon. Dst stands for Disturbance Storm Time. The value of the Dst index is in the range of about +100 to -500, and the stronger the geomagnetic disturbance, the larger the negative direction. The Dst index is an hourly unit and has been officially calculated since 1957 and has been continuously available.

Hereinafter, the Kp index will be named Kp, the AE index will be named AE, and the Dst index will be named Dst.

At this time, the plurality of different index DB may include a Kp-related first index DB, AE-related second index DB and Dst-related third index DB.

In addition, the corona hole DB, Kp-related first index DB, AE-related second index DB and Dst-related third index DB may be included in the data collection unit 110.

In the above-described configuration, the data collection unit 110 may collect the plurality of index information for each index for a predetermined period in relation to the corona hole generation, and then store the index information in the first to third index DB.

For example, the data collection unit 110 stores Kp related index information in the first index DB, AE related index information in the second index DB, and Dst related index information in the third index DB. Can be stored.

In this case, the preset period may be a period up to a point where seven days have elapsed based on the occurrence time of the corona hole.

On the other hand, the preprocessor 120 may generate matching information for learning the occurrence information and index information for each index collected by the data collection unit 110 through a neural network, the configuration of FIG. With reference to Figures 2 to 4 will be described in detail.

First, FIG. 2 is an exemplary diagram of the distribution of corona holes of the sun. As illustrated, the occurrence information may include information on the date and time of occurrence of the corona hole and a corona hole-related generation region on the sun surface.

In addition, as shown in FIG. 3, the pretreatment unit 120 may be set in advance with respect to the sun, and the sun model 10 may be set in advance, and the preprocessor 120 may adjust the surface of the sun model 10. It can be divided into a plurality of grids (unit grids) A having a preset size. In addition, the preprocessor 120 may set a number for each grid A to distinguish each of the plurality of grids.

Through this, the pretreatment unit 120 is divided into a plurality of gratings of the surface, the solar surface grating system of the solar model 10 is set to the grid number for each of the plurality of gratings (A) is the pre-treatment The unit 120 may be set in advance.

At this time, the plurality of gratings (A) may be configured to have a constant area and isotropic between each other, for example, the grating (A) may be configured as a regular hexagon.

On the other hand, as shown in FIG. 4, the preprocessor 120 receives specific occurrence information of corona holes corresponding to a specific occurrence date and time collected by the data collector 110 from the data collector 110 or The corona hole DB may be extracted, and a generation area of the corona hole according to the specific generation information may be matched to the solar surface grid system.

For example, the solar model constituting the solar surface lattice system has a corresponding relationship with a size (or surface size) of a real sun, and the preprocessing unit 120 determines the actual sun according to the specific generation information. The specific region of the solar model corresponding to the generation region of the corona hole generated on the surface may be identified in the solar model constituting the solar surface lattice system according to the correspondence relationship, and the identified region corresponding to the specific occurrence information. By setting a specific area of the model as the generation area of the corona hole according to the specific generation information, the generation area of the corona hole according to the specific generation information may be applied to the solar model of the solar surface grid system and matched.

That is, the preprocessing unit 120 may apply the generation region of the corona hole according to the specific occurrence information to the solar surface grating system through matching between the above-described solar surface grating system and the specific occurrence information.

Accordingly, the preprocessor 120 presets an area occupied by the corona hole according to the specific occurrence information in the unit grid for each unit grid of the solar surface grid system based on the specific occurrence information matched with the solar surface grid system. It may be determined whether the reference value is greater than or equal to the reference value, and according to the determination, it may be determined whether the corona hole is generated for each unit grid.

For example, when the reference value is 25% and the area occupied by the corona hole in the specific lattice exceeds 25%, the preprocessing unit 120 may determine that the corona hole is generated in the specific lattice.

In addition, the preprocessor 120 may generate corona hole state information related to a corona hole generation state including whether corona holes are generated for each unit grid.

In this case, the preprocessor 120 may include the generation date and time of occurrence information corresponding to the corona hole state information in the corona hole state information.

