1. A city waterlogging ponding process prediction method based on machine learning is characterized by comprising the following steps:

s1, establishing an interpolation topological relation between a ponding monitoring site and an approaching rainfall monitoring site by using an improved K neighborhood method;

s2, acquiring historical ponding process data of a ponding monitoring site and historical rainfall process data of a rainfall monitoring site nearby, obtaining historical rainfall process data of the ponding monitoring site by adopting an inverse distance weight interpolation method according to the interpolation topological relation established in the step S1, and constructing a city waterlogging ponding process prediction model related to the ponding process and the rainfall process;

s3, acquiring real-time rainfall process data of a nearby rainfall monitoring site, and acquiring the real-time rainfall process data of the ponding monitoring site by adopting an inverse distance weight interpolation method according to the interpolation topological relation established in the step S1; fusing rainfall process data of each product according to the time-space characteristics of the multi-source quantitative rainfall forecast data products to obtain quantitative rainfall forecast process data of the accumulated water monitoring site;

s4, splicing the real-time rainfall process data of the ponding monitoring site with the quantitative rainfall forecasting process data to obtain the real-time rainfall forecasting process data of the ponding monitoring site;

and S5, forecasting the water accumulation process of the urban waterlogging water accumulation point according to the real-time and forecast rainfall process data of the water accumulation monitoring station by utilizing the urban waterlogging water accumulation point water accumulation process forecasting model.

2. The method for predicting urban waterlogging ponding process based on machine learning of claim 1, wherein the step S1 specifically includes the following sub-steps:

s11, judging whether a rainfall monitoring station exists at the same time at the position of the ponding monitoring station; if yes, forming a point set of the area close to the rainfall monitoring site for the point, and entering the step S17; otherwise, go to step S12;

s12, acquiring longitude and latitude coordinates of a ponding monitoring site and all rainfall monitoring sites in an urban area;

s13, constructing a two-dimensional coordinate system by taking the ponding monitoring sites as the origin of the coordinate system, and calculating the relative coordinates of each rainfall monitoring site;

s14, calculating a model of a vector from the original point to each rainfall monitoring site on a coordinate axis, and selecting a point corresponding to the minimum model as an adjacent rainfall monitoring site;

s15, selecting points meeting set screening conditions in each quadrant as nearby rainfall monitoring stations;

s16, forming a field point set of the nearby rainfall monitoring sites selected in the steps S14 and S15;

s17, establishing an interpolation topological relation between the ponding monitoring site and the point set in the area of the rainfall approaching monitoring site.

3. The method for predicting urban waterlogging ponding process based on machine learning of claim 2, wherein the step S15 specifically includes the following sub-steps:

s151, judging whether a rainfall monitoring station does not exist in the current quadrant; if yes, entering the next quadrant; otherwise, go to step S152;

s152, judging whether only one rainfall monitoring station exists in the current quadrant; if so, selecting the rainfall monitoring site; otherwise, go to step S153;

s153, calculating a model from an original point to a rainfall monitoring site vector; calculating the difference value between all the modes and the minimum mode, and judging whether the number of the stations of which the difference value is not more than a preset threshold value is 1 or not; if so, selecting a rainfall monitoring station with the smallest module; otherwise, go to step S154;

s154, calculating an included angle which is smaller than 90 degrees and is formed between the relative coordinate of each rainfall monitoring station meeting the preset threshold condition and a coordinate axis along the anticlockwise direction, and sorting according to the size;

s155, judging whether the minimum included angle is larger than or equal to 45 degrees; if so, selecting a rainfall monitoring station corresponding to the included angle; otherwise, go to step S156;

s156, judging whether the maximum included angle is smaller than or equal to 45 degrees; if so, selecting a rainfall monitoring station corresponding to the included angle; otherwise, go to step S157;

s157, calculating the difference value between the maximum included angle and the minimum included angle, and judging whether the difference value is larger than or equal to 30 degrees; if so, selecting a rainfall monitoring station corresponding to the maximum included angle and the minimum included angle; otherwise, selecting a rainfall monitoring site corresponding to the minimum value of the difference value between each included angle and 45 degrees.

4. The method for predicting urban waterlogging ponding process based on machine learning of claim 1, wherein the step S2 specifically includes the following sub-steps:

s21, acquiring historical ponding process data of each field of ponding monitoring site and historical rainfall process data of a nearby rainfall monitoring site;

s22, utilizing the historical rainfall process data of the nearby rainfall monitoring site, and utilizing an inverse distance weight interpolation method to interpolate the historical rainfall process data of the nearby rainfall monitoring site according to the interpolation topological relation established in the step S1, so as to generate the historical rainfall process data of the ponding monitoring site;

s23, traversing historical ponding process data and historical rainfall process data of the ponding monitoring site, and calculating the correlation between the ponding process data and the sliding rainfall process data;

s24, selecting a sliding time period with the maximum correlation corresponding to the maximum correlation, and establishing a sliding rainfall process and an accumulated water process data set of historical rainfall accumulated water of each occasion;

s25, performing machine learning training and verification by using the established data set, and constructing a city waterlogging ponding process prediction model related to a ponding process and a rainfall process.

