JP2019067074A - Wind environment learning device, wind environment evaluation system, wind environment learning method and wind environment evaluation method - Google Patents

Abstract

To provide a wind environment learning device and a wind environment learning method, a wind environment evaluation system, and a wind environment evaluation method capable of machine learning of calculation for evaluating a wind environment at a construction time of a high-rise building or the like.SOLUTION: A wind environment learning device 1 includes: a past survey target data capture unit 20 having a past survey target CAD data capturing unit 21 for capturing CAD data of a past survey target building and its surroundings, a past survey target wind information capturing unit 22 for capturing wind data, and a past survey target position data capture unit 23; a past survey result data capture unit 30 having at least one of a past survey result wind speed data capture unit 31 and a past survey result rank data capture unit 32; and a machine learning unit 40 provided to learn a relationship between the past survey target data and the past survey result data from the past survey target data captured by the past survey target data capture unit 20 and the past survey result data captured by the past survey result data capture unit 30 corresponding to the past survey target data captured by the past survey target data capture unit 20.SELECTED DRAWING: Figure 1

Description

  The present invention provides a wind environment learning device and a wind environment learning method for machine learning a calculation for evaluating wind environment at the time of construction of a high-rise building etc., and a wind environment at the design stage at the construction stage of a high-rise building etc. The present invention relates to a wind environment evaluation system and a wind environment evaluation method for performing evaluation.

  In the vicinity of high-rise buildings etc., strong wind such as building wind may occur. Therefore, at the time of construction of a high-rise building or the like, it is necessary to evaluate changes in the wind environment and to confirm that the wind speed falls within a predetermined speed.

  Conventionally, as a method of predicting a change in wind environment, a technique for calculating a wind speed by taking a model similar to a building to which a pressure-sensitive paint is applied in a wind tunnel test and analyzing the pressure-sensitive paint is disclosed (Patent Document 1).

JP, 2014-48120, A

Wind Engineering Research Institute ed., "Basic knowledge of building wind", Kashima Press, December 2005

  However, as a result of conducting the wind tunnel test, if the wind speed does not fall within the predetermined speed, it is necessary to return to the design stage of the building and make a model of the building again. Therefore, there was a risk that it would cost a great deal of cost and construction time.

  The present invention provides a wind environment learning device and method that allows machine learning of computations to evaluate wind environment at the time of construction of a high-rise building etc., and data at an early stage of design without using an experimental model etc. It is an object of the present invention to provide a wind environment evaluation system and a wind environment evaluation method capable of quickly and inexpensively evaluating a wind environment by a designer himself.

The wind environment learning device according to the present invention is
Past survey target CAD data capture unit that captures CAD data of the past survey target building and its surroundings, past survey target wind information capture unit that captures wind data of the past survey target building and its surroundings, and past survey target A past survey target data capture unit having a past survey target position capture unit that captures location data of a building and its surroundings;
The past survey result wind speed importer taking in the distribution of the wind velocity ratio for each wind direction around the building under the past survey, and the past research result rank fetching unit taking in the result of ranking the wind environment around the building under the past survey A past survey result data capture unit having at least one of them;
Past survey result data captured by the past survey target data capture unit and past survey result data captured by the past survey result data capture unit corresponding to past survey target data captured by the past survey target data capture unit And a machine learning unit for learning the relationship between the past survey target data and the past survey result data,
And the like.

The wind environment evaluation system according to the present invention is
An input unit for inputting data,
The machine learning unit according to claim 1, wherein the data sent from the input unit is calculated by machine learning.
An output unit that outputs a result calculated by the machine learning unit;
Equipped with
The input unit is
CAD data input unit for inputting CAD data of the target building and its surroundings,
A wind information input unit for inputting wind data of the target building and its surroundings;
A target position input unit for inputting position data of the target building;
Have
The machine learning unit is characterized in that calculation is performed on data input from the input unit on the basis of machine learning learned in advance, and a wind environment of the target building and its surroundings are predicted.

The wind environment evaluation system according to the present invention is
The machine learning unit ranks the predicted wind environment based on predetermined criteria.