For example, the preprocessing unit 120 determines that the corona hole is generated in the specific unit grid when the area occupied by the corona hole according to the specific occurrence information is greater than or equal to a preset reference value in the entire area of the specific unit grid. 1 may be set to the grid, and if the area occupied by the corona hole in the specific unit grid is less than a predetermined reference value, it may be determined that no corona hole is generated in the specific unit grid, and 0 may be set to the specific unit grid.

That is, the preprocessor 120 may determine whether or not corona holes are generated for each unit grid, and set whether or not corona holes are generated for each unit grid as a binary value (binary value).

Through this, as illustrated in FIG. 5, the preprocessor 120 sets a binary value according to whether or not a corona hole is generated for each unit grid, and includes a binary value for each unit grid based on the solar surface grid system. Corona hole state information about the corona hole generation state may be generated. That is, the preprocessing unit 120 standardizes (shown) the corona hole state information of the area occupied by the corona hole in the sun based on the solar surface lattice system, which is a sun model divided into a plurality of unit grids, by a binary value per unit grid. May be calculated (generated) corresponding to the specific occurrence information.

In this case, the preprocessing unit 120 determines a sequence number according to a preset number for each unit grid, and sorts binary values (binary values) for whether or not corona holes are generated for each unit grid according to the sequence number for each unit grid. Corona hole state information including binary values for each unit grid may be generated.

In addition, the preprocessing unit 120 is collected by the data collection unit 110 for a predetermined period based on a specific occurrence time of the specific occurrence information, the first index DB, the second index DB and the third index DB After extracting the exponent information for each index stored in the first index DB, the second index DB and the third index DB, the matching information matched to the corona hole state information may be generated for each index, and the matching information for each of the plurality of indexes May be provided to the analysis unit 130.

In this case, the matching information for each index includes a plurality of Kp related index information for a predetermined period based on the specific occurrence time point according to the corona hole state information, and a plurality of AE related indexes for a predetermined period based on the specific occurrence time point. The information may include any one of a plurality of Dst-related index information for a predetermined period based on the information and the specific occurrence time point and the corona hole state information.

The preset period may be a period up to a point in which seven days have elapsed based on the occurrence point of the corona hole.

In addition, the preprocessing unit 120 may generate the matching information for each index whenever the corona hole is generated in association with the data collection unit 110 and provide the matching information to the analysis unit 130.

Meanwhile, the analysis unit 130 may include a plurality of artificial intelligence based neural networks (first to third neural networks) for each exponent, and the analysis unit 130 may include the preprocessor 120 for a plurality of exponential neural networks. The matching information for each index provided from) can be learned.

In this case, the neural network may include a deep learning algorithm including an input layer, one or more hidden layers, and an output layer. The neural network may include a recurrent neural network (RNN), Various types of neural networks such as a convolutional neural network (CNN) and a support vector machine (SVM) may be applied.

In addition, the analysis unit 130 applies the plurality of index-specific matching information received from the preprocessor 120 each time the corona hole is generated to the neural network for the preset period based on the generation time of the corona hole. Correlation between a plurality of different exponential geomagnetic disturbance strengths and the corona hole generation state may be learned.

In this case, the period set in advance in the analysis unit 130 may be the same period as the period set in advance for the collection of index information for each index on the basis of the corona hole generation time of the data collection unit 110, for example The period set in advance by the analyzer 130 may be a period of up to 7 days after the generation of the corona hole.

As an example of the operation configuration of the above-described analysis unit 130, as shown in FIG. 6, the analysis unit 130 is a plurality of neural networks (corresponding to a plurality of indices including Kp, AE, Dst, respectively) 1 to 3 neural network), wherein the analysis unit 130 may be configured to a neural network corresponding to an index of exponential information included in matching information received from the preprocessor 120 among a plurality of preset neural networks. The matching information can be learned.

That is, when the index according to the index information included in the matching information is Kp, the analyzing unit 130 may train the matching information including the Kp-related index information to the first Kp-related neural network among the plurality of neural networks. .

In addition, when the index according to the index information included in the matching information is AE, the analyzing unit 130 may train the AE-related second neural network among the plurality of neural networks, the matching information including the AE-related index information. .

