A method for predicting wind pressure inside and outside low-rise buildings with openings

Abstract

The present invention relates to a method for predicting the internal and external wind pressures of low-rise buildings with openings. Firstly, based on wind tunnel tests, a database of internal and external wind pressures of low-rise buildings with layout rules that considers both building interference and the influence of openings is established; secondly, based on the database, the low-rise buildings with openings are divided into three categories of openings; when there is no sufficient wind tunnel test data, the category of openings is judged according to working condition parameters and a reasonable method for determining the internal and external pressures is selected; finally, based on the obtained internal and external wind pressure data, machine learning training is performed to obtain a prediction model for the internal and external wind pressures of low-rise buildings with arbitrary layouts; the established method for predicting the internal and external wind pressures of low-rise buildings can avoid the inaccurate disaster loss assessment caused by blindly using an unsteady Bernoulli equation to calculate the internal and external pressures of buildings, and can also avoid the waste of resources caused by blindly using wind tunnel tests or numerical simulation methods to determine the internal and external pressures of buildings. By introducing machine learning, it is helpful to efficiently realize the wind disaster loss analysis of low-rise buildings.

Description

Technical Field

[0001] The invention belongs to the technical field of urban resilience and disaster assessment, and relates to a method for predicting wind pressure inside and outside a low-rise building complex with openings. Background Art

[0002] Wind-induced damage to the enclosure system of low-rise buildings is a gradual process, which can be roughly divided into four categories according to the damage process and final state (such as Figure 1 shown):

[0003] 1. Wind pressure or wind-induced flying objects cause holes in the windward wall, but the roof is intact;

[0004] 2. Wind pressure causes holes in the roof, but the wall is intact;

[0005] 3. Wind pressure or wind-induced flying objects cause the windward wall to be damaged first and holes to appear. Then the internal pressure increases to positive pressure. The roof with negative pressure on the outer surface is damaged under the combined action of internal and external pressures, and holes are formed on the wall and roof at the same time.

[0006] 4. The wind pressure causes the roof to be damaged first and holes to appear, followed by a large negative internal pressure. The outer surface is under positive pressure and the windward wall is damaged under the combined action of internal and external pressures, or the windward wall is damaged due to debris falling from upstream, and finally holes appear on the wall and roof at the same time.

[0007] At present, the research on the internal pressure of buildings is mainly focused on single buildings with single openings on the wall or roof. When the opening area is not large, the unsteady Bernoulli equation can be used to predict the internal pressure based on the external pressure and the orifice parameters. However, when the roof is severely damaged (i.e., the opening area is large), the applicability of this method needs further study. In addition, current research lacks research on the internal pressure of buildings with openings on both the wall and the roof. However, when there are multiple openings on the wall and the roof at the same time, there are complex fluid dynamics phenomena at the openings, including flow separation, multi-path airflow, local vortex, and sudden increase in wind speed. The uniformity and spatial correlation of the internal pressure are greatly reduced, making it difficult to directly predict the internal pressure using the internal pressure control equation.

[0008] On the other hand, low-rise buildings usually exist in groups, and their internal and external surface wind pressures are often disturbed by surrounding buildings, which may be reduced or amplified. However, this interference effect is affected by many factors such as the layout, spacing and building size of the building group, and it is difficult to draw a generally applicable quantitative conclusion. At present, the research on the interference effect of low-rise building groups mainly focuses on the wind pressure on the outer surface of closed low-rise building groups, but there is no research on the wind pressure on the inner and outer surfaces of open low-rise building groups under the dual factors of interference and opening. However, in the assessment of wind disaster losses, the dual effects of building group interference and openings should be considered at the same time. Therefore, the determination of the wind pressure on the inner and outer surfaces of open low-rise building groups usually requires analysis through wind tunnel tests. However, given the complex interference factors of wind loads on the inner and outer surfaces of low-rise building groups and the diverse opening conditions, it will take a lot of time and economic costs to conduct wind tunnel tests or numerical simulations for each possible condition. Therefore, how to quickly, efficiently and accurately determine the wind pressure distribution law on the inner and outer surfaces of each building in a low-rise building group under various interference and opening conditions is an urgent problem to be solved. Summary of the invention

[0009] In view of this, the present invention provides a method for predicting internal and external wind pressures of low-rise buildings with openings in order to solve the problems that the existing wind pressure prediction methods cannot predict the internal and external pressures of low-rise buildings with openings under the combined influence of wind interference and openings; and cannot predict the internal and external pressures in complex situations such as large openings on walls and roofs, and multiple openings at the same time, and it is necessary to carry out wind tunnel tests on low-rise buildings with openings on the surface, which has high economic and time costs.

[0010] In order to achieve the above object, the present invention provides the following technical solutions:

[0011] A method for predicting wind pressure inside and outside a low-rise building complex with openings comprises the following steps:

[0012] S1. Method for determining internal and external pressure of low-rise buildings considering interference and opening effects under the condition of regular layout buildings;

[0013] S11. Construct a database of internal and external pressure wind tunnel tests on low-rise buildings with openings on the surface;

[0014] Based on the random combination wind tunnel test of low-rise buildings with opening layout rules, the internal and external pressure wind tunnel test database of low-rise buildings with different opening parameters and different building layout parameters is obtained, which takes into account both interference and opening effects.

