KR101736918B1 - System and Method for Predicting Sea Wave Using Sea Surface Wind Numerical Model Forecast Data - Google Patents

Abstract

The present invention relates to an apparatus and a method for a direct prediction of waves using offshore wind numeric model predicting data which can improve the accuracy by directly predicting waves from offshore wind predicting data of a weather numeric model without a calculating process for a numeric prediction model, and reduce a consumed time. The apparatus comprises: a calculating wave information for each offshore wind pattern and a DB which uses hindcast data of offshore winds in the past to make a category of offshore wind patterns, and calculates waves information for each offshore wind pattern to build a DB; and a real-time sea wave predicting information generating unit which determines a pattern type of offshore wind predicting data in real time, and performs a calculation of the wave information for each the offshore wind pattern and a calculation of the wave information on the corresponding pattern from the DB.

Description

TECHNICAL FIELD The present invention relates to a method and apparatus for direct prediction of a sea wave using a numerical model predicted from an oceanic wind model,

More particularly, the present invention relates to a method for directly predicting a sea wave, and more particularly, to a method for predicting a sea wave from a sea-level forecast data of a meteorological numerical model without calculation of a numerical prediction model, And an apparatus and method for direct ocean wave prediction using prediction data.

Coastal inundation, oil pollution accidents, and other coastal disasters are deeply related to sea level and seawater flow.

Therefore, accurate prediction of sea level and flow rate in the coastal area is very important in the search structure and estimation of oil diffusion in marine industry as well as coastal protection and marine accidents.

The flow of seawater is a result of complex action of various factors such as periodic astronomical movement, storm surge, tsunami, low pressure,

In coastal engineering, the wave of the sea is handled as a regular wave equivalent to the wave of the wave, and deformation by wave refraction, diffraction, shaking, breaking wave, friction dissipation is handled.

However, as is well known, the actual ocean wave generated in the ocean is a very irregular natural phenomenon as a complex of waves with wave, cycle, and wave, so the concept of the so-called wave spectrum Has been introduced and is now being used for wave prediction.

In order to predict the wave (or wind) generated in the sea in advance, it is necessary to calculate the forecast data using the wave numerical model.

Also, before calculating the wave forecast data, the forecast data of the sea wind should be calculated first from the meteorological numerical model. This is because the main external force that generates waves at sea level is the wind.

In general, the wave numerical model not only takes a lot of time in the calculation process but also requires waiting time until the forecast data of the weather is produced.

For this reason, in order to shorten the numerical calculation time for the timeliness of the pre-blu forecast, the wave generation mechanism expressed in the wave numerical model is simplified and computed every part (for example, nonlinear interaction mechanism of wave).

This means that the wave numerical model has inherent limitations and errors in accurately predicting the wave phenomenon occurring at sea.

In addition, the error that occurs in predicting the wave is also caused by the fact that the mechanism (or mechanism) of wave generation and propagation has not yet been accurately identified.

For example, the phenomena of waviness in the east coast and in the west coast are not well explained by the wave generation mechanism employed in the current wave numerical model.

From the past long-term experience and observation data, the sea wind is repeating with maintaining a certain pattern according to year and season, and the sea wave generated at this time likewise has to keep a certain pattern in the past as well as the future.

Therefore, it is required to develop a new technique for categorizing and quantifying the relationship between the background pattern of the sea wind and the background pattern of the sea wave, which has occurred in the past, and directly forecasting the future wave from the forecast / forecast data.

Korean Patent Publication No. 10-2015-0117971 Korean Patent Publication No. 10-2015-0117972

The present invention solves the problem of the conventional method of predicting a sea wave, and it directly predicts the sea wave from the sea-wind prediction data of the meteorological numerical model without the calculation process of the numerical prediction model, The present invention provides an apparatus and method for direct ocean wave prediction using ocean-wind numerical model prediction data.

An object of the present invention is to provide a device and a method for directly predicting future waves from the prediction / forecasting data of the sea-winds by categorizing and quantifying the relationship between the pattern of the sea-wind and the pattern of the sea- have.

The present invention relates to a method and a system for estimating a sea-wind numerical model using prediction data of an ocean-wind numerical model which can minimize a prediction error by using observation data of a wave measured at a certain point, which is an end result through all the mechanisms of wave generation and propagation There is provided an apparatus and method for direct wave prediction.