Also, when the index according to the index information included in the matching information is Dst, the analyzing unit 130 may train the matching information including the Dst related index information to the Dst related neural network among the plurality of neural networks.

In this way, the analysis unit 130 may learn by applying the matching information for each index to the plurality of neural networks, so that each of the plurality of neural networks to the corona hole state information for any one of the plurality of indices Correlation between the occurrence time and the corona hole generation state (corona hole generation mode) related to the occurrence of the corona hole and the specific index-related geomagnetic disturbance intensity during the predetermined period can be learned.

On the other hand, based on the above-described configuration, the analysis unit 130 corresponds to a recent specific corona hole that occurred at present to the geomagnetic disturbance intensity for each of a plurality of exponents predicted for a predetermined period of time to come based on the time of occurrence of the specific corona hole. Prediction information may be provided, which will be described in detail.

As shown in FIG. 6, the analysis unit 130 may be generated recently (or presently), and when a specific corona hole is generated in which index information for each index is not collected by the data collecting unit 110, the specific corona hole may be generated. Corona hole state information may be received from the preprocessor 120.

In addition, the analysis unit 130 may apply corona hole state information for the specific corona hole to each of the plurality of neural networks (first to third neural networks), and the specific corona hole state through the plurality of neural networks. Prediction information on the geomagnetic disturbance intensity predicted for each index may be provided for a preset prediction period based on the occurrence time according to the information.

In this case, the prediction period may be a period according to a correlation that the analyzer learns each of the plurality of neural networks.

That is, the analysis unit 130 corresponds to a plurality of indices, respectively, and a plurality of neural networks learned from correlations between the geomagnetic disturbance intensity of a specific index and the corona hole generation state during a preset period according to the corona hole generation state. When the corona hole state information for the specific corona hole is applied to each of the first to third neural networks, the point of time of occurrence of the specific corona hole corresponds to an area (including the occurrence position) generated on the sun surface of the specific corona hole through the plurality of neural networks. The geomagnetic disturbance intensity (exponent value for each index) predicted during a predetermined period (prediction period) for each index can be calculated based on.

Through the above-described configuration, the analysis unit 130 by learning the correlation between the index value change for each exponent for a predetermined period on the basis of the time when the corona hole occurs in the plurality of neural networks and the corona hole generation state, When applying a corona hole generation state of a specific corona hole recently generated to the plurality of neural networks that the correlation has been learned, the subject of the study of geomagnetic disturbance intensity by index when the corona hole occurs based on the correlation time point It is possible to derive the geomagnetic disturbance intensity by index for the future prediction period by the preset period which is the period.

That is, when each of the plurality of neural networks receives a corona hole occurrence state that is not a learning target, the neural network derives a change in geomagnetic disturbance intensity for each exponent during a predetermined period associated with the corona hole occurrence state, wherein the correlation The period for which the geomagnetic disturbance intensity derived from the relationship is derived is the future forecast period.

In addition, the analysis unit 130 is predicted to include the geomagnetic disturbance intensity calculated for each index during a preset prediction period based on the generation time point of the specific corona hole through the plurality of neural networks corresponding to the corona hole state information of the particular corona hole. Information can be generated and provided.

In addition, the information processing unit 140 may receive the prediction information from the analysis unit 130, and based on the corona hole state information of the specific corona hole based on the prediction information received from the analysis unit 130. The geomagnetic disturbance intensity for each prediction period belonging to the prediction period is divided by dividing the preset prediction period based on the occurrence time point into prediction periods of a predetermined interval (for example, +1 day to +7 days based on the occurrence time point). It is possible to generate and output forecast information for geomagnetic disturbance forecasts including category values for indexes that are categorized by averaging index values).

In addition, the information processing unit 140 divides the prediction period based on the occurrence time according to the corona hole state information of the specific corona hole based on the prediction information into prediction periods of a predetermined interval, and belongs to the prediction period. It is also possible to generate and output forecast information for geomagnetic disturbance forecasts, including category values for indexes, categorized based on whether the geomagnetic disturbance intensity by index exceeds the reference value for each forecast period.