[0015] S12. Classification of three types of openings and corresponding operating parameter ranges based on the wind tunnel test database;

[0016] According to the opening classification method proposed in the present invention, the building opening categories are classified. First, based on the internal and external pressure wind tunnel test database of low-rise building complexes established in S11, the number of openings on a target building under a certain working condition is determined. When there is only a single opening, the calculation result of the unsteady Bernoulli equation is compared with the wind tunnel test data. If the comparison result shows that the percentage error is less than 5%, the opening category 1 can be obtained, which is suitable for the opening category and the corresponding working condition parameter range determined by the unsteady Bernoulli equation for the internal pressure of the building. If the comparison result shows that the percentage error is greater than 5%, the opening category 3 is obtained, which is suitable for using wind tunnel tests or numerical simulations. The quasi-direct simulation method is used to determine the opening category and the corresponding operating parameter range of the building internal pressure. When there are multiple openings on the building, the calculation results of the multi-point average unsteady Bernoulli equation are compared with the wind tunnel test data. If the comparison result shows that the percentage error is less than 5%, the opening category 2 can be obtained, which is suitable for determining the opening category and the corresponding operating parameter range of the building internal pressure by the multi-point average unsteady Bernoulli equation. If the comparison result shows that the percentage error is greater than 5%, the opening category 3 is also obtained, which is suitable for determining the opening category and the corresponding operating parameter range of the building internal pressure by direct simulation methods such as wind tunnel tests or numerical simulations.

[0017] S13. Rationally select the internal and external pressure determination method based on the three types of openings for buildings with regular layout under unknown working conditions;

[0018] For buildings with regular layouts and unknown internal pressure or unknown external pressure in the absence of sufficient wind tunnel test data (unknown operating conditions), the opening type is determined based on the operating condition parameters and the operating condition parameter range divided by S12, and a reasonable method for determining the internal and external pressures is selected. Opening Category 1 determines the internal and external pressures of the building by solving the unsteady Bernoulli equation; Opening Category 2 determines the internal and external pressures of the building by solving the multi-point average unsteady Bernoulli equation; Opening Category 3 determines the internal and external pressures of the building through wind tunnel tests or numerical simulations. Among them, opening category 1 and opening category 2 obtain the internal and external pressure data of the building under unknown operating conditions based on analytical calculations, and opening category 3 obtains the internal and external pressure data of the building under unknown operating conditions based on direct simulations;

[0019] S2. Prediction method of internal and external pressure of low-rise buildings considering interference and opening effects under the working condition of arbitrary building groups;

[0020] Based on the internal and external pressure data of the regularly laid out building complex obtained in S11 and S13, according to the neural network proxy model optimization iterative method proposed in the present invention, a unique internal and external wind pressure prediction model for low-rise building complexes that takes into account the interference and opening effects under the working conditions of arbitrary building complexes is constructed, which is not currently studied; the basic parameters of the regularly laid out building complexes in steps S11 and S13 are used as the machine learning training input layer, and the mean and standard deviation of the internal and external wind pressure coefficients of the regularly laid out building complexes obtained in steps S11 and S13 are used as the machine learning training output layer, thereby training to obtain a prediction model for the internal and external wind pressure of low-rise building complexes with openings that can predict the working conditions of arbitrary building complexes and takes into account the interference and opening effects at the same time.

[0021] Furthermore, in step S11, based on the wind tunnel test, a database of internal and external wind pressures of low-rise building groups under the working conditions of layout rule building groups is established, which considers both the interference effects of building groups and the effects of building openings. In the prior art, generally, only the influence of the interference effect between group buildings on the wind pressure distribution of buildings is considered unilaterally while ignoring the influence of building openings, or only the influence of openings on the wind pressure distribution of a single building is considered unilaterally while ignoring the interference effect between group buildings. The database of internal and external wind pressures of low-rise building groups established by the present invention considers both the interference of group buildings and the influence of openings on the internal and external wind pressures of buildings. This database is unique and not available in current research, filling the gap in the current research on the internal and external wind pressure data of low-rise building groups with openings in the field of wind disaster assessment, and meeting the urgent needs in wind disaster assessment.

[0022] Furthermore, in step S12, based on the wind tunnel test database of internal and external pressures of low-rise building complexes established in S11, the low-rise building complexes with openings are divided into three opening categories and the corresponding operating parameter ranges are determined according to the opening category classification method proposed in the present invention; the three opening categories are opening category 1: an opening category suitable for determining the internal pressure of the building by the unsteady Bernoulli equation; opening category 2: an opening category suitable for determining the internal pressure of the building by the multi-point average unsteady Bernoulli equation; opening category 3: an opening category suitable for determining the internal pressure of the building by direct simulation methods such as wind tunnel tests or numerical simulations.