The present invention relates to an ocean wave direct prediction method using ocean-wind numerical model predictive data suitable for only a certain scale of closure only or semi-closure, which is easy to categorize and quantify the sea-level wind instrument and seasonal pattern as in Korea, And an object of the present invention is to provide an apparatus and method.

The present invention relates to a method for predicting a wave pattern of a wave pattern from a database by determining a pattern type of real-time sea wind prediction data by categorizing the off-wind patterns and calculating wave information for each sea wind pattern, And to provide a device and method for direct ocean wave prediction using data.

The objects of the present invention are not limited to the above-mentioned objects, and other objects not mentioned can be clearly understood by those skilled in the art from the following description.

In order to accomplish the above object, there is provided an apparatus for an ocean wave direct prediction using a sea-wind numerical model predictive data according to the present invention includes an object region segmentation unit for setting a region of interest and performing a segmentation in a two- Dimensional sea-wind vector sequence generating unit for generating a one-dimensional sea-wind vector sequence of a 3D marine-type weather pattern, and a one-dimensional sea-water vector sequence generating unit for building a one- An ocean wind pattern scenario deriving unit for deriving a sea wind pattern scenario by performing weighting, similar pattern classification, and similar pattern categorization for each region by searching a wind vector sequence DB and a one-dimensional sea wind vector sequence DB, Based on the comparison results of the observation data comparing section and the observation data comparing section, And a blue information calculation unit for calculating blue information. The categorization of the off-wind patterns and the calculation of wave information for each of the off-wind patterns using the past background data of the off- ; And a real-time wave prediction information generation unit for determining a pattern type of the sea-wind prediction data in real time, calculating wave information for each of the sea-wind patterns, and calculating wave information for the corresponding pattern from the DB.

Here, the real-time wave prediction information generator may generate a one-dimensional real-time wind vector sequence using a predictive data providing unit that provides two-dimensional real-time wind vector field predictive data in real time and two- An optimal scenario detector for searching for an ocean wind pattern scenario DB and detecting an optimal similar scenario using the ocean wind pattern vector generation unit and the wave information calculation DB based on the ocean wind pattern and the wave prediction data for the detected ocean wind scenario, And a blue prediction data generating unit for generating a blue prediction data.

delete

The one-dimensional wind-based vector sequence generation unit includes a one-dimensional vectorization unit for performing one-dimensional vectorization of the detailed in-band two-dimensional sea-like weather pattern, a mid-vectorization unit for performing each mid- And a global vectorization unit that performs global vector calculation from wide-area representative vectors in the global area. The global vectorization unit performs global vector calculation from global representative vectors in a wide area.

The resolution pattern scenario derivation unit may include a one-dimensional vector sequence DB search unit for searching the one-dimensional vector sequence DB, a weighting unit for each region for weighting the searched one-dimensional vector sequence, A similar pattern classifying unit for performing similar pattern categorization, and a scenario deriving unit for deriving an off-wind pattern scenario.

The similar pattern categorizing unit is characterized by obtaining one representative vector sequence by averaging the values of the respective element elements of the one-dimensional vector sequences having similar arrangement per scenario in the classified sea wind pattern scenario.

In order to accomplish the other object, a method for an ocean wave direct prediction using a sea-level numerical model predictive data according to the present invention comprises the steps of setting an area of interest and performing a zoning in a two- A step of generating a one-dimensional sea-wind pattern vector sequence in a two-dimensional sea-side wind-shaped field at a sea-wind pattern vector generation section and generating a one-dimensional sea-wind vector sequence database for a past sea- A step of retrieving a one-dimensional sea-wind vector sequence DB at an ocean-wind pattern scenario derivation unit and performing weighting, similar pattern classification, and similar-pattern categorization for each region to obtain a sea-wind pattern scenario and DB The data comparison section compares the blue observation data with the ocean-wind-side data, Converting offshore wind vector sequence generating section as a one-dimensional vector sequence of a real-time two-dimensional pungjang marine and offshore wind pattern detection scenario DB search and best similar scenario in the best scenario, detection; step of the calculation, and blue information DB screen;
And generating wave predictive data from the wave information for the sea-wind scenario detected by the wave predictive data generation unit.

delete

The step of generating the one-dimensional wind-based vector sequence database for the past marine wind backward data includes a step of performing one-dimensional vectorization of the in-depth two-dimensional marine wind tunnel and a step of extracting, from the detailed inverse one- Performing a global vector calculation from wide-area representative vectors in a global range, performing each global vector calculation from global representative vectors in a wide area, and performing global vector calculation from wide-area representative vectors in a global area.