The wind environment learning method according to the present invention is
A first step of capturing a past surveyed building and its surrounding CAD data, a past surveyed building and its surrounding wind data, and a past surveyed building and its surrounding location data;
As a result of ranking the wind environment around the building under investigation in the past, and a second step of taking in at least one of the distribution of the wind speed ratio for each wind direction around the building under investigation in the past,
The past survey target data and the past survey result data from the input data fetched in the first step and the past survey result data fetched in the second step corresponding to the past survey target data fetched in the first step The third step of learning the relationship with
It is characterized by having.

The wind environment evaluation method according to the present invention is
Input CAD data of the target building and its surroundings,
Enter wind data around the target building,
Enter the location data of the target building and its surroundings,
An operation is performed based on machine learning learned by the wind environment learning method according to claim 4 for each input data,
It is characterized by outputting the evaluation result which estimated the wind environment of the said object building and its circumference which were calculated.

The wind environment evaluation method according to the present invention is
The predicted evaluation results are classified into ranks based on predetermined criteria.

  According to the wind environment learning device and the wind environment learning method according to the present invention, it is possible to perform machine learning of calculation for evaluating the wind environment at the time of construction of a high-rise building or the like. Moreover, according to the wind environment evaluation system and the wind environment evaluation method according to the present invention, it is possible to evaluate the wind environment at a low price and quickly from the data at an early stage of design without using an experimental model. It becomes.

The control block of the wind environment learning apparatus of this embodiment is shown. The flowchart of the wind environment learning method using the wind environment learning apparatus of this embodiment is shown. The object building used for the wind environment learning apparatus of this embodiment, and CAD data of the periphery of it are shown. The two-dimensional information showing the presence or absence of the building used for the wind environment learning apparatus of this embodiment is shown. The three-dimensional information showing the existence and height of a building used for the wind environment learning device of this embodiment is shown. The wind information used for the wind environment learning apparatus of this embodiment is shown. The data showing the statistical information regarding the wind used for the wind environment learning apparatus of this embodiment are shown. The observation data of the wind of the predetermined | prescribed point used for the wind environment learning apparatus of this embodiment are shown. The control block of the wind environment evaluation system of this embodiment is shown. An example of the wind environment evaluation result in the wind environment evaluation system of this embodiment is shown. The other example of the wind environment evaluation result in the wind environment evaluation system of this embodiment is shown. The flowchart of the wind environment evaluation method using the wind environment evaluation system of this embodiment is shown.

  Hereinafter, a wind environment learning apparatus according to an embodiment of the present invention will be described with reference to the drawings.

  FIG. 1 shows a control block of the wind environment learning device 1 of the present embodiment.

  The wind environment learning device 1 according to the present embodiment includes a past data acquisition unit 10 that inputs data, and a learning unit 40 that performs machine learning on data sent from the past data acquisition unit 10.

  The past data acquisition unit 10 of the present embodiment has a past investigation target data acquisition unit 20 for acquiring data of a past investigation target, and a past investigation result data acquisition unit 30 for acquiring data of a past investigation result. .

  The past investigation target data acquisition unit 20 incorporates a past investigation target CAD data acquisition unit 21 for acquiring CAD (Computer-Aided Design) data of a past object building and its surroundings, and wind data of the past object building and its surroundings. It has a past investigation target wind information taking-in unit 22 to take in, and a past investigation target position taking-in unit 23 to take in position data of a past object building and its surroundings.

  The past survey result data capture unit 30 receives the past survey target CAD data capture unit 21 captured in the past survey target data capture unit 20, the past survey target wind information capture unit 22, and the past survey target position capture unit 23. Capture the results obtained by past experiments or calculations. The past survey result data capture unit 30 of this embodiment ranks the past survey result wind velocity capture unit 31 that captures the distribution of the wind velocity ratio for each wind direction around the target building in the past, and the wind environment around the target building in the past. It has at least one of the past survey result rank capture unit 32 that captures the divided result.

  The machine learning unit 40 includes a past survey result data capture unit 30 corresponding to past survey target data captured by the past survey target data capture unit 20 and past survey target data captured by the past survey target data capture unit 20. From the past survey result data taken in, the relationship between the past survey target data and the past survey result data is learned. This learning is preferred as much as possible.

  FIG. 2 shows a flow chart of a wind environment learning method using the wind environment learning device of the present embodiment.

  In the wind environment learning method, machine learning is performed based on the flowchart as shown in FIG.