5. The method for predicting urban waterlogging ponding process based on machine learning of claim 1, wherein the step S3 specifically includes the following sub-steps:

s31, acquiring real-time rainfall process data of a rainfall monitoring site;

s32, interpolating the real-time rainfall process data close to the rainfall monitoring site by using an inverse distance weight interpolation method according to the interpolation topological relation established in the step S1 to generate the real-time rainfall process data of the ponding monitoring site;

s33, obtaining various weather quantitative rainfall forecast products at the positions of the ponding monitoring sites, and selecting the weather quantitative rainfall forecast product with the highest time-space precision to interpolate according to the time-space precision of different products to obtain quantitative rainfall forecast process data of the ponding monitoring sites.

6. The method for predicting urban waterlogging ponding process based on machine learning of claim 5, wherein the step S33 specifically comprises the following substeps:

s331, carrying out spatial superposition on the ponding monitoring station and a weather quantitative forecasting grid to obtain a grid number corresponding to the ponding monitoring station;

s332, reading rainfall process data of various weather quantitative forecast products of the grid numbers;

s333, selecting rainfall process data with highest time-space precision from the rainfall process data read in the step S332;

and S334, interpolating the rainfall process data selected in the step S333 into rainfall forecasting process data at equal intervals by adopting an average distribution method or a constructed arithmetic progression method.

7. The utility model provides a city waterlogging ponding point ponding process prediction system based on machine learning which characterized in that includes:

the interpolation topological relation building module is used for building an interpolation topological relation between a ponding monitoring site and an approaching rainfall monitoring site by utilizing an improved K neighborhood method;

the data acquisition module is used for acquiring water accumulation process data of a water accumulation monitoring site and rainfall process data of a nearby rainfall monitoring site;

the data interpolation module is used for generating rainfall process data of the ponding monitoring station by utilizing the rainfall process data of the nearby rainfall monitoring station and utilizing an inverse distance weight interpolation method according to the established interpolation topological relation;

the data fusion module is used for fusing multi-source quantitative rainfall forecast data products to obtain quantitative rainfall forecast process data of the accumulated water monitoring station;

the data splicing module is used for splicing the quantitative rainfall forecasting process data and the real-time rainfall process data to obtain the real-time rainfall forecasting process data of the accumulated water monitoring site;

the model construction module is used for constructing a city waterlogging accumulated water spot accumulated water process prediction model related to the rainfall process according to the historical accumulated water process data of the accumulated water monitoring site and the historical rainfall process data of the nearby rainfall monitoring site;

and the ponding prediction module is used for predicting the ponding process of the urban waterlogging ponding point based on real-time and forecast rainfall process data of the ponding monitoring station by using the constructed urban waterlogging ponding process prediction model.

8. An urban waterlogging ponding process prediction device based on machine learning, comprising:

a memory for storing a computer program;

a processor for implementing the steps of the machine learning based urban waterlogging ponding process prediction method of any one of claims 1 to 6 when executing the computer program.

9. A computer-readable storage medium, having stored thereon a computer program which, when executed by a processor, implements the steps of the machine learning-based urban waterlogging ponding process prediction method of any one of claims 1 to 6.

Background

Along with the rapid development of urbanization, the urban impervious area is increased, the urban hydrological effect is changed, and the urban rainstorm waterlogging problem begins to be highlighted due to the reasons of insufficient urban drainage facility capacity and the like. In recent years, the phenomenon of 'city seeing sea' in China is frequent, a series of problems such as urban water supply, power supply, communication and the like, traffic paralysis, flooding of business areas, production areas and residential areas and the like are caused, the life and production of residents are influenced and interfered, and the property of people is greatly lost. Urban rainstorm and waterlogging become important factors threatening urban safety and interfering urban operation.

The reason why urban waterlogging frequently occurs in recent years is mainly as follows through comprehensive analysis: the global warming causes frequent extreme weather, and the frequency and intensity of extreme rainfall increase; the accelerated urbanization process increases the hardening rate of the underlying surface, which leads to the increase of surface runoff; the drainage capacity of the urban drainage system is insufficient, the rainfall cannot be discharged in time, and accumulated water is easily generated in low-lying areas.