The searching of the sea-wind pattern scenario DB and the searching of the optimal similar scenario are characterized in that the current vector sequence and the vector sequences in the past DB are matched with the nearest neighbor method.

Then, an inner product is obtained between the current vector sequence and the array vector of vector sequences in the past DB, and when the result is maximized with respect to the inner values with other array vectors, the two vector sequences are matched to each other as the most similar .

In the step of generating wave predictive data from the detected wave information for the ocean-wind scenario, the wave information corresponding to a plurality of consecutive sea wind scenarios detected at a certain point or a plurality of points in the object area is stored in the past wave information DB And generates predicted wave data of a plurality of consecutive time points and filters the predicted values of the wave information of the plurality of points by a smoothing method in order to reduce the error in the detection process of the past DB .

The apparatus and method for direct ocean wave direct prediction using the sea-level numerical model prediction data according to the present invention have the following effects.

First, it is possible to directly predict the sea wave from the sea wind forecast data of the meteorological model without calculation process of the numerical prediction model, to improve the accuracy and reduce the time required.

Second, the present invention provides an apparatus and method for directly predicting future waves from the sea-wind forecast / forecast data by categorizing and quantifying the relationship between the sea-wind pattern and the sea-wave occurrence pattern that occurred in the past.

Third, the prediction error can be minimized by using the observation data of the wave measured at a certain point, which is the final result that has already been displayed through all the mechanism processes related to the generation and propagation of the wave.

Fourth, we provide ocean wave direct prediction technology suitable only for a certain scale of closure only or semi-closure by categorizing and quantifying the ocean-style wind instrument and seasonal pattern as in Korea's East Sea.

Fifth, it is possible to reduce the errors caused by the inherent limitations of the wave numerical model and the lack of understanding of the generation mechanism of the wind waves, and shorten the calculation time for the advance prediction of waves.

1 is a block diagram of an apparatus for an ocean wave direct prediction using a sea-wind numerical model prediction data according to the present invention
FIG. 2 is a block diagram for calculating wave information and building a DB according to the present invention; FIG.
3 is a detailed configuration diagram of a one-dimensional off-axis wind vector sequence generating unit according to the present invention
4 is a detailed configuration diagram of a sea wind pattern scenario deriving unit according to the present invention
5 is a flow chart illustrating a method for direct ocean wave prediction using the sea-wind numerical model prediction data according to the present invention
6 is a diagram showing an example of a method of setting an ocean-
7 is a diagram showing an example of a four-step zoning method of the ROI
8 is a diagram showing a one-dimensional vectorization of a two-dimensional sea wind in one zone
9 is a diagram showing the generation of a one-dimensional vector sequence from a two-
10 is a diagram showing a DB generation of a one-dimensional vector sequence from the past backwash data DB
11 is a diagram showing a process of deriving a sea wind pattern scenario from the past one-dimensional vector sequence DB

Hereinafter, a preferred embodiment of an apparatus and method for direct ocean wave direct prediction using the sea-wind numerical model prediction data according to the present invention will be described in detail.

The features and advantages of the apparatus and method for direct ocean wave direct prediction using the sea-wind numerical model prediction data according to the present invention will be apparent from the following detailed description of each embodiment.

FIG. 1 is a block diagram of an apparatus for an ocean wave direct prediction using a sea-level numerical model prediction data according to the present invention.

And FIG. 2 is a block diagram for calculating wave information and building a DB according to the present invention.

The present invention is to categorize and quantify the relationship between a pattern of a sea-wind which has occurred in the past and a pattern of occurrence of a sea-wave, and to directly predict a future wave from the sea-wind prediction / forecast data from the relationship.

Since the observation data of a wave measured at a certain point is an end result which is already displayed through all the mechanism processes related to generation and propagation of the wave, the present invention eliminates the error caused by the method of predicting through the current wave numerical model .

In particular, the apparatus and method for the direct ocean wave direct prediction using the sea-wind numerical model prediction data according to the present invention can be applied to a closure of only a certain scale, which is easy to categorize and quantify the ocean- It can be applied only to semi-close.

As shown in FIG. 1, the apparatus for the direct prediction of the sea wave using the sea-wind numerical model prediction data according to the present invention for the purpose of the present invention includes a categorization of the sea-wind pattern using past sea- And determines the pattern type of the sea wind prediction data in real time and calculates the wave information by the sea wind pattern and the corresponding pattern from the DB 100 And a real-time-blurred-prediction-information generating unit 200 for performing calculation of the wave-information for the user.