[0023] Further, in step S12, the hole category 1 is a hole category applicable to the unsteady Bernoulli equation, and the unsteady Bernoulli equation is calculated according to the following formula:

[0024]

[0025]

[0026] Where X represents the oscillation displacement of the orifice gas column, P e Indicates the external pressure at the orifice, P i represents the internal pressure, ρ a Indicates air density, L eIndicates the effective length of the opening, C L represents the loss coefficient, l0 represents the hole depth, A represents the opening area, C I represents the coefficient of inertia;

[0027] Opening Category 2: Applicable to the opening category of multi-point average unsteady Bernoulli equation. In this case, the internal and external pressures at each opening are calculated according to the single-hole Bernoulli equation. Finally, the internal and external pressures of the entire building are determined by multi-point averaging. The multi-point average unsteady Bernoulli equation is calculated according to the following formula:

[0028]

[0029] Where L en and P en represent the effective length of the nth opening and the corresponding external pressure, X n represents the oscillation displacement of the gas column at the nth orifice;

[0030] Opening Category 3: The opening category is suitable for wind tunnel tests or numerical simulations. In this case, there are complex fluid dynamics phenomena at the openings on the building walls or roofs, including flow separation, multi-path airflow, local vortices, sudden increases in wind speed, etc. In this case, a direct simulation method based on wind tunnel tests or numerical simulations is selected to determine the internal pressure of the building.

[0031] Furthermore, in step S13, the regular layout building complex reasonably selects the internal and external pressure determination method based on the three types of opening categories under unknown working conditions. For the regular layout building complex in the case of unknown internal pressure or unknown external pressure without sufficient wind tunnel test data (unknown working conditions), the opening type is determined according to the working condition parameter range divided by S12 according to the working condition parameters, and a reasonable internal and external pressure determination method is selected. In the wind disaster assessment, it can avoid blindly using the unsteady Bernoulli equation to calculate the internal and external pressure of the building, resulting in inaccurate disaster loss assessment, and it can also avoid blindly using direct simulation methods such as wind tunnel tests or numerical simulations to determine the internal pressure of the building, resulting in waste of resources.

[0032] Furthermore, in step S13, the building internal and external pressure data of opening category 1 and opening category 2 of the regular layout building group under unknown working conditions are obtained based on analytical calculation, and the building internal and external pressure data of opening category 3 are obtained based on direct simulation. The building internal and external pressure data of the regular layout low-rise building group under unknown working conditions and the building internal and external pressure data under known working conditions established in S11 together constitute the training database and test database of machine learning.

[0033] Furthermore, in step S2, based on machine learning, a prediction model for internal and external wind pressures of low-rise buildings that is not currently studied and that takes into account both interference and the effects of openings under the working conditions of arbitrary building complexes is constructed; the machine learning prediction model is trained on the basis of the internal and external wind pressure database of low-rise buildings under the working conditions of regular layout building complexes that takes into account both the interference effects of building complexes and the effects of building openings obtained in S11 and S13. The training database is unique and has a unique target, and can quickly and accurately predict the internal and external wind pressures of low-rise buildings with openings under arbitrary working conditions, greatly improving the efficiency of regional wind disaster assessment.

[0034] Furthermore, in step S2, the basic parameters of the regular layout building complex in steps S11 and S13 are used as the machine learning training input layer, and the mean and standard deviation of the internal and external wind pressure coefficients of the regular layout building complex obtained in steps S11 and S13 are used as the machine learning training output layer. The number of hidden layers L and the number of neurons K on each hidden layer of the neural network proxy model based on machine learning are optimized and iterated according to the following method: First, the number of hidden layers l and the number of neurons k of the neural network are assigned to 1. At this time, the basic database of internal and external wind pressure coefficients of open low-rise buildings in the training samples is sampled and trained in the neural network proxy model to obtain a temporary neural network. The internal and external pressures of the open building predicted by the temporary neural network are combined with the actual internal and external pressures of the open building to calculate the goodness of fit R 2 If the goodness of fit R is not met 2 >0.95, the number of hidden layers l and the number of neurons k of the neural network are assigned to 2 for iteration, and this process is repeated until the goodness of fit R is satisfied. 2 When the number of hidden layers l and the number of neurons k reach the limit values ​​L and K, the maximum goodness of fit R is output. 2 The corresponding hidden layer number l and neuron number k, the wind pressure prediction model for open low-rise buildings is obtained. The layer number limit L = 5 and the neuron number limit K = 2 / 3 (m + n), n and m are the number of neurons in the input layer and output layer respectively, and the goodness of fit R 2 =0.95, ensuring that the neural network prediction model has a certain accuracy without overfitting.

[0035] Goodness of fit R 2 The numerical characteristics that represent the relationship between the predicted variable and the input sample, and the statistical indicators that reflect the reliability of the predicted variable, can be calculated as follows:

[0036]

[0037] Where y is the sample value of the internal and external wind pressure coefficient of the wind tunnel test of low-rise buildings with openings, is the average value of the input wind pressure coefficient sample, It is the predicted value of the internal and external wind pressure coefficient of the low-rise building with openings predicted by the temporary neural network. 2 The closer the value is to 1, the better the predicted value fits the sample value.

[0038] Furthermore, in step S11, a database of internal and external wind pressures of low-rise building complexes with regular layouts taking into account interference and the influence of openings is obtained through wind tunnel tests for wind disaster assessment. In actual projects, for unknown internal pressures or unknown external pressures of buildings with regular layouts without sufficient wind tunnel test data, the type of opening is determined according to the operating parameters and the range of operating parameters divided by S12, and further according to the internal pressure determination method of step S13, a reasonable method for determining the internal pressure of low-rise building complexes is quickly selected to avoid inaccurate calculations caused by blindly using the unsteady Bernoulli equation and waste of resources caused by blindly conducting wind tunnel tests. For the situation where the internal and external pressures of buildings are unknown under the working conditions of arbitrary building complexes, the machine learning prediction model established by S2 can quickly predict the internal and external pressures of low-rise building complexes with openings under arbitrary layout conditions, thereby improving the efficiency of wind disaster assessment.