  First, in step 1 as the first step, the past investigation target data acquisition unit 20 acquires data of the past investigation target (ST1). Subsequently, in step 2 as the second step, the past investigation result data taking-in unit 30 takes in the past investigation result data obtained by the past experiment or calculation, etc. corresponding to the data to be examined in the past (ST2) . Next, in step 3 as the third step, the machine learning unit 40 performs machine learning on the correspondence between the data to be surveyed in the past and the survey result data in the past (ST3).

  Here, the contents of each data will be described. First, CAD data will be described.

  FIG. 3: shows an example of the object building used for the wind environment learning apparatus 1 of this embodiment, and CAD data of the periphery of it.

  The past survey target CAD data acquisition unit 21 first reads image information regarding the shape of the urban area such as the topography, the shape of the building, and the arrangement of plantings. The image information is preferably STL (Stereolithography) file data or the like, which represents a three-dimensional three-dimensional shape by a set of small triangles. The three-dimensional data may be two-dimensional data because of its large capacity.

  FIG. 4: shows an example of the two-dimensional information showing the presence or absence of the building used for the wind environment learning apparatus 1 of this embodiment. FIG. 5 shows an example of three-dimensional information representing the presence / absence and height of a building used for the wind environment learning device 1 of the present embodiment.

  The past survey object CAD data acquisition unit 21 divides the image information shown in FIG. 3 into blocks of the examination range at equal intervals, and places “1” in a building location and “0” in a location without a building. It may be converted into numerical data as shown in FIG. 4 displayed. In the image information of this embodiment, for example, an area of 1 km to 500 m is divided into orthogonal grids of 1 m. From this data, the presence or absence of surrounding buildings and roads can be understood centering on the target building.

  In addition, as shown in FIG. 5, the CAD data acquisition unit 21 for past investigation may be added with data representing the height of the building, as shown in FIG. 4. For example, as shown in FIG. 5, it can be seen that the target building is 250 m and the surrounding buildings above the page of the target building are 120 m.

  Next, wind information will be described.

  FIG. 6 shows wind information used for the wind environment learning device 1 of the present embodiment. FIG. 6 (a) shows the wind direction frequency distribution, and FIG. 6 (b) shows the wind speed intensity frequency distribution. FIG. 7 shows data representing statistical information on wind used in the wind environment learning device 1 of the present embodiment. However, C is a scale constant and K is a shape constant. 6 and 7 are cited from Non-Patent Document 1.

  The past survey target wind information capture unit 22 captures wind information around the target building. The past investigation target wind information taking-in unit 22 may use, as wind information, a wind direction frequency distribution and a wind speed intensity frequency distribution based on wind observation records of weather stations as shown in FIG. The wind direction frequency distribution shown in FIG. 6A indicates how often the wind from which direction blows around the target building. For example, in FIG. 6 (a), it can be seen that there are many winds from north-northwest. The wind speed intensity frequency distribution shown in FIG. 6 (b) indicates which wind speed the wind blows around and around the target building. For example, in FIG. 6 (b), it can be seen that the frequency of the wind speed 2.0-2.9 (m / s) is high.

  The example shown in FIG. 7 is a probability distribution that statistically represents the wind intensity at each point. The past survey target wind information taking-in unit 22 may use data obtained by processing wind data as shown in FIG. 7 as wind information. In order to know the wind conditions at a predetermined point, a local wind condition map or the like of the New Energy and Industrial Technology Development Organization (NEDO) may be used.

  FIG. 8 shows observation data of wind at a predetermined point used for the wind environment learning device 1 of the present embodiment.

  The example shown in FIG. 8 is data of time history waveforms of wind speed and wind direction observed at a predetermined point. The past survey target wind information taking-in unit 22 may use actually observed data as shown in FIG. 8 as wind information.

  Next, position data will be described. The past survey target position capture unit 23 captures data of the location of the survey target. The data of the target position may use latitude and longitude or predetermined coordinate information. One piece of coordinate information is set for one target position. The target position is preferably wind information at the height of a pedestrian around the target building.

  Next, the past survey result data capture unit 30 for capturing past survey result data corresponding to the past survey target data fetched from the past survey target data capture unit 20 will be described. The past investigation result data capture unit 30 of the present embodiment takes in the past investigation result wind speed taking in the distribution of the wind speed ratio for each wind direction in the vicinity of the target building obtained by calculation, observation or experiment corresponding to the past investigation target data. At least one of the past survey result rank capture unit 32 that incorporates the wind environment around the target building obtained by calculation, observation or experiment corresponding to the data to be surveyed in the past, and the like. Have one.