At present, the urban inland inundation prediction generally adopts a method of combining weather forecast, using numerical model simulation or carrying out speculation according to experience. The method for predicting based on the numerical model is limited by the reasons of insufficient grasping of basic data of the city, low efficiency of model construction and operation, capability of professionals and the like, so that the urban inland inundation is difficult to predict accurately and efficiently; the method for predicting urban inland inundation based on experience is more limited by the knowledge and experience reserve of professionals, and the accuracy of prediction results is different.

Disclosure of Invention

Aiming at the defects in the prior art, the invention provides a method for predicting the urban waterlogging and water accumulation point water accumulation process based on machine learning.

In order to achieve the purpose of the invention, the invention adopts the technical scheme that:

in a first aspect, the invention provides a method for predicting an urban inland inundation water accumulation point water accumulation process based on machine learning, which comprises the following steps:

s1, establishing an interpolation topological relation between a ponding monitoring site and an approaching rainfall monitoring site by using an improved K neighborhood method;

s2, acquiring historical ponding process data of a ponding monitoring site and historical rainfall process data of a rainfall monitoring site nearby, obtaining historical rainfall process data of the ponding monitoring site by adopting an inverse distance weight interpolation method according to the interpolation topological relation established in the step S1, and constructing a city waterlogging ponding process prediction model related to the ponding process and the rainfall process;

s3, acquiring real-time rainfall process data of a nearby rainfall monitoring site, and acquiring the real-time rainfall process data of the ponding monitoring site by adopting an inverse distance weight interpolation method according to the interpolation topological relation established in the step S1; fusing rainfall process data of each product according to the time-space characteristics of the multi-source quantitative rainfall forecast data products to obtain quantitative rainfall forecast process data of the accumulated water monitoring site;

s4, splicing the real-time rainfall process data of the ponding monitoring site with the quantitative rainfall forecasting process data to obtain the real-time rainfall forecasting process data of the ponding monitoring site;

and S5, forecasting the water accumulation process of the urban waterlogging water accumulation point according to the real-time and forecast rainfall process data of the water accumulation monitoring station by utilizing the urban waterlogging water accumulation point water accumulation process forecasting model.

Further, the step S1 specifically includes the following sub-steps:

s11, judging whether a rainfall monitoring station exists at the same time at the position of the ponding monitoring station; if yes, forming a point set of the area close to the rainfall monitoring site for the point, and entering the step S17; otherwise, go to step S12;

s12, acquiring longitude and latitude coordinates of a ponding monitoring site and all rainfall monitoring sites in an urban area;

s13, constructing a two-dimensional coordinate system by taking the ponding monitoring sites as the origin of the coordinate system, and calculating the relative coordinates of each rainfall monitoring site;

s14, calculating a model from the original point to each rainfall monitoring site vector on a coordinate axis, and selecting the minimum point of the model on each coordinate axis as an adjacent rainfall monitoring site;

s15, selecting points meeting set screening conditions in each quadrant as nearby rainfall monitoring stations;

s16, forming the nearby rainfall monitoring sites selected in the steps S14 and S15 into a nearby rainfall monitoring site field point set

S17, establishing an interpolation topological relation between the ponding monitoring site and the point set in the area of the rainfall approaching monitoring site.

Further, the step S15 specifically includes the following sub-steps:

s151, judging whether a rainfall monitoring station does not exist in the current quadrant; if yes, entering the next quadrant; otherwise, go to step S152;

s152, judging whether only one rainfall monitoring station exists in the current quadrant; if so, selecting the rainfall monitoring site; otherwise, go to step S153;

s153, calculating a model from an original point to a rainfall monitoring site vector; calculating the difference value between all the modes and the minimum mode, and judging whether the number of the stations of which the difference value is not more than a preset threshold value is 1 or not; if so, selecting a rainfall monitoring station with the smallest module; otherwise, go to step S154;

s154, calculating an included angle which is smaller than 90 degrees and is formed between the relative coordinate of each rainfall monitoring station meeting the preset threshold condition and a coordinate axis along the anticlockwise direction, and sorting according to the size;

s155, judging whether the minimum included angle is larger than or equal to 45 degrees; if so, selecting a rainfall monitoring station corresponding to the included angle; otherwise, go to step S156;

s156, judging whether the maximum included angle is smaller than or equal to 45 degrees; if so, selecting a rainfall monitoring station corresponding to the included angle; otherwise, go to step S157;

s157, calculating the difference value between the maximum included angle and the minimum included angle, and judging whether the difference value is larger than or equal to 30 degrees; if so, selecting a rainfall monitoring station corresponding to the maximum included angle and the minimum included angle; otherwise, selecting a rainfall monitoring site corresponding to the minimum value of the difference value between each included angle and 45 degrees.