[0039] The beneficial effects of the present invention are:

[0040] 1. In the method for predicting the internal and external wind pressure of low-rise buildings with openings disclosed in the present invention, first, based on wind tunnel tests, a database of internal and external wind pressure of low-rise buildings under the working conditions of regular layout buildings that considers both the interference effect of group buildings and the influence of openings is established; secondly, based on the database, low-rise buildings with openings are divided into three types of openings and the corresponding working condition parameter ranges are determined; when there is no sufficient wind tunnel test data for regular layout buildings in actual projects, the opening category can be judged according to the working condition parameters and a reasonable method for determining the internal and external pressures can be selected; finally, based on the internal and external wind pressure data of regular layout buildings under working conditions that have been obtained, machine learning training is performed to obtain a prediction model for the internal and external wind pressure of low-rise buildings that considers interference and the influence of openings under working conditions of arbitrary layout buildings, which is not currently studied. The method for predicting the internal and external wind pressure of low-rise buildings with openings proposed by the present invention can realize the rapid and accurate determination of the internal and external wind pressure distribution law of each building in the low-rise building group under arbitrary building parameters and opening parameters. The problem of being unable to predict the internal and external wind pressure of low-rise buildings with openings under the combined influence of wind disturbance and openings is solved; the problem of time-consuming and labor-intensive direct simulation methods such as wind tunnel tests of buildings with openings on the surface is solved for arbitrary building parameters and opening parameters. The method for predicting the internal and external wind pressure of low-rise buildings with openings proposed in the present invention can greatly improve the efficiency of regional wind disaster assessment.

[0041] 2. In the method for predicting the internal and external wind pressures of low-rise buildings with openings disclosed in the present invention, the established method for classifying opening categories can quickly determine the opening categories of low-rise buildings with regular layouts, and select a reasonable method for determining internal and external pressures, thereby improving the efficiency of wind disaster assessment; the classification of opening categories and the selection of corresponding methods for determining internal and external pressures solve the problem that when the data in the wind tunnel test database is insufficient, researchers and designers blindly use wind tunnel tests to determine the internal pressure of buildings, which is time-consuming and labor-intensive; it also solves the problem that researchers and designers blindly use the unsteady Bernoulli equation to determine the internal pressure of buildings, resulting in inaccurate wind disaster assessment.

[0042] 3. In the method for predicting the internal and external wind pressures of low-rise building complexes with openings disclosed in the present invention, the internal and external wind pressure database of low-rise building complexes under the layout rule building complex working conditions that takes into account the interference effects of the building complexes and the effects of the building openings is unique and fills the gap in the current research on the internal and external wind pressure data of low-rise building complexes with openings in the field of wind disaster assessment, and meets the urgent needs in wind disaster assessment.

[0043] Other advantages, objectives and features of the present invention will be described in the following description to some extent, and to some extent, will be obvious to those skilled in the art based on the following examination and study, or can be taught from the practice of the present invention. The objectives and other advantages of the present invention can be realized and obtained through the following description. BRIEF DESCRIPTION OF THE DRAWINGS

[0044] In order to make the purpose, technical solutions and advantages of the present invention more clear, the present invention will be described in detail below in conjunction with the accompanying drawings, wherein:

[0045] Figure 1 It is a structural schematic diagram of the opening types of the wall and the roof being gradually destroyed in the prior art;

[0046] Figure 2 It is a flow chart of the method for predicting wind-induced internal and external pressures of low-rise buildings with openings according to the present invention;

[0047] Figure 3 It is a wind tunnel test arrangement diagram of a low-rise building with holes in the method for predicting wind-induced internal and external pressures of a low-rise building with holes of the present invention;

[0048] Figure 4 A diagram showing a method for classifying opening categories in a method for predicting wind-induced internal and external pressures of low-rise buildings with openings according to the present invention;

[0049] Figure 5 It is an opening condition diagram suitable for analytical equations in the method for predicting wind-induced internal and external pressures of low-rise buildings with openings of the present invention;

[0050] Figure 6It is a network structure diagram of a neural network proxy model in the method for predicting wind-induced internal and external pressures of low-rise buildings with openings of the present invention;

[0051] Figure 7 This is a diagram of the optimization iteration method of the neural network proxy model in the method for predicting wind-induced internal and external pressures of low-rise buildings with openings of the present invention. DETAILED DESCRIPTION

[0052] The following describes the embodiments of the present invention by specific examples, and those skilled in the art can easily understand other advantages and effects of the present invention from the contents disclosed in this specification. The present invention can also be implemented or applied through other different specific embodiments, and the details in this specification can also be modified or changed in various ways based on different viewpoints and applications without departing from the spirit of the present invention. It should be noted that the illustrations provided in the following embodiments only illustrate the basic concept of the present invention in a schematic manner, and the following embodiments and features in the embodiments can be combined with each other without conflict.