The wind speed distribution and the ranking according to the present embodiment are performed based on set values determined as shown in Table 1 below, as an example. Table 1 is cited from Non-Patent Document 1.

  The past investigation result wind speed capture unit 31 takes in past investigation result data such as the distribution of the wind speed ratio for each wind direction around the target building obtained by calculation, observation or experiment corresponding to the past investigation target data. The past survey result rank capture unit 32 takes in the past survey result data ranked and the like by the wind speed with a cumulative frequency of 55% and the wind speed with a cumulative frequency of 95% corresponding to the past survey target data.

  For example, in the present embodiment, region A is considered as rank A as a wind environment seen in a residential area, region B is considered as rank B as a wind environment seen in an intermediate block between regions A and C, and region C in an office district Rank C as the wind environment to be seen, and rank D as the undesirable wind environment.

  Next, another example of the past result data in the wind environment learning device 1 of the present embodiment will be shown.

Another example of the past survey result data is data in which the past survey result rank capture unit 32 divides the wind environment around the target building into other ranks. The ranking of this example is performed based on setting values determined as shown in Table 2 below, as an example. Table 2 is cited from Non-Patent Document 1.

  In Table 2, the daily maximum instantaneous wind speed is an evaluation time of 2-3 seconds, and the daily maximum average wind speed is an average wind speed of 10 minutes, and the wind speed is 1.5 m above the ground. In addition, at 10 m / s, the phenomenon occurs that dust flies and dried matter flies at the maximum instantaneous wind speed, and at 15 m / s, the signboard and bicycle fall down and walking becomes difficult, and at 20 m / s the wind blows away The phenomenon that is likely to occur is generated.

  Gust factor G. F (Gust Factor) is a gust factor representing the ratio of the maximum instantaneous wind speed to the maximum average wind speed. The gust factor in Table 2 is 1.5 m on the ground, and the evaluation time is 2 to 3 seconds. The gust factor is 2.5 to 3.0 in a dense urban area where the disturbance is strong but the average wind speed is not so high, 2.0 to 2.5 in a normal urban area, 1.5 to 2.0 in a particularly high wind speed area such as a speedup zone near high buildings. is there.

  In this example, the past investigation result wind speed capture unit 31 takes in the distribution of the wind speed ratio for each wind direction in the vicinity of the target building obtained by calculation, observation or experiment corresponding to the data to be surveyed in the past. The rank taking-in unit 32 takes in past data ranked according to the daily maximum average wind speed.

  In Table 2, for example, it can be read that in the case of application of rank 1, it is acceptable if the frequency at which the daily maximum instantaneous wind speed exceeds 10 m / s is 10% or less (about 37 days a year).

  Next, the machine learning unit 40 will be described. The machine learning unit 40 of the present embodiment receives the target building and its surrounding CAD data, the wind data around the target building, and the position data of the target building and its surrounding, the wind of the target building and its surrounding It mechanically learns operations that output evaluation results to predict the environment.

The machine learning unit 40 may learn from information such as wind speed ratio evaluation for each wind direction when analyzing with a wind tunnel test or a numerical fluid analysis model with an existing experimental model.

  As described above, the wind environment learning device 1 according to the present embodiment receives results from the past investigation result data acquisition unit 30 corresponding to the input data from the input data acquired from the past investigation target data acquisition unit 20. The machine learning unit 40 is made to learn a relationship such that data is output.

  Therefore, when input data is input, it is possible to manage data such that result data is immediately output.

  FIG. 9 shows a control block of the wind environment evaluation system 100 of this embodiment.

  The wind environment evaluation system 100 of this embodiment includes an input unit 50 for inputting data of a target building and its surroundings, a machine learning unit 40 for calculating data sent from the input unit 50 by machine learning, and a machine learning unit 40. And an output unit 60 that outputs the result of the calculation.

  The input unit 50 of the present embodiment includes a CAD data input unit 51 for inputting CAD (Computer-Aided Design) data of a target building and its surroundings, a wind information input unit 52 for reading wind data of the target building and its surroundings, And an object position input unit 53 for inputting position data of the object building and its surroundings.

  First, CAD data will be described. The CAD data input from the CAD data input unit 51 is CAD data as shown in FIG. 3 which represents a target building to be evaluated and its surroundings.