Further, the step S2 specifically includes the following sub-steps:

s21, acquiring historical ponding process data of each field of ponding monitoring site and historical rainfall process data of a nearby rainfall monitoring site;

s22, utilizing the historical rainfall process data of the nearby rainfall monitoring site, and utilizing an inverse distance weight interpolation method to interpolate the historical rainfall process data of the nearby rainfall monitoring site according to the interpolation topological relation established in the step S1, so as to generate the historical rainfall process data of the ponding monitoring site;

s23, traversing historical ponding process data and historical rainfall process data of the ponding monitoring site, and calculating the correlation between the ponding process data and the sliding rainfall process data;

s24, selecting a sliding time period with the maximum correlation corresponding to the maximum correlation, and establishing a sliding rainfall process and an accumulated water process data set of historical rainfall accumulated water of each occasion;

s25, performing machine learning training and verification by using the established data set, and constructing a city waterlogging ponding process prediction model related to a ponding process and a rainfall process.

Further, the step S3 specifically includes the following sub-steps:

s31, acquiring real-time rainfall process data of a rainfall monitoring site;

s32, interpolating the real-time rainfall process data close to the rainfall monitoring site by using an inverse distance weight interpolation method according to the interpolation topological relation established in the step S1 to generate the real-time rainfall process data of the ponding monitoring site;

s33, obtaining various weather quantitative rainfall forecast products at the positions of the ponding monitoring sites, and selecting the weather quantitative rainfall forecast product with the highest time-space precision to interpolate according to the time-space precision of different products to obtain quantitative rainfall forecast process data of the ponding monitoring sites.

Further, the step S33 specifically includes the following sub-steps:

s331, carrying out spatial superposition on the ponding monitoring station and a weather quantitative forecasting grid to obtain a grid number corresponding to the ponding monitoring station;

s332, reading rainfall process data of various weather quantitative forecast products of the grid numbers;

s333, selecting rainfall process data with highest time-space precision from the rainfall process data read in the step S332;

and S334, interpolating the rainfall process data selected in the step S333 into rainfall forecasting process data at equal intervals by adopting an average distribution method or a constructed arithmetic progression method.

In a second aspect, the present invention further provides a system for predicting a water spot-ponding process of urban waterlogging based on machine learning, including:

the interpolation topological relation building module is used for building an interpolation topological relation between a ponding monitoring site and an approaching rainfall monitoring site by utilizing an improved K neighborhood method;

the data acquisition module is used for acquiring water accumulation process data of a water accumulation monitoring site and rainfall process data of a nearby rainfall monitoring site;

the data interpolation module is used for generating rainfall process data of the ponding monitoring station by utilizing the rainfall process data of the nearby rainfall monitoring station and utilizing an inverse distance weight interpolation method according to the established interpolation topological relation;

the data fusion module is used for fusing multi-source quantitative rainfall forecast data products to obtain quantitative rainfall forecast process data of the accumulated water monitoring station;

the data splicing module is used for splicing the quantitative rainfall forecasting process data and the real-time rainfall process data to obtain the real-time rainfall forecasting process data of the accumulated water monitoring site;

the model construction module is used for constructing a city waterlogging accumulated water spot accumulated water process prediction model related to the rainfall process according to the historical accumulated water process data of the accumulated water monitoring site and the historical rainfall process data of the nearby rainfall monitoring site;

and the ponding prediction module is used for predicting the ponding process of the urban waterlogging ponding point based on real-time and forecast rainfall process data of the ponding monitoring station by using the constructed urban waterlogging ponding process prediction model.

In a third aspect, the present invention further provides a device for predicting an urban inland inundation water-accumulation process based on machine learning, including:

a memory for storing a computer program;

a processor for implementing the steps of the method for predicting a water accumulation process of urban waterlogging water accumulation point based on machine learning as described above when executing the computer program.

In a fourth aspect, the present invention further provides a computer-readable storage medium, on which a computer program is stored, where the computer program, when executed by a processor, implements the steps of the above-mentioned urban waterlogging ponding process prediction method based on machine learning.

The invention has the following beneficial effects:

the method comprises the steps of establishing an interpolation topological relation between a ponding monitoring site and an adjacent rainfall monitoring site by using an improved K neighborhood method, and constructing an urban inland waterlogging ponding process prediction model related to a rainfall process according to acquired historical ponding process data of the ponding monitoring site and historical rainfall process data of the ponding monitoring site obtained through interpolation; and splicing the real-time rainfall process data of the ponding monitoring site with the quantitative rainfall forecasting process data to obtain the real-time and rainfall forecasting process data of the ponding monitoring site, and predicting the ponding process of the urban inland waterlogging ponding point by using the constructed urban inland waterlogging ponding point ponding process prediction model. The method has the advantages of simple operation, high modeling speed, high calculation efficiency, high prediction precision and the like, can realize accurate and effective ponding prediction, and is favorable for emergency prevention and treatment of urban waterlogging.