[0053] like Figures 2 to 7 A method for predicting the wind pressure inside and outside a low-rise building with openings is shown, comprising the following steps:

[0054] S1. Method for determining internal and external pressure of low-rise buildings considering interference and opening effects under the condition of regular layout buildings;

[0055] S11. Construct a database of internal and external pressure wind tunnel tests on low-rise buildings with openings on the surface;

[0056] Wind tunnel tests of random combination conditions of low-rise buildings with holes were carried out to obtain a basic database of wind pressure coefficients on the outer surface of low-rise buildings with holes under different opening parameters (opening position, opening size, number of openings), different building layout parameters (number of buildings, building orientation, building density, roof slope), and different atmospheric parameters (wind speed, wind direction) considering interference and opening effects. In order to more reasonably determine the wind tunnel test conditions, hypercube sampling was performed on the basic parameters of the low-rise building complex with holes to reasonably reduce the test conditions; the wind tunnel test selected Class B landforms to simulate the atmospheric boundary layer, set volume compensation for the low-rise building model with holes, used high-speed electronic scanning valves to perform multi-point synchronous pressure measurement of the building, and corrected the pressure measurement pipeline. Based on the results of the wind tunnel pressure measurement test, a database of internal and external wind pressures of low-rise buildings under different building complex conditions was established by taking into account the layout rules of the interference effects of the building complex and the effects of the building openings. The designed wind tunnel test model is as follows: Figure 3 As shown;

[0057] S12. Classification of three types of openings and corresponding operating parameter ranges based on the wind tunnel test database;

[0058] Press Figure 4The opening classification method shown divides low-rise buildings with openings into three opening categories, namely, opening category 1: an opening category suitable for determining the internal pressure of a building using the unsteady Bernoulli equation, opening category 2: an opening category suitable for determining the internal pressure of a building using the multi-point average unsteady Bernoulli equation, and opening category 3: an opening category suitable for determining the internal pressure of a building using direct simulation methods such as wind tunnel tests or numerical simulations.

[0059] Figure 4 The specific iterative process of the opening classification process shown is as follows: first, the wind tunnel test condition i in S11 is assigned a value of 1, and the number of openings on the target building under this condition is determined. When there is only a single opening, the calculation result of the unsteady Bernoulli equation is compared with the wind tunnel test data. If the comparison result shows that the percentage error R is less than 5%, the condition is classified as opening category 1, which is suitable for the unsteady Bernoulli equation to determine the opening category of the building internal pressure and record the parameter conditions of the condition. If the comparison result shows that the percentage error R is greater than 5%, the condition is classified as opening category 3, and the wind tunnel test must be used to determine the opening category of the building internal pressure and record the parameter conditions of the condition. When the building When there are multiple openings on the building, the calculation results of the multi-point average unsteady Bernoulli equation are compared with the wind tunnel test data. If the comparison result shows that the percentage error R is less than 5%, the working condition is classified as opening category 2, and the multi-point average unsteady Bernoulli equation is applicable to determine the opening category of the building internal pressure and record the working condition parameter conditions. If the comparison result shows that the percentage error R is greater than 5%, the working condition is also classified as opening category 3, and the wind tunnel test must be used to determine the opening category of the building internal pressure and record the working condition parameter conditions. After the above comparison is completed for working condition 1, continue to assign it to working condition 2 for iteration until the opening categories and corresponding working condition parameter ranges of all working conditions I are output. The percentage error R calculation formula is:

[0060]

[0061] The theoretical value is the building internal pressure determined by the unsteady Bernoulli equation or the multi-point average unsteady Bernoulli equation, and the experimental value is the building internal pressure determined by the wind tunnel test;

[0062] Opening category 1: Opening category applicable to the unsteady Bernoulli equation, which is calculated as follows:

[0063]

[0064] Where X represents the oscillation displacement of the orifice gas column, P e Indicates the external pressure at the orifice, P i represents the internal pressure, ρ a Indicates air density, L e Indicates the effective length of the opening, C Lrepresents the loss coefficient, l0 represents the hole depth, A represents the opening area, C I Indicates the inertia coefficient, the opening condition is as follows Figure 5 (a)

[0065] Opening Category 2: Applicable to the opening category of multi-point average unsteady Bernoulli equation. In this case, the internal and external pressures at each opening are calculated according to the single-hole Bernoulli equation. Finally, the internal and external pressures of the entire building are determined by multi-point averaging. The multi-point average unsteady Bernoulli equation is calculated according to the following formula:

[0066]

[0067] Where L en and P en represent the effective length of the nth opening and the corresponding external pressure, X n It represents the oscillation displacement of the gas column at the nth orifice. The opening condition is as follows: Figure 5 (b)

[0068] Opening Category 3: The opening category is suitable for wind tunnel tests or numerical simulations. In this case, there are complex fluid dynamics phenomena at the openings on the building walls or roofs, including flow separation, multi-path airflow, local vortices, sudden increases in wind speed, etc. In this case, a direct simulation method based on wind tunnel tests or numerical simulations is selected to determine the internal pressure of the building.