  First, the CAD data input unit 51 reads image information regarding the shape of the urban area, such as topography, building shape, and planting arrangement. The image information is preferably STL (Stereolithography) file data or the like, which represents a three-dimensional three-dimensional shape by a set of small triangles. Note that two-dimensional data may be used.

  The image information shown in FIG. 3 is the numerical data as shown in FIG. 4 in which the block of the examination range is divided at equal intervals, the place where the building is located is indicated by “1” and the place where there is no building is indicated by “0”. Converted to In the image information of this embodiment, for example, an area of 1 km to 500 m is divided into orthogonal grids of 1 m. From this data, the presence or absence of surrounding buildings and roads can be understood centering on the target building.

  Moreover, you may add the information which displayed the height of the building shown in FIG. 5 with respect to FIG. For example, as shown in FIG. 5, it can be seen that the target building is 250 m and the surrounding buildings above the page of the target building are 120 m.

  Next, wind information will be described. The wind information input unit 52 inputs information on wind around the target building. The wind information input unit 52 may use, as the wind information, a wind direction frequency distribution and a wind speed intensity frequency distribution based on the wind observation record of the weather station or the like shown in FIG.

  The wind information input unit 52 may use data obtained by processing the wind data shown in FIG. 7 as wind information. In order to know the wind conditions at a predetermined point, a local wind condition map or the like of the New Energy and Industrial Technology Development Organization (NEDO) may be used.

  The example shown in FIG. 8 is data of time history waveforms of wind speed and wind direction observed for two months at a predetermined point. The wind information input unit 52 may use the actually observed data shown in FIG. 8 as wind information.

  Next, position data will be described. The target position input unit 53 inputs data of the target position. The data of the target position may use latitude and longitude or predetermined coordinate information. One piece of coordinate information is set for one target position.

  Next, the machine learning unit 40 of the present embodiment will be described. The wind environment evaluation system 100 of this embodiment uses the machine learning unit 40 learned by the wind environment learning device 1 shown in FIG. 1. From the input data input from the input unit 50, the machine learning unit 40 causes the output unit 60 to output output data corresponding to the input data.

  The machine learning unit 40 according to the present embodiment performs calculation by machine learning based on the data input from the CAD data input unit 51, the wind information input unit 52, and the target position input unit 53, and calculates the relationship between the surroundings of the target building and the wind. Output information to the output unit 60. The target position to be output is wind information at the height of the pedestrian around the target building.

The machine learning unit 40 may learn information such as an evaluation of the wind speed ratio for each wind direction when analyzed by a wind tunnel experiment or a numerical fluid analysis model with an existing experimental model, and may use the information for calculation.

  The output unit 60 of the present embodiment outputs the distribution of the wind speed ratio for each wind direction around the target building calculated by the machine learning unit 40, and the wind environment around the target building calculated by the machine learning unit 40. And at least one of the rank output unit 62 that outputs the result of the rank division.

  FIG. 10 shows an example of the wind environment evaluation result in the wind environment evaluation system 100 of the present embodiment.

  The example illustrated in FIG. 10 represents the wind environment around the target building by rank. In the present embodiment, the ranking may be performed based on the setting values determined as shown in Table 1.

  The machine learning unit 40 calculates the wind speed with a cumulative frequency of 55% and the wind speed with a cumulative frequency of 95%. And rank division is performed by the calculated wind speed. For example, in the present embodiment, region A is considered as rank A as a wind environment seen in a residential area, region B is considered as rank B as a wind environment seen in an intermediate block between regions A and C, and region C in an office district Rank C as the wind environment to be seen, and rank D as the undesirable wind environment.

  The output unit 60 displays the result of the ranking on the map as shown in FIG. The circle on the map represents the target position, and the type of filled circle represents the rank around the target building. For example, it can be seen that the darkened building in the center shown in FIG. 10 is the target building, and the position of rank C is at the upper left of the target building. As described above, according to the wind environment evaluation system 100 of the present embodiment, it is possible to easily understand the wind information of the target building and its surroundings by ranking.

  Also, the wind environment around the target building may be divided into other ranks. In the present embodiment, the ranking may be performed based on the setting values determined as shown in Table 2. It is preferable that the output part 60 displays the result of rank division on a map similarly to FIG. As described above, according to the wind environment evaluation system 100 of the present embodiment, it is possible to easily understand the wind information of the target building and its surroundings by ranking.