Drawings

Fig. 1 is a flowchart of a method for predicting an urban inland inundation water accumulation point water accumulation process based on machine learning according to an embodiment of the present invention;

fig. 2 is a structural diagram of a prediction system for a water spot-ponding process of urban waterlogging based on machine learning according to an embodiment of the present invention;

fig. 3 is a structural diagram of a prediction device for urban inland inundation water accumulation process based on machine learning according to an embodiment of the present invention.

Detailed Description

The following description of the embodiments of the present invention is provided to facilitate the understanding of the present invention by those skilled in the art, but it should be understood that the present invention is not limited to the scope of the embodiments, and it will be apparent to those skilled in the art that various changes may be made without departing from the spirit and scope of the invention as defined and defined in the appended claims, and all matters produced by the invention using the inventive concept are protected.

As shown in fig. 1, a method for predicting a water accumulation process of urban inland inundation and waterlogging based on machine learning according to an embodiment of the present invention includes the following steps S1 to S5:

s1, establishing an interpolation topological relation between a ponding monitoring site and an approaching rainfall monitoring site by using an improved K neighborhood method;

how to construct an interpolation topological relation between an adjacent rainfall monitoring site and an accumulated water monitoring site based on a spatial position improved K neighborhood method is explained below.

Step S1 specifically includes the following substeps:

s11, judging whether a rainfall monitoring station exists at the same time at the position of the ponding monitoring station; if yes, directly establishing an interpolation topological relation between the accumulated water monitoring station and the rainfall monitoring station; otherwise, go to step S12;

s12, acquiring longitude and latitude coordinates of a ponding monitoring site and all rainfall monitoring sites in an urban area;

in this step, a ponding monitoring site S is obtained₀And all rainfall monitoring sites S (S)₁,S₂,S₃,…,S_j) Latitude and longitude coordinates (x)₀,y₀) And [ (x)₁,y₁),(x₂,y₂),(x₃,y₃),…,(x_j,y_j)]；

S13, constructing a two-dimensional coordinate system by taking the ponding monitoring sites as the origin of the coordinate system, and calculating the relative coordinates of each adjacent rainfall monitoring site

In this step, the station S is monitored by accumulated water₀Constructing a two-dimensional coordinate system as the origin of the coordinate system, and calculating the relative coordinates [ (x) of all rainfall monitoring stations₁-x₀,y₁-y₀),(x₂-x₀,y₂-y₀),(x₃-x₀,y₃-y₀),…,(x_j-x₀,y_j-y₀)]

S14, calculating a model from the original point to each rainfall monitoring site vector on a coordinate axis, and selecting the rainfall monitoring site with the smallest model on each coordinate axis;

in this step, the invention automatically selects the relative coordinate satisfying the condition (x)_n*y_n0| n ∈ (1,2,3, …, j)) of nearby rainfall monitoring sites, constituting a point set S_Z(ii) a If 2 or more points exist on each coordinate axis, only the nearest rainfall monitoring station with the smallest modulus is selected on each coordinate axis.

S15, selecting rainfall monitoring stations meeting the set screening conditions in each quadrant, and specifically comprising the following steps:

s151, judging whether a rainfall monitoring station does not exist in the current quadrant; if yes, entering the next quadrant; otherwise, go to step S152;

s152, judging whether only one rainfall monitoring station exists in the current quadrant; if so, selecting the rainfall monitoring site; otherwise, go to step S153;

s153, calculating a model from an original point to a rainfall monitoring site vector; calculating the difference value between all the modes and the minimum mode, and judging whether the number of the stations of which the difference value is not more than a preset threshold value is 1 or not; if so, selecting a rainfall monitoring station with the smallest module; otherwise, go to step S154;

s154, calculating an included angle which is smaller than 90 degrees and is formed between the relative coordinate of each rainfall monitoring station meeting the preset threshold condition and a coordinate axis along the anticlockwise direction, and sorting according to the size;

s155, judging whether the minimum included angle is larger than or equal to 45 degrees; if so, selecting a rainfall monitoring station corresponding to the included angle; otherwise, go to step S156;

s156, judging whether the maximum included angle is smaller than or equal to 45 degrees; if so, selecting a rainfall monitoring station corresponding to the included angle; otherwise, go to step S157;

s157, calculating the difference value between the maximum included angle and the minimum included angle, and judging whether the difference value is larger than or equal to 30 degrees; if so, selecting a rainfall monitoring station corresponding to the maximum included angle and the minimum included angle; otherwise, selecting a rainfall monitoring site corresponding to the minimum value of the difference value between each included angle and 45 degrees.

Specifically, the selection is divided into four quadrants for points that do not fall on the coordinate axis, namely: ((x)_n>0,y_n>0),(x_n>0,y_n<0),(x_n<0,y_n<0),(x_n<0,y_n>0) I n ∈ (1,2,3, …, j)) four cases are classified as S₁,S₂,S₃,S₄；

With S₁((x_1n,y_1n)|(n∈(1,2,3,…,j)||x_1n,y_1n>0) Take the example) as follows:

if there is no point in the quadrant, the quadrant is skipped to the next quadrant.