[0069] S13. Rationally select the internal and external pressure determination method based on the three types of openings for buildings with regular layout under unknown working conditions;

[0070] For buildings with regular layouts and unknown internal pressure or unknown external pressure in the absence of sufficient wind tunnel test data (unknown operating conditions), the opening type is determined based on the operating condition parameters and the operating condition parameter range divided by S12, and a reasonable method for determining the internal and external pressures is selected. Opening Category 1 determines the internal and external pressures of the building by solving the unsteady Bernoulli equation; Opening Category 2 determines the internal and external pressures of the building by solving the multi-point average unsteady Bernoulli equation; Opening Category 3 determines the internal and external pressures of the building through wind tunnel tests or numerical simulations. Among them, opening category 1 and opening category 2 obtain the internal and external pressure data of the building under unknown operating conditions based on analytical calculations, and opening category 3 obtains the internal and external pressure data of the building under unknown operating conditions based on direct simulations;

[0071] S2. Prediction method of internal and external pressure of low-rise buildings considering interference and opening effects under the working condition of arbitrary building groups;

[0072] Based on the internal and external pressure data of the regularly laid out building complex obtained in S11 and S13, according to the neural network proxy model optimization iterative method proposed in the present invention, a unique internal and external wind pressure prediction model for low-rise building complexes that takes into account interference and opening effects under the working conditions of arbitrary building complexes is constructed, which is not currently studied; the basic parameters of the regularly laid out building complexes in steps S11 and S13 are used as the machine learning training input layer, and the mean and standard deviation of the internal and external wind pressure coefficients of the regularly laid out building complexes obtained in steps S11 and S13 are used as the machine learning training output layer, thereby training to obtain a prediction model for the internal and external wind pressure of low-rise building complexes with openings that can predict the working conditions of arbitrary building complexes and takes into account interference and opening effects at the same time.

[0073] The neural network agent model consists of an input layer, a hidden layer, and an output layer. The network structure is as follows: Figure 6 As shown. Data is input into the neural network through the forward transmission process, and the data passes through the hidden layer and the output layer in turn to obtain the error between the output value and the target value; then, the error is transmitted backward to adjust the weights and thresholds of the neural network, and this process is repeated to make the output result of the neural network constantly approach the true value. The number of neurons in the input layer n is equal to the number of basic parameters. The input layer data includes the coordinates of the center point of the opening position, the size of the opening, the number of openings, the number of buildings, the building orientation, the building density, the roof slope, the wind speed, the wind direction, and the coordinates of the measuring point. The mean and standard deviation of the wind pressure coefficients on the inner and outer surfaces of the low-rise building complex with openings are used as the output layer, so the number of neurons in the output layer m is 4. The present invention adopts Figure 7 The iterative method shown in FIG. 1 optimizes the number of hidden layers L and the number of neurons K in each hidden layer, and sets the limit of the number of layers L = 5 and the limit of the number of neurons K = 2 / 3 (m + n), as well as the goodness of fit R 2 =0.95, ensuring that the neural network prediction model has a certain accuracy without overfitting.

[0074] Goodness of fit R 2 The numerical characteristics that represent the relationship between the predicted variable and the input sample, and the statistical indicators that reflect the reliability of the predicted variable, can be calculated as follows:

[0075]

[0076] Where y is the sample value of the internal and external wind pressure coefficient of the wind tunnel test of low-rise buildings with openings, is the average value of the input wind pressure coefficient sample, It is the predicted value of the internal and external wind pressure coefficient of the low-rise building with openings predicted by the temporary neural network. 2 The closer the value is to 1, the better the predicted value fits the sample value.

[0077] The optimization iterative method for the number of hidden layers and neurons in each hidden layer of the deep learning neural network is as follows: first, the number of hidden layers l and the number of neurons k of the neural network are assigned to 1. At this time, the basic database of internal and external wind pressure coefficients of low-rise buildings with openings in the training samples is trained in the neural network proxy model to obtain a temporary neural network. The internal and external pressures of the open-hole buildings predicted by the temporary neural network are combined with the actual internal and external pressures of the open-hole buildings to calculate the goodness of fit R 2 If the goodness of fit R is not met 2 >0.95, the number of hidden layers l and the number of neurons k of the neural network are assigned to 2 for iteration, and this process is repeated until the goodness of fit R is met. 2 When the number of hidden layers l and the number of neurons k reach the limit values ​​L and K, the maximum goodness of fit R is output. 2 The corresponding number of hidden layers l and the number of neurons k are used to obtain the external wind pressure prediction model for low-rise buildings with openings.

[0078] The machine learning prediction model constructed in step S2 is trained on the basis of the internal and external wind pressure database of low-rise buildings with regular layouts established in S1 while taking into account the interference effects of building complexes and the effects of building openings. Its training database is unique and unique in terms of the target. It can quickly and accurately predict the internal and external wind pressures of low-rise buildings with openings under any working condition, greatly improving the efficiency of regional wind disaster assessment. In addition, the machine learning prediction model for the internal and external pressures of low-rise buildings under the working condition of any building complex established in step S2 is an extension of the method for determining the internal and external pressures of low-rise buildings under the working condition of regular building complexes in S1. This machine learning prediction model can be used to calibrate the accuracy of the calculation results of the internal and external wind pressures of low-rise buildings under the working condition of regular building complexes, and can also predict the distribution of internal and external wind pressures of low-rise buildings under unknown working conditions.

[0079] Finally, it should be noted that the above embodiments are only used to illustrate the technical solution of the present invention rather than to limit it. Although the present invention has been described in detail with reference to the preferred embodiments, those skilled in the art should understand that the technical solution of the present invention can be modified or replaced by equivalents without departing from the purpose and scope of the technical solution, which should be included in the scope of the claims of the present invention.