  Next, the other example of the wind environment evaluation result in the wind environment evaluation system 100 of this embodiment is shown.

  FIG. 11 shows another example of the wind environment evaluation result in the wind environment evaluation system 100 of the present embodiment.

  The wind environment evaluation system 100 may process and output an image using the wind environment evaluation result for the CAD data shown in FIG. 3. For example, in the example illustrated in FIG. 11, the wind environment evaluation system 100 outputs a white circle part where the wind is strong in the building in a darker color. The display of the wind environment evaluation result is not limited to the output of dark color, but may be to change the color or insert a symbol.

  When the wind environment evaluation system 100 evaluates the wind speed and the wind direction, the wind speed is used as input data at the time of learning, and the wind speed rank is used as output data at the time of calculation. On the other hand, when the wind environment evaluation system 100 evaluates the wind pressure, the wind pressure may be used as input data at the time of learning, and the wind load may be used as output data at the time of calculation.

  FIG. 12 shows a flowchart of a wind environment evaluation method using the wind environment evaluation system 100 of the present embodiment.

  In the wind environment evaluation method using the wind environment evaluation system 100 of the present embodiment, first, in step 11, the input unit 50 inputs CAD data of a target building and its surroundings (ST11). Then, in step 12. The input unit 50 inputs wind information around the target building (ST12). Subsequently, in step 13, the input unit 50 inputs the target building and the position around it (ST13).

  Next, in step 14, the machine learning unit 40 performs an evaluation operation based on the information input from the input unit 50 in steps 1, 2 and 3 and the information from the machine learning unit 40 (ST14). Subsequently, at step 15, the output unit 60 outputs the evaluation result calculated by the machine learning unit 40 at step 14 (ST15).

  As described above, according to the wind environment evaluation method using the wind environment evaluation system 100 of the present embodiment, the designer himself can quickly obtain the wind environment from the data at an early stage of design without using an experimental model or the like. It is possible to evaluate.

  As described above, the wind environment learning device 1 according to the present embodiment incorporates the past survey target CAD data capture unit 21 that captures CAD data of a past survey target building and its surroundings, the past that captures wind data of a past survey target building and its surroundings A past survey target data capture unit 20 having a survey target wind information capture unit 22 and a past survey target position capture unit 23 that captures location data of a past survey target building and its surroundings, and around the past survey target building At least one of the past investigation result rank acquisition unit 32 that incorporates the distribution of the wind velocity ratio for each of the wind directions, and the past investigation result rank acquisition unit 32 that incorporates the results of ranking the wind environment around the building under investigation in the past. Past survey result data capture unit 30 having the following data, past survey target data captured by the past survey target data capture unit 20, and past survey target data capture unit 20. And a machine learning unit 40 for learning the relationship between the past survey target data and the past survey result data from the past survey result data captured by the past survey result data capturing unit 30 corresponding to the past survey target data. . Therefore, the wind environment learning device 1 according to the present embodiment can machine-learn operations to evaluate the wind environment at the design stage when constructing a high-rise building or the like.

  In the wind environment evaluation system 1 of the present embodiment, the input unit 50 for inputting data, the machine learning unit 40 for calculating data sent from the input unit 50 by machine learning, and the machine learning unit 40 The input unit 50 includes a CAD data input unit 51 for inputting CAD data of the target building and its surroundings, and a wind information input unit for inputting wind data of the target building and its surroundings 52 and a target position input unit 53 for inputting position data of a target building, and the machine learning unit 40 performs an operation on the data input from the input unit 50 based on machine learning learned in advance, Predict the target building and the wind environment around it. Therefore, the wind environment evaluation system 1 of the present embodiment can quickly evaluate the wind environment at a low price and quickly from the data at an early stage of design without using an experimental model or the like.

  Further, in the wind environment evaluation system 1 of the present embodiment, the machine learning unit 40 ranks the predicted wind environment based on a predetermined reference. Therefore, the wind environment evaluation system 1 of the present embodiment can easily understand the wind information of the target building and its surroundings.