Here, the real-time wave prediction information generator 200 includes a prediction data providing unit 201 for providing real-time two-dimensional sea-wind field vector field prediction data, and a one-dimensional sea-wind vector sequence An optimal scenario detection unit 203 for searching for an ocean wind pattern scenario DB and detecting an optimal similar scenario using the ocean wave pattern-based blue wave information DB 100, And a blue predictive data generation unit 204 for generating a blue predictive data from the detected blue-colored information of the ocean-wind scenario.

The details of the wave information calculation and DB 100 according to the background pattern are as follows.

As shown in FIG. 2, an object region segmentation unit 10 for setting a region of interest and performing zoning in a two-dimensional sea-wind grid region in order to categorize the sea-wind patterns systematically and make it easy to DB, A one-dimensional off-axis wind vector sequence generating unit 20 for generating a one-dimensional off-axis wind vector sequence, and a one-dimensional off-wind vector sequence database for all backside data using the back- Dimensional wind direction vector sequence DB 30 and a one-dimensional wind direction vector sequence DB to perform weighting, similar pattern classification, and similar pattern categorization for each region to derive a sea wind pattern scenario, 50), an observation data comparing unit (60) for comparing the blue observation data and the backwash data, and a comparison unit (60) And a blue information calculation section 70 for calculating.

As shown in FIG. 3, the one-dimensional resolution-based vector sequence generator 20 includes a one-dimensional vectorization unit 21 for performing one-dimensional vectorization of a detailed in- A wide vectorization unit 23 for performing each wide vector calculation from the central representative vectors in the wide area, and a global vector calculation unit 23 for calculating global vector calculation from the wide-area representative vectors in the global area. And a global vectorization unit 24 that performs vectorization.

4, the resolution pattern scenario deriving unit 50 includes a one-dimensional vector sequence DB search unit 51 for searching a one-dimensional vector sequence DB, and a one-dimensional vector sequence search unit 51 for searching the one- A similar pattern classifying unit 53 for classifying a similar pattern, and a one-dimensional vector sequence having similar arrangement per scenario in the classified sea wind pattern scenario are averaged among the values of the respective element elements, A similar pattern categorization unit 54 for performing similar pattern categorization for obtaining a representative vector sequence, and a scenario derivation unit 55 for deriving an off-wind pattern scenario.

A method for the direct ocean wave direct prediction using the sea-wind numerical model prediction data according to the present invention will now be described.

FIG. 5 is a flowchart illustrating a method for an ocean wave direct prediction using the sea-wind numerical model prediction data according to the present invention.

First, a region of interest is set in the two-dimensional sea wind grid region and zoning is performed (S501)

Next, a one-dimensional sea-wind pattern vector sequence is generated in a two-dimensional sea wind field (S502), and a one-dimensional sea-wind vector sequence database is generated for the past sea-wind rear data (S503)

Then, the one-dimensional wind direction vector sequence DB is searched for, weighting by region, similar pattern classification, and similar pattern categorization are performed to derive an ocean wind pattern scenario and DB (S504)

Next, the blue observation data and the ocean-wind rear data are compared to calculate the wave information for each scenario and DB (S505)

Then, it is converted into a one-dimensional vector sequence of a real-time two-dimensional sea-like weather pattern (S506), and an ocean-like pattern scenario DB is searched for and an optimal similar scenario is detected (S507)

Next, wave prediction data is generated from the detected wave information for the ocean-wind scenario (S508)

Here, the step S503 of generating the one-dimensional sea-wind vector sequence database for the past sea-wind rear-side data includes a step of performing one-dimensional vectorization of the detailed in- And performing global vector calculations from global wide-area representative vectors in the global range.

A preferred embodiment of an apparatus and a method for direct ocean wave direct prediction using the sea-wind numerical model prediction data according to the present invention will be described in detail.

FIG. 6 is a block diagram showing an example of a method of setting an ocean-wind-related ROI, and FIG. 7 is a diagram illustrating an example of a four-zone zoning method of ROI.

And FIG. 8 is a diagram showing a one-dimensional vectorization of the two-dimensional sea wind in one zone.

The present invention relates to a method and system for estimating the sea-wind-related information (hereinafter, referred to as " East Sea ") from the past (for example, about 35 years) Categorize patterns and derive representative scenarios.