Claims

1. A method for predicting wind pressure inside and outside low-rise buildings with openings, characterized in that: The following steps are involved: S1. Method for determining internal and external pressure of low-rise buildings considering interference and opening effects under the condition of regular layout buildings; S11. Construct a database of internal and external pressure wind tunnel tests on low-rise buildings with openings on the surface; Based on the random combination wind tunnel test of low-rise buildings with opening layout rules, the internal and external pressure wind tunnel test database of low-rise buildings with different opening parameters and different building layout parameters is obtained, which takes into account both interference and opening effects. S12. Classification of three types of openings and corresponding operating parameter ranges based on the wind tunnel test database; According to the opening classification method proposed in this step, the building opening categories are classified, that is, based on the internal and external pressure wind tunnel test database of low-rise building complexes established in S11, the number of openings on a target building under a certain working condition is determined. When there is only a single opening, the calculation result of the unsteady Bernoulli equation is compared with the wind tunnel test data. If the comparison result shows that the percentage error is less than 5%, the opening category 1 is obtained, which is suitable for the opening category and corresponding working condition parameter range determined by the unsteady Bernoulli equation for the internal pressure of the building. If the comparison result shows that the percentage error is greater than 5%, the opening category 3 is obtained, which is suitable for using wind tunnel tests or data to determine the opening category and corresponding working condition parameter range. The opening category and the corresponding operating parameter range of the building internal pressure are determined by the numerical simulation method; when there are multiple openings on the building, the calculation results of the multi-point average unsteady Bernoulli equation are compared with the wind tunnel test data. If the comparison result shows that the percentage error is less than 5%, the opening category 2 is obtained, which is suitable for determining the opening category and the corresponding operating parameter range of the building internal pressure by the multi-point average unsteady Bernoulli equation. If the comparison result shows that the percentage error is greater than 5%, the opening category 3 is also obtained, which is suitable for determining the opening category and the corresponding operating parameter range of the building internal pressure by direct simulation methods such as wind tunnel tests or numerical simulations; S13. Rationally select the internal and external pressure determination method based on the three types of openings for buildings with regular layout under unknown working conditions; For buildings with regular layouts and unknown internal pressure or external pressure without sufficient wind tunnel test data, the opening type is determined according to the operating parameters and the operating parameter range divided in step S12, and a reasonable method for determining the internal and external pressures is selected; Opening Category 1 determines the internal and external pressures of the building by solving the unsteady Bernoulli equation; Opening Category 2 determines the internal and external pressures of the building by solving the multi-point averaged unsteady Bernoulli equation; Opening Category 3 determines the internal and external pressure of the building through wind tunnel tests or numerical simulations; among them, opening category 1 and opening category 2 obtain the internal and external pressure data of the building under unknown working conditions based on analytical calculations, and opening category 3 obtains the internal and external pressure data of the building under unknown working conditions based on direct simulations; S2. Prediction method of internal and external pressure of low-rise buildings considering interference and opening effects under the working condition of arbitrary building groups; Based on the internal and external pressure data of the regularly laid out building complex obtained in step S11 and step S13, a prediction model for the internal and external wind pressure of low-rise building complexes taking into account the interference and the influence of openings under the working conditions of any building complex is constructed according to the following neural network proxy model optimization iterative method; that is: the basic parameters of the regularly laid out building complexes in steps S11 and S13 are used as the machine learning training input layer, and the mean and standard deviation of the internal and external wind pressure coefficients of the regularly laid out building complexes obtained in steps S11 and S13 are used as the machine learning training output layer, thereby training to obtain a prediction model for the internal and external wind pressure of low-rise building complexes with openings that can predict the working conditions of any building complex and take into account the interference and the influence of openings.

2. The method for predicting the wind pressure inside and outside the low-rise buildings with openings according to claim 1, characterized in that: In step S11, based on the wind tunnel test, a database of internal and external wind pressures of low-rise buildings under the working conditions of regular layout buildings is established, which takes into account both the interference effects of the buildings and the effects of the building openings.

3. The method for predicting wind pressure inside and outside low-rise buildings with openings according to claim 1, characterized in that: In step S12, based on the internal and external pressure wind tunnel test database of low-rise building complex established in step S11, the low-rise building complex with openings is divided into three opening categories according to the opening category classification method and the corresponding operating condition parameter ranges are determined; the three opening categories are opening category 1: an opening category applicable to determining the internal pressure of a building by the unsteady Bernoulli equation; Opening category 2: Opening category applicable to the determination of building internal pressure by multi-point average unsteady Bernoulli equation; Opening Category 3: Opening category suitable for determining the internal pressure of a building by direct simulation methods such as wind tunnel testing or numerical simulation.