In this case, the information processing unit 140 may interoperate with the analysis unit 130 to include corona hole state information corresponding to the forecast information in the forecast information, and thereby output the corona hole state information. By providing the daily category value related to the geomagnetic disturbance intensity by index in addition to the occurrence state of the hole, it is possible to provide a user to determine the daily geomagnetic disturbance intensity according to the corona hole generation state in the sun.

Through the above-described configuration, the geomagnetic disturbance intensity prediction system using the solar corona hole information according to the embodiment of the present invention, the geomagnetic disturbance intensity prediction system using the solar corona hole information according to an embodiment of the present invention is a plurality of units of the solar model Based on the solar surface grid system divided into grids, the generation area according to the corona hole generation can be judged for each unit grid of the solar surface grid system to standardize the generation state of the corona hole and at the time of occurrence according to the generation of the corona hole. By matching exponential values of geomagnetic disturbance intensities with a normalized corona hole occurrence state for a predetermined period, the neural network is trained to assist in deriving the correlation between corona hole occurrence states and geomagnetic disturbance intensities by exponent. When a particular corona hole occurs through a learned neural network The specific corona hole corresponds to the sun-generated region so that the geomagnetic disturbance intensity is predicted for a predetermined period from the time of the corona hole generation. The accuracy and reliability of the geomagnetic disturbance intensity can be greatly increased.

Meanwhile, as shown in FIG. 1, the geomagnetic disturbance intensity prediction system using solar corona hole information according to an embodiment of the present invention may further include a verification unit 150 for verifying the accuracy of the prediction information. have.

The verification unit 150 may receive prediction information from the analysis unit 130, and may identify a time point of generation of a corona hole included in the prediction information.

In addition, the verification unit 150 is collected by the data collection unit 110 for the prediction period of the prediction information (during the prediction period) on the basis of the occurrence time of the corona hole according to the prediction information, and the first to the first to the second. Index information for each index stored in the third index DB may be collected as actual information from the first to third index DB.

Further, the verification unit 150 may determine the geomagnetic disturbance intensity for each index (during the preset prediction period) for the preset prediction period according to the prediction information and the preset period (corresponding to the prediction period) according to the measurement information ( Error information for each index may be generated by comparing the geomagnetic disturbance intensities for each index (for a predetermined period).

Accordingly, the verification unit 150 may provide the analysis unit 130 with error information in which an error equal to or greater than a preset reference value is generated among a plurality of different error information corresponding to a plurality of indices, respectively.

When the error information is received from the verification unit 150, the analyzer 130 applies the error information to the neural network of the exponent corresponding to the error information among the plurality of neural networks configured in the analyzer 130. The error may be corrected by varying a weight of a connection strength between an input layer, one or more hidden layers, and an output layer constituting a neural network, and a bias of units configured in the input layer, hidden layer, and output layer.

Meanwhile, as shown in FIG. 7, the analysis unit 130 learns correlations between a plurality of different indices representing geomagnetic disturbance intensities based on a corona hole generation state, and based on the correlations, It may further comprise a correlation neural network for generating a correction value for increasing the accuracy.

That is, the analysis unit 130 is collected by the data collection unit 110 for a predetermined time based on the occurrence information and the corona hole generation time according to the generation information for each occurrence of the corona hole is the first to third Index information for each index stored in the index DB can be learned in the correlation neural network as measured information, and through this, the correlation neural network can learn correlations between a plurality of exponents corresponding to the corona hole generation state.

Accordingly, the analyzer 130 may apply prediction information obtained by applying the corona hole state information of the specific corona hole to the first to third neural networks and the corona hole state information of the specific corona hole to the correlation neural network. And a pattern according to a correlation between geomagnetic disturbance intensities for each index included in the prediction information through the correlation neural network, in which correlations between the indexes are learned based on the corona hole generation state based on the measured information, and the specific coronas. Prediction information-related pattern correction information for correcting a mutual difference between patterns of correlations between geomagnetic disturbance intensities for each of the measured information-based indexes learned in the correlation neural network corresponding to the corona hole generation state of the hole may be calculated.