  Furthermore, the wind environment learning method according to the present embodiment includes CAD data of a past surveyed building and its surroundings, wind data of a past surveyed building and its surroundings, and location data of a past surveyed building and its surroundings. A second step for taking in at least one of a first step for taking in, a result of ranking the wind environment around the building under investigation in the past, and a distribution of wind speed ratios for each wind direction around the building under investigation in the past; Learn the relationship between past survey target data and past survey result data from the input data fetched in the first step and past survey result data fetched in the second step corresponding to the past survey target data fetched in the first step And the third step of Therefore, according to the wind environment learning method of the present embodiment, it is possible to perform machine learning of calculation for evaluating the wind environment at the design stage when constructing a high-rise building or the like.

  Further, in the wind environment evaluation method of the present embodiment, CAD data of the target building and its periphery are input, wind data around the target building is input, and position data of the target building and its periphery are input and input. A calculation is performed on each data based on machine learning learned by the wind environment learning method according to claim 4, and an evaluation result in which the calculated target building and the wind environment around it are predicted is output. Therefore, in the wind environment evaluation method of the present embodiment, the designer can evaluate the wind environment inexpensively and quickly from the data at an early stage of design without using an experimental model or the like.

  Further, in the wind environment evaluation method of the present embodiment, the predicted evaluation results are divided into ranks based on predetermined criteria. Therefore, the wind environment evaluation method of this embodiment can easily understand the wind information of the target building and its surroundings.

  The present invention is not limited by this embodiment. That is, although a lot of specific detailed contents are included for illustration in describing the embodiment, various variations and changes may be added to the detailed contents by those skilled in the art.

1. Wind environment learning device 10: Past data acquisition unit 20: Past investigation object data acquisition unit 21: Past investigation object CAD data acquisition unit 22: Past investigation object wind information acquisition unit 23: Past investigation object position acquisition Part 30 ... past survey result data capture part 31 ... past survey result wind speed capture part 32 ... past survey result rank capture part 40 ... machine learning part 50 ... input part 51 ... CAD data input part 52 ... wind information input part 53 ... target position input unit 60 ... output unit 61 ... wind speed output unit 62 ... rank output unit 100 ... wind environment evaluation system

Claims (6)

Past survey target CAD data capture unit that captures CAD data of the past survey target building and its surroundings, past survey target wind information capture unit that captures wind data of the past survey target building and its surroundings, and past survey target A past survey target data capture unit having a past survey target position capture unit that captures location data of a building and its surroundings;
The past survey result wind speed importer taking in the distribution of the wind velocity ratio for each wind direction around the building under the past survey, and the past research result rank fetching unit taking in the result of ranking the wind environment around the building under the past survey A past survey result data capture unit having at least one of them;
Past survey result data captured by the past survey target data capture unit and past survey result data captured by the past survey result data capture unit corresponding to past survey target data captured by the past survey target data capture unit And a machine learning unit for learning the relationship between the past survey target data and the past survey result data,
Wind environment learning device characterized by having. An input unit for inputting data,
The machine learning unit according to claim 1, wherein the data sent from the input unit is calculated by machine learning.
An output unit that outputs a result calculated by the machine learning unit;
Equipped with
The input unit is
CAD data input unit for inputting CAD data of the target building and its surroundings,
A wind information input unit for inputting wind data of the target building and its surroundings;
A target position input unit for inputting position data of the target building;
Have
The wind environment evaluation system, wherein the machine learning unit performs an operation based on machine learning learned in advance on data input from the input unit, and predicts a wind environment of the target building and its surroundings. The wind environment evaluation system according to claim 2, wherein the machine learning unit ranks the predicted wind environment based on a predetermined standard. A first step of capturing a past surveyed building and its surrounding CAD data, a past surveyed building and its surrounding wind data, and a past surveyed building and its surrounding location data;
As a result of ranking the wind environment around the building under investigation in the past, and a second step of taking in at least one of the distribution of the wind speed ratio for each wind direction around the building under investigation in the past,
The past survey target data and the past survey result data from the input data fetched in the first step and the past survey result data fetched in the second step corresponding to the past survey target data fetched in the first step The third step of learning the relationship with
Wind environment learning method characterized by having. Input CAD data of the target building and its surroundings,
Enter wind data around the target building,
Enter the location data of the target building and its surroundings,
An operation is performed based on machine learning learned by the wind environment learning method according to claim 4 for each input data,
A wind environment evaluation method characterized by outputting an evaluation result obtained by predicting the calculated wind environment of the target building and its surroundings. The wind environment evaluation method according to claim 5, wherein the predicted evaluation results are ranked based on predetermined criteria.

Images