4. The method for predicting wind pressure inside and outside low-rise buildings with openings according to claim 3, characterized in that: The hole type 1 in step S12 is a hole type applicable to the unsteady Bernoulli equation. The unsteady Bernoulli equation is calculated according to the following formula: Where X represents the oscillation displacement of the orifice gas column, P e Indicates the external pressure at the orifice, P i represents the internal pressure, ρ a Indicates air density, L e Indicates the effective length of the opening, C L represents the loss coefficient, l0 represents the hole depth, A represents the opening area, C I represents the coefficient of inertia; Opening Category 2: Applicable to the opening category of multi-point average unsteady Bernoulli equation. In this case, the internal and external pressures at each opening are calculated according to the single-hole Bernoulli equation. Finally, the internal and external pressures of the entire building are determined by multi-point averaging. The multi-point average unsteady Bernoulli equation is calculated according to the following formula: Where L en and P en represent the effective length of the nth opening and the corresponding external pressure, X n represents the oscillation displacement of the gas column at the nth orifice; Opening Category 3: The opening category is suitable for wind tunnel tests or numerical simulations. In this case, there are complex fluid dynamics phenomena such as flow separation, multi-path airflow, local vortices, and sudden increases in wind speed at the openings on the building walls or roofs. A direct simulation method based on wind tunnel tests or numerical simulations is selected to determine the internal pressure of the building.

5. The method for predicting wind pressure inside and outside low-rise buildings with openings according to claim 1, characterized in that: In step S13, for buildings with regular layouts, a reasonable method for determining internal and external pressures is selected based on the three types of opening categories under unknown working conditions; for buildings with regular layouts, in the absence of sufficient wind tunnel test data and in the case of unknown internal pressure or unknown external pressure, the type of opening is determined based on the working condition parameters and the range of working condition parameters divided in step S12, and a reasonable method for determining internal and external pressures is selected.

6. The method for predicting the wind pressure inside and outside the low-rise buildings with openings according to claim 5, characterized in that: In step S13, the opening category 1 and the opening category 2 of the regular layout building complex under unknown working conditions obtain the building internal and external pressure data based on analytical calculation, and the opening category 3 obtains the building internal and external pressure data based on direct simulation; the building internal and external pressure data of the regular layout low-rise building complex under unknown working conditions in step S13 and the building internal and external pressure data under known working conditions established in step S11 together constitute the training database and test database for machine learning.

7. The method for predicting wind pressure inside and outside low-rise buildings with openings according to claim 1, characterized in that: In step S2, based on machine learning, a prediction model for the internal and external wind pressure of low-rise buildings is constructed, which takes into account both interference and opening effects under the working conditions of arbitrary building complexes. The machine learning prediction model is trained on the basis of the internal and external wind pressure database of low-rise buildings under the working conditions of regular layout building complexes that takes into account both the interference effects of building complexes and the effects of building openings obtained in steps S11 and S13.

8. The method for predicting the wind pressure inside and outside the low-rise buildings with openings according to claim 7, characterized in that: In step S2, the number of hidden layers L and the number of neurons K on each hidden layer of the neural network proxy model based on machine learning are optimized and iterated according to the following method: first, the number of hidden layers l and the number of neurons k of the neural network are assigned to 1. At this time, the basic database of internal and external wind pressure coefficients of low-rise buildings with openings in the training samples is trained in the neural network proxy model to obtain a temporary neural network. The internal and external pressures of the open-hole buildings predicted by the temporary neural network are combined with the actual internal and external pressures of the open-hole buildings to calculate the goodness of fit R 2 If the goodness of fit R is not met 2 >0.95, the number of hidden layers l and the number of neurons k of the neural network are assigned to 2 for iteration, and this process is repeated until the goodness of fit R is satisfied. 2 When the number of hidden layers l and the number of neurons k is greater than 0.95, the iteration is stopped and the corresponding hidden layer number l and the number of neurons k are output to obtain the prediction model of internal and external wind pressure of open low-rise buildings; if the number of hidden layers l and the number of neurons k iteratively reach the limit values ​​L and K, the maximum goodness of fit R is output. 2 The corresponding hidden layer number l and the number of neurons k are used to obtain the prediction model of internal and external wind pressure of open low-rise buildings; the layer number limit L = 5 and the neuron number limit K = 2 / 3 (m + n), n and m are the number of neurons in the input layer and output layer respectively, and the goodness of fit R 2 =0.95, ensuring that the neural network prediction model has a certain degree of accuracy without overfitting; Goodness of fit R 2 The numerical characteristics that represent the relationship between the predicted variable and the input sample, and the statistical indicators that reflect the reliability of the predicted variable, can be calculated as follows: Where y is the sample value of the internal and external wind pressure coefficient of the wind tunnel test of low-rise buildings with openings, is the average value of the input wind pressure coefficient sample, is the predicted value of the internal and external wind pressure coefficient of the low-rise building complex with openings predicted by the temporary neural network; R 2 The closer the value is to 1, the better the predicted value fits the sample value.

9. The method for predicting wind pressure inside and outside low-rise buildings with openings according to claim 1, characterized in that: In step S11, a database of internal and external wind pressures of low-rise buildings with regular layouts that takes into account interference and the influence of openings is obtained through wind tunnel tests for wind disaster assessment. In actual projects, for buildings with regular layouts that do not have sufficient wind tunnel test data for unknown internal pressure or unknown external pressure, the opening type is determined based on the operating parameters and the operating parameter range divided in S12, and further based on the internal pressure determination method in step S13, a reasonable method for determining the internal pressure of low-rise buildings is quickly selected. For buildings with arbitrary layouts and unknown internal and external pressures, the machine learning prediction model established in S2 can quickly predict the internal and external pressures of low-rise buildings with openings under arbitrary layout conditions.

Images