In this case, the pattern correction information may include a plurality of different index-specific correction values or a plurality of different index-specific weights for pattern correction between the measured information and the prediction information.

That is, the analysis unit 130 may be configured to the correlation neural network based on a plurality of exponentially measured information corresponding to a plurality of different corona hole generation states generated in the past similar to the corona hole generation state corresponding to the prediction information. Correlation pattern correction information for correcting a pattern and a difference according to a plurality of exponential correlations obtained by applying the prediction information to the correlation neural network has a pattern according to the correlation between a plurality of different exponentials learned Can be calculated through neural network.

In addition, the analysis unit 130 may apply the pattern correction information calculated through the correlation neural network to the prediction information to correct the geomagnetic disturbance intensity for each index of the prediction information or set a weight for each index. The prediction information may be corrected according to the pattern correction information to increase the accuracy of the prediction information.

Meanwhile, in the above-described configuration, the generation information may further include polarity information related to magnetic polarity of the corona hole.

In addition, the preprocessor 120 may determine the polarity (magnetic field polarity) for each unit grid based on the polarity information included in the generation information to generate corona hole state information in which the polarity for each unit grid is reflected (set). Can be.

Accordingly, the analyzer 130 may determine a correlation between the corona hole generation state according to the corona hole state information in which the polarity is reflected (set) and the geomagnetic disturbance intensity for each exponent for the predetermined period and a plurality of indices, respectively. A plurality of corresponding neural networks can be trained.

In addition, the analyzer 130 receives the corona hole state information in which the polarity information is set in relation to the specific corona hole for predicting the geomagnetic disturbance intensity for each exponent according to the occurrence of a specific corona hole from the preprocessor 120. The polarity distribution of the specific corona hole included in the hole state information and the arrangement state (arrangement distribution) of the specific corona hole may provide prediction information about the geomagnetic disturbance intensity for a predetermined period through the neural network for each index. .

That is, the analysis unit 130 may learn the change of the geomagnetic disturbance intensity according to the magnetic field polarity of the corona hole along with the generation region of the corona hole to the plurality of neural networks. Prediction information about the geomagnetic disturbance intensity for each index that is expected to be observed during a preset period of time may be provided based on the occurrence time point of the specific corona hole in consideration of the magnetic field polarity distribution of the corona hole.

Through this, the present invention can support to accurately predict the change of the geomagnetic disturbance intensity according to the polar distribution of the corona hole as well as the region of the corona hole on the sun surface can support to increase the reliability of the geomagnetic disturbance intensity.

8 is a flowchart illustrating a method for predicting geomagnetic disturbance strength using solar corona hole information of the service providing apparatus 100 that predicts geomagnetic disturbance strength using solar corona hole information according to an embodiment of the present invention.

As shown, the service providing apparatus 100 includes a plurality of different indexes each including occurrence information including the occurrence date and time and occurrence region and geomagnetic disturbance intensity indexes Kp, AE, and Dst each time a corona hole is generated in the sun. Geomagnetic disturbance intensity can be collected and stored (S1).

In addition, the service providing apparatus 100 matches the generation information to a preset solar surface grid system obtained by dividing a surface of the preset solar-related solar model into a plurality of grids having a preset size, thereby generating corona holes for each grid. It is determined whether corona holes are generated according to whether an area occupied is greater than or equal to a predetermined reference value, and corona hole state information related to corona hole generation states including whether corona holes are generated for each lattice is generated (S2), and the specific information according to the generation information Based on the occurrence date and time, matching information obtained by mutually matching the exponential information of the geomagnetic disturbance intensity and the corona hole state information corresponding to the specific occurrence date and time may be generated for each index (S3). .

The service providing apparatus 100 may apply the matching information generated each time the corona hole is generated to a neural network corresponding to the matching information among a plurality of preset neural networks corresponding to the plurality of indices, respectively, to generate the corona hole. The correlation between the geomagnetic disturbance intensity and the corona hole occurrence state during the reference preset period may be learned (S4).