In addition, through comparison of observed scenarios with observed blue data during this past period, blue information is generated when representative scenarios occur. The calculated sea-wind pattern scenarios and the wave information are converted into a DB and stored.

Then, the scenario type of the ocean-wind pattern is determined from the ocean-wind prediction data produced in real-time through the process of [generating the real-time wave prediction information], and the corresponding And extracts the wave information for the scenario type to generate the prediction data.

[Calculation of information and DB for each sea pattern] The process is as follows.

There are sea-wind forecasts or rear / reanalysis data produced for a considerable period (for example, 35 years) for the surrounding waters including the East Sea of Korea (semi-closed only).

The back side data of this sea wind is bar and it is not useful to search and use.

Therefore, the present invention simplifies the rear-side data of the past sea-wind, categorizes and systematizes the data, and makes it easy to search and utilize it.

Thus, before categorizing offshore wind patterns from past sea wind forecast data, the following two tasks should be preceded.

First, the area of interest needs to be set.

From the meteorological model lattice system in which the offshore wind forecast data is produced, the area of interest is set in consideration of the range that can affect the wave generation in the target sea area (half-closed only).

6 shows an example of setting a region of interest with respect to the East Sea within a lattice system (lattice spacing: 9 km) of a weather model (e.g., a WRF model).

For convenience of calculation, the region of interest is set so that the number of lattices in the horizontal and vertical directions is 128 x 128 so as to be a multiple of 4. [

And a four-stage zoning of the area of interest.

In order to systematically categorize the ocean-like patterns and make it easy to DB, it is necessary to zigzag into four levels such as global, wide-area, mid-range and detailed stations as shown in FIG.

The size of each stepwise zone is for convenience, and may be set differently depending on the physical or computational needs, such as the area size, the zoning step, and so on.

The categorization process of the offshore wind pattern can facilitate the classification and categorization of the pattern when the four zone systems shown above are stepped through.

In the process of categorizing the patterns of the oceanic winds, it is required to simplify and convert the two-dimensional sea-wind vector fields into one-dimensional vectors.

One-dimensional vectorization of a two-dimensional vector field in a region in the present invention is described as follows.

The sea wind is represented by a two-dimensional velocity vector (u, v) with u-direction and v-direction components in the horizontal plane. Therefore, the two-dimensional wind-swept vector field in one area can be spatially expressed in a two-dimensional plane, and when stored as a DB of digital data, it is generally stored in a two-dimensional array form.

In order to reduce the size of the 2D vector field data and simplify the characteristics of the wind field, the present invention includes a configuration for converting the vector data into a one-dimensional vector.

For example, as shown in FIG. 8, there are vectors of 64 grid cells in a region of 8 × 8 grid.

In order to characterize the wind field in this zone as a one-dimensional vector, only the direction of the wind speed is considered for the eight directions. For these eight directions, the vectors of all 64 grid cells are decomposed in eight directions, and a cumulative sum is obtained for each direction.

Then the cumulative sum of the wind velocity magnitudes for each direction is divided by 64, the number of grid cells. From this it is possible to obtain the average magnitude of the wind speed for the eight directions in this zone.

In addition, the sum of the average wind velocities in the eight directions becomes the representative average wind velocity of this zone.

In the embodiment of the present invention, the direction of the wind speed is considered only for eight directions at intervals of 45 degrees, but it is natural that the direction of the wind speed can be considered for 16 directions at intervals of 22.5 degrees.

Next, a process of categorizing the off-wind patterns and deriving the off-wind pattern scenarios from the past off-wind prediction (or rear) data will be described.

FIG. 9 is a diagram showing the generation of a one-dimensional vector sequence from a two-dimensional marine raincoat.

FIG. 10 is a diagram illustrating a DB generation process of a one-dimensional vector sequence from the past background data of the past sea-wind; and FIG. 11 is a diagram illustrating a process of deriving a sea-wind pattern scenario from a past one-dimensional vector sequence database.

The process of deriving the scenario of the offshore wind pattern is as follows: (i) the generation of a one-dimensional sea-wind vector sequence from one two-dimensional sea wind vector field (or sea fringe), (ii) And (iii) generation of a sea-wind pattern scenario through grouping of similar patterns from the DB of the past one-dimensional sea-wind vector sequence.

(i) The generation of a one-dimensional vector sequence from a two-dimensional marine rainfall pattern is as follows.

The area of interest of the East Sea maritime festival is divided into four levels, namely, global, regional, intermediate and detailed.