If there is only 1 point in the quadrant, the point is automatically selected and the next quadrant is entered.

If there are more than 2 points in the quadrant, then choose the origin to the point with the smallest modulus of the vector of these points, and mark as S₁’(x’,y’)。

Calculating the modulo sum of the origin to the vectors of other points in the quadrant₁' Difference z of the moduli (d) of the vector_n|(n∈(1,2,3,…,j)；

Presetting a proportion k% (generally taking 10%), judging z_nWhether the number of points corresponding to not more than k%. d is 1; if yes, selecting S₁' is the adjacent rainfall monitoring station of the quadrant, and enters the next quadrant; otherwise, entering the next step;

respectively calculating the Z satisfying by using an included angle formula_nThe counterclockwise direction of the vector formed by the points not greater than k%. multidot.d forms an included angle less than pi/2 with the coordinate axis, such as the vector (x)₁₁,y₁₁) The included angles are as follows:

θ₁₁＝arccos(y₁₁/sqrt(x₁₁^₂+y₁₁^₂))

the included angles are ordered from small to large, and are assumed to be theta₁₁、θ₁₂、θ₁₃。

If theta is greater than theta₁₁>If pi/4, then select θ₁₁The corresponding point is used as an adjacent rainfall monitoring site;

if theta is greater than theta₁₃<If pi/4, then select θ₁₃The corresponding point is used as an adjacent rainfall monitoring site;

if theta is greater than theta₁₁<π/4<θ₁₃Then, calculate Δ θ ═ θ₁₃-θ₁₁；

If Δ θ>If pi/6, then select θ₁₁And theta₁₃The corresponding point is used as an adjacent rainfall monitoring site;

if Δ θ<Pi/6, then calculate min (| theta)₁₁-π/4|,|θ₁₂-π/4|,|θ₁₃And-pi/4) selecting a corresponding point as an adjacent rainfall monitoring station.

S₁And after the selection of the space point is finished, entering the next quadrant.

S16, forming the nearby rainfall monitoring sites selected in the steps S14 and S15 into a nearby rainfall monitoring site field point set;

s17, establishing an interpolation topological relation between the ponding monitoring site and the point set in the area of the rainfall approaching monitoring site.

S2, acquiring historical ponding process data of a ponding monitoring site and historical rainfall process data of a rainfall monitoring site nearby, obtaining historical rainfall process data of the ponding monitoring site by adopting an inverse distance weight interpolation method according to the interpolation topological relation established in the step S1, and constructing a city waterlogging ponding process prediction model related to the ponding process and the rainfall process;

in this step, step S2 specifically includes the following sub-steps:

s21, acquiring historical ponding process data of each field of ponding monitoring site and historical rainfall process data of a nearby rainfall monitoring site;

s22, utilizing the historical rainfall process data of the nearby rainfall monitoring site, and utilizing an inverse distance weight interpolation method to interpolate the historical rainfall process data of the nearby rainfall monitoring site according to the interpolation topological relation established in the step S1, so as to generate the historical rainfall process data of the ponding monitoring site;

s23, traversing historical ponding process data and historical rainfall process data of the ponding monitoring site, and calculating the correlation between the ponding process data and the sliding rainfall process data;

s24, selecting a sliding time period with the maximum correlation corresponding to the maximum correlation, and establishing a sliding rainfall process and an accumulated water process data set of historical rainfall accumulated water of each occasion;

s25, performing machine learning training and verification by using the established data set, and constructing a city waterlogging ponding process prediction model related to a ponding process and a rainfall process.

Specifically, the method utilizes the maximum water accumulation depth and the water accumulation process of the sliding rainfall process and the water accumulation monitoring site to establish a training set, a verification set and a test set, adopts a machine learning model, and further establishes a prediction model related to the water accumulation process-rainfall process, and comprises the following specific processes:

reading historical water accumulation process data (time interval is generally 5 minutes) of each field of a water accumulation monitoring station, and smoothing the water accumulation process by utilizing a Newton interpolation method or a Lagrange interpolation method.

Reading historical rainfall process data (time interval is generally 5 minutes) of a nearby rainfall monitoring site, and calculating the historical rainfall process data of each time of the ponding monitoring site by adopting an inverse distance weight interpolation method according to the interpolation topological relation established in S17.

Thirdly, traversing the historical ponding process data and the historical rainfall process data of the ponding monitoring station, and calculating the minimum time period T minutes (generally times of 5 minutes) so as to ensure that the T time period of the ponding monitoring station slides the rainfall process P_TThe correlation with the corresponding water accumulation process H is the best.