In addition, the service providing apparatus 100 applies the corona hole state information generated according to the generation of a specific corona hole to the plurality of neural networks to determine the occurrence time point corresponding to the corona hole state information of the specific corona hole through the plurality of neural networks. As a reference, prediction information about the geomagnetic disturbance intensity predicted for each index with respect to the preset prediction period may be generated and provided (S5).

The various devices and components described herein may be implemented by hardware circuitry (eg, CMOS based logic circuitry), firmware, software, or a combination thereof. For example, it may be implemented using transistors, logic gates, and electronic circuits in the form of various electrical structures.

The above description may be modified and modified by those skilled in the art without departing from the essential characteristics of the present invention. Therefore, the embodiments disclosed in the present invention are not intended to limit the technical idea of the present invention but to describe the present invention, and the scope of the technical idea of the present invention is not limited by these embodiments. The protection scope of the present invention should be interpreted by the following claims, and all technical ideas within the equivalent scope should be interpreted as being included in the scope of the present invention.

100: service providing device 110: data collection unit
120: preprocessing unit 130: analysis unit
140: information processing unit 150: verification unit

Claims (8)

A data collector configured to collect and store a plurality of different geomagnetic disturbance intensities for each index including Kp, AE, and Dst, which are related to occurrence information including the occurrence date and time and occurrence region of the corona hole in the sun;
Corona is determined according to whether or not the area occupied by the corona hole for each of the grids is equal to or larger than a predetermined reference value by matching the generation information to a preset solar surface grid system that divides the surface of the sun-related solar model into a plurality of grids having a predetermined size. Corona hole state information related to a corona hole generation state including whether a corona hole occurs for each lattice is determined by determining whether a hole is generated, and based on the geomagnetic disturbance intensity for a predetermined period based on a specific occurrence date and time according to the occurrence information. A pre-processing unit generating matching information for each index by matching the plurality of index information with respect to the corona hole state information corresponding to the specific date and time; And
And a plurality of neural networks respectively corresponding to the plurality of indices, wherein the matching information received from the preprocessing unit is applied to a corresponding neural network every time the corona hole is generated, based on the generation time point of the corona hole in the neural network. Correlation between geomagnetic disturbance intensity and corona hole generation state, and corona hole of the specific corona hole through the plurality of neural networks when corona hole state information corresponding to a specific corona hole is received through the preprocessing unit according to a specific corona hole generation An analysis unit that generates and provides prediction information on the geomagnetic disturbance intensity predicted for each index for a preset prediction period based on the occurrence time corresponding to the state information.
Including;
The correlation unit is a correlation neural network preset with the exponential index information collected and stored by the data collection unit for a predetermined time on the basis of the occurrence information and the corona hole generation time according to the occurrence information each time corona hole occurs as the measurement information. Learning correlation between a plurality of exponentials corresponding to a corona hole generation state in the correlation neural network, and predicting information obtained by applying corona hole state information of the specific corona hole to the plurality of neural networks and the corona hole of the specific corona hole. Actual information learned in the correlation neural network corresponding to the corona hole generation state of the specific corona hole and the pattern according to the correlation between the geomagnetic disturbance intensity for each index included in the prediction information by applying the hole state information to the correlation neural network. Geomagnetic Disturbance Intensity by Base Index Computing the prediction information-related pattern correction information for correcting the difference between the patterns for the correlation between the correlation, and geomagnetic disturbance intensity using the solar corona hole information, characterized in that for correcting the prediction information based on the pattern correction information Prediction system.
The method according to claim 1,
On the basis of the prediction information received from the analysis unit, the prediction period preset based on the generation time is divided into prediction periods of a predetermined interval, and the geomagnetic disturbance intensity exceeds the reference value by index for each prediction period belonging to the prediction period. The geomagnetic disturbance intensity prediction system using solar corona hole information, characterized by further comprising an information processing unit for generating and outputting forecast information for geomagnetic disturbance forecasting, including category values for each index categorized based on whether or not.
The method according to claim 2,