The one-dimensional vector sequence generated from one two-dimensional marine rainbow span combines the one-dimensional vectors converted from the wind speed vectors (u, v) belonging to the global, each wide area, each middle area and each detailed area .

In the region of interest, the one-dimensional vector sequence is combined with a total of 85 one-dimensional vectors including 1 global, 4 wide, 16 intermediate, and 64 detailed vectors, and 85 × 8 = 680 vector elements. .

In order to reduce the time required to generate a one-dimensional vector sequence from one two-dimensional marine floodplain, a one-dimensional vector can be obtained from the detailed station to the mid-range, wide-area, and global order in the following procedure.

(b) one-dimensional vectorization of the detailed in-depth two-dimensional vector field, (b) representative vector calculation from the detailed inverse 1-dimensional vectors in each midpoint, (c) wide-area representative vector calculation from the central representative vectors in the wide area, The vector is obtained from the wide-area representative vectors in the order of the global representative vector calculation.

(ii) DB generation of a one-dimensional vector sequence from the past back-side data DB is as follows.

In the past 35 years, if one ocean fringe is produced every three hours, the total number of 35 × 365 × 8 = 102,200 marine fringes is present in the database. Therefore, a DB having a total of 102,200 one-dimensional vector sequences can be generated as shown in FIG. 10 from the DB of the rear side data.

(iii) The evolution of the sea wind pattern scenario from the past one-dimensional vector sequence DB is as follows.

There are a total of 102,200 marine floodplain one - dimensional vector sequences in the backwater database of the past sea - wind.

There may not be a perfect match between the two, but there is a duplicate of similar patterns in the sea-wind pattern. It is more difficult to find elements similar to the elements of a vector in a global or wide area in a one-dimensional vector sequence, and to find similar arrangements in the middle and detailed regions.

On the other hand, one two-dimensional sea flume or one-dimensional sea-wind vector sequence determines one particular wave information (i.e., wave, cycle, and wave) at any one point. In the past, since the number of sea flares in the DB is 102,200 in total, 102,200 pieces of wave information are determined.

Considering that past sea-like DBs have overlapping patterns of similar sea-wind patterns, the number of different wave information will be smaller than that of the other.

Generally, in the East Sea, a wave phenomenon occurs in a range of 0 to 12 m, a period of 3 to 15 s, and a wave direction of 0 to 360 °. If the resolution of the wave information to be obtained from the ocean-wind prediction data is set to 0.2 m, the period is 0.5 s, and the wave direction is 15 °, the number of wave information is 60 × Period) 24 占 (wave direction) 24 = 34,560. That is, in order to generate 34,560 wave information cases, 34,560 ocean wind pattern scenarios are necessary.

Therefore, reclassification is required from a total of 102,200 in the case of the one-dimensional vector sequence of the past sea-wind data to 34,560 in the case of the sea-wind pattern.

As a result, it can be said that an average of 102,200 ÷ 34,560≈3 similar 1-dimensional vector sequences per ocean-wind scenario is obtained.

A method of supervising and supervising machine learning can be used as a method of reclassifying 102,200 1-dimensional vector sequences of past sea-wind DBs into similar 34,560 sea-wind scenarios by combining similar ones.

For example, as shown in FIG. 11, it is possible to use a k -mean cluster algorithm ( k- mean algorithm) in a supervised manner.

In addition, it is possible to reclassify the similarity in the global and wide regions by assigning a higher weight to the similarity.

There are differences in the calculation methods used for classification / clustering according to the applied machine learning technique, but the characteristics (number and pattern) of the classified / clustered groups will be similar.

One-dimensional vector sequences having similar arrangement per scenario in a classified ocean pattern pattern scenario are averaged among the values of each element element to obtain one representative vector sequence.

The calculation of the wave information for each scenario of the sea-wind pattern is derived from observed wave observations in the past when the corresponding scenarios occurred.

In the process of deriving the offshore scenarios, it is expected that there will be an average of three similar 1-dimensional vector sequences clustered per scenario. In other words, this means that similar sea-wind patterns occurred about three times in each scenario, and also that blue observations were collected about three times at the same time.

Therefore, the calculation of the wave information at any point in the ROI is averaged from the wave information data observed at that point in the past.

Calculation of the wave information outside the wave observation point in the area of interest can be derived from the wave modeling interpolation technique using the surrounding wave observation data.