Selecting the sliding rainfall process and the corresponding ponding process in the period of 2T as a verification set and the sliding rainfall process and the corresponding ponding process in the period of T in the rest fields as a training set according to the determined T.

And fifthly, training and verifying by using models such as a linear regression model, a vector machine model, a neural network and the like according to the training set and the verification set. And selecting the model and the parameters according to the principle that the mean value and the variance of the regression residual of the training set are minimum and the mean value and the variance of the prediction residual of the verification set are minimum, thereby establishing the urban waterlogging ponding process prediction model related to the ponding process-rainfall process.

S3, acquiring real-time rainfall process data of a nearby rainfall monitoring site, and acquiring the real-time rainfall process data of the ponding monitoring site by adopting an inverse distance weight interpolation method according to the interpolation topological relation established in the step S1; fusing rainfall process data of each product according to the time-space characteristics of the multi-source quantitative rainfall forecast data products to obtain quantitative rainfall forecast process data of the accumulated water monitoring site;

in this step, step S3 specifically includes the following sub-steps:

s31, acquiring real-time rainfall process data of a rainfall monitoring site;

s32, interpolating the real-time rainfall process data close to the rainfall monitoring site by using an inverse distance weight interpolation method according to the interpolation topological relation established in the step S1 to generate the real-time rainfall process data of the ponding monitoring site;

s33, obtaining various weather quantitative rainfall forecast products at the positions of the ponding monitoring sites, and selecting the weather quantitative rainfall forecast product with the highest time-space precision to interpolate according to the time-space precision of different products to obtain quantitative rainfall forecast process data of the ponding monitoring sites.

In this step, step S33 specifically includes the following sub-steps:

s331, carrying out spatial superposition on the ponding monitoring station and a weather quantitative forecasting grid to obtain a grid number corresponding to the ponding monitoring station;

s332, reading rainfall process data of various weather quantitative forecast products with the grid numbers;

s333, selecting rainfall process data with highest time-space precision from the rainfall process data read in the step S332;

and S334, interpolating the rainfall process data selected in the step S333 into forecast rainfall process data with equal intervals t (generally 5 minutes) by adopting an average distribution method or a constructed arithmetic progression method.

Specifically, the weather quantitative forecast rainfall data is hourly rainfall forecast data such as 12 hours and 24 hours in the future of a kilometer grid. Taking the 24-hour hourly rainfall process data as an example:

the average distribution method comprises the following steps:

the rainfall is distributed equally over the hour by hour, in t minutes (typically 5 minutes);

second, structure arithmetic sequence method:

for the ith hour, the rainfall P is compared_i-1、P_i、P_i+1When i is 1 or 24, P_i-1Or P_i+1Take 0.

P_i-1、P_i、P_i+1The following two cases exist in the magnitude relationship of (1):

(P_i-1<＝P_i<＝P_i+1or P_i-1>＝P_i>＝P_i+1) Or (P)_i-1<P_i>P_i+1Or P_i-1>P_i<P_i+1)

For case 1:

rainfall p 'for the last t minutes (generally 5 minutes) in i-1 hour'_(i-1)Is foundation (i ═ 1, p'_(i-1)0), an arithmetic progression F (p'_(i-1)+ d j | d is a tolerance, j ═ 1,2,3, …,60/t), and the sum of the arithmetic progression and the forecast rainfall P at the i-th hour are required_iAnd (5) equaling, solving the tolerance d, and further calculating the rainfall process with the time interval of t minutes in the ith hour.

For case 2:

rainfall p 'for the last t minutes (generally 5 minutes) in i-1 hour'_(i-1)Is foundation (i ═ 1, p'_(i-1)0), an arithmetic progression F (p'_(i-1)+ d j | d is the tolerance, j 1,2,3, …,30/t), the sum of the series of arithmetic numbers is required to be equal to the forecast rainfall P at the ith hour_i1/2, a rainfall process in which the time interval of the first 30 minutes within the ith hour is t minutes was constructed by determining the tolerance d, and the reverse of this rainfall process was taken as the rainfall process in which the time interval of the last 30 minutes is t minutes.

S4, splicing the real-time rainfall process data of the ponding monitoring site with the quantitative rainfall forecasting process data to obtain the real-time rainfall forecasting process data of the ponding monitoring site;

and S5, forecasting the water accumulation process of the urban waterlogging water accumulation point according to the real-time and forecast rainfall process data of the water accumulation monitoring station by utilizing the urban waterlogging water accumulation point water accumulation process forecasting model.

In the step, according to the real-time and forecast rainfall process with the time interval of t minutes, the ponding point ponding process can be directly forecasted by utilizing the constructed urban inland inundation ponding process forecasting model.