And the information processing unit outputs the corona hole state information corresponding to the forecast information in the forecast information and outputs it.
The method according to claim 1,
The generation information further includes polarity-related polarity information of the corona hole,
The preprocessing unit determines the polarity for each lattice based on the polarity information included in the generation information and generates corona hole state information reflecting the polarity of the unit lattice.
The analyzer correlates the correlation between the corona hole generation state according to the corona hole state information reflecting the polarity and the geomagnetic disturbance intensity for each exponent for the predetermined time period in the plurality of neural networks by index. Geomagnetic Disturbance Strength Prediction System Using Hall Information.
delete The method according to claim 1,
The preprocessing unit determines a sequence number according to a preset number for each unit grid, and sorts binary values for whether or not corona holes are generated for each unit grid according to the sequence number for each unit grid to include a binary value for each unit grid. Geomagnetic disturbance intensity prediction system using solar corona hole information, characterized in that for generating the hole state information.
The method according to claim 1,
Upon receiving the prediction information from the analysis unit, index information for each index collected and stored by the data collection unit for the preset prediction period based on the occurrence time of the corona hole included in the prediction information is collected as actual information, After comparing the geomagnetic disturbance intensity by index for the prediction period according to the prediction information and the geomagnetic disturbance intensity by index for the period according to the measured information, error information is generated for each index and a plurality of indexes corresponding to the plurality of indexes, respectively, The apparatus may further include a verification unit configured to provide error information in which an error equal to or greater than a predetermined reference value among different error information is generated to the analysis unit to correct an error of any one of the plurality of neural networks based on the error information. Geomagnetic Disturbance Intensity Using Solar Corona Hole Information Prediction system.
In the method of predicting the geomagnetic disturbance intensity using the solar corona hole information of the service providing device for predicting the geomagnetic disturbance intensity using the solar corona hole information,
Collecting and storing the occurrence information including the occurrence date and time and the geomagnetic disturbance intensity indexes including Kp, AE, and Dst, each time the corona hole occurs in the sun;
Corona is determined according to whether or not the area occupied by the corona hole for each of the grids is equal to or larger than a predetermined reference value by matching the generation information to a preset solar surface grid system that divides the surface of the sun-related solar model into a plurality of grids having a predetermined size. Corona hole state information related to a corona hole generation state including whether a corona hole occurs for each lattice is determined by determining whether a hole is generated, and based on the geomagnetic disturbance intensity for a predetermined period based on a specific occurrence date and time according to the occurrence information. Generating matching information for each index by matching the plurality of index information with respect to the corona hole state information corresponding to the specific date and time;
Geomagnetic disturbance intensity during a preset period of time based on the generation time point of the corona hole by applying the matching information generated each time the corona hole is generated to the neural network corresponding to the matching information among a plurality of preset neural networks corresponding to the plurality of indices, respectively; Learning correlations between corona hole occurrence states;
Corona hole state information generated according to a specific corona hole generation is applied to the plurality of neural networks, and for each index for a preset prediction period based on an occurrence time corresponding to the corona hole state information of the specific corona hole through the plurality of neural networks. Generating and providing prediction information on the predicted geomagnetic disturbance intensity;
When the corona hole is generated, the index information for each exponent collected and stored for a predetermined time on the basis of the occurrence of the corona hole according to the occurrence information and the occurrence information is learned by the correlation neural network set as the actual measurement information. Learning a correlation between a plurality of exponentials corresponding to a hole occurrence state; And
Correlation between prediction information obtained by applying the corona hole state information of the specific corona hole to the plurality of neural networks and geomagnetic disturbance intensity for each index included in the prediction information by applying the corona hole state information of the specific corona hole to the correlation neural network Correcting the pattern related to the prediction information to correct the difference between patterns for the correlation between the pattern according to the relationship and the geomagnetic disturbance intensity for each index based on the measured information based on the correlation neural network corresponding to the corona hole generation state of the specific corona hole Calculating information and correcting the prediction information based on the pattern correction information;
Geomagnetic disturbance intensity prediction method using solar corona hole information comprising a.

Images