The process of generating the wave prediction information using the above-described back-side wind data and the wave information observation data is as follows.

That is, the present invention can detect wave information in an arbitrary point of interest in a short period of time in a short time using the ocean wave prediction data produced in the future, based on the wave information DB of the established ocean-wind scenario.

[Real-time wave prediction information generation process] is as follows.

The procedure for predicting wave information from arbitrary point of interest from the pre-produced ocean-wind forecast data in real time is summarized as follows.

(i) converted into a one-dimensional vector sequence of a real-time two-dimensional marine rainfall pattern

(ii) Retrieval of the ocean wind pattern scenario DB and optimal similar scenario detection

(iii) Generate wave predictive data from the detected wave information for the ocean-wind scenario

In the procedure for predicting the wave information at an arbitrary point of interest from the previously produced sea-wind prediction data in real time,

(i) The conversion into a one-dimensional vector sequence of a real-time two-dimensional sea flume is performed as follows.

The two-dimensional sea-wind vector field produced for real-time pre-prediction is converted into a one-dimensional sea-wind vector sequence.

Assuming that real-time marine flares for forecasting are generally predicted at 3-hour intervals for the next 3 days, time series data of total 3 x 8 = 24 1-dimensional sea-wind vector sequences are generated.

(ii) The retrieval of the sea wind pattern scenario DB and the detection of the optimal similar scenario are as follows.

The one-dimensional vector sequence is an array vector having an element element length of 860 as described above.

A method of matching the current vector sequence and the vector sequences in the past DB to detect the most similar scenarios from the past sea-wind scenario DB for a single sea-wind vector sequence (i.e., array vector) produced for future forecasting Is used.

Since the one-dimensional vector sequence is an array vector, a nearest neighbor scheme can be used as a method of matching the same array vector.

The neighboring approach is to find the inner product between two array vectors, and when the result is maximized relative to the inner values with other array vectors, the two vector sequences are matched to each other most closely.

In the same way, 24 consecutive offshore wind scenarios are detected for the 24 offshore wind vector sequences produced for future forecasting.

(iii) The generation of the wave prediction data from the detected wave information for the offshore wind scenario is as follows.

Wave information corresponding to 24 consecutive detected sea-wind scenarios at any one point or a plurality of points in the object area is also detected from the past wave information DB to generate wave prediction data of 24 consecutive points .

Since the prediction values of the blurring information of 24 consecutive blur points are a series of trends that gradually change, it is possible to filter the prediction values of the blurring information by a smoothing method or the like in order to reduce an error that may be caused from the detection process of the past DB .

As described above, the apparatus and method for the direct ocean wave direct prediction using the sea wind numerical model predictive data according to the present invention categorize and quantify the relationship between the pattern of the sea wind that has occurred in the past and the occurrence pattern of the sea wave, To predict the future waves directly from the forecast / forecast data of the sea wind.

The prediction error can be minimized by using the observation data of the wave measured at a certain point, which is the final result through all the mechanism processes related to the generation and propagation of the wave, and it is possible to minimize the prediction error by using the sea- It can be effectively applied only to a certain scale of closing or semi-closing which is easy to categorize and quantify.

As described above, it will be understood that the present invention is implemented in a modified form without departing from the essential characteristics of the present invention.

It is therefore to be understood that the specified embodiments are to be considered in an illustrative rather than a restrictive sense and that the scope of the invention is indicated by the appended claims rather than by the foregoing description and that all such differences falling within the scope of equivalents thereof are intended to be embraced therein It should be interpreted.

100. Wave information calculation and DB by sea pattern
200. The real-time blur prediction information generation unit

Claims (12)