The above detailed description is provided for the embodiment of the urban waterlogging and water accumulation point water accumulation process prediction method based on machine learning, and the invention also provides a system, a device and a computer readable storage medium for urban waterlogging and water accumulation point water accumulation process prediction based on machine learning, which correspond to the method. Since the embodiments of the system, the apparatus, and the computer-readable storage medium portion correspond to the embodiments of the method portion, please refer to the description of the embodiments of the method portion for the embodiments of the system, the apparatus, and the computer-readable storage medium portion, which is not repeated here.

Fig. 2 is a system for predicting a water spot-waterlogging process of an urban area based on machine learning, which includes:

the interpolation topological relation building module is used for building an interpolation topological relation between a ponding monitoring site and an approaching rainfall monitoring site by utilizing an improved K neighborhood method;

the data acquisition module is used for acquiring water accumulation process data of a water accumulation monitoring site and rainfall process data of a nearby rainfall monitoring site;

the data interpolation module is used for generating rainfall process data of the ponding monitoring station by utilizing the rainfall process data of the nearby rainfall monitoring station and utilizing an inverse distance weight interpolation method according to the established interpolation topological relation;

the data fusion module is used for fusing multi-source quantitative rainfall forecast data products to obtain quantitative rainfall forecast process data of the accumulated water monitoring station;

the data splicing module is used for splicing the quantitative rainfall forecasting process data and the real-time rainfall process data to obtain the real-time rainfall forecasting process data of the accumulated water monitoring site;

the model construction module is used for constructing a city waterlogging ponding process prediction model related to a rainfall process according to historical ponding process data of a ponding monitoring site and historical rainfall process data close to a rainfall monitoring site;

and the ponding prediction module is used for predicting the ponding process of the urban waterlogging ponding point based on real-time and forecast rainfall process data of the ponding monitoring station by using the constructed urban waterlogging ponding process prediction model.

The urban waterlogging ponding process prediction system based on machine learning provided by the embodiment of the invention has the beneficial effects of the urban waterlogging ponding process prediction method based on machine learning.

Fig. 3 is a device for predicting urban inland inundation water accumulation point water accumulation process based on machine learning, which includes:

a memory for storing a computer program;

and the processor is used for realizing the steps of the urban waterlogging ponding process prediction method based on machine learning when executing the computer program.

The urban waterlogging and water-accumulating point water accumulation process prediction device based on machine learning provided by the embodiment of the invention has the beneficial effects of the urban waterlogging and water-accumulating point water accumulation process prediction method based on machine learning.

The embodiment of the invention also provides a computer readable storage medium, wherein a computer program is stored on the computer readable storage medium, and when being executed by a processor, the computer program realizes the steps of the urban waterlogging ponding process prediction method based on machine learning.

The computer-readable storage medium provided by the embodiment of the invention has the beneficial effects of the urban waterlogging and water accumulation point water accumulation process prediction method based on machine learning.

The present invention is described with reference to flowchart illustrations and/or block diagrams of methods, apparatus (systems) and computer program products according to embodiments of the invention. It will be understood that each flow and/or block of the flow diagrams and/or block diagrams, and combinations of flows and/or blocks in the flow diagrams and/or block diagrams, can be implemented by computer program instructions. These computer program instructions may be provided to a processor of a general purpose computer, special purpose computer, embedded processor, or other programmable data processing apparatus to produce a machine, such that the instructions, which execute via the processor of the computer or other programmable data processing apparatus, create means for implementing the functions specified in the flowchart flow or flows and/or block diagram block or blocks.

These computer program instructions may also be stored in a computer-readable memory that can direct a computer or other programmable data processing apparatus to function in a particular manner, such that the instructions stored in the computer-readable memory produce an article of manufacture including instruction means which implement the function specified in the flowchart flow or flows and/or block diagram block or blocks.

These computer program instructions may also be loaded onto a computer or other programmable data processing apparatus to cause a series of operational steps to be performed on the computer or other programmable apparatus to produce a computer implemented process such that the instructions which execute on the computer or other programmable apparatus provide steps for implementing the functions specified in the flowchart flow or flows and/or block diagram block or blocks.

The principle and the implementation mode of the invention are explained by applying specific embodiments in the invention, and the description of the embodiments is only used for helping to understand the method and the core idea of the invention; meanwhile, for a person skilled in the art, according to the idea of the present invention, there may be variations in the specific embodiments and the application scope, and in summary, the content of the present specification should not be construed as a limitation to the present invention.

It will be appreciated by those of ordinary skill in the art that the embodiments described herein are intended to assist the reader in understanding the principles of the invention and are to be construed as being without limitation to such specifically recited embodiments and examples. Those skilled in the art can make various other specific changes and combinations based on the teachings of the present invention without departing from the spirit of the invention, and these changes and combinations are within the scope of the invention.