A one-dimensional off-axis wind vector sequence generation unit for generating a one-dimensional off-axis wind vector sequence of a two-dimensional marine airfoil, and a one-dimensional on- Dimensional sea-wind vector sequence database and a one-dimensional sea-wind vector sequence database, which construct a one-dimensional sea-wind vector sequence database for all past rear data using DB, A windfall pattern scenario derivation unit for deriving a sea wind pattern scenario by performing categorization, an observation data comparison unit for comparing the blue observation data and the rear wind data, and the wave information for each scenario based on the comparison result of the observation data comparison unit And calculates the categorization of the offshore wind pattern using the past backward data of the offshore wind, A wave information calculation and DB for an ocean wind pattern for calculating a wave information per pattern and DBization;
And a real-time wave prediction information generation unit for determining the pattern type of the sea-wind prediction data in real time, calculating the wave information for each of the off-wind patterns, and calculating the wave information for the corresponding pattern from the DB, An apparatus for direct ocean wave prediction using numerical model predictive data. The apparatus of claim 1, wherein the real-
A prediction data providing unit for providing two-dimensional sea-wind vector field prediction data in real time,
A sea-wind vector sequence generator for generating a one-dimensional sea-wind vector sequence using the two-dimensional sea-wind vector sheath prediction data,
An optimal scenario detecting unit for searching for a sea wind pattern scenario DB and detecting an optimal similar scenario by using a blue information calculation for a sea wind pattern and a DB,
And a wave prediction data generation unit for generating wave prediction data from the wave information of the detected sea-wind scenario. delete The method as claimed in claim 1, wherein the one-
A one-dimensional vectorization unit for performing one-dimensional vectorization of the in-depth two-
An intermediate vectorization unit for performing each of the intermediate vector calculations from the detailed inverse one-dimensional vectors in each of the intermediate vectors,
A wide vectorization unit for performing each wide area vector calculation from the central representative vectors in the wide area,
And a global vectorization unit for performing a global vector calculation from wide-area representative vectors in the entire region. The apparatus for the marine wave direct prediction using the sea-level numerical model prediction data. The wind power generation system according to claim 1,
A one-dimensional vector sequence DB search unit for searching a one-dimensional vector sequence DB,
Whether or not each one-dimensional vector sequence to be weighted is weighted for each region,
A similar pattern classifying unit for classifying similar patterns;
A similar pattern categorizing unit for performing similar pattern categorization,
And a scenario derivation unit for deriving a sea-wind pattern scenario based on the sea-wind numerical model prediction data. 6. The apparatus according to claim 5, wherein the similar-
Wherein the one-dimensional vector sequences having similar arrangements per scenario in the classified sea-wind pattern scenario are obtained by averaging the values of the respective element elements to obtain one representative vector sequence. In the case of the marine wave direct prediction using the sea- Device. Setting a region of interest in the two-dimensional sea wind grid region and performing zoning;
The one-dimensional sea-wind vector sequence generator generates a one-dimensional sea-wind pattern vector sequence in a two-dimensional sea flume and generates a one-dimensional sea-wind vector sequence database for a past sea- step;
A step of retrieving a one-dimensional sea-wind pattern vector sequence DB from a sea-wind pattern scenario derivation unit and performing weighting, similar pattern classification, and similar pattern categorization for each region;
Comparing the blue observation data with the backward data of the ocean in the observation data comparing section, calculating the wave information and DB of the scenario in the wave information calculating section,
A step of converting the one-dimensional vector sequence of the real-time two-dimensional marine rainfall pattern into the one-dimensional vector sequence of the real-time two-dimensional marine weather pattern from the sea-wind pattern vector sequence generating unit,
And generating wave predictive data from the wave information of the sea-wind scenario detected by the wave predictive data generation unit. delete The method as claimed in claim 7, wherein the step of generating the one-dimensional wind-based vector sequence database for the past-
Performing one-dimensional vectorization of the in-depth two-dimensional marine rainfall field;
Performing each rank vector calculation from detailed inverse one-dimensional vectors in each executive;
Performing a wide-area vector calculation from the central representative vectors in the wide area,
And performing global vector calculation from wide-area representative vectors within the entire range. 8. The method as claimed in claim 7, wherein the retrieval of the sea wind pattern scenario DB and the detection of the optimal similar scenario include:
A method for direct ocean wave prediction using ocean-wind numerical model predictive data, characterized in that the current vector sequence and the vector sequences in the past DB are matched with a nearest neighbor method. 11. The method of claim 10, further comprising: obtaining an inner product between the current vector sequence and an array vector of vector sequences in the past DB, and when the result is maximized relative to the inner values with other array vectors, A method for direct ocean wave prediction using a sea-wind numerical model prediction data characterized by matching the most similar ones to each other. 8. The method of claim 7, wherein, in the step of generating the blue prediction data from the blue information for the detected ocean-wind scenario,
Detecting wave information corresponding to a plurality of consecutive sea wind scenarios detected at any one point or a plurality of points in the object area from the past wave information DB to generate wave prediction data at a plurality of consecutive points,
A method for an ocean wave direct prediction using a sea-wind numerical model prediction data, characterized in that prediction values of wave information of a plurality of points are filtered by a smoothing method in order to reduce an error in a detection process of a past DB.

Images